1. Introduction

Microsystems and nanosystems for biomedical fields, such as bio-MEMS and Lab-on-a-Chip, are areas of intensive research [1, 2]. In particular, microarrays represented by DNA chips, which are well established for bio-sensing applications such as DNA detection and SNP genotyping, are valuable tools for fundamental studies in biology and medicine. A DNA chip using micro-spots of biomolecules on a static solid support is a typical static microarray. Compared with static microarrays such as a DNA chip, dynamic microarrays using laser-trapped mobile substrates, usually consisting of microbeads coated with biomolecules, biochemicals, or cells, offer greater flexibility and potential as platforms for sensing/manipulation applications in various scientific fields as well as biomedical fields [3–7]. For assembling dense dynamic arrays, several approaches based on multi-beam optical tweezers techniques, namely the holographic optical tweezers (HOT) [8–10] and generalized phase contrast (GPC) methods [11, 12], have been demonstrated with impressive results. Unlike the time-shared scanning (TSS) approach [6], the HOT/GPC-based approaches are suitable for manipulating a large number of microbeads to form dense dynamic microbead arrays, because no dwell time is required for stable multiple optical traps [12]. However, these approaches require a spatial light modulator (SLM), which is an expensive device and needs complicated control procedures for not only optical engineers but also expert programmers. Therefore, a simple and inexpensive configuration of multi-beam optical tweezers is essential for the dissemination of dynamic microarrays to various scientific fields, especially biomedical fields.

In this paper, we present a simple method based on TSS optical tweezers with a commercially available microlens array for assembling and manipulating dense dynamic arrays that consist of more than a hundred microbeads. Three typical examples of handling the dense microbead arrays, including the dynamic and simultaneous bonding of microbeads, are demonstrated in real-time. The optical design and inexpensive hardware setup for our demonstrations are also described.

2. Optical design and developed system

In our previous paper [12], we developed a hybrid optical tweezers system consisting of two kinds of multi-beam optical tweezers techniques: the GPC method and the galvano-mirror scanning method. This hybrid system was extremely useful for assembling and manipulating dynamic arrays that consist of more than a hundred microbeads, namely 10 × 10 lattice array. However, the part of the hybrid system based on the GPC method required both an expensive SLM and a custom-made phase contrast filter (PCF), which is not easily available to the scientists in biomedical fields or to optical engineers. On the other hand, the part based on TSS with a 2-axis gimbal mirror has a simple optical configuration, and was also extremely useful for testing the new control algorithms of dynamic arrays [6], although the size of dynamic arrays was limited to ten-odd microbeads because of the scanning speed and the dwell time necessary for stable trapping.

Thus, with an aim to promote the use and application of dense dynamical arrays in various scientific fields, we developed a dual TSS-based optical tweezers system with a simple and inexpensive optical conformation using a single commercially available microlens array and two 2-axis scanning mirrors. Although this dual TSS system without an SLM offers limited flexibility as compared to our previous hybrid system with an SLM, it has not only a simple and inexpensive optical conformation but also suitable properties for handling dense dynamical arrays as compared to the traditional scanning optical tweezers [13]; it can arrange numerous microbeads into the desired geometrical patterns, independent of the scanning rate as demonstrating in Section 3.2; it can also assemble and control the several tens of micro-objects at once as demonstrating in Section 3.3.

Figure 1(a) shows the schematic of the optical and control system configurations, and Fig. 1(b) shows a detailed layout of the microlens array L_A and other optical components (namely, 2-axis gimbal mirror GM₁, lens L₁, relay lens L_R, and the microscope’s objective lens L_O) in the TSS-based optical tweezers with the microlens array. This optical structure is linked to the inverted microscope (Olympus, IX70) via its epi-fluorescence port. The single laser source is a continuous wave (cw) Nd:YAG laser (Spectron SL902T, 1064 nm, TEM₀₀), and its expanded beam, upon passing through a half-wave plate (λ/2), is split into two beams (p- and s-polarized beams), the diameters of which are roughly 3 mm, by a polarized beam splitter (PBS₁). One beam (colored orange in Fig. 1(a)) irradiates the microlens array (Newport, MALS14) installed in a rotatable folder (θ_Z) on a Z-axis linear stage (δ_Z), where the microlenses are arranged in a grid pattern with the pitch g_A = 500 ± 0.25 μm. The focal length of each plano-convex microlens is 46.7 mm [14]. After passing through the microlens array, this beam, which has a diameter of 3 mm, is split into small beams, referred to as beamlets, and focused into an array of 6 × 6 foci at the first prefocusing plane (PFP₁). The beamlets pass intermediate optical components (namely, lens L₁ (f₁ = 100 mm), 2-axis gimbal mirror GM₁ (Newport, FSM-300) and relay lens L_R (f_R = 170 mm)) and are directed into the objective lens, L_O (Olympus, UPlanSApo × 100, 1.40 NA, IR, f_O = 1.8 mm), via the epi-fluorescence port.

 

Fig. 1 (a) Optical and control system configurations of a dual TSS optical tweezers using a microlens array. (b) Detailed layout of the microlens array and other optical components for the orange beam in figure (a).

Download Full Size | PDF

For the optimal setup of the TSS optical tweezers with the microlens array, we designed the layout of the intermediate optical components between L_A and L_O in Fig. 1(b), based on an approach similar to that used in previous papers [15–17]. (i) The role of lens L₁ is to ensure that an array of foci in PFP₁, where the foci are formed in steps of g_A, is imaged into L_R’s focal plane in object space (namely, the second prefocussing plane PFP₂) and that the subsequent beamlets passing through L_R are collimated to re-form the array of foci in the imaging plane of L_O, where the foci are formed in steps of g_O. (ii) The role of the relay lens, L_R, is to ensure that the image of the beamlets’ size, h_G, at the GM₁ position is invariantly transferred to L_O’s entrance aperture and that the transmission of the beamlets’ trapping power is, consequently, independent of the 2D-tilt angles of GM₁. (iii) The beamlets’ size, h_G, at the GM₁ position can be independent of the lens separation, d_A1, when the distance between L₁ and GM₁ is chosen to be equal to L₁’s focal length, f₁ [16]. Hence, in the case of the layout in Fig. 1(b), where we postulate that this layout performs roles (i) and (ii), the lens separation d_A1 can be derived from the thin-lens formulas as

On the other hand, for practical utilization of the dynamical arrays, the initial loading of many microbeads without undesired stacking of the microbeads along the beam axis is important but difficult and tiresome for operators. Therefore, in order to load/unload the desired microbeads into/from the dynamic arrays generated by the multi-beam optical tweezers with the microlens array, we employed a set of 3D optical tweezers using the other beam (colored red in Fig. 1(a)). 3D optical tweezers with an electrical focus-tunable lens L_Z (Optotune, EL-10-30-NIR-LD) and galvano mirrors GM₂ were constructed based on the same design as in our previous paper [15]. The focus of this 3D optical tweezers, namely the trapping position in XYZ-coordinates, can be strictly specified by digital-analog (DA) control signals and can also be controlled interactively by a computer’s 3-button mouse. In the developed system in Fig. 1(a), multi-beam optical tweezers with a microlens array and 3D optical tweezers with an electrical focus-tunable lens are coaxially combined by PBS₂ in front of the relay lens, L_R. The power ratio of these two kinds of TSS optical tweezers can be adjusted by the half-wave plate λ/2 to optimally load/unload microbeads into/from the arrays.

3. Demonstrations

3.1 Dynamic microbead array in prime lattice

Here, we demonstrate the interactive rotation of the prime lattice array so as to verify the basic performance of dynamic arrays generated with a microlens array. The sample is polystyrene microbeads (Polysciences Inc, Polybead®, 2μm) dispersed in water and sandwiched between two cover glasses for microscope observation. The laser power at the entrance aperture of an objective lens was 270 mW; that is, the laser power of each beamlet was 7.5 mW. The 4 × 4 part of the prime lattice array in the microscope’s imaging plane is illustrated in Fig. 2(a), where g_O is the grid size, θ_Z is the rotation angle of the array, and point ‘O’ is the rotation center. θ_Z can be adjusted to the desired value by the manual-rotating folder of the microlens. Figures 2(b)–2(d) are snapshots captured with a CCD camera. They present the results of interactive rotation of the prime lattice array. As shown in Fig. 2(b), thirty microbeads were loaded, one by one, into some part of the prime array (a 5 × 6 array) with grid size g_O ≈9 μm, where the 4 × 4 part corresponding to that in Fig. 1(a) is indicated by the short-dashed square. As shown in Figs. 2(c)–2(d), this 5 × 6 microbead array could be rotated easily by turning the rotatable lens folder, maintaining the stable trap of one microbead at each grid. Note that since the whole of the prime lattice array could provide trap points for more than 6 × 6 grids, the microbeads trapped at the grids on the outside of the CCD camera’s view appeared in its images when the lattice array was rotated around the point ‘O’; they are indicated by the white arrows in Figs. 2(c)–2(d).

 

Fig. 2 (a) Schematic of the prime lattice array in objective imaging plane. (b–d) Rotation of the prime array about the z-axis by turning the angle θ_z of a microlens folder: (b) θ_z = 0°, (c) θ_z = 30°, and (d) θ_z = 45°.

Download Full Size | PDF

3.2 Dense dynamic microbead arrays

Here we demonstrate the simple generation of dense arrays by means of the TSS technique with a microlens array. The sample is the same one mentioned in Section 3.1, and the dwell time of scanning at each trap point was 10 msec. The laser power of the multi-beam tweezers with the microlens array was 1400 mW; that is, the laser power of each beamlet was 39 mW, and that of the 3D tweezers was 50 mW. Figure 3(a) illustrates the grids of the prime array and the dense array generated by their time-shared scanning, where δ_x and δ_y are the scanning intervals. Here, the dense array, expressed by yellow circles, is nine times denser than the prime array, expressed by red circles; that is, g_O = 3δ_x = 3δ_y. The scanning path for nine trap points is also drawn by a red solid line with arrows. Figures 3(b)–3(e) present the results of the TSS generation of dense optical trap arrays and the subsequent loadings of microbeads into them using the 3D optical tweezers, where the areas of the 3 × 3 part of the prime lattice corresponding to that in Fig. 3(a) are overlapped by the pastel-yellow squares. In Fig. 3(b), that is, in the case of no scanning, the dynamic microbead array is formed in the prime lattice, which is similar to that in Fig. 2(b). Figure 3(c) shows the dynamic array, which is four times denser than the prime array in Fig. 3(b). The microbeads, which numbered more than a hundred, namely 10 × 10, were arranged tidily in the grids without their undesired stacking along the beam axis; in this case, a laser power of 9.7 mW may have irradiated each grid. Figure 3(d) shows the dynamic array, which is nine times denser than the prime array; in this case, a laser power of 4.3 mW may have been irradiated at each grid. Although all of the 15 × 18 grids spread around the whole area of observation were not loaded tidily with microbeads because of the deficient or non-uniform trapping force at the grids, roughly 200 microbeads could be arranged in these dense grids. These results are, to the best of our knowledge, one of the largest microbead arrays assembled by optical tweezers, including holographic and GPC tweezers.

 

Fig. 3 (a) Schematic of the prime trap points (shown as red circles) generated by a microlens array and the dense trap points (shown as yellow circles) generated by the time-shared scanning. (b–d) Dynamic microbead arrays: (b) in the prime lattice, (c) in the four times denser grids, and (d) in the nine times denser grids. (e) Dynamic pattern of microbeads forming the letters ‘AIST’ in the nine times denser grids.

Download Full Size | PDF

In these demonstrations, using the computer’s 3-button mouse, we employed the 3D optical tweezers of the dual system to trap a microbead and release it at the desired location, while controlling the status (open and closed) of the shutter. In this way, moreover, we could also selectively fill the array of trapping grids with microbeads. Figure 3(e) shows a dynamic pattern of microbeads, forming the alphabetic letters ‘AIST’, where the microbeads are trapped in the same lattice array as shown in Fig. 3(d). Note that the microbead indicated by a white arrow was mis-located into the grid. It was never removed to any other grids because of its strong adhesion on the coverglass’s surface. However, such adhesion arising from the mis-location can be practically suppressed by a surfactant or treatment of coverglass. On the other hand, in the case of a system based on the traditional time-averaged scanning method [13], nuisances such as undesired stacking and overloading could often happen for the patterning of more than a hundred microbeads, and its formable geometrical patterns were also strongly dependent on the scanning rate and laser power.

3.3 Dynamic bonding of multiple microbeads

Here we demonstrate the dynamic and simultaneous bonding of multiple microbeads by means of the TSS technique with a microlens array. The sample and the dwell time at each trap point are the same mentioned in Section 3.2. The laser power at the entrance aperture of the objective lens was 900 mW; that is, the laser power of each beamlet was 25 mW. Figure 4(a) illustrates the original grids, which are represented by small red circles, and their own TSS grids generated by the time-shared scanning, where yellow circles denote the TSS grids with a microbead and white circles denote those without microbeads. One set of the TSS grids is composed of four trapping points around an original grid, with scanning intervals δ_x and δ_y. In the demonstration, since the TSS array is four times denser than the prime lattice, a laser power of 6.3 mW may have irradiated one trapping point.

 

Fig. 4 Dynamic bonding of microbeads in the time-shared lattice and interactive rotation control of the bonded microbeads. (a) Illustration of original lattice points without scanning (shown as small red circles) and trapped beads (show as large yellow circles) by the time-shared scanning tweezers, the density of which is four times denser than that of the original lattice. (b–d) (Visualization 1) Video frame sequence of the dynamic bonding and subsequent interactive rotation. The accompanying movie is in real-time, not accelerated.

Download Full Size | PDF

Figures 4(b)–4(d) (Visualization 1) are video frame sequences showing the processes of dynamic bonding of four microbeads and subsequent rotation of the bonded microbeads, where the positions of original grids corresponding to those in Fig. 4(a) are indicated by the 3 × 3 small red circles. First, as shown in Fig. 4(b), roughly 100 microbeads were loaded into the TSS grids using the 3D optical tweezers by the same manner as mentioned in Section 3.2, where the area for the 3 × 3 part of the original grids corresponding to that in Fig. 4(a) is overlapped by a pastel-pink square. In order to avoid the undesired collisions of the trapped microbeads during the subsequent bonding process, the initial scanning intervals δ_x and δ_y were specified to satisfy the condition 2δ_x = 2δ_y ≤ g_O. Note that not every set of four trapping points generated from an original grid was filled with four microbeads so as to clearly demonstrate their dynamical bonding in the next process, although the filling with four microbeads for all sets of the TSS grids, as in Fig. 3(c), could be easily performed. Secondly, in Fig. 4(c), while gradually reducing the intervals δ_x and δ_y, the microbeads trapped at each set of the TSS grids were approaching each other to form their clusters. In the first process, we can choose the number of microbeads loaded into each set of four TSS grids, ranging from two to four, and selectively load them, one by one, into the desired grids. Therefore, the dynamical bonding in the second process can generate eleven kinds of clusters with different numbers and orientations; two, three, and four microbeads make ₄C₂ ( = 6), ₄C₁ ( = 4), and one variety of orientation, respectively, where expression _nC_m represents the combination of m elements selected from n elements. Note that we can find three kinds of clusters in the 3 × 3 part of the original grids. Finally, in Fig. 4(d), all of the clusters could be simultaneously rotated around their own original grid. As shown in the accompanying movie, the bonding of microbeads that are forming clusters by means of the TSS technique is not permanent; therefore, we can easily reform the arrays of clusters into the initial arrays of microbeads, if it is necessary, while increasing the intervals δ_x and δ_y.

4. Conclusion

We have developed a simple, TSS-based dual optical tweezers system with a commercially available microlens array and 2D scanning mirrors. This system, without an SLM, enables us to easily assemble and manipulate dense dynamic microbead arrays that consist of more than a hundred microbeads. We have also demonstrated the feasibility of simultaneous bonding of microbeads to make the clusters more than several tens at once and the subsequent controlled rotation of these clusters. The simple and inexpensive optical configuration of this multi-beam optical tweezers is, we believe, a valuable approach for dynamic microarrays in various scientific fields, especially in biomedical fields [7]. Furthermore, it is expected that batch processing of multiple single cells as well as microbeads, that is, manipulating many micro-objects at one time, will provide a powerful and versatile tool for automating tiresome tasks such as the mechanical stretching of living cells [18], and will open up new possibilities in single-cell biophysics.