Highly flexible elastomer microfluidic chip for single cell manipulation

Abstract

New materials and fabrication technologies have significantly boosted the development of lab-on-a-chip technologies and functionalities. In this work, we developed a highly flexible elastomer microfluidic chip with a microchannel with a minimum width of ∼5 μm manufactured by imprinting onto an SU-8 template. We found that the deformation induced in the microstructures by manual stretching of the chip is higher than that for the chip itself, which we attribute to the stress concentration of microstructures. Here, we demonstrate that the elastomer enables the manipulation of single cells, such as dynamic trapping–releasing operations, by simply stretching and releasing the elastomer chip.

I. INTRODUCTION

The emergence of new materials and microfabrication technology has significantly expanded the applications and functionalities of lab-on-a-chip technologies.^1–9 The recent rapid development of flexible electronics and organ-on-a-chip has again attracted attention for elastomer materials, which have proved their successful functionalities in many fields, such as bioseparation¹⁰ and flow manipulations.¹¹ In addition to the widely used elastic material Sylgard 184 PDMS, other silicone rubbers such as Ecoflex and Dragon Skin have been chosen to fabricate microfluidic devices due to their high elasticity.¹² High elasticity is critical for the performance of flexible materials in applications as one can produce microfluidic valves,¹³ separation devices,¹⁴ pressure sensors,¹⁵ electronic skin,¹⁶ wearable devices,¹⁷ optics,¹⁸ chemical engineering,¹⁹ and so on.^20–22

The development of soft materials leads to more functionalities and possibilities for microfluidic chips. As the most widely used material, PDMS is important in stretchable devices.^23–27 However, compared to the maximal strain of human tissue (skin) of up to 80%,²⁸ the typical tensile strain of a PDMS chip in the elastic deformation stage is less than 50%,²⁹ which can lead to constraints for mimicking tissue mechanics, especially, for organ-on-a-chip studies. Hydrogels are another type of elastomer with high biocompatibility and elasticity that are commonly used in biological studies and are excellent for 3D-cultured cells,^30,31 but the opacity of some kinds of highly elastic hydrogels can reduce the device transparency.³² In addition, polyurethane, Ecoflex, and other compounds have been used to fabricate stretchable devices;²⁸ however, they have low biocompatibility, transmittance, limited elongation, incompatibility with plasma bonding, etc.³³

An appropriate increase of elasticity possibly helps us to enhance the performance of the devices³⁴ like organ-on-a-chip technology,³⁵ enabling tunable microstructures, thus mimicking the mechanical activity in the physiology of biological tissues as in vitro models.^36–42 Hereby, we report highly flexible microfluidic devices, including the use of materials and fabrication processes. Our results show that microstructures can be prototyped down to less than 5 μm in our current manufactured structures. We found that the tensile strain of the microstructures is apparently higher than that of the whole chip, which we attribute to stress concentration. Numerical results show that the local stress near microstructures is highly enhanced, inducing a larger strain than that in bulk materials. This can be an advantage since a small strain in a chip can induce large deformation of the microstructures and can thus be used for heart model, vessel model, skin expansion in vitro, and so on. As such, we successfully demonstrated single cell capture and release by simply stretching the elastomer chip. We developed a function-switchable microfluidic chip and manipulated liquids/objects at the microscale on our highly elastic chip, which can be useful for organ-on-a-chip applications.

II. MATERIAL AND METHOD

Here, we used two types of silicon rubbers (Silastic® BioMedical Grade ETR Elastomers Q7-4720 and Q7-4750) obtained from Dow Corning Corporation, but neither of them was found to be individually suitable for chip manufacturing because the former rubber (Q7-4720) is too sticky to be delaminated from the mold, while the latter (Q7-4750) has a high elastic modulus that resists to deform. According to our experiments, mixing these two types of rubbers with a mass ratio of 1:2 (Q7-4720:Q7-4750) enables these materials to be used for fabricating highly elastic microfluidic chips, guaranteeing the process of fine microstructures and good delamination from the mold. Unlike liquid rubber such as Sylgard 184 PDMS, both components of Q7-series materials are dough-like, which requires a new fabrication procedure including preparations and fabrications. Figure 1(a) illustrates the fabrication process for our highly elastic chip.

FIG. 1.

The fabrication procedure of highly elastic microfluidic chip and features of the technology and materials. (a) The scheme of the process to fabricate high elastic chips. (b) and (c) The micropillar array structures fabricated by imprinting technology before and after stretching. (d) The smallest single channel fabricated in our current experiments. (e) The optical transmission spectrum for the Q7-mixture and PDMS.

Two types of silicon rubber (Q7-4720 and Q7-4750) were separately prepared by mixing prepolymer and curing agent at a mass ratio of 1:1. Then, we mixed Q7-4720 with two parts of Q7-4750 with a mass ratio of 1:2 for a mixed rubber. We placed the two components of the silicon rubbers into a laminator and fold the squeezed sheets for the second round of lamination. The rolling–folding cycle was repeated over 10 times until the mixture surface turned smooth, which also increased the transparency of the bulk material. Air bubbles in mixtures need to be avoided when operating the rolling–folding process.

After the mixing process, we stamped the Q7-mixed rubber onto a template fabricated using standard SU-8 photolithography to manufacture microstructures on an elastomer chip. Before stamping, the template surface was hydrophobized with 1H, 1H, 2H, 2H-perfluorooctyltrichlorosilane (FOTS) for better demolding. More details about the fabrication of the template can be found in S1 in the supplementary material.

The thickness of the highly elastic chip can be adjusted by setting the distance between the two rollers in the laminator, which we set to 1 mm for the following experiments. A Q7-mixture sheet of the same size as the template was well aligned with the template for stamping. The slow rolling of the press rollers can help us to avoid trapping air bubbles between the Q7-mixture rubber and template. We cured the imprinted rubber together with the template by placing them both onto a 110 °C hot plate for 10 min before peeling off the elastomer from the template as demolding uncured material can induce damage to the imprinted microstructure.

Finally, we prepared another piece of Q7 elastomer sheet with a flat surface and punched holes for chip bonding. After ultrasonic cleaning (acetone, ethanol, and water for 5 min) and a drying process (heating to 100 °C for 1 h), we treated the sample with oxygen plasma (30 W for 40 s) (HARRICK PLASMA CLEANER PDC-002) for a permanent bonding, which is the same operational process as that of PDMS-glass chip fabrications. This bonding is strong enough to resist the pressure of over 5 bar, at least performed in channels in our experiments.

Here, we successfully fabricated standard microstructures such as micro-square pillars with a side length of 30 μm on the surface as shown in Fig. 1(b). We manually stretched the chip [Fig. 1(c)] and found that strain in the micropillars is ∼200%, which is apparently higher than that of the whole chip (∼100%). More microstructures were investigated that presented obviously different strains on the chips, as shown in Fig. S2 in the supplementary material. We demonstrate the finest structures we fabricated in our experiments, a microchannel with a width of 5 μm [Fig. 1(d)], although a smaller structure is still possible. We did not find any collapsed channels after bonding with these fine channels.

We characterized the tensile stress–strain curve for our cured rubber Q7-mixture using a servo control tensile tester (ZHONGNUO INSTRUMENTS INC. WDW-500), as shown in Fig. S3 in the supplementary material, and derived some critical performance properties of the Q7-mixture for comparison with Sylgard 184 PDMS, as listed in Table I. The maximal elongation of our mixed Q7 rubber reach over 1200% before breaking. Moreover, the Q7-series of materials have excellent biocompatibility and are widely used for medical implants.^43–45 According to the Product Information for Silastic® BioMedical Grade ETR Elastomers, saline extracts of the full range of products have biocompatibility of USP Class V, including Q7-4720 and Q7-4750, and elastomers show reactions equivalent to or lesser than the negative control in a 90-day implant test. Thus, we speculate the mixture of Q7-4720 and Q7-4750 has high biocompatibility as well. These advantages show the promising potential for use in organ-on-a-chip studies. Although the results show that the transparency of the Q7-mixture is not as good as that of Sylgard 184 PDMS [Fig. 1(e)] with a slight jade-white color observed for the Q7-mixture, it still enables clear optical observation, as described below in the section on single cell manipulations.

TABLE I.

. Mechanical Characters of Q7-mixture and Sylgard 184 PDMS.

Properties	Q7-mixture	Sylgard 184 PDMS
Prepolymer state	Solid	Liquid
Shore A hardness (HA)	∼47	∼4046
Stretch strength (MPa)	∼10.7	∼6.0247
Elastic deformation range (%)	∼100	<5029
Biocompatibility	Non-irritating and non-toxic	Inflammatory reaction48

III. CHARACTERIZATIONS

We fabricated channels with various widths ranging from 5 to 150 μm to investigate the impact of the width on the deformation of the microfluidic channel under stretching. All the lengths of the channels were much larger than the widths and heights. Here, we define the tensile strain ɛ_w= (w_s − w₀)/w₀, ɛ_h= (h_s − h₀)/h₀, and ɛ_l= (l_s − l₀)/l₀ along the direction of width, height, and length, respectively, where the subscripts s and 0 indicate the typical length after and before stretching, respectively, and $\alpha$ indicates the strain of the whole chip. We set the strain of the chip $\alpha$ as the y-axis and compare it to the strain of the microchannel ɛ in the y-axis with the chip stretched along the direction of the width [Figs. 2(a) and 2(b)]. The results shown in Fig. 2(c) show that the elongation of the channel width reaches a maximum of ∼600% in the 5 μm-wide channel when the chip is elongated to ∼150%. With an increased channel width of 70 μm, the maximal local elongation of the channel width gradually decreases to ∼200% and nearly becomes saturated at the 150 μm microchannel, which was close to the elongation of the chip of ∼150%. The strain of the channel length on the x-axis remains at the same strain value of ∼ − 30% as that for the chip length in the x-direction. Our results indicate that structures with a small typical length are likely to show a stronger deformation, as the stress is inversely proportional to the typical length σ ∼ 1/w. A stronger deformation for microscale structures can help manipulate fluids or objects, with only a reasonable stretching ratio required for chips.

FIG. 2.

Characterization of microfluidic channels prepared using Q7-mixtures. (a) A sketch of a single straight channel and (b) the channel stretched along the y-direction. (c) The strains of the channel length (ɛ_l) and width (ɛ_w) compared with the strain of the whole chip (α). (d) The strain of the channel height compared to the strain in the chip. (e) Typical I–V curves of measured for the microchannel before (blue curve, square symbols) and after (green curve, circle symbols) stretching with a chip strain of ∼150%.

We could optically measure the strain along the width and length directions of the microchannel by taking microscope images before and after stretching; however, it was impossible to characterize the height change. We, therefore, fabricated a microfluidic chip with only a single straight channel to characterize the height change by the measurement of the electrical resistance under stretching [Figs. 2(c) and 2(d)]. We filled these channels with electrolyte solutions consisting of 0.1M KCl connected to two AgCl electrodes with an ammeter (Keithley 2410), and the height change h was calculated from the electrical resistance measured for the solution filled channels R,

$$\begin{array}{c}R={\displaystyle \frac{1}{\kappa }{\displaystyle \frac{l}{wh}}}\end{array},$$ (1)

where κ, l, and w are the conductivity of the electrolyte solution, and the length and width of the channel, respectively.

As we stretched the whole chip, the strain was inhomogeneous near the edge but was homogeneous at the center section of the elastomer. Both our experimental results and numerical simulations confirm that the stress is homogeneous at the center of the stretched elastomer [more details are shown in Figs. S4 and S5(c) in the supplementary material]. Thus, in the following discussion, we took the value of the strain at the center of a standard chip size (2 mm thick, 26 mm wide, and 76 mm long) to evaluate the strain in the microstructures.

Typical I–V curves for unstretched (blue curve, square symbols) and stretched (green curve, circle symbols) microchannels are shown in Fig. 2(e). We derived the channel conductance by linear fits to the measured I–V curves, thus enabling the calculation of the height change using Eq. (1). We found that the calculated height remains nearly constant at various microchannel widths when stretching the chip. The constant height after stretching enables the manipulation of cells, avoiding the squeezing or sticking of cells in the channel.

IV. APPLICATION

Microfluidic devices provide an important platform for single cell studies.^49,50 Multiple physiochemical approaches, such as chemical modification,⁵¹ dielectrophoresis,^52,53 optical tweezers,⁵⁴ ultrasonic fields,⁵⁵ magnetic fields,^56,57 and hydrodynamic trap,⁵⁸ can be integrated into microfluidic devices for the single cell trapping. Hydrodynamic trapping is widely used because it does not need an external force field and results in less cellular injury.^59–62 Trapping can be generated by stagnation flow⁶³ or changes in flow resistance produced by microstructures.^64,65 However, the cell release for many hydrodynamic trapping events still relies on an external force field. A single cell trapping–releasing hydrodynamic trap needs to be developed.

Here, we demonstrate a highly elastic chip for the application of single cell trapping and release as an example. We designed a microfluidic chip with a U-trap array microstructure, according to the principle of hydrodynamic trapping.⁶⁶ Once a single cell is trapped within the interspace, the increase in hydrodynamic resistance avoids a second cell or more cell clusters from being trapped in the same unit. However, the dynamic release of cells is important for single cell studies. The highly flexible elastomer shows a great advantage for the dynamic cell release by simply stretching the elastomer chip, avoiding the use of more complex microvalves or other facilities. When the distance at the rear of the trapping structure is larger than cell sizes, the trapped cells can be released with the flowing solutions, as schematically illustrated in Figs. 3(a) and 3(b).

FIG. 3.

(a) Schematic illustration of the principle of the hydrodynamic trapping of single cells (yellow spheres) and (b) the release of cells by chip stretching. (c) and (d) Images of real chip before and after stretching, with the corresponding deformation of the trap before and after stretching shown in (e) and (f). (g)–(j) The dynamic release process for a single cell triggered by stretching the microfluidic chip along the width direction.

We fabricated cell-trap arrays in our elastomer chip by imprinting by following the above process. Each unit matched the size of a single cell, with a 20 μm-wide opening at the front and a 7 μm-wide gap at the rear. Hundreds of U-shaped cell traps were placed into a single microchannel with a width of 1 mm and a height of 20 μm. We injected solutions with the dispersion of lymphoid cancer cells (EK Bioscience, CC-Y1658) into the microchannel and found that the cells are well trapped inside the interspace of the U-shape array, as shown in Fig. 3(g). As we manually stretched the elastomer chip along the width direction of the channel, the gap distance at the rear of the U-trap increased to 35 μm [Fig. 3(f)], which is larger than the diameter of the trapped cell (∼20 μm). Due to the pressure difference applied at the two ends of the microchannel, the hydrodynamic flow drives cells to escape from the trap [Fig. 3(i)]. Withdrawing the stress from the elastomer chips helps the microstructures to recover again to the cell-trapping mode [Fig. 3(j)]. Finally, we achieved single cell trapping–releasing functions by simply stretching the elastomer microfluidic chip.

To analyze the deformation of the cell-trap structure, we built a linear elastic material numerical simulation model. A microtrap was set at the center of an elastic block that deformed with the block when stretching. Under stretching, the strain of the micro-trap is two orders of magnitude higher than that of the block. The stress distribution diagram and strain–stress curve shown in Fig. S5 in the supplementary material illustrate that the strain between the narrow gaps increases due to the stress concentration. More details for the numerical simulation and analysis can be found in Fig. S5 in the supplementary material.

Our dynamic cell release is feasible for the manipulation of suspension cells, while trypsinization may be needed before the release of adherent cells. High elasticity can induce mechanical stress to the cells, which can help one to respond to the mechanical activity; however, unnecessary mechanical strain is a side effect. In addition, the transparency of the Q7-mixture toward visible light is lower than that of PDMS, which can be improved by reducing the thickness of the device. Although we have only demonstrated a simple function of cell trapping and release for the Q7-mixture based chip, we believe that this highly elastic chip can be useful for more complex manipulation of objects or fluids in lab-on-a-chip systems. Furthermore, the Q7-mixture can be a good option for constructing organ-on-a-chip systems due to its elasticity.

V. CONCLUSION

In this work, we presented the fabrication of a highly flexible elastomer chip and the application of dynamic cell trapping release. We developed fabrication procedures, including the preparation of silicon rubbers and the manufacturing of microstructures by stamping them onto a template. In addition, we characterized the change in the geometry after stretching and found that the strain in small structures is likely to be larger than that in the chip. Finally, we use a highly elastic microfluidic chip for the application of single cell manipulation. Our results show that cells can be released by simply stretching the chip. We believe our elastomer chip will be useful in organ-on-a-chip studies due to its high elasticity and compatibility.

SUPPLEMENTARY MATERIAL

See the supplementary material for (1) the preparation of Q7-mixture chip fabrication (S1); (2) strain of concave and convex microstructures (Fig. S2); (3) tensile stress–strain diagram of the Q7-mixture (Fig. S3); (4) homogeneous strain on chip center (Fig. S4); and (5) numerical simulation of microstructure unconventional deformation caused by stress concentration (S5 and Fig. S5).

AUTHOR DECLARATIONS

Conflict of Interest

The authors have no conflicts to disclose.

Author Contributions

Xi Zhou and Yanbo Xie conceived and designed the experiments; Miao Sun performed the experiments and analyzed the data. Miao Sun, Xi Zhou, and Yanbo Xie wrote the manuscript; Yi Quan and Lianbing Zhang supported the cell-trapping experiments. All authors actively took part in all scientific discussions.

DATA AVAILABILITY

The data that support the findings of this study are available from the corresponding author upon reasonable request.