1. Introduction

Optical imaging techniques have become indispensable for biomedical research and clinical applications. Recent advances have enabled both structural and functional imaging of biological tissues [1]. Optical microscopy, an icon of science and technology for its history and versatility, has been one of the most imperative and powerful imaging tools. Numerous optical imaging techniques have been developed, such as confocal microscopy [2,3], near infrared fluorescence imaging [4,5], multiphoton microscopy [6], nanoscopy [7]. Moreover, by combining other progress in illumination, detection, as well as the usage of genetically encoded fluorescent proteins such as green fluorescent protein (GFP) and its spectral and functional variants [8,9], microscopy imaging techniques have demonstrated subcellular resolution and molecular specificity, providing detailed information on structures and functional dynamics of biological systems.

Owing to the bulky size, bench-top microscopes encounter challenges for in vivo applications. Miniature optical imaging systems have been proposed and actively pursued for the past few decades to overcome the limitations. Various miniscopes have been invented and proved suitable for real-time in vivo imaging. Recent progress in optical fiber technologies further promotes the development of all-fiber-optic endomicroscopy [10,11], which can be more cost-effective and compact. The flexibility of optical fibers allows for broader applications in diagnostics and neuroscience research.

Additionally, volumetric imaging has been a longstanding goal for obtaining more comprehensive information at different depths and better perception of relative structural orientations. It is well known that the transverse resolution deteriorates quickly when the image plane is outside the confocal region [12]. Depth scanning at multiple focal planes is thus required to achieve high-resolution volumetric images. For a bench-top system, this can be realized (and has been demonstrated) by mechanically moving the optics to scan the focused light beam or by directly moving the object [13]. However, depth-scanning in miniature systems is more challenging.

Generally, two types of depth scanning mechanisms have been developed for miniature optical imaging systems. The first type is by physically scanning the lens [14,15], adjusting the separation between optical elements [16,17], or shifting the entire miniature imaging device along the optical axis on a translation stage [18]. These systems may require complicated structural design and suffer from insufficient accuracy or limited scanning range governed by the miniature mechanical actuator. The second type is by optically changing the focal length of the imaging optics. This can be realized by employing focus tunable microlens [19], whose shape or refractive index can be controlled, leading to variable focus positions. It is noted that depth scanning or depth selection can also be achieved by utilizing the coherence or temporal properties of light such as using low coherence gating in OCT [20,21] and temporal focusing in multiphoton microscopy [22]. But the lateral resolution would be compromised during this type of depth scanning/selection unless the geometric beam focus is co-scanned along with the coherence gate or temporal focusing. It is also noted that some computational methods (such as deconvolution and synthetic aperture) have been developed to mitigate out-of-focus image blurring and partially restore the lateral resolution over the imaging depth [23,24]. These topics are beyond the scope of this review.

Thanks to the emergence of unconventional materials that can be used to fabricate lenses on the microscale [19], recently a variety of tunable microlenses have been demonstrated, which can be divided into three main categories: fluid-based microlens, deformable elastomeric microlenses, and liquid-crystal-based microlenses. Different novel actuation mechanisms have been utilized to tune the focal length of microlenses, such as electrically-driven mechanism and mechanically-driven mechanism. There are commercial products in all three categories, among which electrowetting-based liquid lens [25] and liquid-crystal-based lens [26] have been widely used due to the convenient electrical actuation, relative fast response, compact structure, and good stability. Other solutions have also been reported for focus tuning, such as acoustic-modulation-based acousto-optical lens [27,28] and phase-modulation-based lensless endoscope [29,30].

This paper is devoted to reviewing the state-of-art depth-scanning technologies suitable for miniature optical imaging systems. In Section 2, the fundamentals of optical imaging systems are briefly discussed. Then, we review two main depth-scanning approaches: 1) mechanical depth scanning which is often complicated to implement in practice (Section 3), and 2) depth scanning by tunable microlens which is more suitable for miniature optical imaging devices (Section 4). Here, we mainly focus on the second type, covering the underlying physics, actuation mechanisms, and performance of the tunable microlenses. Finally, we briefly introduce the applications of depth scanning in clinical and neuroscience research in Section 5 and discuss the advantages limitations of different technologies and prospects in Section 6.

2. Fundamentals of the optical imaging system

For an optical imaging system, the transverse resolution (i.e., the diameter of the ${e^{ - 2}}$ beam intensity profile) is given by $2{w_0} = \lambda /({\pi \cdot NA} )$ based on Gaussian optics where $\lambda $ is the imaging wavelength and NA is the effective numerical aperture of the imaging optics. The depth of focus, which is longitudinal range over which the lateral imaging beam size is acceptable (i.e., no greater than 140% of the transverse resolution), is given by the confocal parameter $b = 2\pi w_0^2/\lambda $. Typically, a high-NA objective is preferred for achieving a high transverse resolution, but it results in a small depth of focus. Once the imaging depth exceeds the confocal region, the loss of transverse resolution becomes obvious as shown in Fig. 1 and features outside the confocal region start appearing blurred. To reduce this out-of-focus blurring and obtain high-resolution volumetric images, focus scanning along imaging depth at multiple planes can be the most effective way.

 

Fig. 1. Illustration of a Gaussian beam with an objective lens of a low (left) and high (right) numerical aperture (NA). b: the confocal parameter.

Download Full Size | PDF

It is highly desirable to perform depth scanning with a device of simple operations, high stability, a compact size, low cost, high repeatability and negligible hysteresis. Other important factors such as the depth-scanning range, scanning speed, scanning resolution, precision, and working distance also need to be considered depending on specific applications. Generally, for tissue micromophology imaging, scanning resolution of tens of micrometers can be acceptable, and the scanning range is expected to reach a few millimeters. While for cellular and subcellular microstructural imaging, a much finer scanning resolution (at a few-micrometer level or finer) is required to match the imaging resolution. In this case, the scanning range can be shorter (e.g., on a sub-millimeter scale) that matches the imaging depth or imaging penetration depth of the high-resolution imaging technology. Table 1 lists several categorical applications, providing the corresponding scanning resolution and depth-scanning range.

Table 1. Scanning range and resolution for different applications

View Table | View all tables in this article

For systems involving focus tunable microlenses, spherical aberration and image distortion are the major issues that would degrade image quality. Transmission spectrum and optical power rating of the materials would determine the applications of the microlenses. Other considerations when implementing depth scanning in miniature optical imaging devices include the challenges associated with the fabrication and packaging of tunable microlenses. In the following sections, we will discuss the details about each method.

3. Mechanical depth scanning by physically moving the optics

The conventional depth scanning method in a bench-top microscope is based on micromechanical scanners. The combination of galvanometric lateral-scanning mirrors with a piezoelectric (PZT) focusing device attached to the microscope objective has been demonstrated in 3D two-photon (2P) microscopy [14,15]. To achieve depth scanning in a miniature device like an endomicroscope, compact scanners are required to move parts of or even the entire device. Mechanical depth scanning by moving the distal-end micro-lens has been demonstrated. For example, a fiber-optic 3D confocal endomicroscope was designed and constructed by Liu et al. with a 2D Microelectromechanical systems (MEMS) scanning micromirror for both lateral and 1D MEMS for microlens axial scan (Fig. 2(a)) [35]. The use of a MEMS not only facilitates imaging probe miniaturization, but also provides continuous beam scan with high stability, uniform coupling and multidimensional scan capability at a small footprint. The reported MEMS-based 3D scan engine achieved a lateral scan of 180 µm × 180 µm with 1.0 µm lateral resolution and a depth scan of over 400 µm with 7.0 µm axial resolution. Electromagnetic direct linear drives (Fig. 2(b)) have also been shown suited for translating the micro-optics in a miniature imaging probe owing to their scalability, small dimensions, fast speed and high tunability [17,36,37]. These actuators can either move the focusing lens continuously or step-wise. Another innovative adjustable-focus microendoscope, presented by Zou et al., was based on two freeform elements according to the improved Alvarez principle [38]. When these two elements were given a slight lateral displacement relative to each other, an optical power can be generated for the light passing through the whole configuration (Fig. 2(c)). By using two slender piezoelectric (PZT) benders to drive the elements, dynamic tuning of optical powers from about 135 to about 205 diopters was realized. The resolution reached about 30 lp/mm without obvious distortion and blurring in the obtained images.

 

Fig. 2. Mechanical depth scanning in miniature imaging devices. (a) 3D model of the MEMS-based endoscopic probe design. (Adapted from Ref. [35] with permission of Elsevier). (b) Schematic of an electromagnetic direct linear drive (Adapted from Ref. [36] with permission of Springer). (c) Schematic of a focus scanner involving two freeform elements based on the Alvarez principle (left) and the proposed endomicroscope (right) (Adapted from Ref. [38] with permission of Optica). (d) Schematic and photo of the varifocal multiphoton rigid needle probe (Adapted from Ref. [16] with permission of Wiley). (e) Schematic (left) and photo (right) of an SMA-based depth scanner used with an endomicroscope (Adapted from Ref. [18] with permission of Optica).

Download Full Size | PDF

Instead of moving optical elements at the distal end of an imaging device, the scanning mechanism can be implemented at the proximal end. Li et al. recently reported a handhold rigid 2P imaging probe [16], which included the whole scanning mechanism in a compact scanning box (3D), and relied on a 15-cm-long compound GRIN objective to deliver light and collect signal. The depth scanning was achieved by using a PZT stage to mechanically translate the input focus at the proximal end of a compound GRIN objective, while the lateral scanning was performed by a MEMS scanner (Fig. 2(d)). The probe could scan the focus (depth) over 200 µm (in a less than 10 µm interval) while maintaining an excellent 2P resolution (0.833 µm × 6.11 µm, lateral × axial) when equipped with a high-NA micro-objective. This depth scanning method does not involve the distal end optics thus a needle probe of a small diameter is possible. The small diameter (1.75 mm) has been demonstrated, allowing for in vivo imaging of internal organs.

In addition, depth scanning can be performed by shifting the entire miniature imaging probe. Wu et al. first presented [39] and later Li et al. improved a depth scanner [18], which utilized shape memory alloy (SMA) wires to move the entire endomicroscope on the moving carriage of a guide rail (Fig. 2(e)). SMA retains a memory of its original shape; however it suffers hysteresis and is susceptible to temperature perturbation. By introducing a feedback loop to actively stabilize the contraction or relaxation of the SMA wire, the SMA-based depth scanner achieved up to a 490-µm travel range (within a few seconds) and a submicron positioning accuracy.

One advantage of focus tuning by using direct mechanical scanning is the large achievable scanning range. But such systems involve moving parts and are relatively complex. The depth-scanning speed is usually constrained by the inertia of the moving elements. Good repeatability can also be challenging, which would require active location encoders or a feedback control loop. Some key parameters of the above-mentioned mechanical depth scanning systems are listed in Table 2.

Table 2. Parameters of the mechanical depth scanning system

View Table | View all tables in this article

4. Optical depth scanning by tuning the optical power of the optics

To overcome the limitations of mechanical focus scanning, and build even more compact imaging devices, tunable optics has attracted growing attention and enjoyed broader adoption. The focus tunability can be achieved by generating a controlled change in the intrinsic property of the tunable optical component itself, including shape deformation (e.g., in fluid-based microlenses and deformable elastomeric microlenses) or motionless variation of the refractive index of the lens material (e.g., in liquid-crystal-based microlens).

4.1 Fluid-based microlenses

Over the past decade, a rich portfolio of materials have been developed and various effects have been investigated, such as the change in surface tension of a liquid, the swelling of a distensible membrane, or the deformation of a soft polymer [40,41], leading to a diverse group of tunable micro-optics. Most of the deformable microlenses involve fluids, among which liquid microlenses employ liquids and the in-between interfaces for varying optical functionality and hydraulic microlenses utilize liquids enclosed in cavities bounded by distensible membranes, while hydrodynamic microlenses rely on liquid interfaces generated by liquid flow [19].

4.1.1 Liquid microlenses

Most liquid tunable microlenses are designed and fabricated with a chamber that stores the liquid optical material. By modulating the shape or the refractive index of the liquid, the optical power of the microlens is tuned. According to the driving mechanism, liquid microlenses can be mainly categorized into electrical and mechanical ones.

For electrically driven microlenses, electrowetting has become one of the most widely used tools for manipulating liquids by surface tensions. This effect was first reported in 1993 by Berge, who artificially introduced a thin dielectric layer between a conducting liquid droplet and a conductive substrate and observed a decrease of the liquid contact angle upon application of a voltage [42]. A variation of this phenomenon, now known as electrowetting-on-dielectrics (EWOD), allows the use of electrowetting with nearly any liquid, provided it is conductive or, at least, polar [43]. EWOD operates on a capacitive setting (Fig. 3), in which a liquid droplet is deposited on an insulating dielectric layer over a conducting substrate. The contact angle ${\theta _V}$ can be altered by a bias applied between the droplet and the substrate as $cos{\theta _V} = cos{\theta _0} + \alpha {V^2}$, where θ₀ is the contact angle in the absence of any bias, $\alpha $ is a parameter related to the materials and the thickness of the dielectric layer, and V is the applied voltage [25].

 

Fig. 3. For a generic electrowetting-on-dielectrics (EWOD) setup, a liquid droplet is separated from a conducting substrate by a thin insulating dielectric layer. A voltage applied between liquid and substrate results in a bias-dependent contact angle ${\theta _V}$. A change in ${\theta _V}$ leads to a change in droplet curvature and thus tuning of the focal length.

Download Full Size | PDF

One of the first demonstrations of liquid lenses utilizing electrowetting was reported by Gorman et al [44]. In their system, a hexadecanethiol drop rested on a gold surface held at 0 V underwater and formed a hydrophobic monolayer under the drop. When the voltage of the gold surface was set at −1.7 V, the liquid drop displayed the ability of light focusing as a lens. Since this remarkable work, more EWOD-based tunable microlenses have been developed. Berge and Peseux reported an optical tunable lens using a water droplet (Fig. 4(a)) [45]. Such a liquid lens contained two immiscible liquids, and a transparent electrode. The applied voltage controlled the gradients of wettability of the surface, which deformed the shape of the drop, and thus changed the focal length. Later, Krupenkin et al. demonstrated a tunable liquid microlens capable of adjusting its focal length and lateral in-plane position [46]. This was achieved by varying the voltage applied to a set of electrodes positioned underneath of the dielectric substrate. Kuiper and Hendriks presented another electrowetting-based tunable lens by changing the curvature of the meniscus between two immiscible liquids (Fig. 4(b)) [47]. They also demonstrated an achromatic liquid lens by tuning the optical properties of the liquids, rather than using costly achromatization methods. The focus tuning speed of such lens was around several milliseconds, and the high repeatability (over a million times) was proved via durability test. More recently, a variety of novel optical configurations have been developed for extensive use. Li and Jiang fabricated a flexible, electrowetting-based liquid microlens using a soft polymer polydimethylsiloxane (PDMS) [48]. The entire lens structure could be smoothly wrapped onto a curvilinear surface. Microlens arrays [49] and electrowetting lens-prism elements can also be fabricated [50]. These designs can be easily integrated in miniature optical devices.

 

Fig. 4. (a) Cross-sectional schematic and photograph of a variable liquid microlens (with a 5-mm aperture diameter and a 12-mm overall diameter) (Adapted from Ref. [45] with permission of Springer). (b) Cross-sectional schematic of an electrowetting-based tunable lens in a cylindrical glass housing (Adapted from Ref. [47] with permission of AIP). (c) The basic design schematics and actuation mechanism of a liquid microlens using a pinned liquid-liquid interface. The shape of the microlens varies with the local temperature. (Adapted from Ref. [55] with permission of Nature Portfolio). (d) Schematics and photos of an IR light-actuated tunable microlens with 18 hydrogel microposts in divergent and convergent statuses. (Adapted from Ref. [56] with permission of AIP).

Download Full Size | PDF

An alternative to electrowetting for actuating tunable liquid microlens is dielectrophoresis. Different from electrowetting which is a surface phenomenon related to conductive liquid, dielectrophoresis is a bulk effect and requires using two nonconducting liquids with different dielectric constants [51]. Cheng and Yeh reported a packaged liquid lens driven by the dielectrophoretic effect [52]. A large focal length tuning range of 12 to 34 mm was realized by a 3-mm microlens with an applied voltage of 200 V. Both the rise time (when the lens was actuated from the rest state to 200 V) and the fall time (when the applied voltage was switched off) were around a few hundred milliseconds. One critical challenge was its hysteresis and spherical aberrations caused by the non-uniform deformation. Additionally, other electrically-driven mechanisms have been proved feasible to activate liquid microlenses [53,54].

Another group of strategies for actuation is to use mechanical drive. Liquid microlenses activated by mechanical drive usually involve a circular chamber with a flexible membrane which deforms with pressure. Others are formed through air-liquid or liquid-liquid interfaces which can be adjusted by pressure.

The pressure on the microlens liquid can be controlled by using hydrogels, a type of highly hydrophilic polymer which can absorb a large volume of water, and thereby undergo significant expansion. Dong et al. demonstrated a tunable liquid lens by integrating a stimuli-responsive hydrogel (temperature, pH) into the system [55]. Served as the container for a liquid droplet, the hydrogel sensed the presence of the stimuli and started varying the droplet shape, thus tuning the focal length (Fig. 4(c)). For a millimeter scale liquid lens, the focal length can be tuned from $- \infty $ to $+ \infty $ (divergent and convergent), with response times of ten to a few tens of seconds. Zeng and Jiang reported liquid tunable microlenses actuated by light-responsive hydrogel [56]. Multiple hydrogel micropost structures were photopatterned around a lens aperture, whose volumetric change was controlled by infrared light and determined the curvature of the liquid-liquid interface at lens aperture (Fig. 4(d)). The focal length of the microlens can be tuned from −17.4 mm to +8.0 mm in seconds under IR irradiation. In addition, the hydrogel-driven tunable liquid microlenses have been demonstrated with a cylindrical shape [57], extended to microlens arrays [58], or fabricated onto flexible polymer substrates for large FOV [59]. The unique feature of hydrogel-driven tunable microlenses is that they respond self-adaptively to the environmental parameters, thus eliminating the need for complicated external control systems and even power supplies. Another advantage is that the fabrication process, mostly by in situ liquid-phase photopolymerization [60,61], facilitates the integration with electronic and opto-electronic systems. Therefore, it is promising to develop functionally complex yet relatively simple lab-on-a-chip applications. However, the response time of the microlenses is in the second regime. Further miniaturization of the hydrogel structures [60] should improve the response speed.

4.1.2 Hydraulic microlenses

In contrast to the above liquid microlenses, hydraulic microlenses utilize fluid-filled membrane to accomplish focus tuning. This type of lens usually works with microfluidic systems where optical fluids are fully enclosed. One of the most frequently used configurations contains a microfluidic chamber covered with a distensible transparent membrane. The membrane can expand and generate a convex profile when the pressure on the fluid in the chamber increases, while a concave profile forms when the pressure decreases. The optical quality of the membrane material, along with the magnitude of the deforming force and the elasticity of the membranes, collectively influence the performance of these tunable microlenses.

Werber and Zappe presented an elastomer membrane-based tunable microlens activated by a microfluidic system (Fig. 5(a)) [62]. The membrane was stable and flexible over a wide temperature range (-50$^{\circ}{C}$ to +200$^{\circ}{C}$), and no hysteresis was observed. Both plano-convex and plano-concave microlenses with several-hundred-micrometer diameters were demonstrated. Zhang and co-authors also presented a fluidic tunable lens, consisting of a PDMS fluidic chamber covered by a thin PDMS membrane [66]. By injecting fluid into the lens chamber through a syringe pump, the focal length of a 2-cm aperture lens can be tuned from infinity to 41 mm via membrane deformation. To reduce the aberration of hydraulic microlenses caused by an imperfect membrane profile, Reichelt and Zappe extended the doublet concept to designing a multichamber hydraulically tunable lens [67]. By combining suitable optical liquids with appropriate radii of the interfaces, the chromatic and spherical aberrations could possibly be corrected. Other attempts have been reported to achieve achromatic performance with only one type of liquid (Fig. 5(b)), which reduced the complexity of the lens systems. And their novel fabrication process also offered a significant cost saving. [63].

 

Fig. 5. (a) Cross-sectional diagram of a membrane-based microfluidic microlens, showing its implementation as plano-convex and plano-concave lenses. The drawing is to scale for a microlens with 400-µm diameter (Adapted from Ref. [62] with permission of Optica). (b) Schematic of a 5-mm diameter hybrid lens consisting of a fluidic lens chamber, a liquid inlet and an outlet (Adapted from Ref. [63] with permission of Optica). (c) Schematic and photograph of a three-layer microlens chip fabricated using soft lithography (Adapted from Ref. [64] with permission of RSC). (d) Detailed view of a fully integrated thermo-pneumatic microlens and photograph of the back of the bonded Si lens chip with microfluidic structures (Adapted from Ref. [65] with permission of Nature Portfolio).

Download Full Size | PDF

One challenge with hydraulically tunable microlenses is the precision control of the tuning pressure. The traditional approach of using macroscopic external pressure controllers defeats the goal of miniaturization. Recently, more efforts have been put into on-chip tuning, which offers advantages of compactness, precision, and the ability to control arrays of optical elements individually. Fei et al. employed the multilayer soft-lithography technology to fabricate optofluidic compound microlenses, which consist of multiple layers of fluid chambers that serve as lens components (Fig. 5(c)) [64]. The refractive index contrast of each lens was controlled by filling the chambers with a specific medium. The chip had multiple independent pneumatic valves that could be digitally controlled for injecting tiny amounts of varied-index liquids into the chambers with high accuracy and consistency. This enabled quick and precise focal length tuning of the compound microlens from centimeter to sub-millimeter. Zhang et al. developed a liquid tunable microlens based on polyacrylate membranes integrated with a compact on-chip thermo-pneumatic actuator fabricated with full-wafer processing (Fig. 5(d)) [65]. In their design, the optical liquid (i.e., silicone oil) was pushed or pulled into the cavity via an extended microfluidic channel without any pumps, valves or other mechanical means. This system allowed focal length tuning from infinity to 4 mm with good repeatability and minimal hysteresis. The maximum power consumption for the heaters was 300 mW while the heating/cooling of the lens took tens of seconds. The wafer-level fabrication implied potentially a lower cost due to an improved yield.

4.1.3 Hydrodynamic microlenses

A completely different liquid microlens configuration has gained increasing recognition for achieving focus tuning within the plane of the substrate, rather than normal to it, which has been the prevailing approach in all the abovementioned examples. Instead of observing and activating light from the top, hydrodynamic microlenses have accomplished light manipulation from the sides of microfluidic channels. By controlling single or multiple fluidic flows, these lenses can guide or control light in micro- or even nano-scale, and serve as efficient tools for coupling light into waveguides [41].

Variable designs have been proposed, usually containing a curved microfluidic channel filled by two fluids with different refractive indices. A curved profile forms in the boundary region between the two liquids, and the curvature of the boundary is a function of the fluid flow rate. This profile leads to controllable focusing of light beams within the traversal plane. Mao and co-authors fabricated a hydrodynamically tunable optofluidic microlens with two co-injected miscible fluids, CaCl₂ solution (n = 1.445), and deionized H₂O (n = 1.335), which flowed through a 90-degree curve in a microchannel (Fig. 6(a)) [68]. Due to the centrifugal effect, the CaCl₂ solution bowed outwards into water, forming a cylindrical microlens. Such an in-plane tunable microlens can be easily governed by changing the flow rate and it can be easily integrated with lab-on-a-chip devices to create fully integrated optical application. Another type of dynamically reconfigurable microfluidic lens was reported by Seow et al., where three laminar flows were injected into an expansion chamber (Fig. 6(b)) [69]. Different interface curvatures were obtained (meniscus, plano-convex and biconvex) by adjusting these three flow rates. Optical aberration was also eliminated by minimizing the diffusive broadening at the interfaces. Rosenauer and Vellekoop presented the first microfluidic planar device to form a transversal multi-convex microlens by using two parallel waveguides filled with the fluid combination [70]. Other novel designs of hydrodynamic microlenses have also been developed, allowing for simple fabrication, easy operation, rapid tuning, and large tunability [71,72]. The lab-on-a-chip technology has reduced hydrodynamic microlenses to sub-millimeter dimensions, making them compatible with existing microfluidic devices and attractive for on-chip optical applications such as on-chip detection and particle manipulation.

 

Fig. 6. (a) Illustration of the working mechanism of a hydrodynamically tunable optofluidic cylindrical microlens, where CaCl₂ solution bows outward into water due to the centrifugal effect induced in the curve (Adapted from Ref. [68] with permission of RSC). (b) Schematic of the formation of a liquid microlenses with different lens curvatures in an expansion chamber by tuning the flow rates of three liquid streams (Adapted from Ref. [69] with permission of AIP).

Download Full Size | PDF

4.2 Deformable elastomeric microlenses

Deformable elastomeric microlenses are fabricated from soft, elastic materials whose shapes change under the application of stress. Generally, these tunable microlenses are solid components made from soft elastomers like PDMS or transparent silicone. Molding techniques are often involved to prepare the lens with high-quality optical surfaces and a desired shape, including spherical or optimized aspherical shapes [73]. By applying refined actuation, the curvature as well as some nonsymmetric aspects of the lens profile can be controlled, allowing aberration correction and high optical quality. Compared with other tunable microlenses, elastomeric lenses exhibit good long-time stability and shock resistance. Also, these lenses can be readily mass-produced, accommodating a wide range of applications.

Early demonstrations of deformable microlenses employed pneumatic actuation. Hoshino and Shimoyama fabricated a pneumatically actuated microlens array on an elastic PDMS film [77]. Campbell et al. presented another pneumatically driven deformable lenses based on transparent flexible elastomer membranes (Fig. 7(a)) [74]. These liquid-free membrane lenses were demonstrated with continuous focus tuning over 4 diopters at 500 Hz. Subsequent implementations have been developed by employing mechanical actuation. For example, Choi et al. designed a variable-focus lens system inspired by the biomechanics of the human eye [75]. In their design, the gel-type PDMS lens was encircled and stretched by an SMA actuator (Fig. 7(b)). Although this design was insensitive to external disturbance, the tunability of focal length was limited (16.8 to 18.0 mm) even for a lens with 8-mm diameter due to the small displacement of the SMA actuator (0 to 1.2 mm with a sub-millimeter interval). Later, Liebetraut et Al. proposed an elastomeric lens that could be deformed asymmetrically [78]. By applying stress azimuthally, the aberrations, particularly astigmatism, could be controlled.

 

Fig. 7. (a) Schematics of a pneumatically driven microlens with a flexible membrane (Adapted from Ref. [74] with permission of AIP). (b) CAD drawing of an assembled deformable lens. The PDMS lens is encircled and stretched by the load arms which are driven by the SMA actuator (Adapted from Ref. [75] with permission of Optica). (c) Cross-sectional schematic of a fully assembled tunable lens with an integrated actuator (Adapted from Ref. [76] with permission of Optica).

Download Full Size | PDF

The abovementioned actuators that required radial actuation usually tend to be large and bulky, and thus may not satisfy the demands of a miniature optical imaging device. One promising alternative is polymer-based electromechanical transduction, also known as dielectric elastomer (DE) actuation [79]. Within the family of electroactive polymers [80], DEs have proved to be a top choice of smart materials for novel soft actuators capable of high strains, high energy density, high efficiency, fast response, high resilience and light weight. Recent demonstrations of compliant, electrically conductive electrodes in combination with DE allowed the use of an electrical signal to alter the shape of a liquid lens without external actuation. For example, Carpi et al. reported on a DE-based electrically tunable lens, where an annular elastomeric actuator worked as an artificial muscle to deform the lens [81]. Shian et al. presented another elastomer-liquid lens system with a transparent DE actuator located directly in the optical path (Fig. 7(c)) [76], enhancing the simplicity and compactness of tunable lenses.

4.3 Liquid-crystal-based microlenses

The term liquid crystal (LC) applies to a range of materials which exhibit unique properties ranging between those of conventional isotropic liquids and solid-state crystals. These materials consist of small linear molecules, whose orientation may be defined and varied easily by an applied electric field or mechanical surface features [82], thus affecting both the distribution of the refractive indices of LC and the polarization of a propagating optical field.

For fluid-based and deformable elastomeric microlenses, the change in the interface shape/curvature often requires a high voltage (e.g., dielectrophoresis), lack resistance to environmental factors (e.g., temperature) and has restrictions in the focus tunability and the lens aperture. The commercial electrowetting lenses have a clear aperture of at least 1.6 mm. Smaller sizes are currently unavailable [84]. By comparison, focus scanning by changing the refractive index of a LC-based microlens has none of these limitations. LC-based microlenses work by introducing a lens-like phase difference (or wavefront change) to the incident light, which causes the waves to converge or diverge. Since the first demonstration by Bricot in 1977 [85], LC lenses with an electrically tunable focal length have stimulated significant research interest, due to their low power consumption, simple fabrication, compact structure, and good stability [83]. They have been expected to have wide applications for 3D display, imaging, zooming, and optical tweezers, etc. [26].

4.3.1 Nematic LC microlenses

Most commonly-used LC-based lenses are made of Nematic LCs. Composed of rod-like molecules, these materials exhibit optical and dielectric anisotropies. Generally, LCs are confined in an LC cell, sandwiched between two substrates coated with electrodes (e.g., indium tin oxide, ITO) and surface alignment layers (e.g., polyimide). When the molecules are properly aligned in the LC cell under the influence of an external electrical field, their long axes are approximately parallel to each other [86]. Light polarized along this averaged alignment direction (termed director) is defined as e-ray (of an extraordinary refractive index n_e), while that polarized perpendicular to the director is o-ray (of an ordinary refractive index n_o). When the incident light polarization is at an angle (θ) with respect to the LC director, the effective refractive index n_eff for the light can be expressed as ${n_{eff}} = \frac{{{n_e}{n_o}}}{{\sqrt {{{({{n_e}sin\theta } )}^2} + {{({{n_o}cos\theta } )}^2}} }}$ [83]. When a voltage higher than the threshold voltage [87] is applied to the electrodes, the angle θ will change, leading to n_eff varying from n_e to n_o. Figure 8 illustrates the operation principle of a nematic LC lens. By adjusting the distribution of molecular orientations or the refractive index of the LC, a lens-like phase difference in the material can be created. Therefore, the curvature of the incident wavefront can be electrically controlled, resulting in a tunable focal length.

 

Fig. 8. Operation principle of a nematic LC lens: (a) LC molecular and polarization-dependent refractive index. (b) Illustration of the structure for a homogeneous LC cell, where the incident light polarization is along the LC director and the effective refractive index equals ${n_e}$. (c) LC directors are reoriented along the electric field and the incident light polarization is perpendicular to the LC director with an effective refractive index of ${n_o}$. (d) LC directors are reoriented by a $\theta $ angle and the effective refractive index for the incident light is ${n_{eff}}(\theta )$. (e) Positive LC lens. (f) Negative LC lens. (Adapted from Ref. [83] with permission of MDPI).

Download Full Size | PDF

Generally, LC lenses are designed with two different structures: either containing an inhomogeneous cell gap, or using a homogeneous cell gap but controlled by applying inhomogeneous electric fields [26]. The first LC lens was invented with inhomogeneous cell gap [88], followed by many other demonstrations of this kind. Choi et al. designed a nematic LC-based microlens with wide focus tuning capability (Fig. 9(a)) [89]. The large birefringence of the LC provided a 10-mm focus tuning range with a 200-µm lens diameter, under a less than 1.5 V applied voltage. Later, Dai et al. reported a simple printing technique to fabricate negative-positive tunable microlens array using UV adhesive and LC [90]. With careful control of printing, the dimension of microlens could be further decreased to satisfy the requirement for a miniature device or an integrated system. LC microlens with a homogeneous cell gap can be even more compact and enjoy lower light scattering due to better LC alignment, while the design of electrodes can be more complicated for generating non-uniform electric fields and lens-like phase variation. Ren and Wu designed a tunable LC lens consisting of a spherical glass shell and a homogeneous LC cell that required easy fabrication [91]. The inner surface of the glass shell and the bottom surface of the LC cell are coated with ITO electrodes while the LC layer is sandwiched between the ITO electrodes (Fig. 9(b)). With this design, a centrosymmetric gradient refractive index within the LC layer was obtained. Another novel LC lens device was demonstrated with multiple-ring electrodes in unequal widths [92]. The number and widths of the ring electrodes were pre-designed and optimized, enabling a smoother refractive index distribution than the conventional LC lens with only one hole-patterned electrode.

 

Fig. 9. (a) Schematic of an LC mircolens structure (inhomogeneous cell gap) and two operating states (Adapted from Ref. [89] with permission of Elsevier). (b) Cross section of an LC lens with a homogeneous LC cell (Adapted from Ref. [91] with permission of Optica). (c) Side view of a BPLC microlens structure with a curved top electrode and planar bottom electrode (Adapted from Ref. [93] with permission of Optica). (d) Cross section of a multi-electrode BPLC lens (Adapted from Ref. [94] with permission of Optica).

Download Full Size | PDF

4.3.2 Blue phase LC microlenses

Although nematic LCs play a dominant role in current LC-based tunable microlenses, they suffer from two technical challenges: polarization dependency and slow response (usually hundreds of milliseconds or even several seconds). Recently, blue phase liquid crystal (BPLC) has emerged as an alternative owing to the fast response (sub-milliseconds) and polarization insensitivity [83]. Blue phase is a type of LC phase existing over a narrow temperature range (only a few Kelvins) between a chiral nematic phase and an isotropic liquid phase [95]. BPLC shows a self-assembled structure consisting of double-twist cylinders, which are arranged in cubic lattice with lattice periods of several hundred nanometers. Due to the short coherent length, BPLC exhibits sub-millisecond response time [96]. The symmetric 3D cubic structure also makes it optically isotropic, i.e., independent of light polarization. Moreover, BPLC does not require any molecular alignment layer, which simplifies the fabrication process.

The early BPLC-based tunable microlens was reported by Lin et al. in 2010 [97]. In their work, a thin cell of polymer-stabilized BPLC was confined between two glass substrates, with a hole-patterned top electrode and a continuous bottom electrode. A polarization-independent optical phase modulation of $\pi $ radian at 150 V_rms was measured, with 3 ms response time. Li and Wu proposed another BPLC microlens structure with a curved ITO top electrode and a planar bottom electrode to improve the phase profile [93]. By optimizing the shape of the electrode, the spherical aberration was largely suppressed (Fig. 9(c)). Lee et al. designed a multi-electrode structure (Fig. 9(d)) [94]. The voltage of each electrode was controlled individually to generate ideal parabolic phase profiles. This structure was easily fabricated by the existing lithography process, and the operating voltage could be reduced by using a material of a higher dielectric constant for the dielectric layer. For a 4-mm diameter lens, the focal length could be tuned from infinity of 180 mm with a maximum 10 V drive voltage. Some key parameters about different types of tunable microlenses for depth scanning in miniature optical imaging systems are listed in Table 3.

Table 3. Parameters of different types of tunable microlenses

View Table | View all tables in this article

4.4 Other optical methods

Apart from the abovementioned microlenses, ultrasonic lenses have been employed for fast depth scanning, though their sizes can be larger. Ultrasonic lenses use the acoustic wave to induce refractive index change and achieve high-speed focus tuning (100 kHz – 1 MHz, faster than most scanners). Tunable acoustic gradient index of refraction lens has been demonstrated as a fast varifocal element. The optical power of the lens can be changed continuously, enabled rapid selection and modification of the effective focal length on a time scale of 1 µs and shorter [28]. Adjustable ultrasonic lens technology has also been integrated into 2P microscopy, showing the ability of switching the depth of field at a kilohertz rate [98] and volumetric 2P imaging at tens of hertz [99]. In addition, ultrasonic lenses can produce complex beam shapes and control the wavefront, making them suitable for spatial control of multiphoton processes and sample processing (such as polymerization). However, ultrasonic lenses with sufficient compactness suitable for miniature imaging devices are currently not available.

Holographic wavefront control [100,101] provides another method for aberration-free depth scanning, where spatial light modulators (SLM) are involved to generate desired wavefront and correct aberrations [100]. High-resolution, multiregional 3D fluorescence imaging with SLM-equipped microscopy has been demonstrated in both transparent and scattering media [102] and in vivo [103]. Although SLM offers a decent update rate, each update is only for a specific focal distance. For continuous axial scanning (e.g., over a range of 200 focal locations), the scan rate can be quite slow (e.g., 1 Hz with a 200 Hz SLM). Recently, digital micromirror devices (DMD) have been developed to modulate the phase and amplitude of light [104]. Fast 3D scanning (${\sim} $ kHz) has been realized in DMD-based 2P microscopy with sub-micrometer step size [105]. Similar to SLM, DMD based (focus) scanners generally have a pretty large size (on the order of a few cm², particularly when taking into account the drive/control substrate). Remote focusing has proved to be another solution for depth scanning [106,107], which employs a pair of identical objective lenses and telecentric tube lenses to relay the focal movement from one lens to the other. Thus, miniaturization can be challenging. Another limitation for remote focusing is its power handling capability, which blocked the applications with high peak laser intensity [108]. In the field of endoscopy, lensless endoscopes also have the ability of focus tuning. These devices can control the spatial phase of the light emitted from a multimode or multi-core fiber, without the need for any elements on the tip of the endoscope fiber, allowing for extremely compact probe design [30,109,110].

5. Applications

The ability to achieve high-quality in vivo volumetric imaging of biological tissues is highly desirable for disease detection. Tunable lens-based systems are attractive for their compact size, light weight, easy fabrication process and low cost. Olsovsky et al. designed a tunable microlens based reflectance confocal endomicroscope which allows rapid depth scanning of oral epithelium (Fig. 10(a)) [34]. In vivo imaging of stomatology patients was also conducted with the endomicroscope, illustrating its ability to resolve nuclear changes associated with inflammation and dysplasia, and its potential to guide tissue biopsy, improve diagnostic yield, or monitor treatment efficacy. Another demonstration of tunable lens-based endoscope was for imaging the auditory system [113]. Recently, a swept source OCT system was developed for versatile visualization of the eye by integrating an electrically tunable lens in the OCT sample arm [32]. Adaptive operational states of the lens enabled precise and dynamic control of the optical beam focus. Such technology can be used to perform quantitative ocular biometry and serve as a volumetric imaging platform with broad ophthalmic applications.

 

Fig. 10. (a) Photograph of the handheld reflectance confocal endomicroscope (Adapted from Ref. [34] with permission of SPIE). (b) The fabricated laparoscopic camera sealed for placement into the abdominal cavity (Adapted from Ref. [111] with permission of SPIE). (c) Photograph of a mouse wearing the miniature 2P microscope (Adapted from Ref. [112] with permission of Nature Portfolio).

Download Full Size | PDF

Recent innovations in motionless depth scanning technologies also contribute to minimally invasive surgeries, allowing much smaller incisions and resulting in less pain, faster recovery, less wound infection, and minimal scarring [114]. Tunable microlens techniques have been preferred for its ability to switch between concave and convex lenses, its large tuning range, scalability, mechanical, and thermal stability. A miniaturized laparoscopic zoom camera has been reported to improve visualization during minimally invasive surgery (Fig. 10(b)) [111]. The key element was a fluid-filled, elastomer-membrane bioinspired fluidic lens, which performed a convertible curvature change with an applied force or pressure. Another fluidic microlens has been integrated in a fluorescence-based surgical guidance setup, and its performance was demonstrated in a laparoscopic surgical environment [115].

In addition, miniature imaging devices capable of depth scanning have been a long technological pursue in neuroscience research. The past decade has witnessed impressive progress on the development of head-mountable miniscopes, which enable real-time functional neuroimaging in freely-behaving rodents [116–118]. The head-mountable designs require ultracompact and ultralight optical elements. Therefore, microlenses-based depth scanning technologies have been highly favored in this field. Focus-tunable endomicroscopes have been developed by employing electrically tunable LC lens [119] and fluid-containing polymer lens [120]. The first fiber coupled 2P miniscope that was capable of axial-scanning has been reported by Ozbay et al. for imaging in a freely-behaving mouse [121], where a commercially available electrowetting microlens was integrated into the system, enabling not only axial shifting (a 180-µm axial scanning) but also tilting of the focal plane at a 1.3 to 2.5-Hz frame acquisition rate. Recently, Zong et al. presented a miniature 2P microscope equipped with a fast electrically tunable lens (Fig. 10(c)) [112]. Its advanced version [122] achieved multiplane brain imaging of over a volume of 500 × 500 × 240 µm³ with a frame rate of 15 Hz in freely behaving mice, expanding the applicability of miniaturized 2P microscopy for research on learning and memory, social interaction, disease progression, etc.

6. Conclusions and future perspectives

Fueled by the fast and innovative advances in bio-photonics, materials science, micro-mechanics and other fields, depth scanning technologies suitable for miniature optical imaging systems have progressed rapidly over the past decade. Mechanical depth scanning techniques typically involve MEMS mirrors, SMA, or PZT components to build axial scanners. These approaches feature a fast scanning speed and easy operation, while they may introduce mechanical and optical complexities, potentially constraining device miniaturization. One way to overcome the challenges associated with mechanical depth scanning is by utilizing the coherence or temporal properties of light for enabling volumetric imaging, such as coherence gating for OCT or temporal focusing in TF microscopy. But it is noted that high imaging quality requires synchronized focus scanning along with coherence gating or temporal focusing. Other solutions exploit holographic and lensless methods. Among all the approaches, tunable microlenses have attracted an increasing interest for their reliability and added functionalities.

Currently, there are three primary categories of tunable micro-optics. The first group is the liquid-based microlenses, which rely on the change in surface tension of a liquid, the swelling of liquid cavities bounded by distensible membranes, or the liquid interfaces generated by microfluidic flow. Manipulating the liquid’s shape or index distribution enables optical power adjustment, which can be accomplished through an electrical or mechanical drive, or a microfluidic system. Such liquid tunable microlenses may suffer from optical aberrations and can be susceptible to the vibration of external environmental factors like temperature. The second type is deformable elastomeric microlenses that are fabricated from soft elastomers. Generally, such microlenses involve only solid-body components which deform by applying stress in a controlled manner. The flexibility of the soft, elastic materials enables the adjustment of the lens surface or refractive index profile to eliminate aberrations; however, these lenses may need complex mechanical actuators, whose size can be several centimeters, posing additional difficulties for miniaturization. The third kind of adaptive microlenses are made of LCs, in which the orientation of LC molecules can be varied to affect both the refractive index and the polarization characteristics of the element. LC-based microlenses exhibit multiple advantages including low power consumption, large continuous tunability, simple fabrication, compact structure, and robust stability. In addition, their optical profile can be made aspherical to correct various aberrations. Efforts have also been made to reduce the polarization dependence, rendering LC microlens a promising candidate for various applications.

Ongoing developments for clinical applications aim to incorporate depth scanning into an imaging device to enable 3D or adaptive imaging (i.e., imaging at selected depths). Moreover, tunable micro-optics is particularly attractive for neuroscience research, where in vivo brain imaging can be carried out in a freely-behaving rodents (or rats) with a head-mounted miniature imaging device. Incorporating ultracompact tunable microlenses into the head-mountable devices makes it possible to perform volumetric brain imaging potentially of neurons in different cortical layers or brain regions in freely-behaving animals, which would open avenues for functional neural circuit studies.

So far, a variety of tunable microlenses have been made commercially available, such as electrowetting-based microlenses [123], hydraulic microlenses [112], elastomeric microlenses [122], LC-based microlenses [124], etc. The costs of these commercial products are highly competitive (around a few hundred to a few thousand US dollars per unit), making them a competitive choice when compared with depth scanners involving mechanically moving optics. However, many challenges associated tunable micro-optics remain to be solved, including the required precision of fabrication, elimination of aberration at different depths, increase of an optical aperture and FOV while maintain a small physical footprint. An increasing research interest and activity are expected in the coming years to increase the tuning speed, range, enhance scanning stability, reduce the size, and improve the integratability of the tunable microlens with miniaturized optical devices. These advancements will bring innovative miniature imaging devices of more functionalities to broad user groups for basic research and clinical translation.