A novel microlens design with tunable double-focus is presented. It is fabricated by adding only one SU-8 photolithography step to the well-developed liquid-filled microlens fabrication process. The thickness of this layer determines the thickness difference between the central and peripheral region of the membrane, the deformation of which is used to define the surface profile of the microlens. The stepped thickness variation is finally manifested as the difference in deformation contour at two different regions of the membrane when subjected to uniform applied pressure, thereby causing two focal lengths to appear. Experimental and simulation results are presented, from which the tunability of the focal lengths of the double-focus microlens is demonstrated to be effective over a wide range through combining the structural design with pressure control. The successful demonstration of this unconventional microlens design concept will potentially extend the application of liquid-filled microlens technology.
© 2009 Optical Society of America
The microlens is an important optical component in the microoptics area and has found wide applications such as wavefront sensing, optical communication, miniaturized imaging systems, laser field, etc . A single microlens can be monolithically used for imaging purposes, such as those in a camera embedded in mobile phones  and endoscopes . In other applications, a number of microlenses are combined together and arranged in a two-dimensional array. For example, in some visible and infrared (IR) detectors, a microlens array is used to focus the incident beam onto the sensing pixels to significantly improve the fill factor of the sensing array as well as the detection sensitivity , while in the case of a Shack-Hartmann type wavefront sensor, a microlens array is used to dissect the incoming wavefront into a number of segments and create a focal spot for each within the assigned sub-aperture on the charge coupled device (CCD). The light wavefront information can be obtained by measuring the lateral displacement of these spots individually [5,6].
Many methods of successfully fabricating microlenses have been reported, e.g. photoresist thermal reflow , dry etching , polymer-jet-printing , ultraviolet (UV) imprinting , hot embossing  and laser fabrication . Almost all of the microlenses fabricated by these methods are in a “solid” format with the surface profile as well as the focal length being determined by the structure and processing parameters and cannot be changed during operation. However, in some imaging applications, such as for lab-on-a-chip cases, the object (such as a cell) under observation either cannot be confined to a fixed position or has a large fluctuation in surface height. Thus, in order to capture real-time information, the optical system has to be adjusted continually. Conventionally, some mechanical moving parts are used to tune the optical components, making the whole system bulky and complicated. Recently, microlenses have been proposed with tunable focal length capability for miniaturization and adaptiveness enhancing purpose. Microlenses constituted by liquid crystal  and two immiscible liquids  (electrowetting-based liquid lens) with their focal lengths controlled by an electric field have been demonstrated. The main drawbacks with these microlenses are the high cost, low optical transmittance (at least 50% light is absorbed in the case of a liquid crystal illuminated with unpolarized light), liquid evaporation and the high voltages required (for the electrowetting method).
By comparison, liquid-filled microlenses based on the combination of optics and microfluidics-namely, optofluidics technology-possess some distinct advantages. Firstly, its fabrication process is very simple and the commonly-used structural material -Polydimethylsiloxane (PDMS) - has good optical transmission properties over a wide spectral range (from near ultraviolet to near infrared) [14,15]. Secondly, the surface profile of the microlens as well as its focal length can be dynamically adjusted by changing the pressure of the liquid, which can be simply realized with a commercial pumping system or an integrated heating structure to yield a more compact configuration. Some prototypes of this type of microlens having large tunability of focal length have been successfully fabricated and demonstrated [16–24].
In this paper, we present a new design based on the standard liquid-filled microlens fabrication process described previously in the literatures. However, unlike the common microlens which has a single focus, the proposed unconventional microlens has double-focus capability: one formed by the inner portion, whilst the other is created by the peripheral region. This type of microlenses may extend the application areas of this technology, such as in read-write DVD pick-up heads (one focus for read beam, the other for write beam) and optical coherence tomography (OCT) (used for extending the depth of focus). At the same time, since the capability of focal length tunability provided by the liquid-filled microlens technology is totally preserved, the applicability and versatility of such lenses is further enhanced.
2. Design and fabrication
The design of the proposed microlens is quite similar to those demonstrated before6. A suspended PDMS membrane together with the substrate forms a closed cavity defining the microlens aperture. It is connected to an external syringe pumping system via an integrated microfluidic microchannel through which fluid can be introduced into the cavity. By changing the fluid volume as well as the cavity pressure, the membrane is forced to deflect and hence define the surface contour of the lens. In contrast with the commonly-used PDMS membrane with uniform thickness for the deformable microlens surface, we adopted a membrane structure with different thicknesses for the center and peripheral regions as shown in Fig.1. From the point of mechanics, it is obvious that the resultant deformations as well as the radii of curvature at these two portions under uniform pressure will be different. As a result, they can serve as two separate lenses with their focal lengths being determined by the corresponding surface deformation contour. At the same time, since the focal length can be continuously adjusted by varying the fluid pressure, as demonstrated in the case of single-focus microlens structure, it is expected that the distance between these two focuses as mentioned above can also be controlled in the same manner. The specific structural parameters adopted in our design are shown in Table 1.
Figure 2 shows the fabrication flow. First, one layer of SU-8 (SU-8 2025, MicroChem Corp.) with 30μm thickness is spun onto a 4-in polished silicon wafer. Through the standard photolithography process, the patterns including the cavity (it also defines the lens aperture), microchannel, inlet and outlet are transferred onto this layer as shown in Fig. 2(a). The same steps are repeated, but now the thickness of SU-8 layer is changed to 60μm. With proper alignment with the former patterns, the inner region of the microlens is finally confined (Fig. 2(b)). This SU-8 layer is used to define the thickness difference between the peripheral and inner regions of the membrane. Both of these SU-8 features are combined together as the master mold for the following membrane coating process, in which the liquid PDMS prepolymer (Sylgard 184 silicone elastomer-a base and curing agent of Dow Corning Corp -mixed in a 10:1 weight ratio) is directly spin-coated onto it as shown in Fig. 2 (c). After complete curing in a furnace at 60 □ for two hours, this PDMS membrane, with inverse structures having been transferred from the SU-8 mold, is carefully peeled from the mold substrate with the assistance of isopropyl alcohol (IPA) solution and then bonded to a transparent substrate (such as thick PDMS plate or glass slide) using an oxygen plasma activation method as given by Fig. 2(d). Finally, via holes are manually drilled at the inlet and outlet positions to allow access for liquid injection.
3. Simulation and experimental results
3.1 Simulation results
In order to estimate the relationship between membrane deformation as well as focal length and the applied pressure, a model of the microlens is established in ANSYS finite element analysis software. The simulation results for the cross-sectional contour of the deformed membrane under pressures ranging from 200Pa to 9.5kPa are as shown in Fig. 3. For comparison, the profile for the case of a microlens having a uniform thickness of 90μm under 5kPa is also presented. It is seen that besides the top deflection being increased from 0.557mm (single-focus) to 0.672mm (double-focus), as expected, there is a distinct transition point on the cross-section profile, the position of which corresponds to the steep change in membrane thickness. By inputting these surface profiles into ZEMAX, an optical simulation tool, the resultant focal lengths can be calculated. The results (Fig. 4) show that the focal length at the central region can be gradually adjusted from about 46.05mm to nearly 10.43mm when the pressure is changed from 200Pa to 9.5kPa, whereas the tunable range is from 57.34mm to 15.08mm in the peripheral region,. As a result, the difference between these two focuses can vary from 11.29mm to 4.65mm with increasing applied pressure.
We also studied the influence of the membrane design on performance by assigning two other thickness values (60μm and 45μm) to the central part of the membrane. Since membrane deflection is inversely proportional to the third order of thickness, we can easily conclude that with smaller differences in the thickness of the two regions of the microlens, the difference in deflection - as well as the resultant focal length - between the two parts will also be reduced. Figure 5 shows the corresponding simulation results in which the result of the previous design (central portion thickness of 30μm) is also provided for comparison. It is clear that for the same pressure range, the difference between the two focal lengths is changed from 4.99mm to 2.93mm and from 2mm to 1.48mm for 45μm and 60μm central thicknesses, respectively. This additional design flexibility, combined with the tunability under pressure, provides the proposed double-focus microlens with greater versatility (as it is able to cover a wide focal length tunable range) to meet the requirements of different potential applications.
3.2 Experimental results
Figure 6 shows the deformation status of the fabricated microlens under four different pressures. Their corresponding focal lengths are measured using a method similar to that described in Ref  and are listed in Table 2 together with the standard deviations between multi-measurements. Compared with the simulation results, the measured focal lengths are higher, with the differences at the peripheral region being smaller than those for thecentral part. This is mainly caused by the deviation of the actual fabricated membrane thickness from the design value. Considering the fact that there exists two SU-8 steps - one of 30μm thickness and the other of 60μm as mentioned above - in the mold structure, the corners of which correspond to the connection boundary of the membrane to substrate and the division boundary between the central and the peripheral regions of membrane, respectively. Since the desired membrane thickness in the peripheral region is about 90μm, a PDMS layer of total thickness 120μm is required to be coated onto the mold. This can be achieved through multiple spin-coating and curing steps. It is well-known that the layer fabricated by the spin-coating process tends to follow the surface profile of the substrate being coated, therefore causing membrane thickness fluctuation at the pattern location. The amplitude of fluctuation is directly affected by the relative thickness of the layer to that of the pattern structure: the larger the ratio between them, the smaller the amplitude will be. It is for this reason that the common solution for planarization as adopted in the integrated circuit industry is to increase the deposition layer thickness relative to the pattern height being coated. In our case, since the membrane thickness is determined to be 120μm, the ratio of it to the mold height of two SU-8 steps are 4 (120/30) and 1.33 (120/90), respectively. It can be estimated that the thickness fluctuates more in the central region than in the peripheral area. Figure 7 shows the crosssection of the fabricated membrane. It is seen that in the central region, the membrane thickness is about 45.2μm at the center and it gradually reduces to 33.5μm at the edge (the desired value is 30μm). In the peripheral region, the thickness value varies from 100.6μm to 91.3μm along the radial direction (the desired value is 90μm), which agrees well with the analysis above. The larger membrane thickness deviation (thicker) in the central region will definitely cause larger deflection deviation from the simulation result, namely the actual membrane deflection is much smaller than the simulation, as a result, causing longer focal length. Compared with this, the deviation in the peripheral region is a bit smaller, the same as that demonstrated in the measurement results.
Figure 8 shows a schematic of the beam transmission through the double-focus microlens and the expected and real beam images captured at three different positions along the optical path when a collimated laser beam (He-Ne laser, wavelength: 632.8nm) is incident onto the microlens under 5kPa operation pressure. At the focal plane of the central region (marked by A), the beam covering the center part (marked as 1) is converged into a spot, while the beam transmitted through the peripheral area (marked as 2) is still converging. Therefore, the light pattern observed is that of a focal spot surrounded by a light ring. After passing through this focal plane, beam 1 begins to diverge and at the same time, beam 2 continues to converge (pattern B) until its focal spot is formed at position C (pattern C). Further along the transmission path, both parts of the beam will be diverging.
From Rayleigh’s criterion, light transmitted through the microlens will be focused into an Airy speckle at the corresponding focal plane with the radius proportional to the f-number (f/D, where f and D are the focal length and aperture diameter of the microlens respectively). For the central region, f is measured to be 15.66mm and D is 2.5mm, therefore the f-number is calculated to be 15.66/2.5=6.264, while the corresponding values in the peripheral region are 19.72mm, 5mm, and 19.72/5=3.944. This indicates that the focal spot at the first focal plane is larger than that at the second focal plane, the two planes being defined by the central and peripheral region respectively. This analytical result agrees with that observed in the experiment.
In the usual case where the microlens has a single focus and has been fabricated with a similar method to that presented here, some spherical aberration is usually reported to occur. In order to quantify optical aberration, the surface profile data of the microlens must first be obtained. Since the commonly-used white-light interferometer was not available at hand, we constructed an optical setup to capture images of the cross-sectional contour (as shown in Fig. 6) from the side of the microlens. Using professional image processing software together with a particular dimension calibration procedure, the surface profile can be extracted. Through fitting the profile using fourth-order polynomials25 and then inputting the polynomial coefficients into ZEMAX, the spherical aberrations of the microlens under three different states as shown in Fig. 6 are obtained and listed in Table 3. From the result, the spherical aberration in the central region is found to be always smaller than that in the peripheral region.
This is because that the membrane is clamped at its outer circumference while the boundary condition around the inner part is more likely to be a simply-supported situation. Mechanical and optical analyses show that the resultant surface profile of a deformable membrane under simple-support condition is much closer to a spherical shape compared to the clamped case. At the same time, in order to verify the accuracy of this treatment for surface profile data extraction, we also used ZEMAX to determine the simulated focal lengths using the same surface profile data. By comparing with the measurement results as shown in Table 2, it can be seen that the difference is less than 3% for all these three pressure settings.
In order to further demonstrate the characteristics of the fabricated double-focus microlens, an optical setup is constructed in which it together with an objective lens is used to image a target onto a CCD camera. Since there are two different focal lengths in the different regions of the aperture of the microlens, a clear image of the target will not be obtained over the whole aperture. From Fig. 9, it is obvious that when the image captured by the peripheral region of the microlens is clear, that from the central region is a bit blurred (left photograph). Conversely when the central portion of the image is clear, the outer part begins to blur (right photograph).
In this paper, a microlens with double-focus implemented using the widely-reported liquid-filled microlens fabrication process is demonstrated. In contrast with the commonly-adopted design that has a membrane of uniform thickness acting as the tunable surface for the microlens, the membrane in the proposed design has a stepped thickness variation along the radial direction but still maintains an axisymmetrical structure. The membrane in the central region is made thinner than that in the peripheral region. This is realized by adding one more SU-8 photolithography step into the process flow, the thickness of this SU-8 layer being used to define the thickness difference in the resultant membrane. This difference in structural parameter is then transferred to the deformed surface profile when a uniform pressure is applied onto the membrane during device operation, thereby resulting in two different focal lengths corresponding to the two microlens regions. Through structural finite element analysis, it is seen that both focal lengths decreased with increasing applied pressure. Similarly, their difference also exhibits monotonously decreasing characteristic. The effect of the designed difference in the membrane thickness on the optical performance of the microlens is also studied. Based on the results, we conclude that the tunable range of the distance between these two focuses can be made to cover a wide range through combining structural design with pressure control. This largely enhances the application perspective for the proposed microlens. Besides simulation, some experimental results are also provided to demonstrate the double-focus characteristic and at the same time, evaluate the optical quality of the fabricated microlens such as its spherical aberration. Although there are some deviations between the experimental and simulation results, the reason for which have been analyzed, discussed and validated with observations, both trends generally agree well with each other.
Financial support by the Ministry of Education (MOE) Singapore AcRF Tier 1 funding under Grant No. R-265-000-235-112 is gratefully acknowledged.
References and links
1. N. F. Borrelli, Microoptics Technology (2nd Edition, Marcel Dekker, 2005), Chap. 1.
2. S. Kuiper and B. H. W. Hendriks, “Variable-focus liquid lens for miniature cameras,” Appl. Phys. Lett. 85, 1128–1130 (2004). [CrossRef]
3. A. Jain and H. Xie, “Endoscopic microprobe with a LVD microlens scanner for confocal imaging,” Optical MEMS and Their Applications Conference, 2006. IEEE/LEOS International Conference on. 168–169, 2006.
4. Y. Li, X. J. Yi, and J. H. Hao, “Design and fabrication of 128×128 diffractive microlens arrays on Si for PtSi FPA,” Proc. SPIE 3553, 132–137 (1998). [CrossRef]
5. G. Yoon, S. Pantanelli, and L. J. Nagy, “Large-dynamic-range Shack-Hartmann wavefront sensor for highly aberrated eyes,” J. Biomed. Opt. 11, 0305021–0305023 (2006). [CrossRef]
6. H. B. Yu, G. Y. Zhou, F. S. Chau, and F. W. Lee, “A tunable Shack-Hartmann wavefront sensor based on a liquid-filled microlens array,” J. Micromech. Microeng. 18,105017 (2008). [CrossRef]
9. H. Choo and R. S. Muller, “Addressable microlens array to improve dynamic range of Shack-Hartmann sensors,” J. Microelectromech. Syst. 15, 1555–1567 (2006). [CrossRef]
10. C. P. Lin, H. Yang, and C. K. Chao, “A new microlens array fabrication method using UV proximity printing,” J. Micromech. Microeng. 13, 748–757 (2003). [CrossRef]
11. N. S. Ong, Y. H. Koh, and Y. Q. Fu, “Microlens array produced using hot embossing process,” Microelectron. Eng. 60, 365–379 (2002). [CrossRef]
14. H. B. Yu, G. Y. Zhou, F. S. Chau, and F. W. Lee, “Optofluidic variable aperture,” Opt. Lett. 33, 548–550 (2008). [CrossRef]
15. H. B. Yu, G. Y. Zhou, F. S. Chau, and F. W. Lee, “A variable optical attenuator based on optofluidic technology,” J. Micromech. Microeng. 18,115016 (2008). [CrossRef]
17. D. Y. Zhang, N. Justis, and Y. H. Lo, “Fluidic adaptive lens of transformable lens type,” Appl. Phys. Lett. 844194–4196 (2004). [CrossRef]
18. M. l. Agarwal, R. A. Gunasekaran, P. Coane, and K. Varahramyan, “Polymer-based variable focal length microlens system,” J. Micromech. Microeng. 14, 1665–1673 (2004). [CrossRef]
19. J. Chen, W. Wang, J. Fang, and K. Varahramyan, “Variable-focusing microlens with microfluidic chip,” J. Micromech. Microeng. 14, 675–680 (2004). [CrossRef]
20. H. Yang, C. Y. Yang, and M. S. Yeh, “Miniaturized variable-focus lens fabrication using liquid filling technique,” Microsyst. Technol. 14, 1067–1072 (2007). [CrossRef]
22. D. Y. Zhang, V. Lien, Y. Berdichevsky, J. Choi, and Y. H. Lo, “Fluidic adaptive lens with high focal length tenability,” Appl. Phys. Lett. 82, 3171–3173 (2003). [CrossRef]
24. D. Y. Zhang, N. Justis, and Y. H. Lo, “Integrated fluidic adaptive zoom lens,” Opt. Lett. 15, 2855–2857 (2004). [CrossRef]
25. X. F. Zeng and H. R. Jiang, “Polydimethylsiloxane microlens arraya fabricated through liquid-phase photopolymerization and molding,” J. Microelectromech Syst. 17, 1210–1217 (2008). [CrossRef]