Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

MEMS Scanning Mirrors for Optical Coherence Tomography

MEMS Scanning Mirrors for Optical Coherence Tomography hv photonics Review 1 , 2 Christophe Gorecki * and Sylwester Bargiel Polish Academy of Sciences, Institute of Physical Chemistry, International Center for Translational Eye Research, Skierniewicka 10A, 01-230 Warsaw, Poland FEMTO-ST Institute (UMR CNRS 6714/UBFC), 15B Avenue des Montboucons, 25030 Besançon, France; sylwester.bargiel@femto-st.fr * Correspondence: cgorecki@ichf.edu.pl Abstract: This contribution presents an overview of advances in scanning micromirrors based on MEMS (Micro-electro-mechanical systems) technologies to achieve beam scanning for OCT (Optical Coherence Tomography). The use of MEMS scanners for miniaturized OCT probes requires appropriate optical architectures. Their design involves a suitable actuation mechanism and an adapted imaging scheme in terms of achievable scan range, scan speed, low power consumption, and acceptable size of the OCT probe. The electrostatic, electromagnetic, and electrothermal actuation techniques are discussed here as well as the requirements that drive the design and fabrication of functional OCT probes. Each actuation mechanism is illustrated by examples of miniature OCT probes demonstrating the effectiveness of in vivo bioimaging. Finally, the design issues are discussed to permit users to select an OCT scanner that is adapted to their specific imaging needs. Keywords: micro-opto-electro-mechanical system; MEMS scanner; optical coherence tomography 1. Introduction Micro-electro-mechanical systems (MEMS) technology enables the building of mi- crooptical scanners that are well suited for low cost manufacturability and scalability as the Citation: Gorecki, C.; Bargiel, S. MEMS processes emanate from the mature semiconductor microfabrication industry [1]. MEMS Scanning Mirrors for Optical For a long time, the potential of MEMS to steer or direct light has been well demonstrated Coherence Tomography. Photonics in the field of free-space optical systems [2]. In the 80s and early 90s, telecommunications 2021, 8, 6. https://doi.org/10.3390/ became the market driver for the optical applications of MEMS, pushing the develop- photonics8010006 ment of scanning micromirror systems for optical switches and network ports [3]. More recently, many types of MEMS scanning mirrors have been developed, covering a wide Received: 2 October 2020 range of applications from micrometer-scale array-type components to large scanners for Accepted: 24 December 2020 high-resolution imaging [4]. Thus, numerous optical imaging techniques such as confo- Published: 30 December 2020 cal microscopy [5,6], multiphoton microscopy [7,8], and Optical Coherence Tomography (OCT) [9–11] have become important diagnostic tools in biomedicine, particularly offering Publisher’s Note: MDPI stays neu- a platform for endoscopic imaging. These MEMS scanners successfully replaced the bulky tral with regard to jurisdictional clai- and high power consuming galvanometer scanners, providing compact, low cost, and low ms in published maps and institutio- power consumption solutions for high speed beam steering. Further, 2D MEMS mirrors nal affiliations. that scan in two axes are a pertinent alternative to the large galvano-scanners [12]. The MEMS scanner ’s performances are closely linked to the size of the selected actuator, carrying the micromirror and the force developed by this actuator. Figure 1 Copyright: © 2020 by the authors. Li- represents a summary of scanning micromirror applications, including the corresponding censee MDPI, Basel, Switzerland. actuation mechanisms and main microfabrication technologies [13]. At the scale level, This article is an open access article going from 1 mm to 1 cm, the MEMS technology combined with fiber optics enables distributed under the terms and con- miniature scanning components to be embedded inside the endoscopic imaging probes ditions of the Creative Commons At- operating at high speed and high resonance frequency. The MEMS scanners are relatively tribution (CC BY) license (https:// easily integrated and adapted for low cost fabrication and low power consumption. The creativecommons.org/licenses/by/ miniaturization performances and subsequent advances in standardized micromachining 4.0/). Photonics 2021, 8, 6. https://doi.org/10.3390/photonics8010006 https://www.mdpi.com/journal/photonics Photonics 2021, 8, x FOR PEER REVIEW 2 of 25 Photonics 2021, 8, 6 2 of 25 high speed and high resonance frequency. The MEMS scanners are relatively easily inte- grated and adapted for low cost fabrication and low power consumption. The miniaturi- zation performances and subsequent advances in standardized micromachining technol- ogies have also offered numerous low cost and disposable OCT probes for the medical technologies have also offered numerous low cost and disposable OCT probes for the industry. Originally adopted by the ophthalmic community [14–16], OCT has been used medical industry. Originally adopted by the ophthalmic community [14–16], OCT has been to image internal organs, such as the gastrointestinal tract [17], and in the diagnosis of used to image internal organs, such as the gastrointestinal tract [17], and in the diagnosis skin pathologies [18,19]. This strong interest for clinical applications pushed several com- of skin pathologies [18,19]. This strong interest for clinical applications pushed several panies to develop endoscopic OCT systems [20]. Examples of commercial products are companies to develop endoscopic OCT systems [20]. Examples of commercial products are the clinical endoscope and catheter-based systems from the NvisionVLE Imaging System the clinical endoscope and catheter-based systems from the NvisionVLE Imaging System (South Jordan, Utah, USA) [21], the intravascular OCT imaging systems from OPTIS™ (St. (South Jordan, Utah, USA) [21], the intravascular OCT imaging systems from OPTIS™ Jude Medical Inc., St. Paul, MN, USA) [22], Santec’s (Komaki, Japan) swept-source OCT (St. Jude Medical Inc., St. Paul, MN, USA) [22], Santec’s (Komaki, Japan) swept-source systems [23], Thorlabs (Newton, NJ, USA) OCT scanners [24], as well as Mirrorcle (Rich- OCT systems [23], Thorlabs (Newton, NJ, USA) OCT scanners [24], as well as Mirrorcle (Richmond, mond, CA, CA, USA) USA) microscanners [25]. microscanners [25]. Figure 1. Applications, actuation mechanisms, and fabrication technologies for scanning micromirrors. Figure 1. Applications, actuation mechanisms, and fabrication technologies for scanning micromirrors. In this paper, we will demonstrate that for OCT imaging applications, the performance In this paper, we will demonstrate that for OCT imaging applications, the perfor- of the MEMS scanner is often limited by optics and intrinsic characteristics of actuation mance of the MEMS scanner is often limited by optics and intrinsic characteristics of ac- mechanisms. Here, optics require a small focused spot and dynamic focusing, imposing tuation mechanisms. Here, optics require a small focused spot and dynamic focusing, im- severe restrictions on scanning lens performances, while the actuation needs a high scan- posing severe restrictions on scanning lens performances, while the actuation needs a high ning speed, a low power consumption, a precise control of motion linearity, and reduced scanning speed, a low power consumption, a precise control of motion linearity, and re- cross-axis coupling, which may distort the scanning patterns [26,27]. The group of OCT duced cross-axis coupling, which may distort the scanning patterns [26,27]. The group of probes to be discussed in this paper do maintain such opto-mechanical performances, OCT probes to be discussed in this paper do maintain such opto-mechanical perfor- using different actuation mechanisms. Our wish is to demonstrate that the breakthrough mances, using different actuation mechanisms. Our wish is to demonstrate that the break- of compactness is obtained when MEMS dual-axis beam-steering micromirrors [28] are through of compactness is obtained when MEMS dual-axis beam-steering micromirrors used to achieve scanning 3D OCT probes. In the case of endoscopic applications, they are [28] are used to achieve scanning 3D OCT probes. In the case of endoscopic applications, small enough to be included into a standard endoscope channel, with an inner diameter they are small enough to be included into a standard endoscope channel, with an inner of 2.8 mm. Further, 2D scanning motion can derive from electrostatic, electromagnetic, diameter of 2.8 mm. Further, 2D scanning motion can derive from electrostatic, electro- electrothermal, or piezoelectric actuation, providing the scanning mirrors for light beam magnetic, electrothermal, or piezoelectric actuation, providing the scanning mirrors for steering, operating at high speed and fully controlled by non-resonant or resonant regimes. light beam steering, operating at high speed and fully controlled by non-resonant or res- However, we intentionally excluded from the present study piezoelectric actuation and we onant regimes. However, we intentionally excluded from the present study piezoelectric consider only the three other actuation mechanisms for MEMS scanning mirrors that are actuation and we consider only the three other actuation mechanisms for MEMS scanning the most widely used in OCT applications. mirrors that are the most widely used in OCT applications. 2. Requirements for MEMS Microscanners From the user point of view, performances of scanning micromirrors are defined by the maximum scan angle, the number of resolvable spots which represents the scan resolution, the resonance frequency, as well as the surface quality vs. the smoothness and flatness of micromirrors. Photonics 2021, 8, x FOR PEER REVIEW 3 of 25 2. Requirements for MEMS Microscanners From the user point of view, performances of scanning micromirrors are defined by the maximum scan angle, the number of resolvable spots which represents the scan reso- lution, the resonance frequency, as well as the surface quality vs. the smoothness and flat- ness of micromirrors. The number of resolvable spots N of a scanning mirror is defined as a function of optical scan angle and beam divergence , as shown in Figure 2: = = , (1) where D is the mirror diameter, a represents the aperture shape factor (a = 1 in the case of Photonics 2021, 8, 6 3 of 25 a square aperture and 1.22 for a circular aperture) and is the illumination wavelength. As the number of resolvable spots is proportional to the product of mirror diameter by the scan angle, the resonant frequency defining the upper limit of the scanner’s re- The number of resolvable spots N of a scanning mirror is defined as a function of sponse depends on the diameter (or inertia) and mechanical rigidity of the mirror suspen- optical scan angle q and beam divergence dq, as shown in Figure 2: opt sion. In the case of a circular mirror with a diameter of 1 mm, illuminated by a He-Ne laser and scanning in the mode of VGA (640 × 480 pixels), θopt = 28.3°, which is two times bigger q D opt than the mechanical deflection. In thi N = s case, the sc = q anning m , irror needs to deliver a full (1) opt dq al mechanical angle θmech = 14.15° or larger. wherTo m e D isathe ke tmirr he com or diameter parison o , a f MEM represents S scan the ners op apertur erat e shape ing at di factor ffere (a nt = freq 1 inuenc the case ies aof lso, a square aperture and 1.22 for a circular aperture) and l is the illumination wavelength. we use the D = 1,22 product, which is the key performance index for scanners. Figure 2. Total optical scan range of a scanning micromirror vs. the beam divergence. Figure 2. Total optical scan range of a scanning micromirror vs. the beam divergence. As the number of resolvable spots is proportional to the product of mirror diameter by Imaging applications for 2D scanning require a dual axis system with fast scanning the scan angle, the resonant frequency defining the upper limit of the scanner ’s response capability. Two categories of more popular MEMS torsional scanners are the uniaxial scanner depends and on t the he b diameter iaxial gim (or bal-m inertia) ount and ed scann mechanical er [29]. Fi rigidity gure 3 of sho the wmirr s boor th t suspension. hese scanning In the case of a circular mirror with a diameter of 1 mm, illuminated by a He-Ne laser and architectures with associated scanning trajectories. Here, two full raster cycles are repre- scanning in the mode of VGA (640  480 pixels), q = 28.3 , which is two times bigger sented while the scanning spots are schematized in g opt reen. The uniaxial scanner has a sin- than the mechanical deflection. In this case, the scanning mirror needs to deliver a full gle axis of rotation and to obtain a 2D scanning, a pair of uniaxial scanners is used (one mechanical angle q = 14.15 or larger. horizontal, one verti mech cal). Here, the rotation of the first scanner causes the optical beam to To make the comparison of MEMS scanners operating at different frequencies also, walk across the second scanner. In this case, the writing of data is obtained only during we use the q D = 1.22Nl product, which is the key performance index for scanners. opt the forward sweep of the horizontal scanner. The biaxial scanner includes two perpendic- Imaging applications for 2D scanning require a dual axis system with fast scanning ular axes of rotation. The oscillation of the inner frame creates a horizontal scan line, while capability. Two categories of more popular MEMS torsional scanners are the uniaxial the outer frame creates the vertical scan line. This is writing a new line of data in both scan scanner and the biaxial gimbal-mounted scanner [29]. Figure 3 shows both these scanning directions. Thus, the scanner writes two lines during one scan cycle. architectures with associated scanning trajectories. Here, two full raster cycles are rep- resented while the scanning spots are schematized in green. The uniaxial scanner has a single axis of rotation and to obtain a 2D scanning, a pair of uniaxial scanners is used (one horizontal, one vertical). Here, the rotation of the first scanner causes the optical beam to walk across the second scanner. In this case, the writing of data is obtained only during the forward sweep of the horizontal scanner. The biaxial scanner includes two perpendicular axes of rotation. The oscillation of the inner frame creates a horizontal scan line, while the outer frame creates the vertical scan line. This is writing a new line of data in both scan directions. Thus, the scanner writes two lines during one scan cycle. Photonics Photonics 2021 2021 , 8,, x FO 8, 6 R PEER REVIEW 4 of 4 of 25 25 Figure 3. Schema of uniaxial scanner in (a) vs. biaxial scanner in (b) with associated scanning Figure 3. Schema of uniaxial scanner in (a) vs. biaxial scanner in (b) with associated scanning tra- trajectories. jectories. The performance of the MEMS optical scanner is often limited by the optical architec- The performance of the MEMS optical scanner is often limited by the optical archi- ture, which requires small, focused spots and dynamic focusing. In addition, the intrinsic tecture, which requires small, focused spots and dynamic focusing. In addition, the intrin- micromechanical characteristics of the actuation mechanism of the MEMS scanners are sic micromechanical characteristics of the actuation mechanism of the MEMS scanners are crucial and dependent upon the requirements of high scanning speed, low power con- crucial and dependent upon the requirements of high scanning speed, low power con- sumption, as well as a precise control of motion linearity. Here, one of the limiting features sumption, as well as a precise control of motion linearity. Here, one of the limiting features in MEMS scanners is their mode of driving, which can be resonant or quasi-static. A reso- in MEMS scanners is their mode of driving, which can be resonant or quasi-static. A res- nant drive mode with a fine control of the beam’s location is more difficult to implement, onant drive mode with a fine control of the beam’s location is more difficult to implement, requiring one to use the close-loop control of beam steering. A quasi-static mode, on the requiring one to use the close-loop control of beam steering. A quasi-static mode, on the other hand, provides easy programmable control of the beam, permitting operation with other hand, provides easy programmable control of the beam, permitting operation with open-loop control of beam steering. However, certain categories of MEMS scanners are not open-loop control of beam steering. However, certain categories of MEMS scanners are able to offer the significant aperture size and scan angle specifications in the quasi-static not able to offer the significant aperture size and scan angle specifications in the quasi- mode. The quasi-statically tilted MEMS scanner only requires open-loop control for beam static mode. The quasi-statically tilted MEMS scanner only requires open-loop control for steering purposes. The main challenges of a bidirectional scan are the precise control of beam steering purposes. The main challenges of a bidirectional scan are the precise control phase between both the scan lines and a high-quality control of motion, avoiding the effects of phase between both the scan lines and a high-quality control of motion, avoiding the of mechanical coupling between the scanner axes. effects of mechanical coupling between the scanner axes. The choice of appropriate scanning techniques is strongly dependent on the selection The choice of appropriate scanning techniques is strongly dependent on the selection of optimal scanning frequency. The most common scanning techniques are raster scanning of optimal scanning frequency. The most common scanning techniques are raster scan- and Lissajous scanning, shown in Figure 4 [30]. In raster scanning, a low frequency, linear ning and Lissajous scanning, shown in Figure 4 [30]. In raster scanning, a low frequency, vertical scan (usually in quasi-static mode) is paired with an orthogonal high frequency, linear vertical scan (usually in quasi-static mode) is paired with an orthogonal high fre- resonant horizontal scan. For the raster scanner, the vertical scan is often assumed to be quency, resonant horizontal scan. For the raster scanner, the vertical scan is often assumed 60 Hz (video frame rate). The raster scanning can be obtained by bidirectional scanning to be 60 Hz (video frame rate). The raster scanning can be obtained by bidirectional scan- waveforms with N = f /f , where N is a positive integer. A raster scanner involves a x y ning waveforms with N = fx/fy, where N is a positive integer. A raster scanner involves a triangular trajectory in the x-axis while shifting the sample position in steps or continuously triangular trajectory in the x-axis while shifting the sample position in steps or continu- in the y-axis. Here, a laser beam is starting at the top left of Figure 4a and goes from left- ously in the y-axis. Here, a laser beam is starting at the top left of Figure 4a and goes from to-right at a scanning frequency of f to the bottom-right corner (1), then rapidly moves left-to-right at a scanning frequency of fx to the bottom-right corner (1), then rapidly moves back to the left and scans the next line (2), then off once again to go back up to the top (3). back to the left and scans the next line (2), then off once again to go back up to the top (3). During the period of one scan cycle, the vertical position increases steadily downward During the period of one scan cycle, the vertical position increases steadily downward at at a frequency of f , which is the slow scanning frequency. The scan resolution and a frequency of fy, which is the slow scanning frequency. The scan resolution and the frame the frame rate are, respectively, defined by the lateral and vertical scanning frequencies. rate are, respectively, defined by the lateral and vertical scanning frequencies. Raster scan- Raster scanning results in a rectangular scanning area. This simple technique of scan ning results in a rectangular scanning area. This simple technique of scan presents several presents several technical limitations for MEMS micromirrors such as the higher driving technical limitations for MEMS micromirrors such as the higher driving voltages due to voltages due to the use of non-resonant motion and lower mechanical stability for the the use of non-resonant motion and lower mechanical stability for the slow axis. Figure slow axis. Figure 4b shows an alternative to raster scanning which is biresonant Lissajous 4b shows an alternative to raster scanning which is biresonant Lissajous scanning. This is scanning. This is achieved by driving the x and y axes with purely sinusoidal signals of Photonics 2021, 8, x FOR PEER REVIEW 5 of 25 Photonics Photonics 2021 2021 , 8 , , x FO 8, 6 R PEER REVIEW 5 of 5 of 25 25 achieved by driving the x and y axes with purely sinusoidal signals of different frequen- cies fx and fy that are x(t) = Ax cos(2πfxt) and y(t) = Ay cos(2πfyt). The corresponding scan- achieved by driving the x and y axes with purely sinusoidal signals of different frequen- different frequencies f and f that are x(t) = A cos(2f t) and y(t) = A cos(2f t). The x y x x y y ning waveforms have a frequency relation of n = fx/fy, where n is a rational number. Lissa- cies fx and fy that are x(t) = Ax cos(2πfxt) and y(t) = Ay cos(2πfyt). The corresponding scan- corresponding scanning waveforms have a frequency relation of n = f /f , where n is a x y jous scanning provides a rectangular scan area too. The scanning trajectory and the frame ning waveforms have a frequency relation of n = fx/fy, where n is a rational number. Lissa- rational number. Lissajous scanning provides a rectangular scan area too. The scanning rate are more complex than in raster scanning trajectories and MEMS mirrors are operat- jous scanning provides a rectangular scan area too. The scanning trajectory and the frame trajectory and the frame rate are more complex than in raster scanning trajectories and ing at high resonant frequencies for both the axes, offering better mechanical stability. rate are more complex than in raster scanning trajectories and MEMS mirrors are operat- MEMS mirrors are operating at high resonant frequencies for both the axes, offering better ing at high resonant frequencies for both the axes, offering better mechanical stability. mechanical stability. Figure 4. Principle of main scanning techniques: (a) raster scanning and (b ) Lissajous scanning. Figure 4. Principle of main scanning techniques: (a) raster scanning and (b) Lissajous scanning. Figure 4. Principle of main scanning techniques: (a) raster scanning and (b) Lissajous scanning. Figure 5a shows a schematic of a rectangular torsional scanning mirror with rectan- Figure 5a shows a schematic of a rectangular torsional scanning mirror with rectan- gular flexure beams [31]. Applied to produce the desired torsional mode, this torsional Figure 5a shows a schematic of a rectangular torsional scanning mirror with rectan- gular flexure beams [31]. Applied to produce the desired torsional mode, this torsional scanning mirror presents three parasitic oscillation modes that we refer to as vertical (or gular flexure beams [31]. Applied to produce the desired torsional mode, this torsional scanning mirror presents three parasitic oscillation modes that we refer to as vertical (or piston), horizontal, and rocking modes. In the ideal design, the mode frequencies should scanning mirror presents three parasitic oscillation modes that we refer to as vertical (or piston), horizontal, and rocking modes. In the ideal design, the mode frequencies should be well separated from the torsional mode frequency and its harmonics to minimize the piston), horizontal, and rocking modes. In the ideal design, the mode frequencies should be well separated from the torsional mode frequency and its harmonics to minimize the power dissipation. Figure 5b shows that the horizontal and vertical modes do not change be well separated from the torsional mode frequency and its harmonics to minimize the power dissipation. Figure 5b shows that the horizontal and vertical modes do not change the scanned beam direction. Rocking mode deflects the incoming beam perpendicular to power dissipation. Figure 5b shows that the horizontal and vertical modes do not change the scanned beam direction. Rocking mode deflects the incoming beam perpendicular to the intended scan axis, creating an undesired off-axis motion. The biaxial scanners should the scanned beam direction. Rocking mode deflects the incoming beam perpendicular to the intended scan axis, creating an undesired off-axis motion. The biaxial scanners should be designed in such a way that cross coupling effects between the torsional and rocking the intended scan axis, creating an undesired off-axis motion. The biaxial scanners should be designed in such a way that cross coupling effects between the torsional and rocking modes of inner and outer scan frames are minimized. be designed in such a way that cross coupling effects between the torsional and rocking modes of inner and outer scan frames are minimized. modes of inner and outer scan frames are minimized. Figure 5. Torsional resonant scan mirror in (a) with its f our fundamental vibration modes in (b). Figure 5. Torsional resonant scan mirror in (a) with its four fundamental vibration modes in (b). Figure 5. Torsional resonant scan mirror in (a) with its four fundamental vibration modes in (b). Actuated MEMS mirrors are subjected to both static and dynamic deformations. Static deformation is induced by intrinsic material stress or thermal stress of mirror material, Photonics 2021, 8, x FOR PEER REVIEW 6 of 25 Photonics 2021, 8, 6 6 of 25 Actuated MEMS mirrors are subjected to both static and dynamic deformations. Static deformation is induced by intrinsic material stress or thermal stress of mirror mate- rial, while dynamic deformation is produced by forces due to mirror oscillation with a while dynamic deformation is produced by forces due to mirror oscillation with a frequency frequency f. The maximum mirror deviation from linearity according to Brosens’s formula f. The maximum mirror deviation from linearity according to Brosens’s formula [32] is: [32] is: 2 5 f D q mech d µ = = , , (2 (2) ) max E t where ρ is the material density, E the modulus of elasticity, and tm represents the mirror where  is the material density, E the modulus of elasticity, and t represents the mirror thickness. thickness. For a fixed D product, an increase of D leads to larger mirror deformation, lower For a fixed q D product, an increase of D leads to larger mirror deformation, lower opt maximum frequency, and increasing footprint. To keep the spot diffraction limited, the maximum frequency, and increasing footprint. To keep the spot diffraction limited, the maximum mirror deformation should not exceed λ/10. Here, the D -dependency deter- maximum mirror deformation should not exceed /10. Here, the D -dependency deter- mines the upper limit of mirror size, while large mirrors must have a large thickness tm. It mines the upper limit of mirror size, while large mirrors must have a large thickness t . It can be roughly estimated that the optimum mirror surface will be around 1 mm . can be roughly estimated that the optimum mirror surface will be around 1 mm . 3. Requirements for OCT Probes 3. Requirements for OCT Probes A typical OCT setup includes a Michelson interferometer and a low-coherence light A typical OCT setup includes a Michelson interferometer and a low-coherence light source. Interference signal, carrying the information about the measured biological object, source. Interference signal, carrying the information about the measured biological ob- is detected and demodulated to produce a map of the light backscattered from the micro- ject, is detected and demodulated to produce a map of the light backscattered from the structure inside the measured tissue. Image reconstruction is obtained by repeated axial microstructure inside the measured tissue. Image reconstruction is obtained by repeated measurements at different transverse positions as the optical beam is scanned by a MEMS axial measurements at different transverse positions as the optical beam is scanned by a scanning mirror. Figure 6 shows the two classes of miniature OCT probes, categorized on MEMS scanning mirror. Figure 6 shows the two classes of miniature OCT probes, catego- the basis of their scan modes: the side-imaging and the forward-imaging [33–35]. Side- rized on the basis of their scan modes: the side-imaging and the forward-imaging [33–35]. imaging probes, schematized in Figure 6a, are the most widely used because they tend to Side-imaging probes, schematized in Figure 6a, are the most widely used because they have a much simpler actuation mechanism than forward-imaging probes and the actuator tend to have a much simpler actuation mechanism than forward-imaging probes and the tends to be far away from the probe output. This category of OCT probes is very flexible actuator tends to be far away from the probe output. This category of OCT probes is very and has a small size that is well appropriate for the miniaturization. Here, a mirror or flexible and has a small size that is well appropriate for the miniaturization. Here, a mirror prism is connected, for example, to a rotation assembly that shifts the emitted light from or prism is connected, for example, to a rotation assembly that shifts the emitted light from the optical the opticalfiber out fiber out of a of a window on window on tthe he side side of the probe. The of the probe. The side-im side-imaging aging OCT probe OCT probe only provi only provides des si side de im imaging aging aro around und t the he pr probe, which obe, which limits limits its its c clinical linical applications, applications, ma making k- ing the timage-based he image-based surgical surgi needle cal need guidance le guidanc difficult. e diffic The ult.forwar The forw d-imaging ard-imagin probe, g probe, o of which f which its size its tends size tends to be in a range of n to be in a range of needle sizes, eedle is sizes, mor is e more suitable sufor itabsur le fo gical r sur guidance gical guidance inside insi thede t body he bo . However dy. How , forwar ever, forw d-imaging ard-ima pr ging probe obes are generally s are gener mor ally e more comple complex in design. x in desi They gn. require the actuator near the probe tips and are quite difficult to miniaturize. Here, a mirror They require the actuator near the probe tips and are quite difficult to miniaturize. Here, or prism assembly shifts the emitted light from the optical fiber out of a window on the a mirror or prism assembly shifts the emitted light from the optical fiber out of a window front side of the probe, as shown in Figure 6b. on the front side of the probe, as shown in Figure 6b. Figure 6. Two categories of miniature OCT probes: (a) side imaging probe; and (b) forward imaging probe. Figure 6. Two categories of miniature OCT probes: (a) side imaging probe; and (b) forward imaging probe. In addition to the categorization of Figure 6, the scanning arrangement exhibits two In addition to the categorization of Figure 6, the scanning arrangement exhibits two distinct arrangements depending on the position of the MEMS scanner and the objective distinct arrangements depending on the position of the MEMS scanner and the objective lens focusing the beam on the sample to be measured: pre-objective scanning and post- lens focusing the beam on the sample to be measured: pre-objective scanning and post- objective scanning [36,37]. In the pre-objective scanner, a MEMS mirror is placed prior to an objective scanning [36,37]. In the pre-objective scanner, a MEMS mirror is placed prior to objective lens, allowing a long working lens-sample distance. However, this configuration can cause significant off-axis aberration for a deflected scanning beam. The aberration can be minimized by using a specialized f-Theta lens, focusing the laser beam on a single Photonics 2021, 8, x FOR PEER REVIEW 7 of 25 Photonics 2021, 8, 6 7 of 25 an objective lens, allowing a long working lens-sample distance. However, this configu- ration can cause significant off-axis aberration for a deflected scanning beam. The aberra- tion can be minimized by using a specialized f-Theta lens, focusing the laser beam on a single plane over the entire scan field while also ensuring that the beam remains perpen- plane over the entire scan field while also ensuring that the beam remains perpendicular dicular to the plane over the entire scan. In this category of probes, the miniaturization to the plane over the entire scan. In this category of probes, the miniaturization becomes becomes difficult. In the post-objective scanner, a MEMS scanner is located after an objec- difficult. In the post-objective scanner, a MEMS scanner is located after an objective lens. tive lens. This configuration can produce a small off-axis aberration as well as high image This configuration can produce a small off-axis aberration as well as high image resolution. resolution. However, the working distance becomes short and thus, the selection of an However, the working distance becomes short and thus, the selection of an objective with objective with a high numerical aperture objective is necessary. As the deflection by a a high numerical aperture objective is necessary. As the deflection by a mirror results in a mirror results in a curved focal plane, it is possible to compensate this curvature by vary- curved focal plane, it is possible to compensate this curvature by varying the focal length ing the focal length of the lens by moving the lens along the beam path. of the lens by moving the lens along the beam path. Finally, the last categorization of OCT probes refers to the location of the scanning Finally, the last categorization of OCT probes refers to the location of the scanning mechanism. The scanning mechanisms may be considered as either proximal or distal to mechanism. The scanning mechanisms may be considered as either proximal or distal to the light source [20]. Proximal scanners are placed in the illumination pathway upstream the light source [20]. Proximal scanners are placed in the illumination pathway upstream of the fiber and are used with a fiber bundle. This configuration offers the benefit of sep- of the fiber and are used with a fiber bundle. This configuration offers the benefit of sepa- arating bulky scanners from a miniaturized imaging head and typically includes cascaded rating bulky scanners from a miniaturized imaging head and typically includes cascaded galvanometer-mounted scanning mirrors, scanning the beam across the proximal end of galvanometer-mounted scanning mirrors, scanning the beam across the proximal end of a fiber bundle. Distal scanners are placed on the fiber side distal to the light source and a fiber bundle. Distal scanners are placed on the fiber side distal to the light source and usu usually ally scan scan illumin illumination ation from a sin from a single gle fifiber ber over over th the e spec specimen. imen. Her Here, the e, the 2D 2D scanning scanning can can be per be performed formed by a by a MEMS MEMS mirror w mirror which hich pivo pivots in ts in two tw angular o anguldir ar direct ections ion ors or by maintaining by main- the resonant vibrating of the fiber extremity via an attached actuator or cantilever. taining the resonant vibrating of the fiber extremity via an attached actuator or cantilever. After comparing the different configurations of OCT probes, Figure 7 summarizes After comparing the different configurations of OCT probes, Figure 7 summarizes the char the characteristics acteristics mamade de for each for each criterion of criterion the of the propr be classific obe classification ation proposed proposed earlier earlier . . Here, the colorful path corresponds to the optimal design of an OCT probe applied for Here, the colorful path corresponds to the optimal design of an OCT probe applied for gastrointestinal tract evaluation [38]. gastrointestinal tract evaluation [38]. Figure 7. Order of criteria of classifications of the endoscopic probe configurations. Figure 7. Order of criteria of classifications of the endoscopic probe configurations. 4. Electrostatic MEMS Scanning Mirrors for OCT 4. Electrostatic MEMS Scanning Mirrors for OCT 4.1. Principles of Electrostatic Microactuators 4.1. Principles of Electrostatic Microactuators The principle of electrostatic actuation is based on Coulomb’s law, using the attraction of two oppositely charged plates. Electrostatic actuation is currently the predominant The principle of electrostatic actuation is based on Coulomb’s law, using the attrac- method used for MEMS scanners because the capacitive actuators draw very little current, tion of two oppositely charged plates. Electrostatic actuation is currently the predominant therefore requiring low operating power despite the need for relatively high applied method used for MEMS scanners because the capacitive actuators draw very little current, voltages. An electrostatic torsional micromirror is in rotation when a driving voltage is therefore requiring low operating power despite the need for relatively high applied volt- applied between the fixed and movable electrodes. The mirror rotates an angle  about the ages. An electrostatic torsional micromirror is in rotation when a driving voltage is ap- torsion axis until the restoring and electrostatic torques are equal. The torque is given by: plied between the fixed and movable electrodes. The mirror rotates an angle θ about the torsion axis until the restoring and electrostatic torques are equal. The torque is given by: V ¶C T () = (3) 2 ¶ (3) θ= 2 ∂θ T () = k, (4) θ= kθ, (4) where V is the driving voltage, C the capacitance of the actuator, and k is the spring constant. Photonics 2021, 8, x FOR PEER REVIEW 8 of 25 Photonics 2021, 8, 6 8 of 25 where V is the driving voltage, C the capacitance of the actuator, and k is the spring con- stant. For a simple parallel plate actuator, the capacitance is given by: For a simple parallel plate actuator, the capacitance is given by: = , # A (5) C = , (5) where is the permittivity of free-space, A is the surface of electrode, and g is the gap where # is the permittivity of free-space, A is the surface of electrode, and g is the gap between the electrodes. between the electrodes. Figure 8 shows the schematic of two main types of electrostatic actuators: the parallel Figure 8 shows the schematic of two main types of electrostatic actuators: the parallel plate actuator [39] and the comb-drive actuator [40,41]. plate actuator [39] and the comb-drive actuator [40,41]. Figure 8. Two main architectures of electrostatic actuation: (a) parallel-plate actuator and (b) Figure 8. Two main architectures of electrostatic actuation: (a) parallel-plate actuator and (b) comb- comb-drive actuator. drive actuator. With a parallel plate gap closing actuator, the zone of the electrode overlap is mainly With a parallel plate gap closing actuator, the zone of the electrode overlap is mainly the area of the fixed electrode. Thus, the gap is a function of the rotation angle. Here, there the area of the fixed electrode. Thus, the gap is a function of the rotation angle. Here, there is a tradeoff as the initial gap distance needs to be large enough to generate the scan angle, is a tradeoff as the initial gap distance needs to be large enough to generate the scan angle, but small enough for a reasonable driving voltage. The linear scan range is limited by but small enough for a reasonable driving voltage. The linear scan range is limited by the the pull-in effect to around 40% of the maximal mechanical scan angle [42]. Electrostatic pull-in effect to around 40% of the maximal mechanical scan angle [42]. Electrostatic actu- actuators are relatively easy to fabricate by micromachining technologies. Parallel-plate ators are relatively easy to fabricate by micromachining technologies. Parallel-plate actu- actuators employ surface micromachining, often based on polysilicon with sacrificial oxide, ators employ surface micromachining, often based on polysilicon with sacrificial oxide, on electroplated metal with sacrificial organic layer or sputtered metal with a sacrificial on electroplated metal with sacrificial organic layer or sputtered metal with a sacrificial organic layer. Comb-drive actuators are typically fabricated on Silicon-on-Insulator (SOI) organic layer. Comb-drive actuators are typically fabricated on Silicon-on-Insulator (SOI) substrates, ensuring a relatively simple fabrication process and easy thickness control of substrates, ensuring a relatively simple fabrication process and easy thickness control of micromechanical structures. micromechanical structures. 4.2. Examples of Electrostatic OCT Probes 4.2. Examples of Electrostatic OCT Probes A series of endoscopic OCT probes based on electrostatic actuation have been pro- A series of endoscopic OCT probes based on electrostatic actuation have been pro- posed, using 2D MEMS scanners that scan in two axes [43–46] and employing a 2D gimbal- posed, using 2D MEMS scanners that scan in two axes [43–46] and employing a 2D gim- less vertical comb-drive structure. An interesting example of an OCT endoscopic MEMS bal-less vertical comb-drive structure. An interesting example of an OCT endoscopic scanner for high resolution OCT with angled vertical comb-drive actuators has been pro- MEMS scanner for high resolution OCT with angled vertical comb-drive actuators has posed by Aguirre et al. [47] at MIT. Figure 9a shows the SEM photography of this MEMS been proposed by Aguirre et al. [47] at MIT. Figure 9a shows the SEM photography of this scanner. The microscanner uses a torsional beam and includes a gimbal-mounting MEMS MEMS scanner. The microscanner uses a torsional beam and includes a gimbal-mounting mirror to scan a dual axis, combining the scan’s x and y axes with a single pivot point. The MEMS mirror to scan a dual axi  s, combining the scan’s x and y axes with a single pivot actuated mirror provides 6 angular scanning at over 100 V of driving voltage. Here, a point. The actuated mirror provides ±6° angular scanning at over 100 V of driving voltage. silicon micromirror is suspended inside a gimbal frame by a pair of polysilicon torsion springs. The scanning mirror has a circular aperture with a diameter of 1 mm within the footprint size of 3  3 mm . Photonics 2021, 8, x FOR PEER REVIEW 9 of 25 Here, a silicon micromirror is suspended inside a gimbal frame by a pair of polysilicon Photonics 2021, 8, 6 9 of 25 torsion springs. The scanning mirror has a circular aperture with a diameter of 1 mm within the footprint size of 3 × 3 mm . Figure 9. SEM photograph of 2D MEMS scanner in (a) and the schema of MEMS catheter packaging in (b) (Figures Figure 9. SEM photograph of 2D MEMS scanner in (a) and the schema of MEMS catheter packag- from [47]). ing in (b) (Figures from [47]). Figure 9b shows the schematic of OCT catheter packaging. The 2D MEMS scanner Figure 9b shows the schematic of OCT catheter packaging. The 2D MEMS scanner is is inclined at 45 and directs the beam in a side scanning configuration, orthogonally inclined at 45° and directs the beam in a side scanning configuration, orthogonally to the to the endoscope axis. The post-objective scanning eliminates off-axis optical aberration endoscope axis. The post-objective scanning eliminates off-axis optical aberration encoun- encountered with pre-objective scanning schemas. The endoscope head is 5 mm in diameter tered with pre-objective scanning schemas. The endoscope head is 5 mm in diameter and and 2.5 cm long. The optics include a graded index fiber collimator followed by an AR- 2.5 cm long. The optics include a graded index fiber collimator followed by an AR-coated coated achromatic focusing lens and which produces a beam spot diameter of 12 m. achromatic focusing lens and which produces a beam spot diameter of 12 µm. Figure 10a Figure 10a represents the resonance characteristics of the MEMS scanner. The mirror represents the resonance characteristics of the MEMS scanner. The mirror resonance is 463 resonance is 463 Hz and the gimbal axis resonance is around 140 Hz. Resonant operation of Hz and the gimbal axis resonance is around 140 Hz. Resonant operation of the mirror the mirror offers high speed raster scanning for en-face microscopy. The MEMS OCT probes offers high speed raster scanning for en-face microscopy. The MEMS OCT probes were were demonstrated in 3D high resolution OCT imaging. The OCT catheter was combined demonstrated in 3D high resolution OCT imaging. The OCT catheter was combined with with an OCT device employing a commercial femtosecond Nd:glass laser, centered on the an OCT device employing a commercial femtosecond Nd:glass laser, centered on the wavelengths of 1.06 m with a bandwidth of more than 200 nm. The light source was wavelengths of 1.06 µm with a bandwidth of more than 200 nm. The light source was coupled with a fiber-optic interferometer. The sample was measured with ~2000 axial scans coupled with a fiber-optic interferometer. The sample was measured with ~2000 axial per second. The interference signal was first amplified, filtered, and then demodulated by scans per second. The interference signal was first amplified, filtered, and then demodu- the detection block including a 12-bit, acquisition card, and a 5 MHz A/D converter and lated by the detection block including a 12-bit, acquisition card, and a 5 MHz A/D con- then processed by PC computer. The obtained axial resolution of images was <4 m in verter and then processed by PC computer. The obtained axial resolution of images was tissue, while transverse resolution was equivalent to the focusing spot of 12 m. Imaging <4 µm in tissue, while transverse resolution was equivalent to the focusing spot of 12 µm. Photonics 2021, 8, x FOR PEER REVIEW was performed at a rate of 4 frames/s over a 3D field of view of 1.8  1  1.3 mm 10 of 25 with Imaging was performed at a rate of 4 frames/s over a 3D field of view of 1.8 × 1 × 1.3 mm 500  500  1000 pixels. Figure 10b illustrates an example of a 3D image that represents a with 500 × 500 × 1000 pixels. Figure 10b illustrates an example of a 3D image that repre- volume data set from the hamster cheek pouch acquired in vitro. sents a volume data set from the hamster cheek pouch acquired in vitro. According to the schema of catheter packaging in Figure 9b, the OCT probe is limited in size by the footprint of the MEMS mirror—this is the main limit in the miniaturization of such OCT devices. High driving voltage is also an issue for in vivo endoscopy. The acquisition of data is made in open-loop mode, which introduces a possible lack of precise control of the MEMS mirror. This may degrade the scanning trajectory stability and data reproducibility. Addressing this concern will require one to produce more sophisticated MEMS technology, including the possibility of close-loop control detection of mirror mo- tion. An alternative based on the use of in-situ capacitive detection will be described in the next paragraph. Figure 10. FigureReso 10. Resonance nance characte characteristics ristics of the ofM the EMS MEMS scanner ( scanner a); and 3D OCT (a); and 3D OCT image acquired in image acquired vivo in vivo set from the set from hamster cheek po the hamster cheek uch ( pouch b) (Figures from [47]) (b) (Figures from [47 . ]). The interest to miniaturize OCT probes is not only limited to the endoscopic config- urations. Existing bulk microscopes or fiber optics devices for early diagnosis of cancer are expensive and are only affordable at the hospital; thus, they are not sufficiently used by physicians or cancer specialists as an early diagnosis tool. Significant reduction of sys- tem cost and size can be achieved by use of opto-mechanical components, fabricated by micromachining. In 2016, an OCT microsystem including an active 4 × 4 array of spectrally tuned Mirau interferometers, including an electrostatic vertical comb-drive actuator car- rying the array of reference mirrors, was proposed for dermatology applications [48]. The architecture of an active Mirau interferometer is shown schematically in Figure 11a. To perform OCT measurements, the device is incorporated within an experimental setup in- cluding a swept-source laser (center wavelength: 850 nm, swept range: 50 nm) and a high speed smart camera. A Mirau interferometer includes a series of vertically stacked components: a doublet of microlens matrices, a vertical comb-drive actuator, a spacer, and a planar beam splitter plate. The diameter of an individual microlens is 1.9 mm and the array pitch is 2 mm, while the equivalent focal length of a microlens doublet is 7.44 mm and the numerical aperture (NA) is 0.1. The assembly of two glass microlens arrays is made by anodic bond- ing. The axial resolution of OCT imaging is 6 µm, while the transverse resolution is limited to 6 µm. The depth of penetration is 0.6 mm. The key element of a Mirau interferometer is the vertical microscanner W2. The microscanner is designed for generating a vertical displacement of a large platform with a 4 × 4 array of reference micromirrors of the Mirau interferometer, as shown in Figure 11b. (a) (b) Figure 11. Cross-sectional view of the multichannel ”active” Mirau microinterferometer (a) with a focus on a 3D view of an electrostatic vertical microscanner with a 4 × 4 array of suspended reference micromirrors (b). Photonics 2021, 8, x FOR PEER REVIEW 10 of 25 Figure 10. Resonance characteristics of the MEMS scanner (a); and 3D OCT image acquired in vivo Photonics 2021, 8, 6 10 of 25 set from the hamster cheek pouch (b) (Figures from [47]). The interest to miniaturize OCT probes is not only limited to the endoscopic config- According to the schema of catheter packaging in Figure 9b, the OCT probe is limited urations. Existing bulk microscopes or fiber optics devices for early diagnosis of cancer in size by the footprint of the MEMS mirror—this is the main limit in the miniaturization are expensive and are only affordable at the hospital; thus, they are not sufficiently used of such OCT devices. High driving voltage is also an issue for in vivo endoscopy. The by physicians or cancer specialists as an early diagnosis tool. Significant reduction of sys- acquisition of data is made in open-loop mode, which introduces a possible lack of precise tem cost and size can be achieved by use of opto-mechanical components, fabricated by control of the MEMS mirror. This may degrade the scanning trajectory stability and data micromachining. In 2016, an OCT microsystem including an active 4 × 4 array of spectrally reproducibility. Addressing this concern will require one to produce more sophisticated tuned Mirau interferometers, including an electrostatic vertical comb-drive actuator car- MEMS technology, including the possibility of close-loop control detection of mirror motion. rying the array of reference mirrors, was proposed for dermatology applications [48]. The An alternative based on the use of in-situ capacitive detection will be described in the architecture of an active Mirau interferometer is shown schematically in Figure 11a. To next paragraph. perform OCT measurements, the device is incorporated within an experimental setup in- The interest to miniaturize OCT probes is not only limited to the endoscopic config- cluding a swept-source laser (center wavelength: 850 nm, swept range: 50 nm) and a high urations. Existing bulk microscopes or fiber optics devices for early diagnosis of cancer speed smart camera. are expensive and are only affordable at the hospital; thus, they are not sufficiently used A Mirau interferometer includes a series of vertically stacked components: a doublet by physicians or cancer specialists as an early diagnosis tool. Significant reduction of of microlens matrices, a vertical comb-drive actuator, a spacer, and a planar beam splitter system cost and size can be achieved by use of opto-mechanical components, fabricated plate. The diameter of an individual microlens is 1.9 mm and the array pitch is 2 mm, by micromachining. In 2016, an OCT microsystem including an active 4  4 array of spec- while the equivalent focal length of a microlens doublet is 7.44 mm and the numerical trally tuned Mirau interferometers, including an electrostatic vertical comb-drive actuator aperture (NA) is 0.1. The assembly of two glass microlens arrays is made by anodic bond- carrying the array of reference mirrors, was proposed for dermatology applications [48]. ing. The axial resolution of OCT imaging is 6 µm, while the transverse resolution is limited The architecture of an active Mirau interferometer is shown schematically in Figure 11a. to 6 µm. The depth of penetration is 0.6 mm. The key element of a Mirau interferometer To perform OCT measurements, the device is incorporated within an experimental setup is the vertical microscanner W2. The microscanner is designed for generating a vertical including a swept-source laser (center wavelength: 850 nm, swept range: 50 nm) and a displacement of a large platform with a 4 × 4 array of reference micromirrors of the Mirau high speed smart camera. interferometer, as shown in Figure 11b. (a) (b) Figure 11. Cross-sectional view of the multichannel ”active” Mirau microinterferometer (a) with a focus on a 3D view of Figure 11. Cross-sectional view of the multichannel ”active” Mirau microinterferometer (a) with a focus on a 3D view of an an electrostatic vertical microscanner with a 4 × 4 array of suspended reference micromirrors (b). electrostatic vertical microscanner with a 4  4 array of suspended reference micromirrors (b). A Mirau interferometer includes a series of vertically stacked components: a doublet of microlens matrices, a vertical comb-drive actuator, a spacer, and a planar beam splitter plate. The diameter of an individual microlens is 1.9 mm and the array pitch is 2 mm, while the equivalent focal length of a microlens doublet is 7.44 mm and the numerical aperture (NA) is 0.1. The assembly of two glass microlens arrays is made by anodic bonding. The axial resolution of OCT imaging is 6 m, while the transverse resolution is limited to 6 m. The depth of penetration is 0.6 mm. The key element of a Mirau interferometer is the vertical microscanner W2. The microscanner is designed for generating a vertical displacement of a large platform with a 4  4 array of reference micromirrors of the Mirau interferometer, as shown in Figure 11b. The vertical motion of the whole 4  4 array of reference micromirrors at the resonance frequency can be controlled precisely by an in-situ differential position sensor measuring the variation of capacitance due to the comb-drive displacement [49]. The vertical actuation of reference mirrors leads to a phase-shifted imaging that enables rapid measurement of the amplitudes and phases of interference signal and improves the signal to noise ratio and sensitivity. The size of an individual micromirror, suspended by a system of spider legs, is 400  400 m . The array of micromirrors is vertically aligned with the lenses, 2 2 forming an 8  8 mm structure. The resulting imager covers the same area of 8  8 mm of the sample, reconstructing the topography in a continuous way by stitching together Photonics 2021, 8, x FOR PEER REVIEW 11 of 25 The vertical motion of the whole 4 × 4 array of reference micromirrors at the reso- nance frequency can be controlled precisely by an in-situ differential position sensor measuring the variation of capacitance due to the comb-drive displacement [49]. The ver- tical actuation of reference mirrors leads to a phase-shifted imaging that enables rapid measurement of the amplitudes and phases of interference signal and improves the signal Photonics 2021, 8, 6 to noise ratio and sensitivity. The size of an individual micromirror, suspended by a s 11 y ofs- 25 tem of spider legs, is 400 × 400 µm . The array of micromirrors is vertically aligned with the lenses, forming an 8 × 8 mm structure. The resulting imager covers the same area of 8 × 8 mm of the sample, reconstructing the topography in a continuous way by stitching 4  4 single-channel interferograms via a system of actuators shifting mechanically the together 4 × 4 single-channel interferograms via a system of actuators shifting mechani- entire Mirau-array over the overlapping region. The 3 mm thick spacer W3 in silicon cally the entire Mirau-array over the overlapping region. The 3 mm thick spacer W3 in adjusts the position of lens focus from the planar beam splitter plate. The beam-splitter silicon adjusts the position of lens focus from the planar beam splitter plate. The beam- W4 has a transmission-reflection ratio of 70/30. Figure 12a shows the assembled Mirau splitter W4 has a transmission-reflection ratio of 70/30. Figure 12a shows the assembled interferometer mounted on the PCB (printed circuit board). The footprint of this chip is Mirau interferometer mounted on the PCB (printed circuit board). The footprint of this chip 15  15 mm , whereas the overall thickness is about 5 mm. is 15 × 15 mm , whereas the overall thickness is about 5 mm. (a) (b) Figure 12. PCB-mounted chip of a Mirau interferometer in (a); and 3D swept-source OCT image of Figure 12. PCB-mounted chip of a Mirau interferometer in (a); and 3D swept-source OCT image of onion slices in (b). onion slices in (b). The original A-scan includes several parasitic terms such as the autocorrelation terms The original A-scan includes several parasitic terms such as the autocorrelation terms due to the beam splitter, reference mirrors images, mirror replica images, and DC noise due to the beam splitter, reference mirrors images, mirror replica images, and DC noise term, making the interpretation of true OCT signal difficult. Help to the use of four-frame term, making the interpretation of true OCT signal difficult. Help to the use of four-frame phase shift algorithm all these signals are removed, improving both the signal-to-noise phase shift algorithm all these signals are removed, improving both the signal-to-noise ratio and the measurements range. Figure 12b shows a volumetric 300 × 300 × 600 µm ratio and the measurements range. Figure 12b shows a volumetric 300  300  600 m OCT image o OCT image f of an on an onion ion slic slice, e, where the where the microsco microscopic pic structure is visible. The structure is visible. The sensitivity of sensitivity of this im this image age is in the ran is in the range ge oof f 80 dB 80 dB. . 5. Electromagnetic MEMS Scanning Mirrors for OCT 5. Electromagnetic MEMS Scanning Mirrors for OCT 5.1. Principles of Electromagnetic Microactuators 5.1. Principles of Electromagnetic Microactuators Electromagnetic MEMS actuators are driven by Lorentz force [50]. In this case, a Electromagnetic MEMS actuators are driven by Lorentz force [50]. In this case, a cur- current-carrying conductor is placed in a static magnetic field. This field produced around rent-carrying conductor is placed in a static magnetic field. This field produced around the conductor interacts with the static field to produce a force. In an electromagnetic the conductor interacts with the static field to produce a force. In an electromagnetic ac- actuator, including the flexure beams and a moving mirror plate, the module of Lorentz tuator, including the flexure beams and a moving mirror plate, the module of Lorentz force is expressed as: force is expressed as: F = BIL sin q, (6) = sin , (6) where B is magnetic flux density of the magnetic field, I is the current flowing through the beam, L is the beam length, and q is the angle between the current and the magnetic field. where B is magnetic flux density of the magnetic field, I is the current flowing through the beam, Ther L is te he beam are numer lengt ous h, variations and is the on an the gle between architectur th ee cur of the rent electr and the m omagnetic agnetic field. actuators: permanent magnets interacting with an external field, permanent magnets interacting with There are numerous variations on the architecture of the electromagnetic actuators: current-carrying coils, and current carrying conductors interacting with an external field. permanent magnets interacting with an external field, permanent magnets interacting A common advantage is the relatively high generated force. The main drawbacks are with current-carrying coils, and current carrying conductors interacting with an external the high-power dissipation as well as a complex fabrication, including severe materials field. A common advantage is the relatively high generated force. The main drawbacks challenges, and difficulty to miniaturize the micromirrors because of the use of external bulk magnets. Figure 13 shows the schematic design of a 2D electromagnetic scanner where the Lorentz force interaction is generated between the micro-coil integrated on the scanner gimbal and the permanent magnets located outside of the scanner. To decrease the size of electromagnetic actuators, a combined electrostatic/electromag- netic 2D scanner has been developed by Microvision for retinal scanning displays [29]. Such a mixed configuration employed electromagnetic actuation to move the outer frame, which provides the slow vertical axis, while electrostatic actuation was used for the inner mirror axis, which provides the fast horizontal axis. Photonics 2021, 8, x FOR PEER REVIEW 12 of 25 are the high-power dissipation as well as a complex fabrication, including severe materi- als challenges, and difficulty to miniaturize the micromirrors because of the use of exter- nal bulk magnets. Figure 13 shows the schematic design of a 2D electromagnetic scanner Photonics 2021, 8, 6 12 of 25 where the Lorentz force interaction is generated between the micro-coil integrated on the scanner gimbal and the permanent magnets located outside of the scanner. Figure 13. Figure 13. 2D e 2D electr lectromagnetic omagnetic actuator actuator generating the generating the Lor Loren entztfor z force ce between betweethe n the micro-co micro-coil integrated il inte- grated on the scanner gimbal and the two permanent magnets. on the scanner gimbal and the two permanent magnets. 5.2. Examples of Electromagnetic OCT Probes To decrease the size of electromagnetic actuators, a combined electrostatic/electro- magnetic 2D scanner has been developed by Microvision for retinal scanning displays Magnetically actuated scanning mirrors made by micromachining were demonstrated [29]. Such a mixed configuration employed electromagnetic actuation to move the outer by Judy et al. [51]. The Olympus Company developed one of the first 1D electromagnetic frame MEMS , which scanners provides t for confocal he slow vert microscopy ical axis [52], , whil while e el an ectrosta early 2D tic electr actuaomagnetic tion was used MEMS for tscanner he inner m was irrpr or ax oposed is, whic byh p Asada rovides et al. the f [53 a]. st hor This izon cat tegory al axis.of MEMS scanning mirrors replaced the electrostatic actuators in situations where it was necessary to increase the 5.2. scanning Examples range of Electroma or lower gn the etic OCT driving Prob voltage. es Serious efforts were performed to make the commercialization of MEMS electromagnetic scanner OCT probes easier for clinical Magnetically actuated scanning mirrors made by micromachining were demon- applications. To overcome the fabrication complexity of earlier electromagnetic actuators, strated by Judy et al. [51]. The Olympus Company developed one of the first 1D electro- for example, a flexible 2-axis polydimethylsiloxane (PDMS)-based electromagnetic MEMS magnetic MEMS scanners for confocal microscopy [52], while an early 2D electromagnetic scanning mirror was developed [54]. The size of such a MEMS scanner remained relatively MEMS scanner was proposed by Asada et al. [53]. This category of MEMS scanning mir- big (15  15  15 mm ). rors replaced the electrostatic actuators in situations where it was necessary to increase An interesting electromagnetic scanning actuator for OCT imaging was demonstrated the scanning range or lower the driving voltage. Serious efforts were performed to make by Kim et al. [55]. Figure 14 shows the photographs of a two-axis gimbaled mirror with the commercialization of MEMS electromagnetic scanner OCT probes easier for clinical folded flexure hinges. The rotation of the mirror is possible in two axes along the flexures: applications. To overcome the fabrication complexity of earlier electromagnetic actuators, one inner axis and one orthogonally placed outer axis. Resonant frequencies for the inner for example, a flexible 2-axis polydimethylsiloxane (PDMS)-based electromagnetic MEMS and outer axes were 450 Hz and 350 Hz, respectively. To generate the magnetic actuation, a scanning mirror was developed [54]. The size of such a MEMS scanner remained relatively permanent magnet glued to the backside of the mirror plate and a pair of coils is placed Photonics 2021, 8, x FOR PEER REVIEW 13 of 25 big (15 × 15 × 15 mm ). inside the probe body for each scan direction. The mirror plate is 0.6  0.8 mm , with a An interesting electromagnetic scanning actuator for OCT imaging was demon- device footprint of 2.4  2.9 mm . strated by Kim et al. [55]. Figure 14 shows the photographs of a two-axis gimbaled mirror with folded flexure hinges. The rotation of the mirror is possible in two axes along the flexures: one inner axis and one orthogonally placed outer axis. Resonant frequencies for the inner and outer axes were 450 Hz and 350 Hz, respectively. To generate the magnetic actuation, a permanent magnet glued to the backside of the mirror plate and a pair of coils is placed inside the probe body for each scan direction. The mirror plate is 0.6 × 0.8 mm , with a device footprint of 2.4 × 2.9 mm . Figure 14. Photograph of electromagnetic micro-scanner in (a) and a SEM image of folded flexure Figure 14. Photograph of electromagnetic micro-scanner in (a) and a SEM image of folded flexure hinges in (b) (Figures from [55]). hinges in (b) (Figures from [55]). Figure 15a shows a schematic of the assembled catheter packaging. The light source Figure 15a shows a schematic of the assembled catheter packaging. The light source is pigtailed by a single mode optical fiber, delivering via a GRIN lens a focused beam is pigtailed by a single mode optical fiber, delivering via a GRIN lens a focused beam redirected by the MEMS mirror on the sample, which is then scanned. Scattered light from the sample returns through the same optical path and is collected by the pigtailed GRIN lens. A glass window with an AR (anti reflective) coating protects the MEMS scanner and eliminates the back reflections. The catheter package has a 2.8 mm diameter and a 12 mm length. The MEMS scanner is fabricated from a SOI wafer with a 50 µm thick device layer on a 350 µm thick handle layer, including a 1 µm thick oxide box layer. Magnet layers are composed of small NdFeB magnets, each one measuring 0.6 × 0.8 × 0.18 mm . Figure 15b represents optical angles of the MEMS scanner in both inner and outer axes as a function of the driving voltage. An optical scan angle of about ±30° was obtained with ±1.2 V and ±4 V driving voltages for the inner and outer axis, corresponding to 50 mA and 100 mA current, respectively. The light refracted by the protection window produces slight nonlinearity in the de- flection angle due to thickness variations of the window. Spurious vibrations are observed for large scan angles at the mirror resonant frequency. To avoid these effects, the working scan angle was reduced to ±20° optical angle for the inner axis and less than ±30° optical angle for the outer axis. Optical resolution was estimated to be 5 µm. In vivo 3D endo- scopic imaging of tissues was made by combining the two-axis scanning catheters and the multifunctional SD-OCT system. 3D images of a fingertip were acquired at 18.5 frames/s, with the scan performed with voltages of ±2.8 V and ±0.8 V applied on the inner and outer axis, covering 1.5 × 1 mm lateral scan range and consuming 150 mW of power. Figure 15. Schematic of the assembled catheter packaging in (a) and optical angles of the MEMS scanner for both axes vs. the driving voltage in (b) (Figures from [55]). Photonics 2021, 8, x FOR PEER REVIEW 13 of 25 Figure 14. Photograph of electromagnetic micro-scanner in (a) and a SEM image of folded flexure hinges in (b) (Figures from [55]). Figure 15a shows a schematic of the assembled catheter packaging. The light source is pigtailed by a single mode optical fiber, delivering via a GRIN lens a focused beam redirected by the MEMS mirror on the sample, which is then scanned. Scattered light from the sample returns through the same optical path and is collected by the pigtailed GRIN lens. A glass window with an AR (anti reflective) coating protects the MEMS scanner and eliminates the back reflections. The catheter package has a 2.8 mm diameter and a 12 mm length. The MEMS scanner is fabricated from a SOI wafer with a 50 µm thick device layer on a 350 µm thick handle layer, including a 1 µm thick oxide box layer. Magnet layers are composed of small NdFeB magnets, each one measuring 0.6 × 0.8 × 0.18 mm . Figure 15b Photonics 2021, 8, 6 13 of 25 represents optical angles of the MEMS scanner in both inner and outer axes as a function of the driving voltage. An optical scan angle of about ±30° was obtained with ±1.2 V and ±4 V driving voltages for the inner and outer axis, corresponding to 50 mA and 100 mA redirected by the MEMS mirror on the sample, which is then scanned. Scattered light from current, respectively. the sample returns through the same optical path and is collected by the pigtailed GRIN The light refracted by the protection window produces slight nonlinearity in the de- lens. A glass window with an AR (anti reflective) coating protects the MEMS scanner and flection angle due to thickness variations of the window. Spurious vibrations are observed eliminates the back reflections. The catheter package has a 2.8 mm diameter and a 12 mm for large scan angles at the mirror resonant frequency. To avoid these effects, the working length. The MEMS scanner is fabricated from a SOI wafer with a 50 m thick device layer scan angle was reduced to ±20° optical angle for the inner axis and less than ±30° optical on a 350 m thick handle layer, including a 1 m thick oxide box layer. Magnet layers are angle for the outer axis. Optical resolution was estimated to be 5 µm. In vivo 3D endo- composed of small NdFeB magnets, each one measuring 0.6  0.8  0.18 mm . Figure 15b scopic imaging of tissues was made by combining the two-axis scanning catheters and the represents optical angles of the MEMS scanner in both inner and outer axes as a function multifunctional SD-OCT system. 3D images of a fingertip were acquired at 18.5 frames/s, of the driving voltage. An optical scan angle of about 30 was obtained with 1.2 V and with the scan performed with voltages of ±2.8 V and ±0.8 V applied on the inner and outer 4 V driving voltages for the inner and outer axis, corresponding to 50 mA and 100 mA axis, covering 1.5 × 1 mm lateral scan range and consuming 150 mW of power. current, respectively. Figure 15. Schematic of the assembled catheter packaging in (a) and optical angles of the MEMS scanner for both axes vs. Figure 15. Schematic of the assembled catheter packaging in (a) and optical angles of the MEMS scanner for both axes vs. the driving voltage in (b) (Figures from [55]). the driving voltage in (b) (Figures from [55]). The light refracted by the protection window produces slight nonlinearity in the de- flection angle due to thickness variations of the window. Spurious vibrations are observed for large scan angles at the mirror resonant frequency. To avoid these effects, the working scan angle was reduced to 20 optical angle for the inner axis and less than 30 optical angle for the outer axis. Optical resolution was estimated to be 5 m. In vivo 3D endo- scopic imaging of tissues was made by combining the two-axis scanning catheters and the Photonics 2021, 8, x FOR PEER REVIEW 14 of 25 multifunctional SD-OCT system. 3D images of a fingertip were acquired at 18.5 frames/s, with the scan performed with voltages of 2.8 V and 0.8 V applied on the inner and outer axis, covering 1.5 1 mm lateral scan range and consuming 150 mW of power. Figure 16 shows a 3D OCT image of fingertip tissue where the fingerprint orienta- Figure 16 shows a 3D OCT image of fingertip tissue where the fingerprint orientations tions are visible. are visible. Figure 16. 3D OCT image of fingertip tissue where the fingerprint orientations are visible (Figure Figure 16. 3D OCT image of fingertip tissue where the fingerprint orientations are visible (Figure from [55]). from [55]). Watanabe et al. [56] have demonstrated an electromagnetic MEMS scanner for OCT Watanabe et al. [56] have demonstrated an electromagnetic MEMS scanner for OCT imaging. The schematic diagram of a fabricated mirror scanner is shown in Figure 17. The imaging. The schematic diagram of a fabricated mirror scanner is shown in Figure 17. The device device incl includes udes a 0. a 0.2 2 mm mm thick thick si silicon licon fr frame ame c carrying arrying tthe he micromir micromirr ror orwit with h it its s ac actuator tuator, a , printed circuit board, and a magnet holder with a magnet inside. The minimum size of 2 2 the magnet is 6 × 6 × 5 mm . A metal coated 1.8 × 1.8 mm mirror and two y-scan coils are formed on the y-frame in the center of the silicon frame. The folded y-scan beams are supported by an x-frame. Two x-scan coils are formed on the x-frame. All the coils have a dimension of 2 × 2 mm . The folded x-scan beams are supported by an external fixed frame. When a current is passed through the y-scan coils, the mirror deflects in the y di- rection. When a current is passed through the four x-scan coils, the mirror tilts in the x direction. Thus, the light beam can be 2D steered. The entire microscanner is mounted on a 15 × 6 mm PCB, which is fixed on the magnet holder. The MEMS microscanner was placed in a Fourier domain OCT setup including a SLD light source operating at 1.55 µm and a fiber optic Michelson interferometer. OCT images of human fingers were obtained, as shown in Figure 18. The scanners discussed here demonstrate that one of the main drawbacks of electro- magnetic mirrors is that an external magnet is required for actuation. Such magnet tech- nology is often not compatible with the process flow of the micro-scanner. A bulky magnet reduces the potential of miniaturization of the probe. Another drawback concerns the rel- atively high-power consumption of electromagnetic scanners. Finally, the relative com- plexity of the fabrication process and high costs are a bottleneck for fast clinical translation of the electromagnetic MEMS scanners. Photonics 2021, 8, 6 14 of 25 a printed circuit board, and a magnet holder with a magnet inside. The minimum size 2 2 of the magnet is 6  6  5 mm . A metal coated 1.8  1.8 mm mirror and two y-scan coils are formed on the y-frame in the center of the silicon frame. The folded y-scan beams are supported by an x-frame. Two x-scan coils are formed on the x-frame. All the coils have a dimension of 2  2 mm . The folded x-scan beams are supported by an external fixed frame. When a current is passed through the y-scan coils, the mirror deflects in the y direction. When a current is passed through the four x-scan coils, the mirror tilts in the x Photonics 2021, 8, x FOR PEER REVIEW 15 of 25 direction. Thus, the light beam can be 2D steered. The entire microscanner is mounted on a 15  6 mm PCB, which is fixed on the magnet holder. Photonics 2021, 8, x FOR PEER REVIEW 15 of 25 Figure 17. Architecture of the MEMS electromagnetic scanner: (a) device schema and (b) the top view of the mirror actuator Figure 17. Architecture of the MEMS electromagnetic scanner: (a) device schema and (b) the top view of the mirror actu- (Figures from [56]). ator (Figures from [56]). The MEMS microscanner was placed in a Fourier domain OCT setup including a SLD Figure 17. Architecture of the MEMS electromagnetic scanner: (a) device schema and (b) the top view of the mirror actu- light source operating at 1.55 m and a fiber optic Michelson interferometer. OCT images ator (Figures from [56]). of human fingers were obtained, as shown in Figure 18. Figure 18. 3 × 3 mm OCT images of human fingers: the epidermis including stratum corneum and the crista cutis (Figures from [56]). Figure 18. 3  3 mm OCT images of human fingers: the epidermis including stratum corneum and the crista cutis (Figures Figure 18. 3 × 3 mm OCT images of human fingers: the epidermis including stratum corneum and the crista cutis (Figures 6. Electrothermal MEMS Scanning Mirrors for OCT from [56]). from [56]). 6.1. Principles of Electrothermal Microactuators The scanners discussed here demonstrate that one of the main drawbacks of elec- The principle of an electrothermal actuator is based on the Joule heating and thermal 6. Electrothermal MEMS Scanning Mirrors for OCT tromagnetic mirrors is that an external magnet is required for actuation. Such magnet expansion principles. The actuation uses the balance between the thermal energy gener- 6.1. Principles of Electrothermal Microactuators technology is often not compatible with the process flow of the micro-scanner. A bulky ated by an electrical current and the heat dissipation through the actuator structure The principle of an electrothermal actuator is based on the Joule heating and thermal magnet reduces the potential of miniaturization of the probe. Another drawback con- [57,58]. The three categories of electrothermal actuators are bimorph actuators, chevron expansion principles. The actuation uses the balance between the thermal energy gener- cerns the relatively high-power consumption of electromagnetic scanners. Finally, the actuators, and hot-and-cold-arm actuators. The structure of the more popular bimorph ated by an electrical current and the heat dissipation through the actuator structure actuator contains two layers of materials with different coefficients of thermal expansion [57,58]. The three categories of electrothermal actuators are bimorph actuators, chevron (CTE). A metallic heater is sandwiched between the active materials, as shown in Fig- actuators, and hot-and-cold-arm actuators. The structure of the more popular bimorph ure 19. The injection of electrical current within the heater layer generates the Joule effect actuator contains two layers of materials with different coefficients of thermal expansion in the active materials and produces the deflection angle. Equation (7) shows that the me- (CTE). A metallic heater is sandwiched between the active materials, as shown in Fig- chanical strain of material is directly proportional to the temperature change ∆ : ure 19. The injection of electrical current within the heater layer generates the Joule effect = ∆. (7) in the active materials and produces the deflection angle. Equation (7) shows that the me- chanical strain of material is directly proportional to the temperature change ∆ : The curvature of generated mechanical bending can be approximated as = ∆. (7) = , (8) The curvature of generated mechanical bending can be approximated as = , (8) Photonics 2021, 8, 6 15 of 25 relative complexity of the fabrication process and high costs are a bottleneck for fast clinical translation of the electromagnetic MEMS scanners. 6. Electrothermal MEMS Scanning Mirrors for OCT 6.1. Principles of Electrothermal Microactuators The principle of an electrothermal actuator is based on the Joule heating and thermal expansion principles. The actuation uses the balance between the thermal energy generated by an electrical current and the heat dissipation through the actuator structure [57,58]. The three categories of electrothermal actuators are bimorph actuators, chevron actuators, and hot-and-cold-arm actuators. The structure of the more popular bimorph actuator contains two layers of materials with different coefficients of thermal expansion a (CTE). A metallic heater is sandwiched between the active materials, as shown in Figure 19. The injection of electrical current within the heater layer generates the Joule effect in the active materials and produces the deflection angle. Equation (7) shows that the mechanical strain of material # is directly proportional to the temperature change DT: # = a DT. (7) The curvature of generated mechanical bending can be approximated as Photonics 2021, 8, x FOR PEER REVIEW 16 of 25 t + t R = , (8) # # 1 2 where t is the thickness of active layers and # represents the thermal strain of ac- where 1,2 is the thickness of active layers and 1,2 represents the thermal strain of active , , tive layers. A wide range of active materials can be used. Thus, the CTE of silicon −6 is layers. A wide range of active materials can be used. Thus, the CTE of silicon is 2.6 × 10 /K, 6 6 6 2.6  10 /K, while that of−6 SiO is 0.35  10 /K, aluminum −6 CTE is 25  10 /K, and while that of SiO2 is 0.35 × 10 /K, aluminum CTE is 25 × 10 /K, and the CTE of polyimide the CTE of polyimide from Amoco Ultradel −6 1414 is 191  10 /K. from Amoco Ultradel 1414 is 191 × 10 /K. Figure 19. Principle of electrothermal actuation. Figure 19. Principle of electrothermal actuation. 6.2. Examples of Electrothermal OCT Probes 6.2. Examples of Electrothermal OCT Probes In electrothermal actuators, the actuation force is typically larger than that of elec- In electrothermal actuators, the actuation force is typically larger than that of electro- trostatic and electromagnetic actuators. They benefit also from high fill factor. Such static and electromagnetic actuators. They benefit also from high fill factor. Such charac- characteristics make the electrothermal bimorph-based MEMS scanners very suitable for teristics make the electrothermal bimorph-based MEMS scanners very suitable for minia- miniaturizing of OCT probes for endoscopic applications. An example of such a probe, turizing of OCT probes for endoscopic applications. An example of such a probe, based based on a 2D thermal bimorph micromirror, has been developed by Sun et al. [59]. The on a 2D thermal bimorph micromirror, has been developed by Sun et al. [59]. The resulting resulting microscanner includes a mirror suspended without the gimbal by four actuators microscanner includes a mirror suspended without the gimbal by four actuators on its on its four sides. The actuator is based on three Al/SiO bimorph beams connected in four sides. The actuator is based on three Al/SiO2 bimorph beams connected in series with series with a Pt heater embedded for electrothermal actuation and two rigid suspended a Pt heater embedded for electrothermal actuation and two rigid suspended silicon silicon frames. This design is called lateral-shift-free (LSF) LVD design [60]. frames. This design is called lateral-shift-free (LSF) LVD design [60]. More recently, the architecture of a swept-source OCT endomicroscope has been demonstrated, including a mirror scanner with a similar mechanism of actuation [61]. The OCT probe contains a spectrally tuned single-channel Mirau micro-interferometer, inte- grated with a two axis MEMS electro-thermal micro-scanner. This optical microsystem operates in the side-imaging mode. Figure 20 shows the schematic of an OCT microsystem where a 4.7 × 4.7 × 5.3 mm Mirau microinterferometer is shown inside the blue square. The size of the microscanner external frame is 4 × 4 mm , with a mirror diameter of 1 mm. The monolithically integrated Mirau microinterferometer [62] includes a silicon base for a GRIN lens assembly port, a glass wafer with a reflowed focusing lens with a focal length of 9 mm, a reference micro- mirror, a silicon separator, and a beam splitter plate. The GRIN lens generates a collimated light beam with a diameter of 1 mm, which illuminates the plano-convex Mirau glass lens of 1.9 mm diameter [63]. The glass beam-splitting plate divides the converging beam into a reference beam and a scanning beam. A silicon separator ensures the position of the beam splitter plate at half of the focal length of the focusing lens. The reference beam is back-reflected from the 150 µm reference micromirror at the backside of the focusing lens, whereas the scanning beam is directed by the MEMS scanner towards the sample to be measured. The schematic of a MEMS scanner, based on a two-axis MEMS electrothermal micro-mirror, is shown in Figure 21a, while the image of a MEMS scanner assembled on top of a Mirau interferometer is shown in Figure 21b [58]. Photonics 2021, 8, 6 16 of 25 More recently, the architecture of a swept-source OCT endomicroscope has been demonstrated, including a mirror scanner with a similar mechanism of actuation [61]. The OCT probe contains a spectrally tuned single-channel Mirau micro-interferometer, integrated with a two axis MEMS electro-thermal micro-scanner. This optical microsystem operates in the side-imaging mode. Figure 20 shows the schematic of an OCT microsystem where a 4.7  4.7  5.3 mm Mirau microinterferometer is shown inside the blue square. The size of the microscanner external frame is 4  4 mm , with a mirror diameter of 1 mm. The monolithically integrated Mirau microinterferometer [62] includes a silicon base for a GRIN lens assembly port, a glass wafer with a reflowed focusing lens with a focal length of 9 mm, a reference micromirror, a silicon separator, and a beam splitter plate. The GRIN lens generates a collimated light beam with a diameter of 1 mm, which illuminates the plano-convex Mirau glass lens of 1.9 mm diameter [63]. The glass beam-splitting plate divides the converging beam into a reference beam and a scanning beam. A silicon separator ensures the position of the beam splitter plate at half of the focal length of the focusing lens. The reference beam is back-reflected from the 150 m reference micromirror at the backside of the focusing lens, whereas the scanning beam is directed by the MEMS scanner towards the sample to be measured. The schematic of a MEMS scanner, based on a two-axis MEMS electrothermal micro-mirror, is shown in Figure 21a, while the image of a MEMS scanner assembled on top of a Mirau interferometer is shown in Figure 21b [58]. The inner mirror plate is connected to a rigid frame via a pair of torsional bars in two diametrically opposite ends located on the rotation axis. A pair of electrothermal bimorphs generates a force onto the perpendicular free ends of the mirror plate in the same angular direction. An array of electrothermal bimorph cantilevers deflects the rigid frame and a mechanical stopper maintains the position of the mirror inclined at 45 from the optical axis. The performed scans reach large mechanical angles of 32 for the frame mirror and 22 for the in-frame mirror. Figure 21c,d shows the deflection amplitude of the micromirror versus the frequency response of the pitch-axis and roll-axis, respectively. Here, a small coupling between both the axes is observed. The fabricated micromirror has a mechanical resonant frequency around 1200 Hz for both axes. Figure 22a shows the real-time reconstruction of a “femto-st” pattern, obtained by Lissajous imaging at a sampling frequency of 1 MHz [64]. The acquisition of data was Photonics 2021, 8, x FOR PEER REVIEW 17 of 25 performed in open-loop mode. The pattern lines of 30 m-wide are well-resolved. Figure 20. Schematic diagram of the MOEMS (Micro-Opto-Electro-Mechanical Systems) probe. Figure 20. Schematic diagram of the MOEMS (Micro-Opto-Electro-Mechanical Systems) probe. Figure 21. Electrothermal mirror: (a) schematic and (b) microphotograph of a MEMS scanner assembled on top of a Mirau interferometer. The frequency response for inner pitch axis (c) and roll-axis (d). The inner mirror plate is connected to a rigid frame via a pair of torsional bars in two diametrically opposite ends located on the rotation axis. A pair of electrothermal bi- morphs generates a force onto the perpendicular free ends of the mirror plate in the same angular direction. An array of electrothermal bimorph cantilevers deflects the rigid frame and a mechanical stopper maintains the position of the mirror inclined at 45° from the optical axis. The performed scans reach large mechanical angles of 32° for the frame mirror and 22° for the in-frame mirror. Figure 21c,d shows the deflection amplitude of the micro- mirror versus the frequency response of the pitch-axis and roll-axis, respectively. Here, a small coupling between both the axes is observed. The fabricated micromirror has a me- chanical resonant frequency around 1200 Hz for both axes. Figure 22a shows the real-time reconstruction of a “femto-st” pattern, obtained by Lissajous imaging at a sampling frequency of 1 MHz [64]. The acquisition of data was performed in open-loop mode. The pattern lines of 30 µm-wide are well-resolved. Photonics 2021, 8, x FOR PEER REVIEW 17 of 25 Photonics 2021, 8, 6 17 of 25 Figure 20. Schematic diagram of the MOEMS (Micro-Opto-Electro-Mechanical Systems) probe. Photonics 2021, 8, x FOR PEER REVIEW 18 of 25 Figure 21. Electrothermal mirror: (a) schematic and (b) microphotograph of a MEMS scanner assembled on top of a Mirau Figure 21. Electrothermal mirror: (a) schematic and (b) microphotograph of a MEMS scanner assembled on top of a Mirau interferometer. The frequency response for inner pitch axis (c) and roll-axis (d). interferometer. The frequency response for inner pitch axis (c) and roll-axis (d). The inner mirror plate is connected to a rigid frame via a pair of torsional bars in two diametrically opposite ends located on the rotation axis. A pair of electrothermal bi- morphs generates a force onto the perpendicular free ends of the mirror plate in the same angular direction. An array of electrothermal bimorph cantilevers deflects the rigid frame and a mechanical stopper maintains the position of the mirror inclined at 45° from the optical axis. The performed scans reach large mechanical angles of 32° for the frame mirror and 22° for the in-frame mirror. Figure 21c,d shows the deflection amplitude of the micro- mirror versus the frequency response of the pitch-axis and roll-axis, respectively. Here, a small coupling between both the axes is observed. The fabricated micromirror has a me- chanical resonant frequency around 1200 Hz for both axes. Figure 22. Reconstruction of a “femto-st” pattern in (a) and B-scan of a multilayer glass sample in (b). Figure 22. Reconstruction of a “femto-st” pattern in (a) and B-scan of a multilayer glass sample in (b). Figure 22a shows the real-time reconstruction of a “femto-st” pattern, obtained by The complete OCT probe is connected to the illumination and detection blocks by Lissajous imaging at a sampling frequency of 1 MHz [64]. The acquisition of data was The complete OCT probe is connected to the illumination and detection blocks by a a single-mode optical fiber. The system is illuminated by a swept source with a central performed in open-loop mode. The pattern lines of 30 µm-wide are well-resolved. single-mode optical fiber. The system is illuminated by a swept source with a central wavelength of 840 nm and A-scan frequency of 110 kHz [65]. The OCT images were wavelength of 840 nm and A-scan frequency of 110 kHz [65]. The OCT images were ob- obtained from a sample made of three cover glasses, each 160 m thick. Figure 22b shows tained from a sample made of three cover glasses, each 160 µm thick. Figure 22b shows the averaged B-scan images of this sample. the averaged B-scan images of this sample. Circumferential scanning for endoscopic OCT with MEMS electrothermal mirrors was Circumferential scanning for endoscopic OCT with MEMS electrothermal mirrors demonstrated in 2018 by S. Luo et al. [66]. This microscanner uses a circular array of six was demonstrated in 2018 by S. Luo et al. [66]. This microscanner uses a circular array of scan units, including electrothermal MEMS mirrors and C-lens collimators with a focal six scan units, including electrothermal MEMS mirrors and C-lens collimators with a focal length greater than 10 mm, as shown in Figure 23. This compact microscanner presented a length greater than 10 mm, as shown in Figure 23. This compact microscanner presented chip size of 1.5  1.3 mm . Each C-lens and a single-mode fiber are packaged inside a glass a chip size of 1.5 × 1.3 mm . Each C-lens and a single-mode fiber are packaged inside a tube with a diameter of 1.4 mm. The full circumferential scans have been demonstrated glass tube with a diameter of 1.4 mm. The full circumferential scans have been demon- with individual micromirrors scanning up to 45 at a voltage of less than 12 V. strated with individual micromirrors scanning up to 45° at a voltage of less than 12 V. Figure 24a shows the MEMS mirrors with a 0.5  0.5 mm mirror plate. Four bimorph actuators are placed symmetrically at the four sides of a central mirror plate. Each bimorph actuator consists of three pairs of double S-shaped bimorph (Al/SiO2) beams, sandwiched with a heater layer made of a thin film of Ti/TiN. A mechanical scan angle of 13 is achieved, resulting in a 26 optical scan angle or a 52 field of view (FOV). Figure 23. The schematic of MEMS OCT probe (Figures from [66]). Figure 24a shows the MEMS mirrors with a 0.5 × 0.5 mm mirror plate. Four bimorph actuators are placed symmetrically at the four sides of a central mirror plate. Each bi- morph actuator consists of three pairs of double S-shaped bimorph (Al/SiO2) beams, sand- wiched with a heater layer made of a thin film of Ti/TiN. A mechanical scan angle of 13° is achieved, resulting in a ±26° optical scan angle or a 52° field of view (FOV). Configured with a swept-source OCT setup, this MEMS array-based circumferential scanning probe was applied to image a swine’s small intestine wrapped on a 20 mm-di- ameter glass tube, as shown in Figure 24b. The OCT imaging result shows that this new MEMS endoscopic OCT has promising applications in large tubular organs. An alternative solution is to move the objective lens directly by an actuated mi- crostage. L. Wu et al. [67] demonstrated a tunable microlens scanner, operating at a 880 nm wavelength. This microscanner uses a lateral-shift-free electrothermal bimorph actu- ator, carrying a 1 mm diameter glass rod lens moving at a resonance frequency of 79 Hz. Later, L. Liu et al. [68] developed another electrothermal MEMS microlens scanner, mov- ing a 2.4 mm plano-convex glass lens with a maximum travel range of 400 µm and a res- onance frequency of 24 Hz. Photonics 2021, 8, x FOR PEER REVIEW 18 of 25 Figure 22. Reconstruction of a “femto-st” pattern in (a) and B-scan of a multilayer glass sample in (b). The complete OCT probe is connected to the illumination and detection blocks by a single-mode optical fiber. The system is illuminated by a swept source with a central wavelength of 840 nm and A-scan frequency of 110 kHz [65]. The OCT images were ob- tained from a sample made of three cover glasses, each 160 µm thick. Figure 22b shows the averaged B-scan images of this sample. Circumferential scanning for endoscopic OCT with MEMS electrothermal mirrors was demonstrated in 2018 by S. Luo et al. [66]. This microscanner uses a circular array of six scan units, including electrothermal MEMS mirrors and C-lens collimators with a focal length greater than 10 mm, as shown in Figure 23. This compact microscanner presented Photonics 2021, 8, 6 a chip size of 1.5 × 1.3 mm . Each C-lens and a single-mode fiber are packaged insi 18 de a of 25 glass tube with a diameter of 1.4 mm. The full circumferential scans have been demon- strated with individual micromirrors scanning up to 45° at a voltage of less than 12 V. Photonics 2021, 8, x FOR PEER REVIEW 19 of 25 Figure 23. The schematic of MEMS OCT probe (Figures from [66]). Figure 23. The schematic of MEMS OCT probe (Figures from [66]). Figure 24a shows the MEMS mirrors with a 0.5 × 0.5 mm mirror plate. Four bimorph actuators are placed symmetrically at the four sides of a central mirror plate. Each bi- morph actuator consists of three pairs of double S-shaped bimorph (Al/SiO2) beams, sand- wiched with a heater layer made of a thin film of Ti/TiN. A mechanical scan angle of 13° is achieved, resulting in a ±26° optical scan angle or a 52° field of view (FOV). Configured with a swept-source OCT setup, this MEMS array-based circumferential scanning probe was applied to image a swine’s small intestine wrapped on a 20 mm-di- ameter glass tube, as shown in Figure 24b. The OCT imaging result shows that this new MEMS endoscopic OCT has promising applications in large tubular organs. An alternative solution is to move the objective lens directly by an actuated mi- crostage. L. Wu et al. [67] demonstrated a tunable microlens scanner, operating at a 880 nm wavelength. This microscanner uses a lateral-shift-free electrothermal bimorph actu- ator, carrying a 1 mm diameter glass rod lens moving at a resonance frequency of 79 Hz. Later, L. Liu et al. [68] developed another electrothermal MEMS microlens scanner, mov- ing a 2.4 mm plano-conv (ex a) ( glass lens with a maximum travel range of 4 b) 00 µm and a res- onance frequency of 24 Hz. Figure 24. An SEM image of the MEMS mirror in (a) and the imaging sample of a swine’s small Figure 24. An SEM image of the MEMS mirror in (a) and the imaging sample of a swine’s small intestine covering a glass tube in (b) (Figures from [66]). intestine covering a glass tube in (b) (Figures from [66]). A more compact microlens scanner was developed in 2020 by L. Zhou et al. [69] Configured with a swept-source OCT setup, this MEMS array-based circumferential where the actuation mechanism is based on a single serpentine inverted-series-connected scanning probe was applied to image a swine’s small intestine wrapped on a 20 mm- (ISC) electrothermal bimorph actuator carrying a microlens. The shape of the entire mi- diameter glass tube, as shown in Figure 24b. The OCT imaging result shows that this new crolens scanner is circular, with an outer diameter of 4.4 mm and a clear optical aperture MEMS endoscopic OCT has promising applications in large tubular organs. of 1.8 m An m alternative , as shown in F solution igure is 25to a. Th move e microst the objective age include lens s a rin directly g-shap by ed an fram actuated e and eimi- ght sets of bimorph actuators. It is loaded with a 2.4 mm plano-convex glass microlens on its crostage. L. Wu et al. [67] demonstrated a tunable microlens scanner, operating at a 880 nm wavelength. ring-shaped frame. The m This microscanner icrolens we uses aig lateral-shift-fr ht is about 8 ee mg. Figure electrothermal 25b shows the bimorph S-sh actuator aped , carrying form of an IS a 1 mm C elect diameter rotherm glass al act rod uatlens or where moving the ro at a und hin resonance ges ar frequency e used to connect all ISC of 79 Hz. Later, L. Liu et al. [68] developed another electrothermal MEMS microlens scanner, moving a structures, minimizing the residual stresses during motion. The resonant frequency of the 2.4 mm plano-convex glass lens with a maximum travel range of 400 m and a resonance MEMS microstage loaded with a lens reaches 140 Hz, which is acceptable for endomicro- frequency of 24 Hz. scopic imaging tasks. A more compact microlens scanner was developed in 2020 by L. Zhou et al. [69] where (a) (b) the actuation mechanism is based on a single serpentine inverted-series-connected (ISC) electrothermal bimorph actuator carrying a microlens. The shape of the entire microlens scanner is circular, with an outer diameter of 4.4 mm and a clear optical aperture of 1.8 mm, as shown in Figure 25a. The microstage includes a ring-shaped frame and eight sets of bimorph actuators. It is loaded with a 2.4 mm plano-convex glass microlens on its ring- shaped frame. The microlens weight is about 8 mg. Figure 25b shows the S-shaped form of an ISC electrothermal actuator where the round hinges are used to connect all ISC structures, minimizing the residual stresses during motion. The resonant frequency of the MEMS microstage loaded with a lens reaches 140 Hz, which is acceptable for endomicroscopic imaging tasks. Figure 25. SEMs of the microlens scanner: (a) top view of complete structure and (b) focus on one of eight inverted-series-connected (ISC) actuators where the silicon beam is suspended 257 µm above the outer silicon frame (Figures from [69]). 7. Discussion on the Microscanner Design and Conclusions The replacement of galvanometer mirrors for OCT beam scanning by 2D MEMS scan- ning mirrors converted the bulk microscope to a compact and light device. However, gal- vanometers demonstrate better performance because they operate in a closed loop, using position feedback to correct the drive waveform instead of an open loop for MEMS scanning *mirrors. When compared to MEMS mirrors, galvanometers are higher cost and relatively larger. This paper demonstrated that current MEMS scanning technologies have advantages and limitations compared to galvanometers. The use of MEMS scanning mirrors in OCT Photonics 2021, 8, x FOR PEER REVIEW 19 of 25 (a) (b) Figure 24. An SEM image of the MEMS mirror in (a) and the imaging sample of a swine’s small intestine covering a glass tube in (b) (Figures from [66]). A more compact microlens scanner was developed in 2020 by L. Zhou et al. [69] where the actuation mechanism is based on a single serpentine inverted-series-connected (ISC) electrothermal bimorph actuator carrying a microlens. The shape of the entire mi- crolens scanner is circular, with an outer diameter of 4.4 mm and a clear optical aperture of 1.8 mm, as shown in Figure 25a. The microstage includes a ring-shaped frame and eight sets of bimorph actuators. It is loaded with a 2.4 mm plano-convex glass microlens on its ring-shaped frame. The microlens weight is about 8 mg. Figure 25b shows the S-shaped form of an ISC electrothermal actuator where the round hinges are used to connect all ISC Photonics 2021, 8, 6 19 of 25 structures, minimizing the residual stresses during motion. The resonant frequency of the MEMS microstage loaded with a lens reaches 140 Hz, which is acceptable for endomicro- scopic imaging tasks. (a) (b) Figure 25. SEMs of the microlens scanner: (a) top view of complete structure and (b) focus on one Figure 25. SEMs of the microlens scanner: (a) top view of complete structure and (b) focus on one of of eight inverted-series-connected (ISC) actuators where the silicon beam is suspended 257 µm eight inverted-series-connected (ISC) actuators where the silicon beam is suspended 257 m above above the outer silicon frame (Figures from [69]). the outer silicon frame (Figures from [69]). 7. Discussion on the Microscanner Design and Conclusions 7. Discussion on the Microscanner Design and Conclusions The replacement of galvanometer mirrors for OCT beam scanning by 2D MEMS scan- The replacement of galvanometer mirrors for OCT beam scanning by 2D MEMS ning mirrors converted the bulk microscope to a compact and light device. However, gal- scanning mirrors converted the bulk microscope to a compact and light device. However, vanometers demonstrate better performance because they operate in a closed loop, using galvanometers demonstrate better performance because they operate in a closed loop, using position feed position back to c feedback orrec to t corr the dri ect the ve waveform drive waveform insteadinstead of an opof enan loo open p for M loop EM for S sc MEMS anning scanning *mirrors. When compared *mirrors. When compar to MEMS mir ed to MEMS rors, ga mirr lvanomete ors, galvanometers rs are higher cos are higher t and relativel cost andy relatively larger. larger. This This p paper aperdemonstrated demonstrated t th h at atcurr current ent MEMS MEMSscanning scanningtechnologies technologies h have ave advantages advantages and and limit limitations ations compared compared to to galv galvanometers. anometers. The use The use of of MEM MEMSSscanning scanning mir mirrr ors ors in OCT in OCT and other biomedical applications reduces the complexity of scan control and offers a lower cost scanner. To build minimally invasive endoscopic probes, scanning micromirrors are required to be compact (<5 mm). The size of the scanning mirror is a crucial parameter because the micromirror dimension should be larger than the laser spot size as well. High speed and large transverse scans can also be achieved, which enables real-time in vivo imaging and a large field of view, respectively. To design the ideal MEMS scanner for OCT, we need to consider the scan angle, the resonance frequency, dynamic, as well as static mirror flatness, and good resonant mode separation. The choice of these parameters influences the image quality, the desired resolution, and the presence of image distortions due to aberrations from the scanning mirror and objective lens and defines the limits for scan speed and total image size. As we demonstrated in Section 2, the specifications of a MEMS mirror determine the number of resolvable spots and the OCT B-scan rate. Increasing the size of the MEMS mirror or scan angle would increase the number of resolvable spots. The repetition frequency of a B-scan is limited by the resonance frequency of the MEMS mirror. Such resonance frequency is defined by the mirror inertia and is proportional to the inverse square of the mirror diameter. Generally, we would select a MEMS scanner that achieves the largest angle and the highest operating speed if large areas are to be imaged quickly. In clinical applications, the minimal line scan lengths must cover approximately a range from 1 mm to 2 mm. To maximize the imaging speed, the solution is to perform the scan at the resonance frequency of the MEMS mirror. However, the scan operation performed at the resonance frequency is a source of image distortions produced by parasitic vibrations of scanner axes. To overcome the parasitic vibrations, the B-scan frequency must be fixed below the resonance frequency of the MEMS mirror. The size of scanning micromirrors included in the OCT probe strongly influence the scanning performances and specifications of the OCT probe. Smaller diameter scanning mirrors facilitate both the integration of the probe within the standard endoscope and probe guidance in the internal organs. Increasing the mirror diameter reduces the resonance frequency, resulting in a slower scan repetition frequency, limiting the number of B-scans. Photonics 2021, 8, 6 20 of 25 In this case, if the suspension is stiffened to maintain a given speed, the torque available from the actuator must be increased. The thickness of the MEMS device is also crucial in the definition of micromechanical features of the MEMS device. Thus, the thicker electrodes of a comb-drive actuator improve the electrostatic force without affecting the deformation of the scanning micromirror. A thinner air gap of the comb-drive, obtained via surface micromachining, would decrease the high driving voltage (often ~100 V), making the device safer for patients. In microactuators where the moving part is a thin membrane, thinner membranes improve the motion range because of the increase of actuator deformation. In conclusion, the advantages of smaller MEMS scanning mirrors include the smaller mass, lower stiffness, and higher imaging speed. All of these parameters must be carefully considered when choosing the appropriate MEMS scanning mirror for a specific application of OCT imaging. Many parameters of the MEMS device and optical probe also must be selected during design, fabrication, or assembly. Others can be adjusted during the OCT experiment. Table 1 compares qualitatively the characteristics of MEMS scanning micromirrors for three types of actuation mechanisms, based on the literature review [70]. This study focuses on the highest performing MEMS scanning mirrors designed for miniaturized displays and optical imaging. The typical diameter of scanning micromirrors ranges from 0.5 mm to 3.5 mm. However, an average diameter of 1 mm is observed for all categories of MEMS scanning mirrors. Average values of micromirror mechanical specifications for all types of actuators are compared. Analyzing the range of motion for electrostatic microactuators, we can see that the average performance of the vertical comb actuators is 14 for an average piston motion of 94 m, which is better than that of linear comb-drives (8.5 for piston motion of 29 m). The average performance of electromagnetic actuators is 15 and 5 m for angular motion and piston motion, respectively. Finally, the electrothermal actuators present the best performances in both rotational motion and out-of-plane motion. Here, the average angular motion is about 27 and 280 m for average piston motion. An important characteristic to be analyzed is the resonance frequency for each category of microactuators. The resonance frequency of electrothermal and electromagnetic micromirrors is in the range from 100 Hz to 1000 Hz, while the resonance frequency of the comb-drives stays within the range of 150 Hz to 10,000 Hz. The required response time for dynamic systems is about 5 ms for scanning micromirrors implemented in OCT probes. Table 1 shows that each actuation principle has advantages in some aspects while having disadvantages in others. Electrostatic actuation has a fast response and the lowest power consumption, but it requires large driving voltage, which may not be safe for endoscopic applications. In addition, the electrostatic scanners have strong nonlinearities, limiting the MEMS displacement. Here, the electrostatic torque is a function of V , not linearly varying with the actuation voltage. This can result in a distortion of the scan pattern when driving with linearly ramped voltages. Several approaches of linearization have been proposed to eliminate the distortion, improving the linearity of scan patterns [71]. Finally, an additional advantage for electrostatic comb-drive actuators is the easy and standardized micromachining technology. The main disadvantage of electrostatic actuators is the pull-in voltage limiting the linear displacement and the relatively high driving voltage. Electromagnetic actuation offers a large scan angle, low driving voltage, and relatively more linear response than the competing actuation mechanisms. Despite the advantages of MEMS electromagnetic scanners, the achievable performance is limited by the large thermal dissipation inside a coil. In addition, magnets strong enough for high performance present significant volume and might require magnetic shielding. This leads to total package sizes that are larger than the competing MEMS actuators. Finally, they are complex and difficult to fabricate, particularly at small scales. Electrothermal actuators present large scan angles at low driving voltages. They offer the largest fill factor compared to other categories of MEMS mirrors. However, the thermal response is relatively slow, but sufficient to perform real-time imaging. Since electrothermal actuators have relatively simple geometries, they are easy to fabricate and Photonics 2021, 8, 6 21 of 25 can be made with a high fill factor. Electrothermal scanners seem to be excellent candidates to satisfy the requirements of OCT endoscopic imaging applications, however significant engineering effort is still needed to limit optical aberrations and to control scanner stability. Electrothermal scanners on silicon mirror plates that are suspended by a pair of torsion bars often present a temperature dependence in their oscillation behavior. This results in mechanical buckling of the beam. In this case, the initial position of mirror deflection can be difficult to control because the center of gravity of the oscillation system can be displaced from the initial position, changing the resonant mode and degrading the amplitude of oscillation when the scanner is driven at a constant frequency. In this case, the control of initial tilt angles can be difficult, causing optical alignment problems. The elastic constant of silicon can be a source of decreasing the temperature rise, generating a drift of resonance frequency—this is the case for all actuation mechanisms. For all three actuation mechanisms, the one considered to have the most appropriate architecture seems to be the resonant scanner with gimbaled orthogonal single-axis mirrors. Here, the design needs to strike a balance where the combination of scan angle, resonance frequency, and mirror size is enough for definition of the desired resolution, while keeping the mirror optically flat to avoid image distortions. One important challenge with gimbaled dual-axis scanners is the possibility of crosstalk between the two axes. Another challenge for MEMS scanners developed for endoscopic applications is compact package size, requiring a package of around 10 mm or less. For all three mechanisms of actuation, a scanning mirror fails to maintain its flatness when it is mechanically oscillated at resonance frequency. The mirror surface should be sufficiently flat if the mirror curvature is less than 1 m, so as to not distort the beam. Mirror deformation leads to unwanted expansion of the reflected beam spot, degrading the image quality by blurring near the left- and right–hand side edges of the projected image. To overcome this problem, Hsu et al. [72] proposed to modify the mirror plate by generating a backside island to improve the rigidity of the mirror. Thus, masses far from the rotation axis are removed, keeping the resonant frequency high, while removing those in the central part. The structure of a gimbaled 2D scanner can be subjected to external vibration as well as intrinsic oscillations of the orthogonal axis. In particular, a scanner spinning about its y-axis at resonance frequency can be forced to tilt in the orthogonal x-axis by the gimbal structure because Coriolis forces are generated [73]. This last effect results in a small coupling of both axes, blurring the image near the edges. The solution is to design the scanner mechanism to have no resonant coupling, making the resonance frequencies of the orthogonal axes distant. It is also critical that the microscanner exhibits excellent accuracy of scan repeatability. The angular accuracy must be better than 1% of the angle step size. One limitation of OCT probes using MEMS scanning mirrors arises because it is necessary to use a sinusoidal scan, as described in Section 2. The sinusoidal scan is less linear compared to linear scanning because the scan speed is not uniform at the center and at the edges of the B-scan because of non-uniformly spaced A-scans. The linearization of the scan requires driving the MEMS with higher harmonics of the scan frequency where parasitic vibrations of the resonance frequency appear since the MEMS scanning mirror does not have a closed loop feedback system. In this case, it is pertinent to implement a closed feedback loop, based on the integration of a piezoresistive strain gauge on the torsional beams of the scanning mirror to sense the beam strain and adjust the microscanner angle. As discussed here, challenges need to be overcome to enable the implementation of MEMS scanners into endoscopic OCT systems, requiring minimally invasive recording of cross-sectional images in vivo with high resolution and high speed. To date, the electrother- mal two-axis and resonant MEMS scanners seem to be the closest candidates to satisfy the requirements of endoscopic imaging applications under conditions to better control scanner stability. In conclusion, we hope to clearly demonstrate that MEMS scanner-based OCT probes offer several significant advantages and we expect them to have a bright future in mobile medical imaging devices. Photonics 2021, 8, 6 22 of 25 Table 1. Comparison of three main actuation mechanisms used for scanning in OCT probes. Mirror Size Angular Deflection Resonance Frequency Advantages Drawbacks (mm)/Data from [46] ( )/Data from [46] (Hz)/Data from [46] Pull-in effect Fast response High driving voltage Electrostatic all Low power consumption (50–100 V) Large scan angle Low force <10 Linear comb ~1 Ave scan: 8.5 250–5000 (ave 1800) Ave piston: 29 m >10 Vertical comb ~1 Ave scan: 14 150–10,000 (ave 5000) Ave piston: 94 m electrostatic attractive requires mechanical a pair of electrodes with Small temperature Silicon DRIE and Characteristics force between resonance to enhance scan air gap dependance metallization conductors angle High power consumption Maximum: 20 Larger driving External magnets Electromagnetic 0.8–3.5 Ave scan: 15 100–1000 Lower driving voltage increasing the size Ave piston: 5 m Large scan angle Electromagnetic interferences Lorentz force coil and permanent Multiple layers of metal and Small temperature Characteristics between current and size limited by magnet magnet insulator for coil dependance magnetic field ~25, maximum 64 Large scan angle High power Electrothermal 0.5–1 Ave scan: 27 117–1000 Low driving voltage consumption Ave piston: 270 m High fill-factor Slow response materials with different thermal expansion by multiple layers of metal and Requires relatively high Bimorph sensitive to Characteristics thermal expansion Joule effect insulator for heater power temperature change coefficients Funding: This work was supported by the collaborative project VIAMOS of the European Commis- sion (FP7, ICT program, grant no. 318542), the ANR Labex Action program (ANR-11-LABX-0001-01), and received a support from Collégium SMYLE. It was also supported by the French RENATECH network and its FEMTO-ST technological facility. Work is supported by the Foundation for Polish Science co-financed by the EU under the European Regional Fund (project no MAB/2019/12). Conflicts of Interest: The authors declare no conflict of interest. References 1. Petersen, K.E. Silicon as a mechanical materiel. IEEE 1982, 70, 420–457. [CrossRef] 2. Motamedi, M.E. MOEMS: Micro-Opto-Electro-Mechanical Systems; SPIE Press Bellingham: Washington, DC, USA, 2005. 3. Wu, M.C.; Solgaard, O.; Ford, J.E. Optical MEMS for Lightwave Communication. J. Lightwave Technol. 2006, 24, 4433–4454. [CrossRef] 4. Neumann, H.; Kiesslich, R.; Wallace, M.B.; Neurath, M.F. Confocal laser endomicroscopy: Technical advances and clinical applications. Gastroenterology 2010, 139, 388–392. [CrossRef] [PubMed] 5. Gmitro, A.F.; Aziz, D. Confocal microscopy through a fiber-optic imaging bundle. Opt. Lett. 1993, 18, 565–567. [CrossRef] [PubMed] 6. Bargiel, S.; Baranski, M.; Wiemer, M.; Frömel, J.; Wang, W.S.; Gorecki, C. Technological Platform for Vertical Multi-Wafer Integration of Microscanners and Micro-Optical Components. Micromachines 2019, 10, 185. [CrossRef] [PubMed] 7. Denk, W.; Strickler, J.H.; Webb, W.W. Two-photon laser scanning fluorescence microscopy. Science 1990, 248, 73–76. [CrossRef] 8. Koehler, M.J.; Vogel, T.; Elsner, P.; König, K.; Bückle, R.; Kaatz, M. In vivo measurement of the human epidermal thickness in different localizations by multiphoton laser tomography. Skin Res. Technol. 1010, 16, 259–264. 9. Huang, D.; Swanson, E.A.; Lin, C.P.; Schuman, J.S.; Stinson, W.G.; Chang, W.; Hee, M.R.; Flotte, T.; Gregory, K.; Puliafito, C.A. Optical coherence tomography. Science 1991, 254, 1178–1181. [CrossRef] 10. Lu, C.D.; Kraus, M.F.; Potsaid, B.; Liu, J.J.; Choi, W.; Jayaraman, V.; Cable, A.E.; Hornegger, J.; Duker, J.S.; Fujimoto, J.G. Handheld ultrahigh speed swept source optical coherence tomography instrument using a MEMS scanning mirror. Biomed. Opt. Express 2014, 5, 293–311. [CrossRef] 11. Cao, T.; Tey, H.L. High-definition optical coherence tomography—An aid to clinical practice and research in dermatology. J. Dtsch. Dermatol. Ges. 2015, 13, 886–890. [CrossRef] 12. Yin, C.; Glaser, A.K.; Leigh, S.Y.; Chen, Y.; Wei, L.; Pillai, P.C.S.; Rosenberg, M.C.; Abeytunge, S.; Peterson, G.; Glazowski, C.; et al. Miniature in vivo MEMS-based line-scanned dual-axis confocal microscope for point-of-care pathology. Biomed. Opt. Express 2016, 2, 251–263. [CrossRef] [PubMed] Photonics 2021, 8, 6 23 of 25 13. Patterson, P.R.; Hahb, D.; Fujino, M.; Piyawattanametha, W.; Wu, M.C. Scanning micromirrors: An overview. Proc. SPIE 2004, 5604, 195–207. 14. Puliafito, C.A.; Hee, M.R.; Lin, C.P.; Reichel, E.; Schuman, J.S.; Duker, J.S.; A Izatt, J.; A Swanson, E.; Fujimoto, J.G. Imaging of macular diseases with optical coherence tomography. Ophthalmology 1995, 102, 217–229. [CrossRef] 15. Schuman, J.S.; Hee, M.R.; Arya, A.V.; Pedut-Kloizman, T.; Puliafito, C.A.; Fujimoto, J.G.; Swanson, E.A. Optical coherence tomography: A new tool for glaucoma diagnosis. Curr. Opin. Ophthalmol. 1995, 6, 89–95. [CrossRef] [PubMed] 16. Hitzenberger, C.K.; Drexler, W.; Leitgeb, R.A.; Findl, O.; Ferche, A.F. Key Developments for Partial Coherence Biometry and Optical Coherence Tomography in the Human Eye Made in Vienna. Investig. Ophthalmol. Vis. Sci. 2016, 57, 460–474. [CrossRef] [PubMed] 17. Sivak, M.V.; Kobayashi, K.; Izatt, J.A.; Rollins, A.M.; Ung-Runyawee, R.; Chak, A.; Wong, R.C.; Isenberg, G.A.; Willis, J. High- resolution endoscopic imaging of the GI tract using optical coherence tomography. Gastrointest. Endosc. 2000, 51, 474–479. [CrossRef] 18. Welzel, J. Optical coherence tomography in dermatology: A review. Ski. Res. Technol. 2001, 7, 1–9. [CrossRef] 19. Pierce, M.C.; Sheridan, R.L.; Park, B.H.; Cense, B.; De Boer, J.F. Collagen denaturation can be quantified in burned human skin using polarization-sensitive optical coherence tomography. Burns 2004, 30, 511–517. [CrossRef] 20. Gora, M.J.; Suter, M.J.; Tearney, G.J.; Li, X. Endoscopic optical coherence tomography: Technologies and clinical applications. Biomed. Opt. Express 2017, 8, 2405–2444. [CrossRef] 21. Nine Point Medical. Available online: https://www.ninepointmedical.com/ (accessed on 30 December 2020). 22. Cardiovascular. Available online: https://www.cardiovascular.abbott/us/en/home.html (accessed on 30 December 2020). 23. Santec. Swept-Source OCT System. Available online: https://www.santec.com/en/products/oct/swept-source-oct/ (accessed on 30 December 2020). 24. Thor Labs. OCTH-1300-Handheld Scanner. Available online: https://www.thorlabs.com/thorproduct.cfm?partnumber=OCTH- 1300 (accessed on 30 December 2020). 25. MirrorcleTechnologies. Biomedical Imaging. Available online: https://www.mirrorcletech.com/wp/applications/#biomed (accessed on 30 December 2020). 26. Wang, D.; Liang, P.; Samuelson, S.; Jia, H.; Ma, J.; Xie, H. Correction of image distortions in endoscopic optical coherence tomography based on two-axis scanning MEMS mirrors. Biomed. Opt. Express 2013, 4, 2066–2077. [CrossRef] 27. Lin, L.Y.; Keeler, E.G. Progress of MEMS Scanning Micromirrors for Optical Bio-Imaging. Micromachines 2015, 6, 1675–1689. [CrossRef] 28. McCormick, D.; Jung, W.; Milanovic, V.; Chen, Z.; Tien, N. 3-D MEMS based real-time minimally invasive endoscopic optical coherence tomography. In Proceedings of the IEEE/LEOS International Conference on Optical MEMS and Their Applications Conference, Oulu, Finland, 25–26 August 2005; pp. 25–26. 29. Urey, H.; Wine, D.W.; Osborn, T.D. Optical performance requirements for MEMS-scanner based microdisplays. Proc. SPIE 2000, 4178, 176–185. 30. Hwang, K.; Seo, Y.-H.; Jeong, K.-H. Microscanners for optical endomicroscopic applications. Micro Nano Syst. Lett. 2017, 5. [CrossRef] 31. Urey, H.; Kan, C.; O Davis, W. Vibration mode frequency formulae for micromechanical scanners. J. Micromech. Microeng. 2005, 15, 1713–1721. [CrossRef] 32. Holmstrom, S.T.S.; Baran, U.; Urey, H. MEMS Laser Scanners: A Review. J. Microelectromech. Syst. 2014, 23, 259–275. [CrossRef] 33. Boppart, S.A.; Bouma, B.E.; Pitris, C.; Tearney, G.J.; Fujimoto, J.G.; Brezinski, M.E. Forward-imaging instruments *for optical coherence tomography. Opt. Lett. 1997, 22, 1618–1620. [CrossRef] 34. Sun, J.; Xie, H. MEMS-Based Endoscopic Optical Coherence Tomography. Int. J. Opt. 2011, 2011, 1–12. [CrossRef] 35. Zaeh, M.; Moesl, J.; Musiol, J.; Oefele, F. Material Processing with Remote Technology—Revolution or Evolution? Phys. Procedia 2010, 5, 19–33. [CrossRef] 36. Pelsue, K. Precision, Post-Objective, Two-Axis, Galvanometer Scanning. Proc. SPIE 1983, 390, 70–79. [CrossRef] 37. A Flusberg, B.; Cocker, E.D.; Piyawattanametha, W.; Jung, J.C.; Cheung, E.L.M.; Schnitzer, M.J. Fiber-optic fluorescence imaging. Nat. Methods 2005, 2, 941–950. [CrossRef] 38. Tanguy, Q. Design and Fabrication of a MEMS Scanner for OCT Imaging Endo-Microscopic Probe. Ph.D. Thesis, Université Bourgogne Franche-Comté, Besançon, France, February 2018. 39. Petersen, K.E. Silicon torsional scanning mirror. IBM J. Res. Dev. 1980, 24, 631–637. [CrossRef] 40. Tang, W.C.; Nguyen, T.C.H.; Juan, W.H.; Howe, R.T. Electrostatic-comb drive of lateral polysilicon resonator. Sens. Actuators A Phys. 1990, 21, 328–331. [CrossRef] 41. Selvakumar, A.; Najafi, K.; Juan, W.; Pang, S. Vertical comb array microactuators. In Proceedings of the IEEE Micro Electro Mechanical Systems, Amsterdam, The Netherlands, 29 January–2 February 1995; pp. 43–48. 42. Degani, O.; Socher, E.; Lipson, A.; Lejtner, T.; Setter, D.; Kaldor, S.; Nemirovsky, Y. Pull-in study of an electrostatic torsion microactuator. J. Microelectromech. Syst. 1998, 7, 373–379. [CrossRef] 43. Jung, W.; McCormick, D.T.; Zhang, J.; Wang, L.; Tien, N.C.; Chen, Z. Three-dimensional endoscopic optical coherence tomography by use of a two-axis microelectromechanical scanning mirror. Appl. Phys. Lett. 2006, 88, 163901. [CrossRef] Photonics 2021, 8, 6 24 of 25 44. Singh, J.; Teo, J.H.S.; Xu, Y.; Premachandran, C.S.; Chen, N.; Kotlanka, R.; Olivo, M.; Sheppard, C.J.R. A two axes scanning SOI MEMS micromirror for endoscopic bioimaging. J. Micromech. Microeng. 2007, 18, 025001. [CrossRef] 45. Kumar, K.; Condit, J.C.; McElroy, A.; Kemp, N.J.; Hoshino, K.; Milner, T.E.; Zhang, X. Fast 3D in vivo swept-source optical coherence tomography using a two-axis MEMS scanning micromirror. J. Opt. A Pure Appl. Opt. 2008, 10, 044013. [CrossRef] 46. Wang, D.; Fu, L.; Wang, X.; Gong, Z.; Samuelson, S.; Duan, C.; Jia, H.; Ma, J.S.; Xie, H. Endoscopic swept source optical coherence tomography based on a two-axis microelectromechanical system mirror. J. Biomed. Opt. 2013, 18, 086005. [CrossRef] 47. Aguirre, A.D.; Hertz, P.R.; Chen, Y.; Fujimoto, J.G.; Piyawattanametha, W.; Fan, L.; Wu, M.C. Two-axis MEMS scanning catheter for ultrahigh resolution three-dimensional and en face imaging. Opt. Express 2007, 15, 2445–2453. [CrossRef] 48. Gorecki, C.; Lullin, J.; Perrin, S.; Bargiel, S.; Albero, J.; Gaiffe, O.; Rutkowski, J.; Cote, J.M.; Krauter, J.; Osten, W.; et al. Micromachined phase-shifted array-type Mirau interferometer for swept-source OCT imaging: Design, microfabrication and experimental validation. Biomed. Opt. Express 2019, 10, 1111–1125. [CrossRef] 49. Lullin, J.; Bargiel, S.; Lemoal, P.; Perrin, S.; Albero, J.; Passilly, N.; Froehly, L.; Lardet-Vieudrin, F.; Gorecki, C. An electrostatic vertical microscanner for phase modulating array-type Mirau microinterferometry. J. Micromech. Microeng. 2015, 25, 115013. [CrossRef] 50. Miyajima, H. Development of a MEMS electromagnetic optical scanner for a commercial laser scanning microscope. J. Micro/ Nanolithograph. MEMS MOEMS 2004, 3, 348. [CrossRef] 51. Asada, N.; Takeuchi, M.; Vaganov, V.; Belov, N.; Hout, S.I.; Sluchak, I. Silicon micro-optical scanner. Sens. Actuators A Phys. 2000, 83, 284–290. [CrossRef] 52. Judy, J.W.; Muller, R.S. Magnetically actuated, addressable microstructures. J. Microelectromech. Syst. 1997, 6, 249–256. [CrossRef] 53. Miyajima, H.; Asaoka, N.; Arima, M.; Minamoto, Y.; Murakami, K.; Tokuda, K.; Matsumoto, K. A Durable, Shock Resistant Electromagnetic Optical Scanner with Polyimide-Based Hinges. J. Microelectromech. Syst. 2001, 10, 418–424. [CrossRef] 54. Kim, S.; Lee, C.; Kim, J.Y.; Lim, G.; Kim, J.; Kim, C. A 2-axis Polydimethylsiloxane (PDMS) based electromagnetic MEMS scanning mirror for optical coherence tomography. Proc. SPIE 2016, 9698, 969812. 55. Kim, K.H.; Park, B.H.; Maguluri, G.N.; Lee, T.W.; Rogomentich, F.J.; Bancu, M.G.; Bouma, B.E.; De Boer, J.F.; Bernstein, J.J. Two-axis magnetically-driven MEMS scanning catheter for endoscopic high-speed optical coherence tomography. Opt. Express 2007, 15, 18130–18140. [CrossRef] [PubMed] 56. Watanabe, Y.; Abe, Y.; Iwamatsu, S.; Kobayashi, S.; Takahashi, Y.; Sato, T. Electromagnetically Driven Two-Axis Optical Beam Steering MEMS Mirror and Its Dependence of Actuation on Magnetic Field. Electron. Commun. Jpn. 2011, 94, 24–31. [CrossRef] 57. Potekhina, A.; Wang, C. Review of Electrothermal Actuators and Applications. Actuators 2019, 8, 69. [CrossRef] 58. Tanguy, Q.A.; Bargiel, S.; Xie, H.; Passilly, N.; Barthès, M.; Gaiffe, O.; Rutkowski, J.; Lutz, P.; Gorecki, C. Design and Fabrication of a 2-Axis Electrothermal MEMS Micro-Scanner for Optical Coherence Tomography. Micromachines 2017, 8, 146. [CrossRef] 59. Sun, J.; Guo, S.; Wu, L.; Liu, L.; Choe, S.-W.; Sorg, B.S.; Xie, H. 3D In Vivo optical coherence tomography based on a low-voltage, large-scan-range 2D MEMS mirror. Opt. Express 2010, 18, 12065–12075. [CrossRef] 60. Wu, L.; Xie, H. A large vertical displacement electrothermal bimorph microactuator with very small lateral shift. Sens. Actuators A Phys. 2008, A 145–146, 371–379. [CrossRef] 61. Struk, P.; Bargiel, S.; Tanguy, Q.A.A.; Ramirez, F.E.G.; Passilly, N.; Lutz, P.; Gaiffe, O.; Xie, H.; Gorecki, C. Swept-source optical coherence tomography microsystem with an integrated Mirau interferometer and electrothermal micro-scanner. Opt. Lett. 2018, 43, 4847–4850. [CrossRef] [PubMed] 62. Struk, P.; Bargiel, S.; Froehly, L.; Baranski, M.; Passilly, N.; Albero, J.; Gorecki, C. Swept source optical coherence tomography endomicroscope based on vertically integrated Mirau micro interferometer: Concept and technology. IEEE Sens. J. 2015, 15, 7061–7070. [CrossRef] 63. Albero, J.; Perrin, S.; Bargiel, S.; Passilly, N.; Baranski, M.; Gauthier-Manuel, L.; Bernard, F.; Lullin, J.; Froehly, L.; Krauter, J.; et al. Dense arrays of millimeter-sized glass lenses fabricated at wafer-level. Opt. Express 2015, 23, 11702–11712. [CrossRef] [PubMed] 64. Tanguy, Q.A.A.; Gaiffe, O.; Passilly, N.; Cote, J.-M.; Cabodevila, G.; Bargiel, S.; Lutz, P.; Xie, H.; Gorecki, C. Real-time Lissajous imaging with a low-voltage 2-axis MEMS scanner based on electrothermal actuation. Opt. Express 2020, 28, 8512–8527. [CrossRef] [PubMed] 65. García-Ramírez, F.E.; Bargiel, S.; Gaiffe, O.; Tanguy, Q.A.; Struk, P.; Cote, J.M.; Passilly, N.; Lutz, P.; Gorecki, C.; Xie, H. Characterization of an integrated MOEMS scanning probe towards real-time Lissajous-based SS-OCT imaging for endoscopic applications. In Proceedings of the OSA Biophotonics Congress—Biomedical Optics, Fort Lauderdale, FL, USA, 20 April 2020. 66. Luo, S.; Wang, D.; Tang, J.; Zhou, L.; Duan, C.; Wang, N.; Liu, H.; Zhu, Y.; Li, G.; Zhao, H.; et al. Circumferential-scanning endoscopic optical coherence tomography probe based on a circular array of six 2-axis MEMS mirrors. Biomed. Opt. Express 2018, 9, 2104–2114. [CrossRef] 67. Wu, L.; Xie, H. A millimeter-tunable-range microlens for endoscopic biomedical imaging applications. IEEE J. Quantum Electron. 2010, 46, 1237–1244. [CrossRef] 68. Liu, L.; Wang, E.; Zhang, X.; Liang, W.; Li, X.; Xie, H. MEMS-based 3D confocal scanning microendoscope using MEMS scanners for both lateral and axial scan. Sens. Actuators A 2014, 215, 89–95. [CrossRef] 69. Zhou, L.; Yu, X.; Feng, P.X.-L.; Li, J.; Xie, H. A MEMS lens scanner based on serpentine electrothermal bimorph actuators for large axial tuning. Opt. Express 2020, 28, 23439. [CrossRef] Photonics 2021, 8, 6 25 of 25 70. Pengwang, E.; Rabenorosoa, M.; Rakotondrabe, N. Andreff, “Scanning Micromirror Platform Based on MEMS Technology for Medical Application”. Micromachines 2016, 7, 24. [CrossRef] 71. Tsai, J.C.; Lu, L.C.; Hsu, W.C.; Sun, C.W.; Wu, M.C. Linearization of a two-axis MEMS scanner driven by vertical comb-drive actuators. J. Micromech. Microeng. 2008, 18, 015015. [CrossRef] 72. Hsu, S.; Klose, T.; Drabe, C.; Schenk, H. Fabrication and characterization of a dynamically flat high resolution micro-scanner. J. Opt. A Pure Appl. Opt. 2008, 10, 044005. [CrossRef] 73. Shaeffer, D.K. MEMS inertial sensors: A tutorial overview. IEEE Commun. Mag. 2013, 51, 100–109. [CrossRef] http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Photonics Multidisciplinary Digital Publishing Institute

MEMS Scanning Mirrors for Optical Coherence Tomography

Photonics , Volume 8 (1) – Dec 30, 2020

Loading next page...
 
/lp/multidisciplinary-digital-publishing-institute/mems-scanning-mirrors-for-optical-coherence-tomography-NhjPk8iId0
Publisher
Multidisciplinary Digital Publishing Institute
Copyright
© 1996-2020 MDPI (Basel, Switzerland) unless otherwise stated Disclaimer The statements, opinions and data contained in the journals are solely those of the individual authors and contributors and not of the publisher and the editor(s). MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. Terms and Conditions Privacy Policy
ISSN
2304-6732
DOI
10.3390/photonics8010006
Publisher site
See Article on Publisher Site

Abstract

hv photonics Review 1 , 2 Christophe Gorecki * and Sylwester Bargiel Polish Academy of Sciences, Institute of Physical Chemistry, International Center for Translational Eye Research, Skierniewicka 10A, 01-230 Warsaw, Poland FEMTO-ST Institute (UMR CNRS 6714/UBFC), 15B Avenue des Montboucons, 25030 Besançon, France; sylwester.bargiel@femto-st.fr * Correspondence: cgorecki@ichf.edu.pl Abstract: This contribution presents an overview of advances in scanning micromirrors based on MEMS (Micro-electro-mechanical systems) technologies to achieve beam scanning for OCT (Optical Coherence Tomography). The use of MEMS scanners for miniaturized OCT probes requires appropriate optical architectures. Their design involves a suitable actuation mechanism and an adapted imaging scheme in terms of achievable scan range, scan speed, low power consumption, and acceptable size of the OCT probe. The electrostatic, electromagnetic, and electrothermal actuation techniques are discussed here as well as the requirements that drive the design and fabrication of functional OCT probes. Each actuation mechanism is illustrated by examples of miniature OCT probes demonstrating the effectiveness of in vivo bioimaging. Finally, the design issues are discussed to permit users to select an OCT scanner that is adapted to their specific imaging needs. Keywords: micro-opto-electro-mechanical system; MEMS scanner; optical coherence tomography 1. Introduction Micro-electro-mechanical systems (MEMS) technology enables the building of mi- crooptical scanners that are well suited for low cost manufacturability and scalability as the Citation: Gorecki, C.; Bargiel, S. MEMS processes emanate from the mature semiconductor microfabrication industry [1]. MEMS Scanning Mirrors for Optical For a long time, the potential of MEMS to steer or direct light has been well demonstrated Coherence Tomography. Photonics in the field of free-space optical systems [2]. In the 80s and early 90s, telecommunications 2021, 8, 6. https://doi.org/10.3390/ became the market driver for the optical applications of MEMS, pushing the develop- photonics8010006 ment of scanning micromirror systems for optical switches and network ports [3]. More recently, many types of MEMS scanning mirrors have been developed, covering a wide Received: 2 October 2020 range of applications from micrometer-scale array-type components to large scanners for Accepted: 24 December 2020 high-resolution imaging [4]. Thus, numerous optical imaging techniques such as confo- Published: 30 December 2020 cal microscopy [5,6], multiphoton microscopy [7,8], and Optical Coherence Tomography (OCT) [9–11] have become important diagnostic tools in biomedicine, particularly offering Publisher’s Note: MDPI stays neu- a platform for endoscopic imaging. These MEMS scanners successfully replaced the bulky tral with regard to jurisdictional clai- and high power consuming galvanometer scanners, providing compact, low cost, and low ms in published maps and institutio- power consumption solutions for high speed beam steering. Further, 2D MEMS mirrors nal affiliations. that scan in two axes are a pertinent alternative to the large galvano-scanners [12]. The MEMS scanner ’s performances are closely linked to the size of the selected actuator, carrying the micromirror and the force developed by this actuator. Figure 1 Copyright: © 2020 by the authors. Li- represents a summary of scanning micromirror applications, including the corresponding censee MDPI, Basel, Switzerland. actuation mechanisms and main microfabrication technologies [13]. At the scale level, This article is an open access article going from 1 mm to 1 cm, the MEMS technology combined with fiber optics enables distributed under the terms and con- miniature scanning components to be embedded inside the endoscopic imaging probes ditions of the Creative Commons At- operating at high speed and high resonance frequency. The MEMS scanners are relatively tribution (CC BY) license (https:// easily integrated and adapted for low cost fabrication and low power consumption. The creativecommons.org/licenses/by/ miniaturization performances and subsequent advances in standardized micromachining 4.0/). Photonics 2021, 8, 6. https://doi.org/10.3390/photonics8010006 https://www.mdpi.com/journal/photonics Photonics 2021, 8, x FOR PEER REVIEW 2 of 25 Photonics 2021, 8, 6 2 of 25 high speed and high resonance frequency. The MEMS scanners are relatively easily inte- grated and adapted for low cost fabrication and low power consumption. The miniaturi- zation performances and subsequent advances in standardized micromachining technol- ogies have also offered numerous low cost and disposable OCT probes for the medical technologies have also offered numerous low cost and disposable OCT probes for the industry. Originally adopted by the ophthalmic community [14–16], OCT has been used medical industry. Originally adopted by the ophthalmic community [14–16], OCT has been to image internal organs, such as the gastrointestinal tract [17], and in the diagnosis of used to image internal organs, such as the gastrointestinal tract [17], and in the diagnosis skin pathologies [18,19]. This strong interest for clinical applications pushed several com- of skin pathologies [18,19]. This strong interest for clinical applications pushed several panies to develop endoscopic OCT systems [20]. Examples of commercial products are companies to develop endoscopic OCT systems [20]. Examples of commercial products are the clinical endoscope and catheter-based systems from the NvisionVLE Imaging System the clinical endoscope and catheter-based systems from the NvisionVLE Imaging System (South Jordan, Utah, USA) [21], the intravascular OCT imaging systems from OPTIS™ (St. (South Jordan, Utah, USA) [21], the intravascular OCT imaging systems from OPTIS™ Jude Medical Inc., St. Paul, MN, USA) [22], Santec’s (Komaki, Japan) swept-source OCT (St. Jude Medical Inc., St. Paul, MN, USA) [22], Santec’s (Komaki, Japan) swept-source systems [23], Thorlabs (Newton, NJ, USA) OCT scanners [24], as well as Mirrorcle (Rich- OCT systems [23], Thorlabs (Newton, NJ, USA) OCT scanners [24], as well as Mirrorcle (Richmond, mond, CA, CA, USA) USA) microscanners [25]. microscanners [25]. Figure 1. Applications, actuation mechanisms, and fabrication technologies for scanning micromirrors. Figure 1. Applications, actuation mechanisms, and fabrication technologies for scanning micromirrors. In this paper, we will demonstrate that for OCT imaging applications, the performance In this paper, we will demonstrate that for OCT imaging applications, the perfor- of the MEMS scanner is often limited by optics and intrinsic characteristics of actuation mance of the MEMS scanner is often limited by optics and intrinsic characteristics of ac- mechanisms. Here, optics require a small focused spot and dynamic focusing, imposing tuation mechanisms. Here, optics require a small focused spot and dynamic focusing, im- severe restrictions on scanning lens performances, while the actuation needs a high scan- posing severe restrictions on scanning lens performances, while the actuation needs a high ning speed, a low power consumption, a precise control of motion linearity, and reduced scanning speed, a low power consumption, a precise control of motion linearity, and re- cross-axis coupling, which may distort the scanning patterns [26,27]. The group of OCT duced cross-axis coupling, which may distort the scanning patterns [26,27]. The group of probes to be discussed in this paper do maintain such opto-mechanical performances, OCT probes to be discussed in this paper do maintain such opto-mechanical perfor- using different actuation mechanisms. Our wish is to demonstrate that the breakthrough mances, using different actuation mechanisms. Our wish is to demonstrate that the break- of compactness is obtained when MEMS dual-axis beam-steering micromirrors [28] are through of compactness is obtained when MEMS dual-axis beam-steering micromirrors used to achieve scanning 3D OCT probes. In the case of endoscopic applications, they are [28] are used to achieve scanning 3D OCT probes. In the case of endoscopic applications, small enough to be included into a standard endoscope channel, with an inner diameter they are small enough to be included into a standard endoscope channel, with an inner of 2.8 mm. Further, 2D scanning motion can derive from electrostatic, electromagnetic, diameter of 2.8 mm. Further, 2D scanning motion can derive from electrostatic, electro- electrothermal, or piezoelectric actuation, providing the scanning mirrors for light beam magnetic, electrothermal, or piezoelectric actuation, providing the scanning mirrors for steering, operating at high speed and fully controlled by non-resonant or resonant regimes. light beam steering, operating at high speed and fully controlled by non-resonant or res- However, we intentionally excluded from the present study piezoelectric actuation and we onant regimes. However, we intentionally excluded from the present study piezoelectric consider only the three other actuation mechanisms for MEMS scanning mirrors that are actuation and we consider only the three other actuation mechanisms for MEMS scanning the most widely used in OCT applications. mirrors that are the most widely used in OCT applications. 2. Requirements for MEMS Microscanners From the user point of view, performances of scanning micromirrors are defined by the maximum scan angle, the number of resolvable spots which represents the scan resolution, the resonance frequency, as well as the surface quality vs. the smoothness and flatness of micromirrors. Photonics 2021, 8, x FOR PEER REVIEW 3 of 25 2. Requirements for MEMS Microscanners From the user point of view, performances of scanning micromirrors are defined by the maximum scan angle, the number of resolvable spots which represents the scan reso- lution, the resonance frequency, as well as the surface quality vs. the smoothness and flat- ness of micromirrors. The number of resolvable spots N of a scanning mirror is defined as a function of optical scan angle and beam divergence , as shown in Figure 2: = = , (1) where D is the mirror diameter, a represents the aperture shape factor (a = 1 in the case of Photonics 2021, 8, 6 3 of 25 a square aperture and 1.22 for a circular aperture) and is the illumination wavelength. As the number of resolvable spots is proportional to the product of mirror diameter by the scan angle, the resonant frequency defining the upper limit of the scanner’s re- The number of resolvable spots N of a scanning mirror is defined as a function of sponse depends on the diameter (or inertia) and mechanical rigidity of the mirror suspen- optical scan angle q and beam divergence dq, as shown in Figure 2: opt sion. In the case of a circular mirror with a diameter of 1 mm, illuminated by a He-Ne laser and scanning in the mode of VGA (640 × 480 pixels), θopt = 28.3°, which is two times bigger q D opt than the mechanical deflection. In thi N = s case, the sc = q anning m , irror needs to deliver a full (1) opt dq al mechanical angle θmech = 14.15° or larger. wherTo m e D isathe ke tmirr he com or diameter parison o , a f MEM represents S scan the ners op apertur erat e shape ing at di factor ffere (a nt = freq 1 inuenc the case ies aof lso, a square aperture and 1.22 for a circular aperture) and l is the illumination wavelength. we use the D = 1,22 product, which is the key performance index for scanners. Figure 2. Total optical scan range of a scanning micromirror vs. the beam divergence. Figure 2. Total optical scan range of a scanning micromirror vs. the beam divergence. As the number of resolvable spots is proportional to the product of mirror diameter by Imaging applications for 2D scanning require a dual axis system with fast scanning the scan angle, the resonant frequency defining the upper limit of the scanner ’s response capability. Two categories of more popular MEMS torsional scanners are the uniaxial scanner depends and on t the he b diameter iaxial gim (or bal-m inertia) ount and ed scann mechanical er [29]. Fi rigidity gure 3 of sho the wmirr s boor th t suspension. hese scanning In the case of a circular mirror with a diameter of 1 mm, illuminated by a He-Ne laser and architectures with associated scanning trajectories. Here, two full raster cycles are repre- scanning in the mode of VGA (640  480 pixels), q = 28.3 , which is two times bigger sented while the scanning spots are schematized in g opt reen. The uniaxial scanner has a sin- than the mechanical deflection. In this case, the scanning mirror needs to deliver a full gle axis of rotation and to obtain a 2D scanning, a pair of uniaxial scanners is used (one mechanical angle q = 14.15 or larger. horizontal, one verti mech cal). Here, the rotation of the first scanner causes the optical beam to To make the comparison of MEMS scanners operating at different frequencies also, walk across the second scanner. In this case, the writing of data is obtained only during we use the q D = 1.22Nl product, which is the key performance index for scanners. opt the forward sweep of the horizontal scanner. The biaxial scanner includes two perpendic- Imaging applications for 2D scanning require a dual axis system with fast scanning ular axes of rotation. The oscillation of the inner frame creates a horizontal scan line, while capability. Two categories of more popular MEMS torsional scanners are the uniaxial the outer frame creates the vertical scan line. This is writing a new line of data in both scan scanner and the biaxial gimbal-mounted scanner [29]. Figure 3 shows both these scanning directions. Thus, the scanner writes two lines during one scan cycle. architectures with associated scanning trajectories. Here, two full raster cycles are rep- resented while the scanning spots are schematized in green. The uniaxial scanner has a single axis of rotation and to obtain a 2D scanning, a pair of uniaxial scanners is used (one horizontal, one vertical). Here, the rotation of the first scanner causes the optical beam to walk across the second scanner. In this case, the writing of data is obtained only during the forward sweep of the horizontal scanner. The biaxial scanner includes two perpendicular axes of rotation. The oscillation of the inner frame creates a horizontal scan line, while the outer frame creates the vertical scan line. This is writing a new line of data in both scan directions. Thus, the scanner writes two lines during one scan cycle. Photonics Photonics 2021 2021 , 8,, x FO 8, 6 R PEER REVIEW 4 of 4 of 25 25 Figure 3. Schema of uniaxial scanner in (a) vs. biaxial scanner in (b) with associated scanning Figure 3. Schema of uniaxial scanner in (a) vs. biaxial scanner in (b) with associated scanning tra- trajectories. jectories. The performance of the MEMS optical scanner is often limited by the optical architec- The performance of the MEMS optical scanner is often limited by the optical archi- ture, which requires small, focused spots and dynamic focusing. In addition, the intrinsic tecture, which requires small, focused spots and dynamic focusing. In addition, the intrin- micromechanical characteristics of the actuation mechanism of the MEMS scanners are sic micromechanical characteristics of the actuation mechanism of the MEMS scanners are crucial and dependent upon the requirements of high scanning speed, low power con- crucial and dependent upon the requirements of high scanning speed, low power con- sumption, as well as a precise control of motion linearity. Here, one of the limiting features sumption, as well as a precise control of motion linearity. Here, one of the limiting features in MEMS scanners is their mode of driving, which can be resonant or quasi-static. A reso- in MEMS scanners is their mode of driving, which can be resonant or quasi-static. A res- nant drive mode with a fine control of the beam’s location is more difficult to implement, onant drive mode with a fine control of the beam’s location is more difficult to implement, requiring one to use the close-loop control of beam steering. A quasi-static mode, on the requiring one to use the close-loop control of beam steering. A quasi-static mode, on the other hand, provides easy programmable control of the beam, permitting operation with other hand, provides easy programmable control of the beam, permitting operation with open-loop control of beam steering. However, certain categories of MEMS scanners are not open-loop control of beam steering. However, certain categories of MEMS scanners are able to offer the significant aperture size and scan angle specifications in the quasi-static not able to offer the significant aperture size and scan angle specifications in the quasi- mode. The quasi-statically tilted MEMS scanner only requires open-loop control for beam static mode. The quasi-statically tilted MEMS scanner only requires open-loop control for steering purposes. The main challenges of a bidirectional scan are the precise control of beam steering purposes. The main challenges of a bidirectional scan are the precise control phase between both the scan lines and a high-quality control of motion, avoiding the effects of phase between both the scan lines and a high-quality control of motion, avoiding the of mechanical coupling between the scanner axes. effects of mechanical coupling between the scanner axes. The choice of appropriate scanning techniques is strongly dependent on the selection The choice of appropriate scanning techniques is strongly dependent on the selection of optimal scanning frequency. The most common scanning techniques are raster scanning of optimal scanning frequency. The most common scanning techniques are raster scan- and Lissajous scanning, shown in Figure 4 [30]. In raster scanning, a low frequency, linear ning and Lissajous scanning, shown in Figure 4 [30]. In raster scanning, a low frequency, vertical scan (usually in quasi-static mode) is paired with an orthogonal high frequency, linear vertical scan (usually in quasi-static mode) is paired with an orthogonal high fre- resonant horizontal scan. For the raster scanner, the vertical scan is often assumed to be quency, resonant horizontal scan. For the raster scanner, the vertical scan is often assumed 60 Hz (video frame rate). The raster scanning can be obtained by bidirectional scanning to be 60 Hz (video frame rate). The raster scanning can be obtained by bidirectional scan- waveforms with N = f /f , where N is a positive integer. A raster scanner involves a x y ning waveforms with N = fx/fy, where N is a positive integer. A raster scanner involves a triangular trajectory in the x-axis while shifting the sample position in steps or continuously triangular trajectory in the x-axis while shifting the sample position in steps or continu- in the y-axis. Here, a laser beam is starting at the top left of Figure 4a and goes from left- ously in the y-axis. Here, a laser beam is starting at the top left of Figure 4a and goes from to-right at a scanning frequency of f to the bottom-right corner (1), then rapidly moves left-to-right at a scanning frequency of fx to the bottom-right corner (1), then rapidly moves back to the left and scans the next line (2), then off once again to go back up to the top (3). back to the left and scans the next line (2), then off once again to go back up to the top (3). During the period of one scan cycle, the vertical position increases steadily downward During the period of one scan cycle, the vertical position increases steadily downward at at a frequency of f , which is the slow scanning frequency. The scan resolution and a frequency of fy, which is the slow scanning frequency. The scan resolution and the frame the frame rate are, respectively, defined by the lateral and vertical scanning frequencies. rate are, respectively, defined by the lateral and vertical scanning frequencies. Raster scan- Raster scanning results in a rectangular scanning area. This simple technique of scan ning results in a rectangular scanning area. This simple technique of scan presents several presents several technical limitations for MEMS micromirrors such as the higher driving technical limitations for MEMS micromirrors such as the higher driving voltages due to voltages due to the use of non-resonant motion and lower mechanical stability for the the use of non-resonant motion and lower mechanical stability for the slow axis. Figure slow axis. Figure 4b shows an alternative to raster scanning which is biresonant Lissajous 4b shows an alternative to raster scanning which is biresonant Lissajous scanning. This is scanning. This is achieved by driving the x and y axes with purely sinusoidal signals of Photonics 2021, 8, x FOR PEER REVIEW 5 of 25 Photonics Photonics 2021 2021 , 8 , , x FO 8, 6 R PEER REVIEW 5 of 5 of 25 25 achieved by driving the x and y axes with purely sinusoidal signals of different frequen- cies fx and fy that are x(t) = Ax cos(2πfxt) and y(t) = Ay cos(2πfyt). The corresponding scan- achieved by driving the x and y axes with purely sinusoidal signals of different frequen- different frequencies f and f that are x(t) = A cos(2f t) and y(t) = A cos(2f t). The x y x x y y ning waveforms have a frequency relation of n = fx/fy, where n is a rational number. Lissa- cies fx and fy that are x(t) = Ax cos(2πfxt) and y(t) = Ay cos(2πfyt). The corresponding scan- corresponding scanning waveforms have a frequency relation of n = f /f , where n is a x y jous scanning provides a rectangular scan area too. The scanning trajectory and the frame ning waveforms have a frequency relation of n = fx/fy, where n is a rational number. Lissa- rational number. Lissajous scanning provides a rectangular scan area too. The scanning rate are more complex than in raster scanning trajectories and MEMS mirrors are operat- jous scanning provides a rectangular scan area too. The scanning trajectory and the frame trajectory and the frame rate are more complex than in raster scanning trajectories and ing at high resonant frequencies for both the axes, offering better mechanical stability. rate are more complex than in raster scanning trajectories and MEMS mirrors are operat- MEMS mirrors are operating at high resonant frequencies for both the axes, offering better ing at high resonant frequencies for both the axes, offering better mechanical stability. mechanical stability. Figure 4. Principle of main scanning techniques: (a) raster scanning and (b ) Lissajous scanning. Figure 4. Principle of main scanning techniques: (a) raster scanning and (b) Lissajous scanning. Figure 4. Principle of main scanning techniques: (a) raster scanning and (b) Lissajous scanning. Figure 5a shows a schematic of a rectangular torsional scanning mirror with rectan- Figure 5a shows a schematic of a rectangular torsional scanning mirror with rectan- gular flexure beams [31]. Applied to produce the desired torsional mode, this torsional Figure 5a shows a schematic of a rectangular torsional scanning mirror with rectan- gular flexure beams [31]. Applied to produce the desired torsional mode, this torsional scanning mirror presents three parasitic oscillation modes that we refer to as vertical (or gular flexure beams [31]. Applied to produce the desired torsional mode, this torsional scanning mirror presents three parasitic oscillation modes that we refer to as vertical (or piston), horizontal, and rocking modes. In the ideal design, the mode frequencies should scanning mirror presents three parasitic oscillation modes that we refer to as vertical (or piston), horizontal, and rocking modes. In the ideal design, the mode frequencies should be well separated from the torsional mode frequency and its harmonics to minimize the piston), horizontal, and rocking modes. In the ideal design, the mode frequencies should be well separated from the torsional mode frequency and its harmonics to minimize the power dissipation. Figure 5b shows that the horizontal and vertical modes do not change be well separated from the torsional mode frequency and its harmonics to minimize the power dissipation. Figure 5b shows that the horizontal and vertical modes do not change the scanned beam direction. Rocking mode deflects the incoming beam perpendicular to power dissipation. Figure 5b shows that the horizontal and vertical modes do not change the scanned beam direction. Rocking mode deflects the incoming beam perpendicular to the intended scan axis, creating an undesired off-axis motion. The biaxial scanners should the scanned beam direction. Rocking mode deflects the incoming beam perpendicular to the intended scan axis, creating an undesired off-axis motion. The biaxial scanners should be designed in such a way that cross coupling effects between the torsional and rocking the intended scan axis, creating an undesired off-axis motion. The biaxial scanners should be designed in such a way that cross coupling effects between the torsional and rocking modes of inner and outer scan frames are minimized. be designed in such a way that cross coupling effects between the torsional and rocking modes of inner and outer scan frames are minimized. modes of inner and outer scan frames are minimized. Figure 5. Torsional resonant scan mirror in (a) with its f our fundamental vibration modes in (b). Figure 5. Torsional resonant scan mirror in (a) with its four fundamental vibration modes in (b). Figure 5. Torsional resonant scan mirror in (a) with its four fundamental vibration modes in (b). Actuated MEMS mirrors are subjected to both static and dynamic deformations. Static deformation is induced by intrinsic material stress or thermal stress of mirror material, Photonics 2021, 8, x FOR PEER REVIEW 6 of 25 Photonics 2021, 8, 6 6 of 25 Actuated MEMS mirrors are subjected to both static and dynamic deformations. Static deformation is induced by intrinsic material stress or thermal stress of mirror mate- rial, while dynamic deformation is produced by forces due to mirror oscillation with a while dynamic deformation is produced by forces due to mirror oscillation with a frequency frequency f. The maximum mirror deviation from linearity according to Brosens’s formula f. The maximum mirror deviation from linearity according to Brosens’s formula [32] is: [32] is: 2 5 f D q mech d µ = = , , (2 (2) ) max E t where ρ is the material density, E the modulus of elasticity, and tm represents the mirror where  is the material density, E the modulus of elasticity, and t represents the mirror thickness. thickness. For a fixed D product, an increase of D leads to larger mirror deformation, lower For a fixed q D product, an increase of D leads to larger mirror deformation, lower opt maximum frequency, and increasing footprint. To keep the spot diffraction limited, the maximum frequency, and increasing footprint. To keep the spot diffraction limited, the maximum mirror deformation should not exceed λ/10. Here, the D -dependency deter- maximum mirror deformation should not exceed /10. Here, the D -dependency deter- mines the upper limit of mirror size, while large mirrors must have a large thickness tm. It mines the upper limit of mirror size, while large mirrors must have a large thickness t . It can be roughly estimated that the optimum mirror surface will be around 1 mm . can be roughly estimated that the optimum mirror surface will be around 1 mm . 3. Requirements for OCT Probes 3. Requirements for OCT Probes A typical OCT setup includes a Michelson interferometer and a low-coherence light A typical OCT setup includes a Michelson interferometer and a low-coherence light source. Interference signal, carrying the information about the measured biological object, source. Interference signal, carrying the information about the measured biological ob- is detected and demodulated to produce a map of the light backscattered from the micro- ject, is detected and demodulated to produce a map of the light backscattered from the structure inside the measured tissue. Image reconstruction is obtained by repeated axial microstructure inside the measured tissue. Image reconstruction is obtained by repeated measurements at different transverse positions as the optical beam is scanned by a MEMS axial measurements at different transverse positions as the optical beam is scanned by a scanning mirror. Figure 6 shows the two classes of miniature OCT probes, categorized on MEMS scanning mirror. Figure 6 shows the two classes of miniature OCT probes, catego- the basis of their scan modes: the side-imaging and the forward-imaging [33–35]. Side- rized on the basis of their scan modes: the side-imaging and the forward-imaging [33–35]. imaging probes, schematized in Figure 6a, are the most widely used because they tend to Side-imaging probes, schematized in Figure 6a, are the most widely used because they have a much simpler actuation mechanism than forward-imaging probes and the actuator tend to have a much simpler actuation mechanism than forward-imaging probes and the tends to be far away from the probe output. This category of OCT probes is very flexible actuator tends to be far away from the probe output. This category of OCT probes is very and has a small size that is well appropriate for the miniaturization. Here, a mirror or flexible and has a small size that is well appropriate for the miniaturization. Here, a mirror prism is connected, for example, to a rotation assembly that shifts the emitted light from or prism is connected, for example, to a rotation assembly that shifts the emitted light from the optical the opticalfiber out fiber out of a of a window on window on tthe he side side of the probe. The of the probe. The side-im side-imaging aging OCT probe OCT probe only provi only provides des si side de im imaging aging aro around und t the he pr probe, which obe, which limits limits its its c clinical linical applications, applications, ma making k- ing the timage-based he image-based surgical surgi needle cal need guidance le guidanc difficult. e diffic The ult.forwar The forw d-imaging ard-imagin probe, g probe, o of which f which its size its tends size tends to be in a range of n to be in a range of needle sizes, eedle is sizes, mor is e more suitable sufor itabsur le fo gical r sur guidance gical guidance inside insi thede t body he bo . However dy. How , forwar ever, forw d-imaging ard-ima pr ging probe obes are generally s are gener mor ally e more comple complex in design. x in desi They gn. require the actuator near the probe tips and are quite difficult to miniaturize. Here, a mirror They require the actuator near the probe tips and are quite difficult to miniaturize. Here, or prism assembly shifts the emitted light from the optical fiber out of a window on the a mirror or prism assembly shifts the emitted light from the optical fiber out of a window front side of the probe, as shown in Figure 6b. on the front side of the probe, as shown in Figure 6b. Figure 6. Two categories of miniature OCT probes: (a) side imaging probe; and (b) forward imaging probe. Figure 6. Two categories of miniature OCT probes: (a) side imaging probe; and (b) forward imaging probe. In addition to the categorization of Figure 6, the scanning arrangement exhibits two In addition to the categorization of Figure 6, the scanning arrangement exhibits two distinct arrangements depending on the position of the MEMS scanner and the objective distinct arrangements depending on the position of the MEMS scanner and the objective lens focusing the beam on the sample to be measured: pre-objective scanning and post- lens focusing the beam on the sample to be measured: pre-objective scanning and post- objective scanning [36,37]. In the pre-objective scanner, a MEMS mirror is placed prior to an objective scanning [36,37]. In the pre-objective scanner, a MEMS mirror is placed prior to objective lens, allowing a long working lens-sample distance. However, this configuration can cause significant off-axis aberration for a deflected scanning beam. The aberration can be minimized by using a specialized f-Theta lens, focusing the laser beam on a single Photonics 2021, 8, x FOR PEER REVIEW 7 of 25 Photonics 2021, 8, 6 7 of 25 an objective lens, allowing a long working lens-sample distance. However, this configu- ration can cause significant off-axis aberration for a deflected scanning beam. The aberra- tion can be minimized by using a specialized f-Theta lens, focusing the laser beam on a single plane over the entire scan field while also ensuring that the beam remains perpen- plane over the entire scan field while also ensuring that the beam remains perpendicular dicular to the plane over the entire scan. In this category of probes, the miniaturization to the plane over the entire scan. In this category of probes, the miniaturization becomes becomes difficult. In the post-objective scanner, a MEMS scanner is located after an objec- difficult. In the post-objective scanner, a MEMS scanner is located after an objective lens. tive lens. This configuration can produce a small off-axis aberration as well as high image This configuration can produce a small off-axis aberration as well as high image resolution. resolution. However, the working distance becomes short and thus, the selection of an However, the working distance becomes short and thus, the selection of an objective with objective with a high numerical aperture objective is necessary. As the deflection by a a high numerical aperture objective is necessary. As the deflection by a mirror results in a mirror results in a curved focal plane, it is possible to compensate this curvature by vary- curved focal plane, it is possible to compensate this curvature by varying the focal length ing the focal length of the lens by moving the lens along the beam path. of the lens by moving the lens along the beam path. Finally, the last categorization of OCT probes refers to the location of the scanning Finally, the last categorization of OCT probes refers to the location of the scanning mechanism. The scanning mechanisms may be considered as either proximal or distal to mechanism. The scanning mechanisms may be considered as either proximal or distal to the light source [20]. Proximal scanners are placed in the illumination pathway upstream the light source [20]. Proximal scanners are placed in the illumination pathway upstream of the fiber and are used with a fiber bundle. This configuration offers the benefit of sep- of the fiber and are used with a fiber bundle. This configuration offers the benefit of sepa- arating bulky scanners from a miniaturized imaging head and typically includes cascaded rating bulky scanners from a miniaturized imaging head and typically includes cascaded galvanometer-mounted scanning mirrors, scanning the beam across the proximal end of galvanometer-mounted scanning mirrors, scanning the beam across the proximal end of a fiber bundle. Distal scanners are placed on the fiber side distal to the light source and a fiber bundle. Distal scanners are placed on the fiber side distal to the light source and usu usually ally scan scan illumin illumination ation from a sin from a single gle fifiber ber over over th the e spec specimen. imen. Her Here, the e, the 2D 2D scanning scanning can can be per be performed formed by a by a MEMS MEMS mirror w mirror which hich pivo pivots in ts in two tw angular o anguldir ar direct ections ion ors or by maintaining by main- the resonant vibrating of the fiber extremity via an attached actuator or cantilever. taining the resonant vibrating of the fiber extremity via an attached actuator or cantilever. After comparing the different configurations of OCT probes, Figure 7 summarizes After comparing the different configurations of OCT probes, Figure 7 summarizes the char the characteristics acteristics mamade de for each for each criterion of criterion the of the propr be classific obe classification ation proposed proposed earlier earlier . . Here, the colorful path corresponds to the optimal design of an OCT probe applied for Here, the colorful path corresponds to the optimal design of an OCT probe applied for gastrointestinal tract evaluation [38]. gastrointestinal tract evaluation [38]. Figure 7. Order of criteria of classifications of the endoscopic probe configurations. Figure 7. Order of criteria of classifications of the endoscopic probe configurations. 4. Electrostatic MEMS Scanning Mirrors for OCT 4. Electrostatic MEMS Scanning Mirrors for OCT 4.1. Principles of Electrostatic Microactuators 4.1. Principles of Electrostatic Microactuators The principle of electrostatic actuation is based on Coulomb’s law, using the attraction of two oppositely charged plates. Electrostatic actuation is currently the predominant The principle of electrostatic actuation is based on Coulomb’s law, using the attrac- method used for MEMS scanners because the capacitive actuators draw very little current, tion of two oppositely charged plates. Electrostatic actuation is currently the predominant therefore requiring low operating power despite the need for relatively high applied method used for MEMS scanners because the capacitive actuators draw very little current, voltages. An electrostatic torsional micromirror is in rotation when a driving voltage is therefore requiring low operating power despite the need for relatively high applied volt- applied between the fixed and movable electrodes. The mirror rotates an angle  about the ages. An electrostatic torsional micromirror is in rotation when a driving voltage is ap- torsion axis until the restoring and electrostatic torques are equal. The torque is given by: plied between the fixed and movable electrodes. The mirror rotates an angle θ about the torsion axis until the restoring and electrostatic torques are equal. The torque is given by: V ¶C T () = (3) 2 ¶ (3) θ= 2 ∂θ T () = k, (4) θ= kθ, (4) where V is the driving voltage, C the capacitance of the actuator, and k is the spring constant. Photonics 2021, 8, x FOR PEER REVIEW 8 of 25 Photonics 2021, 8, 6 8 of 25 where V is the driving voltage, C the capacitance of the actuator, and k is the spring con- stant. For a simple parallel plate actuator, the capacitance is given by: For a simple parallel plate actuator, the capacitance is given by: = , # A (5) C = , (5) where is the permittivity of free-space, A is the surface of electrode, and g is the gap where # is the permittivity of free-space, A is the surface of electrode, and g is the gap between the electrodes. between the electrodes. Figure 8 shows the schematic of two main types of electrostatic actuators: the parallel Figure 8 shows the schematic of two main types of electrostatic actuators: the parallel plate actuator [39] and the comb-drive actuator [40,41]. plate actuator [39] and the comb-drive actuator [40,41]. Figure 8. Two main architectures of electrostatic actuation: (a) parallel-plate actuator and (b) Figure 8. Two main architectures of electrostatic actuation: (a) parallel-plate actuator and (b) comb- comb-drive actuator. drive actuator. With a parallel plate gap closing actuator, the zone of the electrode overlap is mainly With a parallel plate gap closing actuator, the zone of the electrode overlap is mainly the area of the fixed electrode. Thus, the gap is a function of the rotation angle. Here, there the area of the fixed electrode. Thus, the gap is a function of the rotation angle. Here, there is a tradeoff as the initial gap distance needs to be large enough to generate the scan angle, is a tradeoff as the initial gap distance needs to be large enough to generate the scan angle, but small enough for a reasonable driving voltage. The linear scan range is limited by but small enough for a reasonable driving voltage. The linear scan range is limited by the the pull-in effect to around 40% of the maximal mechanical scan angle [42]. Electrostatic pull-in effect to around 40% of the maximal mechanical scan angle [42]. Electrostatic actu- actuators are relatively easy to fabricate by micromachining technologies. Parallel-plate ators are relatively easy to fabricate by micromachining technologies. Parallel-plate actu- actuators employ surface micromachining, often based on polysilicon with sacrificial oxide, ators employ surface micromachining, often based on polysilicon with sacrificial oxide, on electroplated metal with sacrificial organic layer or sputtered metal with a sacrificial on electroplated metal with sacrificial organic layer or sputtered metal with a sacrificial organic layer. Comb-drive actuators are typically fabricated on Silicon-on-Insulator (SOI) organic layer. Comb-drive actuators are typically fabricated on Silicon-on-Insulator (SOI) substrates, ensuring a relatively simple fabrication process and easy thickness control of substrates, ensuring a relatively simple fabrication process and easy thickness control of micromechanical structures. micromechanical structures. 4.2. Examples of Electrostatic OCT Probes 4.2. Examples of Electrostatic OCT Probes A series of endoscopic OCT probes based on electrostatic actuation have been pro- A series of endoscopic OCT probes based on electrostatic actuation have been pro- posed, using 2D MEMS scanners that scan in two axes [43–46] and employing a 2D gimbal- posed, using 2D MEMS scanners that scan in two axes [43–46] and employing a 2D gim- less vertical comb-drive structure. An interesting example of an OCT endoscopic MEMS bal-less vertical comb-drive structure. An interesting example of an OCT endoscopic scanner for high resolution OCT with angled vertical comb-drive actuators has been pro- MEMS scanner for high resolution OCT with angled vertical comb-drive actuators has posed by Aguirre et al. [47] at MIT. Figure 9a shows the SEM photography of this MEMS been proposed by Aguirre et al. [47] at MIT. Figure 9a shows the SEM photography of this scanner. The microscanner uses a torsional beam and includes a gimbal-mounting MEMS MEMS scanner. The microscanner uses a torsional beam and includes a gimbal-mounting mirror to scan a dual axis, combining the scan’s x and y axes with a single pivot point. The MEMS mirror to scan a dual axi  s, combining the scan’s x and y axes with a single pivot actuated mirror provides 6 angular scanning at over 100 V of driving voltage. Here, a point. The actuated mirror provides ±6° angular scanning at over 100 V of driving voltage. silicon micromirror is suspended inside a gimbal frame by a pair of polysilicon torsion springs. The scanning mirror has a circular aperture with a diameter of 1 mm within the footprint size of 3  3 mm . Photonics 2021, 8, x FOR PEER REVIEW 9 of 25 Here, a silicon micromirror is suspended inside a gimbal frame by a pair of polysilicon Photonics 2021, 8, 6 9 of 25 torsion springs. The scanning mirror has a circular aperture with a diameter of 1 mm within the footprint size of 3 × 3 mm . Figure 9. SEM photograph of 2D MEMS scanner in (a) and the schema of MEMS catheter packaging in (b) (Figures Figure 9. SEM photograph of 2D MEMS scanner in (a) and the schema of MEMS catheter packag- from [47]). ing in (b) (Figures from [47]). Figure 9b shows the schematic of OCT catheter packaging. The 2D MEMS scanner Figure 9b shows the schematic of OCT catheter packaging. The 2D MEMS scanner is is inclined at 45 and directs the beam in a side scanning configuration, orthogonally inclined at 45° and directs the beam in a side scanning configuration, orthogonally to the to the endoscope axis. The post-objective scanning eliminates off-axis optical aberration endoscope axis. The post-objective scanning eliminates off-axis optical aberration encoun- encountered with pre-objective scanning schemas. The endoscope head is 5 mm in diameter tered with pre-objective scanning schemas. The endoscope head is 5 mm in diameter and and 2.5 cm long. The optics include a graded index fiber collimator followed by an AR- 2.5 cm long. The optics include a graded index fiber collimator followed by an AR-coated coated achromatic focusing lens and which produces a beam spot diameter of 12 m. achromatic focusing lens and which produces a beam spot diameter of 12 µm. Figure 10a Figure 10a represents the resonance characteristics of the MEMS scanner. The mirror represents the resonance characteristics of the MEMS scanner. The mirror resonance is 463 resonance is 463 Hz and the gimbal axis resonance is around 140 Hz. Resonant operation of Hz and the gimbal axis resonance is around 140 Hz. Resonant operation of the mirror the mirror offers high speed raster scanning for en-face microscopy. The MEMS OCT probes offers high speed raster scanning for en-face microscopy. The MEMS OCT probes were were demonstrated in 3D high resolution OCT imaging. The OCT catheter was combined demonstrated in 3D high resolution OCT imaging. The OCT catheter was combined with with an OCT device employing a commercial femtosecond Nd:glass laser, centered on the an OCT device employing a commercial femtosecond Nd:glass laser, centered on the wavelengths of 1.06 m with a bandwidth of more than 200 nm. The light source was wavelengths of 1.06 µm with a bandwidth of more than 200 nm. The light source was coupled with a fiber-optic interferometer. The sample was measured with ~2000 axial scans coupled with a fiber-optic interferometer. The sample was measured with ~2000 axial per second. The interference signal was first amplified, filtered, and then demodulated by scans per second. The interference signal was first amplified, filtered, and then demodu- the detection block including a 12-bit, acquisition card, and a 5 MHz A/D converter and lated by the detection block including a 12-bit, acquisition card, and a 5 MHz A/D con- then processed by PC computer. The obtained axial resolution of images was <4 m in verter and then processed by PC computer. The obtained axial resolution of images was tissue, while transverse resolution was equivalent to the focusing spot of 12 m. Imaging <4 µm in tissue, while transverse resolution was equivalent to the focusing spot of 12 µm. Photonics 2021, 8, x FOR PEER REVIEW was performed at a rate of 4 frames/s over a 3D field of view of 1.8  1  1.3 mm 10 of 25 with Imaging was performed at a rate of 4 frames/s over a 3D field of view of 1.8 × 1 × 1.3 mm 500  500  1000 pixels. Figure 10b illustrates an example of a 3D image that represents a with 500 × 500 × 1000 pixels. Figure 10b illustrates an example of a 3D image that repre- volume data set from the hamster cheek pouch acquired in vitro. sents a volume data set from the hamster cheek pouch acquired in vitro. According to the schema of catheter packaging in Figure 9b, the OCT probe is limited in size by the footprint of the MEMS mirror—this is the main limit in the miniaturization of such OCT devices. High driving voltage is also an issue for in vivo endoscopy. The acquisition of data is made in open-loop mode, which introduces a possible lack of precise control of the MEMS mirror. This may degrade the scanning trajectory stability and data reproducibility. Addressing this concern will require one to produce more sophisticated MEMS technology, including the possibility of close-loop control detection of mirror mo- tion. An alternative based on the use of in-situ capacitive detection will be described in the next paragraph. Figure 10. FigureReso 10. Resonance nance characte characteristics ristics of the ofM the EMS MEMS scanner ( scanner a); and 3D OCT (a); and 3D OCT image acquired in image acquired vivo in vivo set from the set from hamster cheek po the hamster cheek uch ( pouch b) (Figures from [47]) (b) (Figures from [47 . ]). The interest to miniaturize OCT probes is not only limited to the endoscopic config- urations. Existing bulk microscopes or fiber optics devices for early diagnosis of cancer are expensive and are only affordable at the hospital; thus, they are not sufficiently used by physicians or cancer specialists as an early diagnosis tool. Significant reduction of sys- tem cost and size can be achieved by use of opto-mechanical components, fabricated by micromachining. In 2016, an OCT microsystem including an active 4 × 4 array of spectrally tuned Mirau interferometers, including an electrostatic vertical comb-drive actuator car- rying the array of reference mirrors, was proposed for dermatology applications [48]. The architecture of an active Mirau interferometer is shown schematically in Figure 11a. To perform OCT measurements, the device is incorporated within an experimental setup in- cluding a swept-source laser (center wavelength: 850 nm, swept range: 50 nm) and a high speed smart camera. A Mirau interferometer includes a series of vertically stacked components: a doublet of microlens matrices, a vertical comb-drive actuator, a spacer, and a planar beam splitter plate. The diameter of an individual microlens is 1.9 mm and the array pitch is 2 mm, while the equivalent focal length of a microlens doublet is 7.44 mm and the numerical aperture (NA) is 0.1. The assembly of two glass microlens arrays is made by anodic bond- ing. The axial resolution of OCT imaging is 6 µm, while the transverse resolution is limited to 6 µm. The depth of penetration is 0.6 mm. The key element of a Mirau interferometer is the vertical microscanner W2. The microscanner is designed for generating a vertical displacement of a large platform with a 4 × 4 array of reference micromirrors of the Mirau interferometer, as shown in Figure 11b. (a) (b) Figure 11. Cross-sectional view of the multichannel ”active” Mirau microinterferometer (a) with a focus on a 3D view of an electrostatic vertical microscanner with a 4 × 4 array of suspended reference micromirrors (b). Photonics 2021, 8, x FOR PEER REVIEW 10 of 25 Figure 10. Resonance characteristics of the MEMS scanner (a); and 3D OCT image acquired in vivo Photonics 2021, 8, 6 10 of 25 set from the hamster cheek pouch (b) (Figures from [47]). The interest to miniaturize OCT probes is not only limited to the endoscopic config- According to the schema of catheter packaging in Figure 9b, the OCT probe is limited urations. Existing bulk microscopes or fiber optics devices for early diagnosis of cancer in size by the footprint of the MEMS mirror—this is the main limit in the miniaturization are expensive and are only affordable at the hospital; thus, they are not sufficiently used of such OCT devices. High driving voltage is also an issue for in vivo endoscopy. The by physicians or cancer specialists as an early diagnosis tool. Significant reduction of sys- acquisition of data is made in open-loop mode, which introduces a possible lack of precise tem cost and size can be achieved by use of opto-mechanical components, fabricated by control of the MEMS mirror. This may degrade the scanning trajectory stability and data micromachining. In 2016, an OCT microsystem including an active 4 × 4 array of spectrally reproducibility. Addressing this concern will require one to produce more sophisticated tuned Mirau interferometers, including an electrostatic vertical comb-drive actuator car- MEMS technology, including the possibility of close-loop control detection of mirror motion. rying the array of reference mirrors, was proposed for dermatology applications [48]. The An alternative based on the use of in-situ capacitive detection will be described in the architecture of an active Mirau interferometer is shown schematically in Figure 11a. To next paragraph. perform OCT measurements, the device is incorporated within an experimental setup in- The interest to miniaturize OCT probes is not only limited to the endoscopic config- cluding a swept-source laser (center wavelength: 850 nm, swept range: 50 nm) and a high urations. Existing bulk microscopes or fiber optics devices for early diagnosis of cancer speed smart camera. are expensive and are only affordable at the hospital; thus, they are not sufficiently used A Mirau interferometer includes a series of vertically stacked components: a doublet by physicians or cancer specialists as an early diagnosis tool. Significant reduction of of microlens matrices, a vertical comb-drive actuator, a spacer, and a planar beam splitter system cost and size can be achieved by use of opto-mechanical components, fabricated plate. The diameter of an individual microlens is 1.9 mm and the array pitch is 2 mm, by micromachining. In 2016, an OCT microsystem including an active 4  4 array of spec- while the equivalent focal length of a microlens doublet is 7.44 mm and the numerical trally tuned Mirau interferometers, including an electrostatic vertical comb-drive actuator aperture (NA) is 0.1. The assembly of two glass microlens arrays is made by anodic bond- carrying the array of reference mirrors, was proposed for dermatology applications [48]. ing. The axial resolution of OCT imaging is 6 µm, while the transverse resolution is limited The architecture of an active Mirau interferometer is shown schematically in Figure 11a. to 6 µm. The depth of penetration is 0.6 mm. The key element of a Mirau interferometer To perform OCT measurements, the device is incorporated within an experimental setup is the vertical microscanner W2. The microscanner is designed for generating a vertical including a swept-source laser (center wavelength: 850 nm, swept range: 50 nm) and a displacement of a large platform with a 4 × 4 array of reference micromirrors of the Mirau high speed smart camera. interferometer, as shown in Figure 11b. (a) (b) Figure 11. Cross-sectional view of the multichannel ”active” Mirau microinterferometer (a) with a focus on a 3D view of Figure 11. Cross-sectional view of the multichannel ”active” Mirau microinterferometer (a) with a focus on a 3D view of an an electrostatic vertical microscanner with a 4 × 4 array of suspended reference micromirrors (b). electrostatic vertical microscanner with a 4  4 array of suspended reference micromirrors (b). A Mirau interferometer includes a series of vertically stacked components: a doublet of microlens matrices, a vertical comb-drive actuator, a spacer, and a planar beam splitter plate. The diameter of an individual microlens is 1.9 mm and the array pitch is 2 mm, while the equivalent focal length of a microlens doublet is 7.44 mm and the numerical aperture (NA) is 0.1. The assembly of two glass microlens arrays is made by anodic bonding. The axial resolution of OCT imaging is 6 m, while the transverse resolution is limited to 6 m. The depth of penetration is 0.6 mm. The key element of a Mirau interferometer is the vertical microscanner W2. The microscanner is designed for generating a vertical displacement of a large platform with a 4  4 array of reference micromirrors of the Mirau interferometer, as shown in Figure 11b. The vertical motion of the whole 4  4 array of reference micromirrors at the resonance frequency can be controlled precisely by an in-situ differential position sensor measuring the variation of capacitance due to the comb-drive displacement [49]. The vertical actuation of reference mirrors leads to a phase-shifted imaging that enables rapid measurement of the amplitudes and phases of interference signal and improves the signal to noise ratio and sensitivity. The size of an individual micromirror, suspended by a system of spider legs, is 400  400 m . The array of micromirrors is vertically aligned with the lenses, 2 2 forming an 8  8 mm structure. The resulting imager covers the same area of 8  8 mm of the sample, reconstructing the topography in a continuous way by stitching together Photonics 2021, 8, x FOR PEER REVIEW 11 of 25 The vertical motion of the whole 4 × 4 array of reference micromirrors at the reso- nance frequency can be controlled precisely by an in-situ differential position sensor measuring the variation of capacitance due to the comb-drive displacement [49]. The ver- tical actuation of reference mirrors leads to a phase-shifted imaging that enables rapid measurement of the amplitudes and phases of interference signal and improves the signal Photonics 2021, 8, 6 to noise ratio and sensitivity. The size of an individual micromirror, suspended by a s 11 y ofs- 25 tem of spider legs, is 400 × 400 µm . The array of micromirrors is vertically aligned with the lenses, forming an 8 × 8 mm structure. The resulting imager covers the same area of 8 × 8 mm of the sample, reconstructing the topography in a continuous way by stitching 4  4 single-channel interferograms via a system of actuators shifting mechanically the together 4 × 4 single-channel interferograms via a system of actuators shifting mechani- entire Mirau-array over the overlapping region. The 3 mm thick spacer W3 in silicon cally the entire Mirau-array over the overlapping region. The 3 mm thick spacer W3 in adjusts the position of lens focus from the planar beam splitter plate. The beam-splitter silicon adjusts the position of lens focus from the planar beam splitter plate. The beam- W4 has a transmission-reflection ratio of 70/30. Figure 12a shows the assembled Mirau splitter W4 has a transmission-reflection ratio of 70/30. Figure 12a shows the assembled interferometer mounted on the PCB (printed circuit board). The footprint of this chip is Mirau interferometer mounted on the PCB (printed circuit board). The footprint of this chip 15  15 mm , whereas the overall thickness is about 5 mm. is 15 × 15 mm , whereas the overall thickness is about 5 mm. (a) (b) Figure 12. PCB-mounted chip of a Mirau interferometer in (a); and 3D swept-source OCT image of Figure 12. PCB-mounted chip of a Mirau interferometer in (a); and 3D swept-source OCT image of onion slices in (b). onion slices in (b). The original A-scan includes several parasitic terms such as the autocorrelation terms The original A-scan includes several parasitic terms such as the autocorrelation terms due to the beam splitter, reference mirrors images, mirror replica images, and DC noise due to the beam splitter, reference mirrors images, mirror replica images, and DC noise term, making the interpretation of true OCT signal difficult. Help to the use of four-frame term, making the interpretation of true OCT signal difficult. Help to the use of four-frame phase shift algorithm all these signals are removed, improving both the signal-to-noise phase shift algorithm all these signals are removed, improving both the signal-to-noise ratio and the measurements range. Figure 12b shows a volumetric 300 × 300 × 600 µm ratio and the measurements range. Figure 12b shows a volumetric 300  300  600 m OCT image o OCT image f of an on an onion ion slic slice, e, where the where the microsco microscopic pic structure is visible. The structure is visible. The sensitivity of sensitivity of this im this image age is in the ran is in the range ge oof f 80 dB 80 dB. . 5. Electromagnetic MEMS Scanning Mirrors for OCT 5. Electromagnetic MEMS Scanning Mirrors for OCT 5.1. Principles of Electromagnetic Microactuators 5.1. Principles of Electromagnetic Microactuators Electromagnetic MEMS actuators are driven by Lorentz force [50]. In this case, a Electromagnetic MEMS actuators are driven by Lorentz force [50]. In this case, a cur- current-carrying conductor is placed in a static magnetic field. This field produced around rent-carrying conductor is placed in a static magnetic field. This field produced around the conductor interacts with the static field to produce a force. In an electromagnetic the conductor interacts with the static field to produce a force. In an electromagnetic ac- actuator, including the flexure beams and a moving mirror plate, the module of Lorentz tuator, including the flexure beams and a moving mirror plate, the module of Lorentz force is expressed as: force is expressed as: F = BIL sin q, (6) = sin , (6) where B is magnetic flux density of the magnetic field, I is the current flowing through the beam, L is the beam length, and q is the angle between the current and the magnetic field. where B is magnetic flux density of the magnetic field, I is the current flowing through the beam, Ther L is te he beam are numer lengt ous h, variations and is the on an the gle between architectur th ee cur of the rent electr and the m omagnetic agnetic field. actuators: permanent magnets interacting with an external field, permanent magnets interacting with There are numerous variations on the architecture of the electromagnetic actuators: current-carrying coils, and current carrying conductors interacting with an external field. permanent magnets interacting with an external field, permanent magnets interacting A common advantage is the relatively high generated force. The main drawbacks are with current-carrying coils, and current carrying conductors interacting with an external the high-power dissipation as well as a complex fabrication, including severe materials field. A common advantage is the relatively high generated force. The main drawbacks challenges, and difficulty to miniaturize the micromirrors because of the use of external bulk magnets. Figure 13 shows the schematic design of a 2D electromagnetic scanner where the Lorentz force interaction is generated between the micro-coil integrated on the scanner gimbal and the permanent magnets located outside of the scanner. To decrease the size of electromagnetic actuators, a combined electrostatic/electromag- netic 2D scanner has been developed by Microvision for retinal scanning displays [29]. Such a mixed configuration employed electromagnetic actuation to move the outer frame, which provides the slow vertical axis, while electrostatic actuation was used for the inner mirror axis, which provides the fast horizontal axis. Photonics 2021, 8, x FOR PEER REVIEW 12 of 25 are the high-power dissipation as well as a complex fabrication, including severe materi- als challenges, and difficulty to miniaturize the micromirrors because of the use of exter- nal bulk magnets. Figure 13 shows the schematic design of a 2D electromagnetic scanner Photonics 2021, 8, 6 12 of 25 where the Lorentz force interaction is generated between the micro-coil integrated on the scanner gimbal and the permanent magnets located outside of the scanner. Figure 13. Figure 13. 2D e 2D electr lectromagnetic omagnetic actuator actuator generating the generating the Lor Loren entztfor z force ce between betweethe n the micro-co micro-coil integrated il inte- grated on the scanner gimbal and the two permanent magnets. on the scanner gimbal and the two permanent magnets. 5.2. Examples of Electromagnetic OCT Probes To decrease the size of electromagnetic actuators, a combined electrostatic/electro- magnetic 2D scanner has been developed by Microvision for retinal scanning displays Magnetically actuated scanning mirrors made by micromachining were demonstrated [29]. Such a mixed configuration employed electromagnetic actuation to move the outer by Judy et al. [51]. The Olympus Company developed one of the first 1D electromagnetic frame MEMS , which scanners provides t for confocal he slow vert microscopy ical axis [52], , whil while e el an ectrosta early 2D tic electr actuaomagnetic tion was used MEMS for tscanner he inner m was irrpr or ax oposed is, whic byh p Asada rovides et al. the f [53 a]. st hor This izon cat tegory al axis.of MEMS scanning mirrors replaced the electrostatic actuators in situations where it was necessary to increase the 5.2. scanning Examples range of Electroma or lower gn the etic OCT driving Prob voltage. es Serious efforts were performed to make the commercialization of MEMS electromagnetic scanner OCT probes easier for clinical Magnetically actuated scanning mirrors made by micromachining were demon- applications. To overcome the fabrication complexity of earlier electromagnetic actuators, strated by Judy et al. [51]. The Olympus Company developed one of the first 1D electro- for example, a flexible 2-axis polydimethylsiloxane (PDMS)-based electromagnetic MEMS magnetic MEMS scanners for confocal microscopy [52], while an early 2D electromagnetic scanning mirror was developed [54]. The size of such a MEMS scanner remained relatively MEMS scanner was proposed by Asada et al. [53]. This category of MEMS scanning mir- big (15  15  15 mm ). rors replaced the electrostatic actuators in situations where it was necessary to increase An interesting electromagnetic scanning actuator for OCT imaging was demonstrated the scanning range or lower the driving voltage. Serious efforts were performed to make by Kim et al. [55]. Figure 14 shows the photographs of a two-axis gimbaled mirror with the commercialization of MEMS electromagnetic scanner OCT probes easier for clinical folded flexure hinges. The rotation of the mirror is possible in two axes along the flexures: applications. To overcome the fabrication complexity of earlier electromagnetic actuators, one inner axis and one orthogonally placed outer axis. Resonant frequencies for the inner for example, a flexible 2-axis polydimethylsiloxane (PDMS)-based electromagnetic MEMS and outer axes were 450 Hz and 350 Hz, respectively. To generate the magnetic actuation, a scanning mirror was developed [54]. The size of such a MEMS scanner remained relatively permanent magnet glued to the backside of the mirror plate and a pair of coils is placed Photonics 2021, 8, x FOR PEER REVIEW 13 of 25 big (15 × 15 × 15 mm ). inside the probe body for each scan direction. The mirror plate is 0.6  0.8 mm , with a An interesting electromagnetic scanning actuator for OCT imaging was demon- device footprint of 2.4  2.9 mm . strated by Kim et al. [55]. Figure 14 shows the photographs of a two-axis gimbaled mirror with folded flexure hinges. The rotation of the mirror is possible in two axes along the flexures: one inner axis and one orthogonally placed outer axis. Resonant frequencies for the inner and outer axes were 450 Hz and 350 Hz, respectively. To generate the magnetic actuation, a permanent magnet glued to the backside of the mirror plate and a pair of coils is placed inside the probe body for each scan direction. The mirror plate is 0.6 × 0.8 mm , with a device footprint of 2.4 × 2.9 mm . Figure 14. Photograph of electromagnetic micro-scanner in (a) and a SEM image of folded flexure Figure 14. Photograph of electromagnetic micro-scanner in (a) and a SEM image of folded flexure hinges in (b) (Figures from [55]). hinges in (b) (Figures from [55]). Figure 15a shows a schematic of the assembled catheter packaging. The light source Figure 15a shows a schematic of the assembled catheter packaging. The light source is pigtailed by a single mode optical fiber, delivering via a GRIN lens a focused beam is pigtailed by a single mode optical fiber, delivering via a GRIN lens a focused beam redirected by the MEMS mirror on the sample, which is then scanned. Scattered light from the sample returns through the same optical path and is collected by the pigtailed GRIN lens. A glass window with an AR (anti reflective) coating protects the MEMS scanner and eliminates the back reflections. The catheter package has a 2.8 mm diameter and a 12 mm length. The MEMS scanner is fabricated from a SOI wafer with a 50 µm thick device layer on a 350 µm thick handle layer, including a 1 µm thick oxide box layer. Magnet layers are composed of small NdFeB magnets, each one measuring 0.6 × 0.8 × 0.18 mm . Figure 15b represents optical angles of the MEMS scanner in both inner and outer axes as a function of the driving voltage. An optical scan angle of about ±30° was obtained with ±1.2 V and ±4 V driving voltages for the inner and outer axis, corresponding to 50 mA and 100 mA current, respectively. The light refracted by the protection window produces slight nonlinearity in the de- flection angle due to thickness variations of the window. Spurious vibrations are observed for large scan angles at the mirror resonant frequency. To avoid these effects, the working scan angle was reduced to ±20° optical angle for the inner axis and less than ±30° optical angle for the outer axis. Optical resolution was estimated to be 5 µm. In vivo 3D endo- scopic imaging of tissues was made by combining the two-axis scanning catheters and the multifunctional SD-OCT system. 3D images of a fingertip were acquired at 18.5 frames/s, with the scan performed with voltages of ±2.8 V and ±0.8 V applied on the inner and outer axis, covering 1.5 × 1 mm lateral scan range and consuming 150 mW of power. Figure 15. Schematic of the assembled catheter packaging in (a) and optical angles of the MEMS scanner for both axes vs. the driving voltage in (b) (Figures from [55]). Photonics 2021, 8, x FOR PEER REVIEW 13 of 25 Figure 14. Photograph of electromagnetic micro-scanner in (a) and a SEM image of folded flexure hinges in (b) (Figures from [55]). Figure 15a shows a schematic of the assembled catheter packaging. The light source is pigtailed by a single mode optical fiber, delivering via a GRIN lens a focused beam redirected by the MEMS mirror on the sample, which is then scanned. Scattered light from the sample returns through the same optical path and is collected by the pigtailed GRIN lens. A glass window with an AR (anti reflective) coating protects the MEMS scanner and eliminates the back reflections. The catheter package has a 2.8 mm diameter and a 12 mm length. The MEMS scanner is fabricated from a SOI wafer with a 50 µm thick device layer on a 350 µm thick handle layer, including a 1 µm thick oxide box layer. Magnet layers are composed of small NdFeB magnets, each one measuring 0.6 × 0.8 × 0.18 mm . Figure 15b Photonics 2021, 8, 6 13 of 25 represents optical angles of the MEMS scanner in both inner and outer axes as a function of the driving voltage. An optical scan angle of about ±30° was obtained with ±1.2 V and ±4 V driving voltages for the inner and outer axis, corresponding to 50 mA and 100 mA redirected by the MEMS mirror on the sample, which is then scanned. Scattered light from current, respectively. the sample returns through the same optical path and is collected by the pigtailed GRIN The light refracted by the protection window produces slight nonlinearity in the de- lens. A glass window with an AR (anti reflective) coating protects the MEMS scanner and flection angle due to thickness variations of the window. Spurious vibrations are observed eliminates the back reflections. The catheter package has a 2.8 mm diameter and a 12 mm for large scan angles at the mirror resonant frequency. To avoid these effects, the working length. The MEMS scanner is fabricated from a SOI wafer with a 50 m thick device layer scan angle was reduced to ±20° optical angle for the inner axis and less than ±30° optical on a 350 m thick handle layer, including a 1 m thick oxide box layer. Magnet layers are angle for the outer axis. Optical resolution was estimated to be 5 µm. In vivo 3D endo- composed of small NdFeB magnets, each one measuring 0.6  0.8  0.18 mm . Figure 15b scopic imaging of tissues was made by combining the two-axis scanning catheters and the represents optical angles of the MEMS scanner in both inner and outer axes as a function multifunctional SD-OCT system. 3D images of a fingertip were acquired at 18.5 frames/s, of the driving voltage. An optical scan angle of about 30 was obtained with 1.2 V and with the scan performed with voltages of ±2.8 V and ±0.8 V applied on the inner and outer 4 V driving voltages for the inner and outer axis, corresponding to 50 mA and 100 mA axis, covering 1.5 × 1 mm lateral scan range and consuming 150 mW of power. current, respectively. Figure 15. Schematic of the assembled catheter packaging in (a) and optical angles of the MEMS scanner for both axes vs. Figure 15. Schematic of the assembled catheter packaging in (a) and optical angles of the MEMS scanner for both axes vs. the driving voltage in (b) (Figures from [55]). the driving voltage in (b) (Figures from [55]). The light refracted by the protection window produces slight nonlinearity in the de- flection angle due to thickness variations of the window. Spurious vibrations are observed for large scan angles at the mirror resonant frequency. To avoid these effects, the working scan angle was reduced to 20 optical angle for the inner axis and less than 30 optical angle for the outer axis. Optical resolution was estimated to be 5 m. In vivo 3D endo- scopic imaging of tissues was made by combining the two-axis scanning catheters and the Photonics 2021, 8, x FOR PEER REVIEW 14 of 25 multifunctional SD-OCT system. 3D images of a fingertip were acquired at 18.5 frames/s, with the scan performed with voltages of 2.8 V and 0.8 V applied on the inner and outer axis, covering 1.5 1 mm lateral scan range and consuming 150 mW of power. Figure 16 shows a 3D OCT image of fingertip tissue where the fingerprint orienta- Figure 16 shows a 3D OCT image of fingertip tissue where the fingerprint orientations tions are visible. are visible. Figure 16. 3D OCT image of fingertip tissue where the fingerprint orientations are visible (Figure Figure 16. 3D OCT image of fingertip tissue where the fingerprint orientations are visible (Figure from [55]). from [55]). Watanabe et al. [56] have demonstrated an electromagnetic MEMS scanner for OCT Watanabe et al. [56] have demonstrated an electromagnetic MEMS scanner for OCT imaging. The schematic diagram of a fabricated mirror scanner is shown in Figure 17. The imaging. The schematic diagram of a fabricated mirror scanner is shown in Figure 17. The device device incl includes udes a 0. a 0.2 2 mm mm thick thick si silicon licon fr frame ame c carrying arrying tthe he micromir micromirr ror orwit with h it its s ac actuator tuator, a , printed circuit board, and a magnet holder with a magnet inside. The minimum size of 2 2 the magnet is 6 × 6 × 5 mm . A metal coated 1.8 × 1.8 mm mirror and two y-scan coils are formed on the y-frame in the center of the silicon frame. The folded y-scan beams are supported by an x-frame. Two x-scan coils are formed on the x-frame. All the coils have a dimension of 2 × 2 mm . The folded x-scan beams are supported by an external fixed frame. When a current is passed through the y-scan coils, the mirror deflects in the y di- rection. When a current is passed through the four x-scan coils, the mirror tilts in the x direction. Thus, the light beam can be 2D steered. The entire microscanner is mounted on a 15 × 6 mm PCB, which is fixed on the magnet holder. The MEMS microscanner was placed in a Fourier domain OCT setup including a SLD light source operating at 1.55 µm and a fiber optic Michelson interferometer. OCT images of human fingers were obtained, as shown in Figure 18. The scanners discussed here demonstrate that one of the main drawbacks of electro- magnetic mirrors is that an external magnet is required for actuation. Such magnet tech- nology is often not compatible with the process flow of the micro-scanner. A bulky magnet reduces the potential of miniaturization of the probe. Another drawback concerns the rel- atively high-power consumption of electromagnetic scanners. Finally, the relative com- plexity of the fabrication process and high costs are a bottleneck for fast clinical translation of the electromagnetic MEMS scanners. Photonics 2021, 8, 6 14 of 25 a printed circuit board, and a magnet holder with a magnet inside. The minimum size 2 2 of the magnet is 6  6  5 mm . A metal coated 1.8  1.8 mm mirror and two y-scan coils are formed on the y-frame in the center of the silicon frame. The folded y-scan beams are supported by an x-frame. Two x-scan coils are formed on the x-frame. All the coils have a dimension of 2  2 mm . The folded x-scan beams are supported by an external fixed frame. When a current is passed through the y-scan coils, the mirror deflects in the y direction. When a current is passed through the four x-scan coils, the mirror tilts in the x Photonics 2021, 8, x FOR PEER REVIEW 15 of 25 direction. Thus, the light beam can be 2D steered. The entire microscanner is mounted on a 15  6 mm PCB, which is fixed on the magnet holder. Photonics 2021, 8, x FOR PEER REVIEW 15 of 25 Figure 17. Architecture of the MEMS electromagnetic scanner: (a) device schema and (b) the top view of the mirror actuator Figure 17. Architecture of the MEMS electromagnetic scanner: (a) device schema and (b) the top view of the mirror actu- (Figures from [56]). ator (Figures from [56]). The MEMS microscanner was placed in a Fourier domain OCT setup including a SLD Figure 17. Architecture of the MEMS electromagnetic scanner: (a) device schema and (b) the top view of the mirror actu- light source operating at 1.55 m and a fiber optic Michelson interferometer. OCT images ator (Figures from [56]). of human fingers were obtained, as shown in Figure 18. Figure 18. 3 × 3 mm OCT images of human fingers: the epidermis including stratum corneum and the crista cutis (Figures from [56]). Figure 18. 3  3 mm OCT images of human fingers: the epidermis including stratum corneum and the crista cutis (Figures Figure 18. 3 × 3 mm OCT images of human fingers: the epidermis including stratum corneum and the crista cutis (Figures 6. Electrothermal MEMS Scanning Mirrors for OCT from [56]). from [56]). 6.1. Principles of Electrothermal Microactuators The scanners discussed here demonstrate that one of the main drawbacks of elec- The principle of an electrothermal actuator is based on the Joule heating and thermal 6. Electrothermal MEMS Scanning Mirrors for OCT tromagnetic mirrors is that an external magnet is required for actuation. Such magnet expansion principles. The actuation uses the balance between the thermal energy gener- 6.1. Principles of Electrothermal Microactuators technology is often not compatible with the process flow of the micro-scanner. A bulky ated by an electrical current and the heat dissipation through the actuator structure The principle of an electrothermal actuator is based on the Joule heating and thermal magnet reduces the potential of miniaturization of the probe. Another drawback con- [57,58]. The three categories of electrothermal actuators are bimorph actuators, chevron expansion principles. The actuation uses the balance between the thermal energy gener- cerns the relatively high-power consumption of electromagnetic scanners. Finally, the actuators, and hot-and-cold-arm actuators. The structure of the more popular bimorph ated by an electrical current and the heat dissipation through the actuator structure actuator contains two layers of materials with different coefficients of thermal expansion [57,58]. The three categories of electrothermal actuators are bimorph actuators, chevron (CTE). A metallic heater is sandwiched between the active materials, as shown in Fig- actuators, and hot-and-cold-arm actuators. The structure of the more popular bimorph ure 19. The injection of electrical current within the heater layer generates the Joule effect actuator contains two layers of materials with different coefficients of thermal expansion in the active materials and produces the deflection angle. Equation (7) shows that the me- (CTE). A metallic heater is sandwiched between the active materials, as shown in Fig- chanical strain of material is directly proportional to the temperature change ∆ : ure 19. The injection of electrical current within the heater layer generates the Joule effect = ∆. (7) in the active materials and produces the deflection angle. Equation (7) shows that the me- chanical strain of material is directly proportional to the temperature change ∆ : The curvature of generated mechanical bending can be approximated as = ∆. (7) = , (8) The curvature of generated mechanical bending can be approximated as = , (8) Photonics 2021, 8, 6 15 of 25 relative complexity of the fabrication process and high costs are a bottleneck for fast clinical translation of the electromagnetic MEMS scanners. 6. Electrothermal MEMS Scanning Mirrors for OCT 6.1. Principles of Electrothermal Microactuators The principle of an electrothermal actuator is based on the Joule heating and thermal expansion principles. The actuation uses the balance between the thermal energy generated by an electrical current and the heat dissipation through the actuator structure [57,58]. The three categories of electrothermal actuators are bimorph actuators, chevron actuators, and hot-and-cold-arm actuators. The structure of the more popular bimorph actuator contains two layers of materials with different coefficients of thermal expansion a (CTE). A metallic heater is sandwiched between the active materials, as shown in Figure 19. The injection of electrical current within the heater layer generates the Joule effect in the active materials and produces the deflection angle. Equation (7) shows that the mechanical strain of material # is directly proportional to the temperature change DT: # = a DT. (7) The curvature of generated mechanical bending can be approximated as Photonics 2021, 8, x FOR PEER REVIEW 16 of 25 t + t R = , (8) # # 1 2 where t is the thickness of active layers and # represents the thermal strain of ac- where 1,2 is the thickness of active layers and 1,2 represents the thermal strain of active , , tive layers. A wide range of active materials can be used. Thus, the CTE of silicon −6 is layers. A wide range of active materials can be used. Thus, the CTE of silicon is 2.6 × 10 /K, 6 6 6 2.6  10 /K, while that of−6 SiO is 0.35  10 /K, aluminum −6 CTE is 25  10 /K, and while that of SiO2 is 0.35 × 10 /K, aluminum CTE is 25 × 10 /K, and the CTE of polyimide the CTE of polyimide from Amoco Ultradel −6 1414 is 191  10 /K. from Amoco Ultradel 1414 is 191 × 10 /K. Figure 19. Principle of electrothermal actuation. Figure 19. Principle of electrothermal actuation. 6.2. Examples of Electrothermal OCT Probes 6.2. Examples of Electrothermal OCT Probes In electrothermal actuators, the actuation force is typically larger than that of elec- In electrothermal actuators, the actuation force is typically larger than that of electro- trostatic and electromagnetic actuators. They benefit also from high fill factor. Such static and electromagnetic actuators. They benefit also from high fill factor. Such charac- characteristics make the electrothermal bimorph-based MEMS scanners very suitable for teristics make the electrothermal bimorph-based MEMS scanners very suitable for minia- miniaturizing of OCT probes for endoscopic applications. An example of such a probe, turizing of OCT probes for endoscopic applications. An example of such a probe, based based on a 2D thermal bimorph micromirror, has been developed by Sun et al. [59]. The on a 2D thermal bimorph micromirror, has been developed by Sun et al. [59]. The resulting resulting microscanner includes a mirror suspended without the gimbal by four actuators microscanner includes a mirror suspended without the gimbal by four actuators on its on its four sides. The actuator is based on three Al/SiO bimorph beams connected in four sides. The actuator is based on three Al/SiO2 bimorph beams connected in series with series with a Pt heater embedded for electrothermal actuation and two rigid suspended a Pt heater embedded for electrothermal actuation and two rigid suspended silicon silicon frames. This design is called lateral-shift-free (LSF) LVD design [60]. frames. This design is called lateral-shift-free (LSF) LVD design [60]. More recently, the architecture of a swept-source OCT endomicroscope has been demonstrated, including a mirror scanner with a similar mechanism of actuation [61]. The OCT probe contains a spectrally tuned single-channel Mirau micro-interferometer, inte- grated with a two axis MEMS electro-thermal micro-scanner. This optical microsystem operates in the side-imaging mode. Figure 20 shows the schematic of an OCT microsystem where a 4.7 × 4.7 × 5.3 mm Mirau microinterferometer is shown inside the blue square. The size of the microscanner external frame is 4 × 4 mm , with a mirror diameter of 1 mm. The monolithically integrated Mirau microinterferometer [62] includes a silicon base for a GRIN lens assembly port, a glass wafer with a reflowed focusing lens with a focal length of 9 mm, a reference micro- mirror, a silicon separator, and a beam splitter plate. The GRIN lens generates a collimated light beam with a diameter of 1 mm, which illuminates the plano-convex Mirau glass lens of 1.9 mm diameter [63]. The glass beam-splitting plate divides the converging beam into a reference beam and a scanning beam. A silicon separator ensures the position of the beam splitter plate at half of the focal length of the focusing lens. The reference beam is back-reflected from the 150 µm reference micromirror at the backside of the focusing lens, whereas the scanning beam is directed by the MEMS scanner towards the sample to be measured. The schematic of a MEMS scanner, based on a two-axis MEMS electrothermal micro-mirror, is shown in Figure 21a, while the image of a MEMS scanner assembled on top of a Mirau interferometer is shown in Figure 21b [58]. Photonics 2021, 8, 6 16 of 25 More recently, the architecture of a swept-source OCT endomicroscope has been demonstrated, including a mirror scanner with a similar mechanism of actuation [61]. The OCT probe contains a spectrally tuned single-channel Mirau micro-interferometer, integrated with a two axis MEMS electro-thermal micro-scanner. This optical microsystem operates in the side-imaging mode. Figure 20 shows the schematic of an OCT microsystem where a 4.7  4.7  5.3 mm Mirau microinterferometer is shown inside the blue square. The size of the microscanner external frame is 4  4 mm , with a mirror diameter of 1 mm. The monolithically integrated Mirau microinterferometer [62] includes a silicon base for a GRIN lens assembly port, a glass wafer with a reflowed focusing lens with a focal length of 9 mm, a reference micromirror, a silicon separator, and a beam splitter plate. The GRIN lens generates a collimated light beam with a diameter of 1 mm, which illuminates the plano-convex Mirau glass lens of 1.9 mm diameter [63]. The glass beam-splitting plate divides the converging beam into a reference beam and a scanning beam. A silicon separator ensures the position of the beam splitter plate at half of the focal length of the focusing lens. The reference beam is back-reflected from the 150 m reference micromirror at the backside of the focusing lens, whereas the scanning beam is directed by the MEMS scanner towards the sample to be measured. The schematic of a MEMS scanner, based on a two-axis MEMS electrothermal micro-mirror, is shown in Figure 21a, while the image of a MEMS scanner assembled on top of a Mirau interferometer is shown in Figure 21b [58]. The inner mirror plate is connected to a rigid frame via a pair of torsional bars in two diametrically opposite ends located on the rotation axis. A pair of electrothermal bimorphs generates a force onto the perpendicular free ends of the mirror plate in the same angular direction. An array of electrothermal bimorph cantilevers deflects the rigid frame and a mechanical stopper maintains the position of the mirror inclined at 45 from the optical axis. The performed scans reach large mechanical angles of 32 for the frame mirror and 22 for the in-frame mirror. Figure 21c,d shows the deflection amplitude of the micromirror versus the frequency response of the pitch-axis and roll-axis, respectively. Here, a small coupling between both the axes is observed. The fabricated micromirror has a mechanical resonant frequency around 1200 Hz for both axes. Figure 22a shows the real-time reconstruction of a “femto-st” pattern, obtained by Lissajous imaging at a sampling frequency of 1 MHz [64]. The acquisition of data was Photonics 2021, 8, x FOR PEER REVIEW 17 of 25 performed in open-loop mode. The pattern lines of 30 m-wide are well-resolved. Figure 20. Schematic diagram of the MOEMS (Micro-Opto-Electro-Mechanical Systems) probe. Figure 20. Schematic diagram of the MOEMS (Micro-Opto-Electro-Mechanical Systems) probe. Figure 21. Electrothermal mirror: (a) schematic and (b) microphotograph of a MEMS scanner assembled on top of a Mirau interferometer. The frequency response for inner pitch axis (c) and roll-axis (d). The inner mirror plate is connected to a rigid frame via a pair of torsional bars in two diametrically opposite ends located on the rotation axis. A pair of electrothermal bi- morphs generates a force onto the perpendicular free ends of the mirror plate in the same angular direction. An array of electrothermal bimorph cantilevers deflects the rigid frame and a mechanical stopper maintains the position of the mirror inclined at 45° from the optical axis. The performed scans reach large mechanical angles of 32° for the frame mirror and 22° for the in-frame mirror. Figure 21c,d shows the deflection amplitude of the micro- mirror versus the frequency response of the pitch-axis and roll-axis, respectively. Here, a small coupling between both the axes is observed. The fabricated micromirror has a me- chanical resonant frequency around 1200 Hz for both axes. Figure 22a shows the real-time reconstruction of a “femto-st” pattern, obtained by Lissajous imaging at a sampling frequency of 1 MHz [64]. The acquisition of data was performed in open-loop mode. The pattern lines of 30 µm-wide are well-resolved. Photonics 2021, 8, x FOR PEER REVIEW 17 of 25 Photonics 2021, 8, 6 17 of 25 Figure 20. Schematic diagram of the MOEMS (Micro-Opto-Electro-Mechanical Systems) probe. Photonics 2021, 8, x FOR PEER REVIEW 18 of 25 Figure 21. Electrothermal mirror: (a) schematic and (b) microphotograph of a MEMS scanner assembled on top of a Mirau Figure 21. Electrothermal mirror: (a) schematic and (b) microphotograph of a MEMS scanner assembled on top of a Mirau interferometer. The frequency response for inner pitch axis (c) and roll-axis (d). interferometer. The frequency response for inner pitch axis (c) and roll-axis (d). The inner mirror plate is connected to a rigid frame via a pair of torsional bars in two diametrically opposite ends located on the rotation axis. A pair of electrothermal bi- morphs generates a force onto the perpendicular free ends of the mirror plate in the same angular direction. An array of electrothermal bimorph cantilevers deflects the rigid frame and a mechanical stopper maintains the position of the mirror inclined at 45° from the optical axis. The performed scans reach large mechanical angles of 32° for the frame mirror and 22° for the in-frame mirror. Figure 21c,d shows the deflection amplitude of the micro- mirror versus the frequency response of the pitch-axis and roll-axis, respectively. Here, a small coupling between both the axes is observed. The fabricated micromirror has a me- chanical resonant frequency around 1200 Hz for both axes. Figure 22. Reconstruction of a “femto-st” pattern in (a) and B-scan of a multilayer glass sample in (b). Figure 22. Reconstruction of a “femto-st” pattern in (a) and B-scan of a multilayer glass sample in (b). Figure 22a shows the real-time reconstruction of a “femto-st” pattern, obtained by The complete OCT probe is connected to the illumination and detection blocks by Lissajous imaging at a sampling frequency of 1 MHz [64]. The acquisition of data was The complete OCT probe is connected to the illumination and detection blocks by a a single-mode optical fiber. The system is illuminated by a swept source with a central performed in open-loop mode. The pattern lines of 30 µm-wide are well-resolved. single-mode optical fiber. The system is illuminated by a swept source with a central wavelength of 840 nm and A-scan frequency of 110 kHz [65]. The OCT images were wavelength of 840 nm and A-scan frequency of 110 kHz [65]. The OCT images were ob- obtained from a sample made of three cover glasses, each 160 m thick. Figure 22b shows tained from a sample made of three cover glasses, each 160 µm thick. Figure 22b shows the averaged B-scan images of this sample. the averaged B-scan images of this sample. Circumferential scanning for endoscopic OCT with MEMS electrothermal mirrors was Circumferential scanning for endoscopic OCT with MEMS electrothermal mirrors demonstrated in 2018 by S. Luo et al. [66]. This microscanner uses a circular array of six was demonstrated in 2018 by S. Luo et al. [66]. This microscanner uses a circular array of scan units, including electrothermal MEMS mirrors and C-lens collimators with a focal six scan units, including electrothermal MEMS mirrors and C-lens collimators with a focal length greater than 10 mm, as shown in Figure 23. This compact microscanner presented a length greater than 10 mm, as shown in Figure 23. This compact microscanner presented chip size of 1.5  1.3 mm . Each C-lens and a single-mode fiber are packaged inside a glass a chip size of 1.5 × 1.3 mm . Each C-lens and a single-mode fiber are packaged inside a tube with a diameter of 1.4 mm. The full circumferential scans have been demonstrated glass tube with a diameter of 1.4 mm. The full circumferential scans have been demon- with individual micromirrors scanning up to 45 at a voltage of less than 12 V. strated with individual micromirrors scanning up to 45° at a voltage of less than 12 V. Figure 24a shows the MEMS mirrors with a 0.5  0.5 mm mirror plate. Four bimorph actuators are placed symmetrically at the four sides of a central mirror plate. Each bimorph actuator consists of three pairs of double S-shaped bimorph (Al/SiO2) beams, sandwiched with a heater layer made of a thin film of Ti/TiN. A mechanical scan angle of 13 is achieved, resulting in a 26 optical scan angle or a 52 field of view (FOV). Figure 23. The schematic of MEMS OCT probe (Figures from [66]). Figure 24a shows the MEMS mirrors with a 0.5 × 0.5 mm mirror plate. Four bimorph actuators are placed symmetrically at the four sides of a central mirror plate. Each bi- morph actuator consists of three pairs of double S-shaped bimorph (Al/SiO2) beams, sand- wiched with a heater layer made of a thin film of Ti/TiN. A mechanical scan angle of 13° is achieved, resulting in a ±26° optical scan angle or a 52° field of view (FOV). Configured with a swept-source OCT setup, this MEMS array-based circumferential scanning probe was applied to image a swine’s small intestine wrapped on a 20 mm-di- ameter glass tube, as shown in Figure 24b. The OCT imaging result shows that this new MEMS endoscopic OCT has promising applications in large tubular organs. An alternative solution is to move the objective lens directly by an actuated mi- crostage. L. Wu et al. [67] demonstrated a tunable microlens scanner, operating at a 880 nm wavelength. This microscanner uses a lateral-shift-free electrothermal bimorph actu- ator, carrying a 1 mm diameter glass rod lens moving at a resonance frequency of 79 Hz. Later, L. Liu et al. [68] developed another electrothermal MEMS microlens scanner, mov- ing a 2.4 mm plano-convex glass lens with a maximum travel range of 400 µm and a res- onance frequency of 24 Hz. Photonics 2021, 8, x FOR PEER REVIEW 18 of 25 Figure 22. Reconstruction of a “femto-st” pattern in (a) and B-scan of a multilayer glass sample in (b). The complete OCT probe is connected to the illumination and detection blocks by a single-mode optical fiber. The system is illuminated by a swept source with a central wavelength of 840 nm and A-scan frequency of 110 kHz [65]. The OCT images were ob- tained from a sample made of three cover glasses, each 160 µm thick. Figure 22b shows the averaged B-scan images of this sample. Circumferential scanning for endoscopic OCT with MEMS electrothermal mirrors was demonstrated in 2018 by S. Luo et al. [66]. This microscanner uses a circular array of six scan units, including electrothermal MEMS mirrors and C-lens collimators with a focal length greater than 10 mm, as shown in Figure 23. This compact microscanner presented Photonics 2021, 8, 6 a chip size of 1.5 × 1.3 mm . Each C-lens and a single-mode fiber are packaged insi 18 de a of 25 glass tube with a diameter of 1.4 mm. The full circumferential scans have been demon- strated with individual micromirrors scanning up to 45° at a voltage of less than 12 V. Photonics 2021, 8, x FOR PEER REVIEW 19 of 25 Figure 23. The schematic of MEMS OCT probe (Figures from [66]). Figure 23. The schematic of MEMS OCT probe (Figures from [66]). Figure 24a shows the MEMS mirrors with a 0.5 × 0.5 mm mirror plate. Four bimorph actuators are placed symmetrically at the four sides of a central mirror plate. Each bi- morph actuator consists of three pairs of double S-shaped bimorph (Al/SiO2) beams, sand- wiched with a heater layer made of a thin film of Ti/TiN. A mechanical scan angle of 13° is achieved, resulting in a ±26° optical scan angle or a 52° field of view (FOV). Configured with a swept-source OCT setup, this MEMS array-based circumferential scanning probe was applied to image a swine’s small intestine wrapped on a 20 mm-di- ameter glass tube, as shown in Figure 24b. The OCT imaging result shows that this new MEMS endoscopic OCT has promising applications in large tubular organs. An alternative solution is to move the objective lens directly by an actuated mi- crostage. L. Wu et al. [67] demonstrated a tunable microlens scanner, operating at a 880 nm wavelength. This microscanner uses a lateral-shift-free electrothermal bimorph actu- ator, carrying a 1 mm diameter glass rod lens moving at a resonance frequency of 79 Hz. Later, L. Liu et al. [68] developed another electrothermal MEMS microlens scanner, mov- ing a 2.4 mm plano-conv (ex a) ( glass lens with a maximum travel range of 4 b) 00 µm and a res- onance frequency of 24 Hz. Figure 24. An SEM image of the MEMS mirror in (a) and the imaging sample of a swine’s small Figure 24. An SEM image of the MEMS mirror in (a) and the imaging sample of a swine’s small intestine covering a glass tube in (b) (Figures from [66]). intestine covering a glass tube in (b) (Figures from [66]). A more compact microlens scanner was developed in 2020 by L. Zhou et al. [69] Configured with a swept-source OCT setup, this MEMS array-based circumferential where the actuation mechanism is based on a single serpentine inverted-series-connected scanning probe was applied to image a swine’s small intestine wrapped on a 20 mm- (ISC) electrothermal bimorph actuator carrying a microlens. The shape of the entire mi- diameter glass tube, as shown in Figure 24b. The OCT imaging result shows that this new crolens scanner is circular, with an outer diameter of 4.4 mm and a clear optical aperture MEMS endoscopic OCT has promising applications in large tubular organs. of 1.8 m An m alternative , as shown in F solution igure is 25to a. Th move e microst the objective age include lens s a rin directly g-shap by ed an fram actuated e and eimi- ght sets of bimorph actuators. It is loaded with a 2.4 mm plano-convex glass microlens on its crostage. L. Wu et al. [67] demonstrated a tunable microlens scanner, operating at a 880 nm wavelength. ring-shaped frame. The m This microscanner icrolens we uses aig lateral-shift-fr ht is about 8 ee mg. Figure electrothermal 25b shows the bimorph S-sh actuator aped , carrying form of an IS a 1 mm C elect diameter rotherm glass al act rod uatlens or where moving the ro at a und hin resonance ges ar frequency e used to connect all ISC of 79 Hz. Later, L. Liu et al. [68] developed another electrothermal MEMS microlens scanner, moving a structures, minimizing the residual stresses during motion. The resonant frequency of the 2.4 mm plano-convex glass lens with a maximum travel range of 400 m and a resonance MEMS microstage loaded with a lens reaches 140 Hz, which is acceptable for endomicro- frequency of 24 Hz. scopic imaging tasks. A more compact microlens scanner was developed in 2020 by L. Zhou et al. [69] where (a) (b) the actuation mechanism is based on a single serpentine inverted-series-connected (ISC) electrothermal bimorph actuator carrying a microlens. The shape of the entire microlens scanner is circular, with an outer diameter of 4.4 mm and a clear optical aperture of 1.8 mm, as shown in Figure 25a. The microstage includes a ring-shaped frame and eight sets of bimorph actuators. It is loaded with a 2.4 mm plano-convex glass microlens on its ring- shaped frame. The microlens weight is about 8 mg. Figure 25b shows the S-shaped form of an ISC electrothermal actuator where the round hinges are used to connect all ISC structures, minimizing the residual stresses during motion. The resonant frequency of the MEMS microstage loaded with a lens reaches 140 Hz, which is acceptable for endomicroscopic imaging tasks. Figure 25. SEMs of the microlens scanner: (a) top view of complete structure and (b) focus on one of eight inverted-series-connected (ISC) actuators where the silicon beam is suspended 257 µm above the outer silicon frame (Figures from [69]). 7. Discussion on the Microscanner Design and Conclusions The replacement of galvanometer mirrors for OCT beam scanning by 2D MEMS scan- ning mirrors converted the bulk microscope to a compact and light device. However, gal- vanometers demonstrate better performance because they operate in a closed loop, using position feedback to correct the drive waveform instead of an open loop for MEMS scanning *mirrors. When compared to MEMS mirrors, galvanometers are higher cost and relatively larger. This paper demonstrated that current MEMS scanning technologies have advantages and limitations compared to galvanometers. The use of MEMS scanning mirrors in OCT Photonics 2021, 8, x FOR PEER REVIEW 19 of 25 (a) (b) Figure 24. An SEM image of the MEMS mirror in (a) and the imaging sample of a swine’s small intestine covering a glass tube in (b) (Figures from [66]). A more compact microlens scanner was developed in 2020 by L. Zhou et al. [69] where the actuation mechanism is based on a single serpentine inverted-series-connected (ISC) electrothermal bimorph actuator carrying a microlens. The shape of the entire mi- crolens scanner is circular, with an outer diameter of 4.4 mm and a clear optical aperture of 1.8 mm, as shown in Figure 25a. The microstage includes a ring-shaped frame and eight sets of bimorph actuators. It is loaded with a 2.4 mm plano-convex glass microlens on its ring-shaped frame. The microlens weight is about 8 mg. Figure 25b shows the S-shaped form of an ISC electrothermal actuator where the round hinges are used to connect all ISC Photonics 2021, 8, 6 19 of 25 structures, minimizing the residual stresses during motion. The resonant frequency of the MEMS microstage loaded with a lens reaches 140 Hz, which is acceptable for endomicro- scopic imaging tasks. (a) (b) Figure 25. SEMs of the microlens scanner: (a) top view of complete structure and (b) focus on one Figure 25. SEMs of the microlens scanner: (a) top view of complete structure and (b) focus on one of of eight inverted-series-connected (ISC) actuators where the silicon beam is suspended 257 µm eight inverted-series-connected (ISC) actuators where the silicon beam is suspended 257 m above above the outer silicon frame (Figures from [69]). the outer silicon frame (Figures from [69]). 7. Discussion on the Microscanner Design and Conclusions 7. Discussion on the Microscanner Design and Conclusions The replacement of galvanometer mirrors for OCT beam scanning by 2D MEMS scan- The replacement of galvanometer mirrors for OCT beam scanning by 2D MEMS ning mirrors converted the bulk microscope to a compact and light device. However, gal- scanning mirrors converted the bulk microscope to a compact and light device. However, vanometers demonstrate better performance because they operate in a closed loop, using galvanometers demonstrate better performance because they operate in a closed loop, using position feed position back to c feedback orrec to t corr the dri ect the ve waveform drive waveform insteadinstead of an opof enan loo open p for M loop EM for S sc MEMS anning scanning *mirrors. When compared *mirrors. When compar to MEMS mir ed to MEMS rors, ga mirr lvanomete ors, galvanometers rs are higher cos are higher t and relativel cost andy relatively larger. larger. This This p paper aperdemonstrated demonstrated t th h at atcurr current ent MEMS MEMSscanning scanningtechnologies technologies h have ave advantages advantages and and limit limitations ations compared compared to to galv galvanometers. anometers. The use The use of of MEM MEMSSscanning scanning mir mirrr ors ors in OCT in OCT and other biomedical applications reduces the complexity of scan control and offers a lower cost scanner. To build minimally invasive endoscopic probes, scanning micromirrors are required to be compact (<5 mm). The size of the scanning mirror is a crucial parameter because the micromirror dimension should be larger than the laser spot size as well. High speed and large transverse scans can also be achieved, which enables real-time in vivo imaging and a large field of view, respectively. To design the ideal MEMS scanner for OCT, we need to consider the scan angle, the resonance frequency, dynamic, as well as static mirror flatness, and good resonant mode separation. The choice of these parameters influences the image quality, the desired resolution, and the presence of image distortions due to aberrations from the scanning mirror and objective lens and defines the limits for scan speed and total image size. As we demonstrated in Section 2, the specifications of a MEMS mirror determine the number of resolvable spots and the OCT B-scan rate. Increasing the size of the MEMS mirror or scan angle would increase the number of resolvable spots. The repetition frequency of a B-scan is limited by the resonance frequency of the MEMS mirror. Such resonance frequency is defined by the mirror inertia and is proportional to the inverse square of the mirror diameter. Generally, we would select a MEMS scanner that achieves the largest angle and the highest operating speed if large areas are to be imaged quickly. In clinical applications, the minimal line scan lengths must cover approximately a range from 1 mm to 2 mm. To maximize the imaging speed, the solution is to perform the scan at the resonance frequency of the MEMS mirror. However, the scan operation performed at the resonance frequency is a source of image distortions produced by parasitic vibrations of scanner axes. To overcome the parasitic vibrations, the B-scan frequency must be fixed below the resonance frequency of the MEMS mirror. The size of scanning micromirrors included in the OCT probe strongly influence the scanning performances and specifications of the OCT probe. Smaller diameter scanning mirrors facilitate both the integration of the probe within the standard endoscope and probe guidance in the internal organs. Increasing the mirror diameter reduces the resonance frequency, resulting in a slower scan repetition frequency, limiting the number of B-scans. Photonics 2021, 8, 6 20 of 25 In this case, if the suspension is stiffened to maintain a given speed, the torque available from the actuator must be increased. The thickness of the MEMS device is also crucial in the definition of micromechanical features of the MEMS device. Thus, the thicker electrodes of a comb-drive actuator improve the electrostatic force without affecting the deformation of the scanning micromirror. A thinner air gap of the comb-drive, obtained via surface micromachining, would decrease the high driving voltage (often ~100 V), making the device safer for patients. In microactuators where the moving part is a thin membrane, thinner membranes improve the motion range because of the increase of actuator deformation. In conclusion, the advantages of smaller MEMS scanning mirrors include the smaller mass, lower stiffness, and higher imaging speed. All of these parameters must be carefully considered when choosing the appropriate MEMS scanning mirror for a specific application of OCT imaging. Many parameters of the MEMS device and optical probe also must be selected during design, fabrication, or assembly. Others can be adjusted during the OCT experiment. Table 1 compares qualitatively the characteristics of MEMS scanning micromirrors for three types of actuation mechanisms, based on the literature review [70]. This study focuses on the highest performing MEMS scanning mirrors designed for miniaturized displays and optical imaging. The typical diameter of scanning micromirrors ranges from 0.5 mm to 3.5 mm. However, an average diameter of 1 mm is observed for all categories of MEMS scanning mirrors. Average values of micromirror mechanical specifications for all types of actuators are compared. Analyzing the range of motion for electrostatic microactuators, we can see that the average performance of the vertical comb actuators is 14 for an average piston motion of 94 m, which is better than that of linear comb-drives (8.5 for piston motion of 29 m). The average performance of electromagnetic actuators is 15 and 5 m for angular motion and piston motion, respectively. Finally, the electrothermal actuators present the best performances in both rotational motion and out-of-plane motion. Here, the average angular motion is about 27 and 280 m for average piston motion. An important characteristic to be analyzed is the resonance frequency for each category of microactuators. The resonance frequency of electrothermal and electromagnetic micromirrors is in the range from 100 Hz to 1000 Hz, while the resonance frequency of the comb-drives stays within the range of 150 Hz to 10,000 Hz. The required response time for dynamic systems is about 5 ms for scanning micromirrors implemented in OCT probes. Table 1 shows that each actuation principle has advantages in some aspects while having disadvantages in others. Electrostatic actuation has a fast response and the lowest power consumption, but it requires large driving voltage, which may not be safe for endoscopic applications. In addition, the electrostatic scanners have strong nonlinearities, limiting the MEMS displacement. Here, the electrostatic torque is a function of V , not linearly varying with the actuation voltage. This can result in a distortion of the scan pattern when driving with linearly ramped voltages. Several approaches of linearization have been proposed to eliminate the distortion, improving the linearity of scan patterns [71]. Finally, an additional advantage for electrostatic comb-drive actuators is the easy and standardized micromachining technology. The main disadvantage of electrostatic actuators is the pull-in voltage limiting the linear displacement and the relatively high driving voltage. Electromagnetic actuation offers a large scan angle, low driving voltage, and relatively more linear response than the competing actuation mechanisms. Despite the advantages of MEMS electromagnetic scanners, the achievable performance is limited by the large thermal dissipation inside a coil. In addition, magnets strong enough for high performance present significant volume and might require magnetic shielding. This leads to total package sizes that are larger than the competing MEMS actuators. Finally, they are complex and difficult to fabricate, particularly at small scales. Electrothermal actuators present large scan angles at low driving voltages. They offer the largest fill factor compared to other categories of MEMS mirrors. However, the thermal response is relatively slow, but sufficient to perform real-time imaging. Since electrothermal actuators have relatively simple geometries, they are easy to fabricate and Photonics 2021, 8, 6 21 of 25 can be made with a high fill factor. Electrothermal scanners seem to be excellent candidates to satisfy the requirements of OCT endoscopic imaging applications, however significant engineering effort is still needed to limit optical aberrations and to control scanner stability. Electrothermal scanners on silicon mirror plates that are suspended by a pair of torsion bars often present a temperature dependence in their oscillation behavior. This results in mechanical buckling of the beam. In this case, the initial position of mirror deflection can be difficult to control because the center of gravity of the oscillation system can be displaced from the initial position, changing the resonant mode and degrading the amplitude of oscillation when the scanner is driven at a constant frequency. In this case, the control of initial tilt angles can be difficult, causing optical alignment problems. The elastic constant of silicon can be a source of decreasing the temperature rise, generating a drift of resonance frequency—this is the case for all actuation mechanisms. For all three actuation mechanisms, the one considered to have the most appropriate architecture seems to be the resonant scanner with gimbaled orthogonal single-axis mirrors. Here, the design needs to strike a balance where the combination of scan angle, resonance frequency, and mirror size is enough for definition of the desired resolution, while keeping the mirror optically flat to avoid image distortions. One important challenge with gimbaled dual-axis scanners is the possibility of crosstalk between the two axes. Another challenge for MEMS scanners developed for endoscopic applications is compact package size, requiring a package of around 10 mm or less. For all three mechanisms of actuation, a scanning mirror fails to maintain its flatness when it is mechanically oscillated at resonance frequency. The mirror surface should be sufficiently flat if the mirror curvature is less than 1 m, so as to not distort the beam. Mirror deformation leads to unwanted expansion of the reflected beam spot, degrading the image quality by blurring near the left- and right–hand side edges of the projected image. To overcome this problem, Hsu et al. [72] proposed to modify the mirror plate by generating a backside island to improve the rigidity of the mirror. Thus, masses far from the rotation axis are removed, keeping the resonant frequency high, while removing those in the central part. The structure of a gimbaled 2D scanner can be subjected to external vibration as well as intrinsic oscillations of the orthogonal axis. In particular, a scanner spinning about its y-axis at resonance frequency can be forced to tilt in the orthogonal x-axis by the gimbal structure because Coriolis forces are generated [73]. This last effect results in a small coupling of both axes, blurring the image near the edges. The solution is to design the scanner mechanism to have no resonant coupling, making the resonance frequencies of the orthogonal axes distant. It is also critical that the microscanner exhibits excellent accuracy of scan repeatability. The angular accuracy must be better than 1% of the angle step size. One limitation of OCT probes using MEMS scanning mirrors arises because it is necessary to use a sinusoidal scan, as described in Section 2. The sinusoidal scan is less linear compared to linear scanning because the scan speed is not uniform at the center and at the edges of the B-scan because of non-uniformly spaced A-scans. The linearization of the scan requires driving the MEMS with higher harmonics of the scan frequency where parasitic vibrations of the resonance frequency appear since the MEMS scanning mirror does not have a closed loop feedback system. In this case, it is pertinent to implement a closed feedback loop, based on the integration of a piezoresistive strain gauge on the torsional beams of the scanning mirror to sense the beam strain and adjust the microscanner angle. As discussed here, challenges need to be overcome to enable the implementation of MEMS scanners into endoscopic OCT systems, requiring minimally invasive recording of cross-sectional images in vivo with high resolution and high speed. To date, the electrother- mal two-axis and resonant MEMS scanners seem to be the closest candidates to satisfy the requirements of endoscopic imaging applications under conditions to better control scanner stability. In conclusion, we hope to clearly demonstrate that MEMS scanner-based OCT probes offer several significant advantages and we expect them to have a bright future in mobile medical imaging devices. Photonics 2021, 8, 6 22 of 25 Table 1. Comparison of three main actuation mechanisms used for scanning in OCT probes. Mirror Size Angular Deflection Resonance Frequency Advantages Drawbacks (mm)/Data from [46] ( )/Data from [46] (Hz)/Data from [46] Pull-in effect Fast response High driving voltage Electrostatic all Low power consumption (50–100 V) Large scan angle Low force <10 Linear comb ~1 Ave scan: 8.5 250–5000 (ave 1800) Ave piston: 29 m >10 Vertical comb ~1 Ave scan: 14 150–10,000 (ave 5000) Ave piston: 94 m electrostatic attractive requires mechanical a pair of electrodes with Small temperature Silicon DRIE and Characteristics force between resonance to enhance scan air gap dependance metallization conductors angle High power consumption Maximum: 20 Larger driving External magnets Electromagnetic 0.8–3.5 Ave scan: 15 100–1000 Lower driving voltage increasing the size Ave piston: 5 m Large scan angle Electromagnetic interferences Lorentz force coil and permanent Multiple layers of metal and Small temperature Characteristics between current and size limited by magnet magnet insulator for coil dependance magnetic field ~25, maximum 64 Large scan angle High power Electrothermal 0.5–1 Ave scan: 27 117–1000 Low driving voltage consumption Ave piston: 270 m High fill-factor Slow response materials with different thermal expansion by multiple layers of metal and Requires relatively high Bimorph sensitive to Characteristics thermal expansion Joule effect insulator for heater power temperature change coefficients Funding: This work was supported by the collaborative project VIAMOS of the European Commis- sion (FP7, ICT program, grant no. 318542), the ANR Labex Action program (ANR-11-LABX-0001-01), and received a support from Collégium SMYLE. It was also supported by the French RENATECH network and its FEMTO-ST technological facility. Work is supported by the Foundation for Polish Science co-financed by the EU under the European Regional Fund (project no MAB/2019/12). Conflicts of Interest: The authors declare no conflict of interest. References 1. Petersen, K.E. Silicon as a mechanical materiel. IEEE 1982, 70, 420–457. [CrossRef] 2. Motamedi, M.E. MOEMS: Micro-Opto-Electro-Mechanical Systems; SPIE Press Bellingham: Washington, DC, USA, 2005. 3. Wu, M.C.; Solgaard, O.; Ford, J.E. Optical MEMS for Lightwave Communication. J. Lightwave Technol. 2006, 24, 4433–4454. [CrossRef] 4. Neumann, H.; Kiesslich, R.; Wallace, M.B.; Neurath, M.F. Confocal laser endomicroscopy: Technical advances and clinical applications. Gastroenterology 2010, 139, 388–392. [CrossRef] [PubMed] 5. Gmitro, A.F.; Aziz, D. Confocal microscopy through a fiber-optic imaging bundle. Opt. Lett. 1993, 18, 565–567. [CrossRef] [PubMed] 6. Bargiel, S.; Baranski, M.; Wiemer, M.; Frömel, J.; Wang, W.S.; Gorecki, C. Technological Platform for Vertical Multi-Wafer Integration of Microscanners and Micro-Optical Components. Micromachines 2019, 10, 185. [CrossRef] [PubMed] 7. Denk, W.; Strickler, J.H.; Webb, W.W. Two-photon laser scanning fluorescence microscopy. Science 1990, 248, 73–76. [CrossRef] 8. Koehler, M.J.; Vogel, T.; Elsner, P.; König, K.; Bückle, R.; Kaatz, M. In vivo measurement of the human epidermal thickness in different localizations by multiphoton laser tomography. Skin Res. Technol. 1010, 16, 259–264. 9. Huang, D.; Swanson, E.A.; Lin, C.P.; Schuman, J.S.; Stinson, W.G.; Chang, W.; Hee, M.R.; Flotte, T.; Gregory, K.; Puliafito, C.A. Optical coherence tomography. Science 1991, 254, 1178–1181. [CrossRef] 10. Lu, C.D.; Kraus, M.F.; Potsaid, B.; Liu, J.J.; Choi, W.; Jayaraman, V.; Cable, A.E.; Hornegger, J.; Duker, J.S.; Fujimoto, J.G. Handheld ultrahigh speed swept source optical coherence tomography instrument using a MEMS scanning mirror. Biomed. Opt. Express 2014, 5, 293–311. [CrossRef] 11. Cao, T.; Tey, H.L. High-definition optical coherence tomography—An aid to clinical practice and research in dermatology. J. Dtsch. Dermatol. Ges. 2015, 13, 886–890. [CrossRef] 12. Yin, C.; Glaser, A.K.; Leigh, S.Y.; Chen, Y.; Wei, L.; Pillai, P.C.S.; Rosenberg, M.C.; Abeytunge, S.; Peterson, G.; Glazowski, C.; et al. Miniature in vivo MEMS-based line-scanned dual-axis confocal microscope for point-of-care pathology. Biomed. Opt. Express 2016, 2, 251–263. [CrossRef] [PubMed] Photonics 2021, 8, 6 23 of 25 13. Patterson, P.R.; Hahb, D.; Fujino, M.; Piyawattanametha, W.; Wu, M.C. Scanning micromirrors: An overview. Proc. SPIE 2004, 5604, 195–207. 14. Puliafito, C.A.; Hee, M.R.; Lin, C.P.; Reichel, E.; Schuman, J.S.; Duker, J.S.; A Izatt, J.; A Swanson, E.; Fujimoto, J.G. Imaging of macular diseases with optical coherence tomography. Ophthalmology 1995, 102, 217–229. [CrossRef] 15. Schuman, J.S.; Hee, M.R.; Arya, A.V.; Pedut-Kloizman, T.; Puliafito, C.A.; Fujimoto, J.G.; Swanson, E.A. Optical coherence tomography: A new tool for glaucoma diagnosis. Curr. Opin. Ophthalmol. 1995, 6, 89–95. [CrossRef] [PubMed] 16. Hitzenberger, C.K.; Drexler, W.; Leitgeb, R.A.; Findl, O.; Ferche, A.F. Key Developments for Partial Coherence Biometry and Optical Coherence Tomography in the Human Eye Made in Vienna. Investig. Ophthalmol. Vis. Sci. 2016, 57, 460–474. [CrossRef] [PubMed] 17. Sivak, M.V.; Kobayashi, K.; Izatt, J.A.; Rollins, A.M.; Ung-Runyawee, R.; Chak, A.; Wong, R.C.; Isenberg, G.A.; Willis, J. High- resolution endoscopic imaging of the GI tract using optical coherence tomography. Gastrointest. Endosc. 2000, 51, 474–479. [CrossRef] 18. Welzel, J. Optical coherence tomography in dermatology: A review. Ski. Res. Technol. 2001, 7, 1–9. [CrossRef] 19. Pierce, M.C.; Sheridan, R.L.; Park, B.H.; Cense, B.; De Boer, J.F. Collagen denaturation can be quantified in burned human skin using polarization-sensitive optical coherence tomography. Burns 2004, 30, 511–517. [CrossRef] 20. Gora, M.J.; Suter, M.J.; Tearney, G.J.; Li, X. Endoscopic optical coherence tomography: Technologies and clinical applications. Biomed. Opt. Express 2017, 8, 2405–2444. [CrossRef] 21. Nine Point Medical. Available online: https://www.ninepointmedical.com/ (accessed on 30 December 2020). 22. Cardiovascular. Available online: https://www.cardiovascular.abbott/us/en/home.html (accessed on 30 December 2020). 23. Santec. Swept-Source OCT System. Available online: https://www.santec.com/en/products/oct/swept-source-oct/ (accessed on 30 December 2020). 24. Thor Labs. OCTH-1300-Handheld Scanner. Available online: https://www.thorlabs.com/thorproduct.cfm?partnumber=OCTH- 1300 (accessed on 30 December 2020). 25. MirrorcleTechnologies. Biomedical Imaging. Available online: https://www.mirrorcletech.com/wp/applications/#biomed (accessed on 30 December 2020). 26. Wang, D.; Liang, P.; Samuelson, S.; Jia, H.; Ma, J.; Xie, H. Correction of image distortions in endoscopic optical coherence tomography based on two-axis scanning MEMS mirrors. Biomed. Opt. Express 2013, 4, 2066–2077. [CrossRef] 27. Lin, L.Y.; Keeler, E.G. Progress of MEMS Scanning Micromirrors for Optical Bio-Imaging. Micromachines 2015, 6, 1675–1689. [CrossRef] 28. McCormick, D.; Jung, W.; Milanovic, V.; Chen, Z.; Tien, N. 3-D MEMS based real-time minimally invasive endoscopic optical coherence tomography. In Proceedings of the IEEE/LEOS International Conference on Optical MEMS and Their Applications Conference, Oulu, Finland, 25–26 August 2005; pp. 25–26. 29. Urey, H.; Wine, D.W.; Osborn, T.D. Optical performance requirements for MEMS-scanner based microdisplays. Proc. SPIE 2000, 4178, 176–185. 30. Hwang, K.; Seo, Y.-H.; Jeong, K.-H. Microscanners for optical endomicroscopic applications. Micro Nano Syst. Lett. 2017, 5. [CrossRef] 31. Urey, H.; Kan, C.; O Davis, W. Vibration mode frequency formulae for micromechanical scanners. J. Micromech. Microeng. 2005, 15, 1713–1721. [CrossRef] 32. Holmstrom, S.T.S.; Baran, U.; Urey, H. MEMS Laser Scanners: A Review. J. Microelectromech. Syst. 2014, 23, 259–275. [CrossRef] 33. Boppart, S.A.; Bouma, B.E.; Pitris, C.; Tearney, G.J.; Fujimoto, J.G.; Brezinski, M.E. Forward-imaging instruments *for optical coherence tomography. Opt. Lett. 1997, 22, 1618–1620. [CrossRef] 34. Sun, J.; Xie, H. MEMS-Based Endoscopic Optical Coherence Tomography. Int. J. Opt. 2011, 2011, 1–12. [CrossRef] 35. Zaeh, M.; Moesl, J.; Musiol, J.; Oefele, F. Material Processing with Remote Technology—Revolution or Evolution? Phys. Procedia 2010, 5, 19–33. [CrossRef] 36. Pelsue, K. Precision, Post-Objective, Two-Axis, Galvanometer Scanning. Proc. SPIE 1983, 390, 70–79. [CrossRef] 37. A Flusberg, B.; Cocker, E.D.; Piyawattanametha, W.; Jung, J.C.; Cheung, E.L.M.; Schnitzer, M.J. Fiber-optic fluorescence imaging. Nat. Methods 2005, 2, 941–950. [CrossRef] 38. Tanguy, Q. Design and Fabrication of a MEMS Scanner for OCT Imaging Endo-Microscopic Probe. Ph.D. Thesis, Université Bourgogne Franche-Comté, Besançon, France, February 2018. 39. Petersen, K.E. Silicon torsional scanning mirror. IBM J. Res. Dev. 1980, 24, 631–637. [CrossRef] 40. Tang, W.C.; Nguyen, T.C.H.; Juan, W.H.; Howe, R.T. Electrostatic-comb drive of lateral polysilicon resonator. Sens. Actuators A Phys. 1990, 21, 328–331. [CrossRef] 41. Selvakumar, A.; Najafi, K.; Juan, W.; Pang, S. Vertical comb array microactuators. In Proceedings of the IEEE Micro Electro Mechanical Systems, Amsterdam, The Netherlands, 29 January–2 February 1995; pp. 43–48. 42. Degani, O.; Socher, E.; Lipson, A.; Lejtner, T.; Setter, D.; Kaldor, S.; Nemirovsky, Y. Pull-in study of an electrostatic torsion microactuator. J. Microelectromech. Syst. 1998, 7, 373–379. [CrossRef] 43. Jung, W.; McCormick, D.T.; Zhang, J.; Wang, L.; Tien, N.C.; Chen, Z. Three-dimensional endoscopic optical coherence tomography by use of a two-axis microelectromechanical scanning mirror. Appl. Phys. Lett. 2006, 88, 163901. [CrossRef] Photonics 2021, 8, 6 24 of 25 44. Singh, J.; Teo, J.H.S.; Xu, Y.; Premachandran, C.S.; Chen, N.; Kotlanka, R.; Olivo, M.; Sheppard, C.J.R. A two axes scanning SOI MEMS micromirror for endoscopic bioimaging. J. Micromech. Microeng. 2007, 18, 025001. [CrossRef] 45. Kumar, K.; Condit, J.C.; McElroy, A.; Kemp, N.J.; Hoshino, K.; Milner, T.E.; Zhang, X. Fast 3D in vivo swept-source optical coherence tomography using a two-axis MEMS scanning micromirror. J. Opt. A Pure Appl. Opt. 2008, 10, 044013. [CrossRef] 46. Wang, D.; Fu, L.; Wang, X.; Gong, Z.; Samuelson, S.; Duan, C.; Jia, H.; Ma, J.S.; Xie, H. Endoscopic swept source optical coherence tomography based on a two-axis microelectromechanical system mirror. J. Biomed. Opt. 2013, 18, 086005. [CrossRef] 47. Aguirre, A.D.; Hertz, P.R.; Chen, Y.; Fujimoto, J.G.; Piyawattanametha, W.; Fan, L.; Wu, M.C. Two-axis MEMS scanning catheter for ultrahigh resolution three-dimensional and en face imaging. Opt. Express 2007, 15, 2445–2453. [CrossRef] 48. Gorecki, C.; Lullin, J.; Perrin, S.; Bargiel, S.; Albero, J.; Gaiffe, O.; Rutkowski, J.; Cote, J.M.; Krauter, J.; Osten, W.; et al. Micromachined phase-shifted array-type Mirau interferometer for swept-source OCT imaging: Design, microfabrication and experimental validation. Biomed. Opt. Express 2019, 10, 1111–1125. [CrossRef] 49. Lullin, J.; Bargiel, S.; Lemoal, P.; Perrin, S.; Albero, J.; Passilly, N.; Froehly, L.; Lardet-Vieudrin, F.; Gorecki, C. An electrostatic vertical microscanner for phase modulating array-type Mirau microinterferometry. J. Micromech. Microeng. 2015, 25, 115013. [CrossRef] 50. Miyajima, H. Development of a MEMS electromagnetic optical scanner for a commercial laser scanning microscope. J. Micro/ Nanolithograph. MEMS MOEMS 2004, 3, 348. [CrossRef] 51. Asada, N.; Takeuchi, M.; Vaganov, V.; Belov, N.; Hout, S.I.; Sluchak, I. Silicon micro-optical scanner. Sens. Actuators A Phys. 2000, 83, 284–290. [CrossRef] 52. Judy, J.W.; Muller, R.S. Magnetically actuated, addressable microstructures. J. Microelectromech. Syst. 1997, 6, 249–256. [CrossRef] 53. Miyajima, H.; Asaoka, N.; Arima, M.; Minamoto, Y.; Murakami, K.; Tokuda, K.; Matsumoto, K. A Durable, Shock Resistant Electromagnetic Optical Scanner with Polyimide-Based Hinges. J. Microelectromech. Syst. 2001, 10, 418–424. [CrossRef] 54. Kim, S.; Lee, C.; Kim, J.Y.; Lim, G.; Kim, J.; Kim, C. A 2-axis Polydimethylsiloxane (PDMS) based electromagnetic MEMS scanning mirror for optical coherence tomography. Proc. SPIE 2016, 9698, 969812. 55. Kim, K.H.; Park, B.H.; Maguluri, G.N.; Lee, T.W.; Rogomentich, F.J.; Bancu, M.G.; Bouma, B.E.; De Boer, J.F.; Bernstein, J.J. Two-axis magnetically-driven MEMS scanning catheter for endoscopic high-speed optical coherence tomography. Opt. Express 2007, 15, 18130–18140. [CrossRef] [PubMed] 56. Watanabe, Y.; Abe, Y.; Iwamatsu, S.; Kobayashi, S.; Takahashi, Y.; Sato, T. Electromagnetically Driven Two-Axis Optical Beam Steering MEMS Mirror and Its Dependence of Actuation on Magnetic Field. Electron. Commun. Jpn. 2011, 94, 24–31. [CrossRef] 57. Potekhina, A.; Wang, C. Review of Electrothermal Actuators and Applications. Actuators 2019, 8, 69. [CrossRef] 58. Tanguy, Q.A.; Bargiel, S.; Xie, H.; Passilly, N.; Barthès, M.; Gaiffe, O.; Rutkowski, J.; Lutz, P.; Gorecki, C. Design and Fabrication of a 2-Axis Electrothermal MEMS Micro-Scanner for Optical Coherence Tomography. Micromachines 2017, 8, 146. [CrossRef] 59. Sun, J.; Guo, S.; Wu, L.; Liu, L.; Choe, S.-W.; Sorg, B.S.; Xie, H. 3D In Vivo optical coherence tomography based on a low-voltage, large-scan-range 2D MEMS mirror. Opt. Express 2010, 18, 12065–12075. [CrossRef] 60. Wu, L.; Xie, H. A large vertical displacement electrothermal bimorph microactuator with very small lateral shift. Sens. Actuators A Phys. 2008, A 145–146, 371–379. [CrossRef] 61. Struk, P.; Bargiel, S.; Tanguy, Q.A.A.; Ramirez, F.E.G.; Passilly, N.; Lutz, P.; Gaiffe, O.; Xie, H.; Gorecki, C. Swept-source optical coherence tomography microsystem with an integrated Mirau interferometer and electrothermal micro-scanner. Opt. Lett. 2018, 43, 4847–4850. [CrossRef] [PubMed] 62. Struk, P.; Bargiel, S.; Froehly, L.; Baranski, M.; Passilly, N.; Albero, J.; Gorecki, C. Swept source optical coherence tomography endomicroscope based on vertically integrated Mirau micro interferometer: Concept and technology. IEEE Sens. J. 2015, 15, 7061–7070. [CrossRef] 63. Albero, J.; Perrin, S.; Bargiel, S.; Passilly, N.; Baranski, M.; Gauthier-Manuel, L.; Bernard, F.; Lullin, J.; Froehly, L.; Krauter, J.; et al. Dense arrays of millimeter-sized glass lenses fabricated at wafer-level. Opt. Express 2015, 23, 11702–11712. [CrossRef] [PubMed] 64. Tanguy, Q.A.A.; Gaiffe, O.; Passilly, N.; Cote, J.-M.; Cabodevila, G.; Bargiel, S.; Lutz, P.; Xie, H.; Gorecki, C. Real-time Lissajous imaging with a low-voltage 2-axis MEMS scanner based on electrothermal actuation. Opt. Express 2020, 28, 8512–8527. [CrossRef] [PubMed] 65. García-Ramírez, F.E.; Bargiel, S.; Gaiffe, O.; Tanguy, Q.A.; Struk, P.; Cote, J.M.; Passilly, N.; Lutz, P.; Gorecki, C.; Xie, H. Characterization of an integrated MOEMS scanning probe towards real-time Lissajous-based SS-OCT imaging for endoscopic applications. In Proceedings of the OSA Biophotonics Congress—Biomedical Optics, Fort Lauderdale, FL, USA, 20 April 2020. 66. Luo, S.; Wang, D.; Tang, J.; Zhou, L.; Duan, C.; Wang, N.; Liu, H.; Zhu, Y.; Li, G.; Zhao, H.; et al. Circumferential-scanning endoscopic optical coherence tomography probe based on a circular array of six 2-axis MEMS mirrors. Biomed. Opt. Express 2018, 9, 2104–2114. [CrossRef] 67. Wu, L.; Xie, H. A millimeter-tunable-range microlens for endoscopic biomedical imaging applications. IEEE J. Quantum Electron. 2010, 46, 1237–1244. [CrossRef] 68. Liu, L.; Wang, E.; Zhang, X.; Liang, W.; Li, X.; Xie, H. MEMS-based 3D confocal scanning microendoscope using MEMS scanners for both lateral and axial scan. Sens. Actuators A 2014, 215, 89–95. [CrossRef] 69. Zhou, L.; Yu, X.; Feng, P.X.-L.; Li, J.; Xie, H. A MEMS lens scanner based on serpentine electrothermal bimorph actuators for large axial tuning. Opt. Express 2020, 28, 23439. [CrossRef] Photonics 2021, 8, 6 25 of 25 70. Pengwang, E.; Rabenorosoa, M.; Rakotondrabe, N. Andreff, “Scanning Micromirror Platform Based on MEMS Technology for Medical Application”. Micromachines 2016, 7, 24. [CrossRef] 71. Tsai, J.C.; Lu, L.C.; Hsu, W.C.; Sun, C.W.; Wu, M.C. Linearization of a two-axis MEMS scanner driven by vertical comb-drive actuators. J. Micromech. Microeng. 2008, 18, 015015. [CrossRef] 72. Hsu, S.; Klose, T.; Drabe, C.; Schenk, H. Fabrication and characterization of a dynamically flat high resolution micro-scanner. J. Opt. A Pure Appl. Opt. 2008, 10, 044005. [CrossRef] 73. Shaeffer, D.K. MEMS inertial sensors: A tutorial overview. IEEE Commun. Mag. 2013, 51, 100–109. [CrossRef]

Journal

PhotonicsMultidisciplinary Digital Publishing Institute

Published: Dec 30, 2020

There are no references for this article.