Fabrication of concentric microarrays for self-assembly and manipulation of particle distribution

Abstract

We demonstrate the use of a micropatterned surface composed of a concentric circular array of SiO₂ and Si layers to distribute particles from edge to centre and reduce the coffee stain effect from a colloidal or suspended solution droplet after natural evaporation by self assembly over the array. The alternating SiO₂/Si layers of the concentric circular array have widths of 3 μm and depths of 400 nm, and its surface is nonhomogeneous due to the presence of the alternating hydrophilic (SiO₂)–hydrophobic (Si) layers and the 400 nm variation in surface roughness. The droplet diameter, contact angle and droplet height change more rapidly from the top of such a surface compared to plain Si/SiO₂ during natural evaporation. The depinning process, capillary force and Marangoni convection are the likely drying mechanisms of particle advection in the droplet. This results in even distribution on the top and side walls of the SiO₂ ring and thus minimizes problems associated with the “coffee-stain” effect, providing better distribution when dispensed from a solution. The particle distribution is restricted to the top of the circular rims and the sides or the floor of the space between the rings depending on the surface energy of the ring surfaces, the structural geometry and the natural evaporation of the droplet.

---

Introduction

Many biological fluids are suspensions or colloids in a base liquid. Cells, vesicles and chemically-functionalized nano-particle solutions fall under this category. A uniform distribution of suspended particles on a surface is desired for many applications including micropatterned arrays of biochips for bioreactions with targeting probes such as antigens,¹ antibodies,² oligonucleotides,³ binding proteins,⁴ aptamers,⁵ receptors,⁶ and quantum dots⁷ immobilized on the surface. Generally, these particles are conjugated with other materials such as proteins and deoxyribonucleic acid (DNA). For assay applications, the particles need to be localized on the substrate. Several methods have been developed to array beads onto a flat surface. Some examples of methods for obtaining well-ordered structures include the use of (i) porous films,⁸ (ii) electrostatic and capillary forces, (iii) dielectrophoresis, (iv) template-assisted self-assembly, (v) electrodeposition of adhesives, (vi) crosslinking, (vii) bead array assays for direct hybridization⁹ and (viii) hydrophobic interactions.¹⁰

Of these methods, hydrophobic interactions are the most explored surface modification strategy for spot probe assay experiments. Various techniques and materials including biologically-inspired fibres and textiles, phase separation, crystal growth, amphiphilic inorganic materials, nanostructured crystals, differential etching, diffusion-limited growth processes, lithographic techniques, aggregation/assembly of particles, and templating have been extensively studied by researchers investigating the hydrophobicity of surfaces.^11–13 Even though microarray assay technology has matured by the adoption of the suitable technologies mentioned above, a number of issues remain unresolved. For example, the signal intensity is higher at the edges and lower at the centre of a probe spot. Therefore, the uniform distribution of the colloidal particles or vesicles is very important to minimizing the saturation of the detector during fluorescence imaging. During drying, the particles/vesicles tend to aggregate at the edge of the droplet, which is commonly known as the “coffee stain” problem¹⁴ and causes doughnut patterns. The presence of coffee stain effect ring-shaped intensity patterns hinders the correct analysis of the probe spot. Very high intra-spot standard deviation occurs, and existing image analysis tools do not contain algorithms capable of minimizing/eliminating such doughnut pattern effects. Microarrays on solid supports are generally fabricated on glass slide supports. In many applications, the surfaces of the glass slides are treated with different chemistries such as epoxy, amino-silane, poly-L-lysine, streptavidin and so on.^15–17 Hydrogel-coated glass slides are also used for immunoassays because the gel itself supports probe binding through its moieties.¹⁸ For polymer/lipid vesicles or other bio-applications, however, physisorption must be used instead of chemical tailoring to bind the spheres covalently to the substrate; the surfaces of the vesicles cannot contain moieties that can be covalently bound to the surface. In this context, we study nonhomogeneous alternating hydrophilic (SiO₂)–hydrophobic (Si) surfaces (each with 3 μm widths and 400 nm trench depths from SiO₂ to Si) obtained using single-mask lithography followed by dry etching of SiO₂ in order to reduce the coffee stain effect and achieve uniform particle distribution after natural evaporation.

Experimental

Two different materials, SiO₂ and Si, were employed to examine the natural evaporation of water droplets from homogeneous and nonhomogeneous surfaces, respectively. The surface of the SiO₂ substrate is inherently hydrophilic and homogenous. The nonhomogeneous surface is an alternating SiO₂ and Si concentric circular structure. The concentric ring structure consists of an SiO₂ ring followed by an Si trough. The width of each SiO₂ ring crest and Si trough is 3 μm, i.e., the pitch is 6 μm. The single device comprises 165 SiO₂ ring crests and Si troughs. The diameter of a single device is ∼1 mm, and the depth of SiO₂ to Si is 400 nm. The concentric SiO₂/Si ring structure was fabricated using a single-mask photolithography and dry etching process. The starting substrate was a 100 mm Si(100) wafer. After standard wafer cleaning, 400 nm of SiO₂ was produced through thermal oxidation. The SiO₂ thickness on top of the Si wafer was verified using an ellipsometer. The starting substrate was 400 nm of thermally-grown SiO₂ on a 100 mm diameter 〈100〉 Si substrate. The top 400 nm thick SiO₂ layer was etched using an RIE in an Oxford PlasmaLab80 system using 25 sccm of Ar and 25 sccm CHF₃ at room temperature with a pressure of 30 mTorr and an RF power of 100 W. After dry etching, the remaining photoresist and etch by-product were cleaned using piranha solution at 140 °C. The substrate with the concentric circular ring structure and a plain Si/SiO₂ substrate were used to study the polystyrene particle distributions after droplet evaporation. The working colloidal fluid was a mixture of fluorescently-loaded polystyrene beads with 200 and 50 nm diameters in deionised (DI) water. The excitation wavelength of the fluorescent-loaded polystyrene particle was 441 nm. After excitation, the particle radiated a 486 nm red emission. The volume of fluid used for droplet evaporation was 0.5 μL. Contact angles, contact diameters and centre heights of the evaporating droplets were measured using an imaging system consisting of a CMOS camera (ARTCAM-300MI, Art-Ray Corporation, VA, USA), micropipettes (P2, Gilson, Inc., WI, USA), a magnifying lens (70 mm F2.8, EX DG, SIGMA Corporation, NY, USA), a halogen lamp (400 W, 3 M Overhead Projectors, MN, USA) and a computer with an installed frame grabber. The distribution of particles was characterized using scanning electron microscopy (SEM, XL-30, Philips Corp.) at an acceleration voltage of 20 kV.

Results and discussions

Fig. 1 shows the schematic of the plain and micropatterned Si/SiO₂ substrates for particle distribution after natural evaporation.

Fig. 1 Schematic of plain and micropatterned Si/SiO₂ substrates for particle distribution.

The micropatterned substrate is a concentric ring structure composed of hydrophilic SiO₂ protrusions followed by hydrophobic Si troughs. The concentric structure has multiple rings with an outer diameter of 1 mm. Droplets (0.5 μL) were dispensed on both substrates. The chosen volume of 0.5 μL was chosen based on several test runs for the confinement of the liquid within the outer ring. The images of droplets on both substrates immediately after being dispensed (without any time for evaporation) are shown in Fig. 2.

Fig. 2 Droplet shape on plain and microring-patterned Si/SiO₂ substrates.

The contact angles, contact line diameters and heights at the centres of the droplets are shown in Table 1. The contact angle measurements show that the patterned substrate (102°) is more hydrophobic than the plain Si/SiO₂ substrate (38°). The experiments were repeated three times, and the tabulated data are mean values with ±5% variation.

Table 1 Contact angle, contact diameter and droplet height at the centre for evaporation time t = 0 min

Substrate	Droplet image at t = 0 min	Con. angle (deg)	Con. diameter (mm)	Height at centre (mm)
Plain Si/SiO2		38.9	1.535	0.199
Micro- patterned Si/SiO2 ring structure		102.6	0.95	0.373

---

The combined effects of the surface topography and the concentric alternating hydrophilic/hydrophobic SiO₂/Si ring structure create such a hydrophobic surface. The contact diameter is shorter for the patterned substrate, and thus the droplet height is higher compared to the plain surface for a droplet of equal volume. The shape of the droplet on the patterned substrate is much more spherical than that on the smooth, flat Si/SiO₂ surface due to the surface topography. A footing at the base of the droplet on the patterned surface can be observed in Fig. 2. In this study, the microstructural dimension was selected to achieve Wenzel sticky states where droplets wet the “valley” patterned area for particle distribution. The base of the droplet seen in Fig. 2 illustrates that the water droplet completely wets the textured surface, which is known as the “Wenzel state.” In this state, since the air pockets are thermodynamically unstable, the liquid begins to nucleate a wetting layer from the base of the droplet on the substrate to form a mushroom-like shape, as seen in Fig. 2.

The natural evaporation of the droplets from both surfaces was investigated using real-time monitoring of the contact angle, contact diameter and droplet height. Fig. 3 shows a comparison of the normalized diameter plotted against the normalized droplet evaporation time on both surfaces.

Fig. 3 The time-dependent contact diameter of naturally-evaporated colloidal fluid droplets on a plain Si/SiO₂ surface and a concentric Si/SiO₂ ring-structured surface.

The diameter remains almost unchanged till approximately 50% of the total evaporation time, which is the initial mode of an evaporating droplet on these two substrates. For the plain Si/SiO₂ substrate, a gradual reduction in diameter is observed, and the initial wetting diameter decreases by ∼15% until the droplet completely dries out. On the other hand, a rapid diameter reduction is observed after 50% of the evaporation time in the case of the Si/SiO₂ ring-patterned substrate.

In order to investigate the contact line movement, the time-dependent edge shrinking velocity, which is related to the rate of decrease in the droplet diameter dD/dt for an evaporating droplet, has been plotted for both substrate surfaces (Fig. 4; the plotted data are dD/dt derived from the mean values of three experimental runs at each temporal point).

Fig. 4 Droplet edge shrinking velocities on plain Si/SiO₂ and concentric Si/SiO₂ ring-structured surfaces.

In the case of the patterned surface, several small peaks appear during the initial mode of evaporation when the diameter remains almost fixed. These peaks are due to the inward movement of the contact line during evaporation; thus, rapid droplet diameter reduction occurs during evaporation. In contrast, the time-dependent edge shrinking velocity on the plain Si/SiO₂ surface exhibits no such dominant peaks. This is likely due to the slow reduction in droplet diameter with evaporation time and the fact that the contact line was fixed during most of the evaporation (Fig. 3). The edge shrinking velocity of the droplet on the ring structure exhibits nonlinear behaviour, and its magnitude becomes extremely large at the end of evaporation. In contrast, the behaviour is more linear for the plain Si/SiO₂ surface. The time evolution of droplet height was investigated on both surfaces to understand the effects of surface homogeneity (topography and hydrophobic/hydrophilic character) during evaporation (Fig. 5).

Fig. 5 Time evolution of droplet height plotted for the plain Si/SiO₂ and concentric Si/SiO₂ ring-structured substrates.

Droplet height reduction is dependent on contact angle as a function of time and the contact-line radius,

where θ is the contact angle and R is the contact-line radius.¹⁴ At r = 0 (i.e., at the centre of the droplet), the height is:

The theoretical values for both the plain Si/SiO₂ and patterned substrate (Fig. 5; the dashed and solid lines, respectively) are quite close to the experimental data. A trivial deviation was observed, likely due to the surface nonhomogenity and the limited accuracy of experimental data collection. The initial droplet height on the Si/SiO₂ ring structure is higher, but it is reduced much faster than on the plain Si/SiO₂ surface. The smaller droplet diameter and larger contact angle on the patterned ring structure result in a larger surface area-to-volume ratio and a larger liquid–air interface. Therefore, both the smaller droplet size and larger liquid–air interface promote faster evaporation on the patterned substrate compared to the plain Si/SiO₂ substrate.

In addition to the contact angle and surface area-to-volume ratio, the capillary forces from the alternating hydrophobic Si and hydrophilic SiO₂ layers of the microring architecture also influence the drying effect and particle distribution on the substrate surface during evaporation. Fig. 6(a–c) demonstrate the particle distribution on the plain and ring- patterned Si/SiO₂ surfaces.

Fig. 6 SEM images of particles after natural evaporation on (a) a plain SiO₂ substrate after the evaporation of a colloidal solution of 200 nm polystyrene beads, (b) an Si/SiO₂ ring structure after the evaporation of a colloidal solution of 200 nm polystyrene beads and (c) an Si/SiO₂ ring structure after the evaporation of a colloidal solution of 50 nm polystyrene beads.

A significantly thick outer stain ring is observed due to the pileup of polystyrene particles near the droplet contact line, similar to the coffee stain effect on the plain Si/SiO₂ substrate (Fig. 6(a)). This phenomenon is due to the development of internal fluid flow within the droplet along the outward radial direction, effectively advecting particles towards the wetting line during evaporation. This flow pattern arises because of pinning and the maximum evaporation flux at the wetting line, as explained by Deegan et al.¹⁴ and Hu and Larson.¹⁹ In contrast, the ring structure shows an even distribution of particles under the liquid droplet (Fig. 6(b) and (c)). Although some particles still accumulate at the edge of the drop during the initial pinned stage, a significant number of them are distributed over the entire patterned structure. The droplet diameter on such a patterned surface is reduced continuously faster after 50% of the total evaporation time compared to the plain Si/SiO₂ surface (Fig. 3). During this phase, the contact line continually switches between a pinned and a depinned state at the alternating hydrophilic (SiO₂) and hydrophobic (Si) rings. When a microdrop, such as the 0.5 μL drops used in this study, is located on a step at the boundary of a hydrophilic SiO₂ region and a hydrophobic Si region, it progressively moves towards the hydrophilic SiO₂ region due to the dominance of capillary forces over gravity.²⁰ During evaporation, the liquid–solid wetting line creates a local increase in surface tension, which reduces surface flow and results in the liquid first moving outward and then turning back due to the Marangoni effect. Generally, the outward liquid flow is parallel to the substrate surface, while the inward flow is nearly parallel to the vapour liquid interface, dragging the particle in the inward direction.²¹ Thus, the combined effects of depinning, capillary force and the Marangoni convection drag act to deposit particles regularly on the concentric hydrophilic SiO₂ rings. During evaporation, the suspended particles move outwards or inwards due to the above-mentioned combined effects. Impeded by the ring side wall, particles are also deposited along the side of the SiO₂ ring, as shown in Fig. 6(b) and (c). The distributions of the 200 and 50 nm beads on the ring-patterned Si/SiO₂ substrate after complete evaporation are similar, indicating that the distribution is determined by the substrate surface texture regardless of bead size. However, the bead size must be smaller than the width of the SiO₂ crest.

The real-time droplet evaporation and saturation effect of the detector were studied using an optical imaging system. The device was mounted on an upright microscope (Olympus, BX51) with a mercury lamp attachment for fluorescence imaging. A charge-coupled device camera (Roper Scientific, Photometrics Cascade 512B and JVC, TK-C1481BEG) was used to capture the images and videos of the experiments for analysis. Fig. 7(a) to (e) show the full structure and zoomed images at different evaporation times. The contact diameter of an evaporating droplet on the concentric array of a ring- textured surface decreases with evaporation time. For example, a substantial reduction in contact diameter occurs after 75 s of evaporation (Fig. 7(c)). The distributions of fluorescently-loaded polystyrene beads on the ring structure are evenly arranged on top of the SiO₂, and some are situated at both the inner and outer peripheries of each SiO₂ ring (Fig. 7(d)), as explained earlier. Generally, special sample/specimen preparation is required for spectral imaging in order to avoid dramatic mismatches in intensity. These mismatches cause detector saturation for the most concentrated probes such as in coffee stain rings. On the other hand, those of lower intensity can easily be lost in the noise floor. In this study, the concentric Si/SiO₂ structure minimizes the coffee stain effect, promotes the self-assembled uniform distribution of particles after natural evaporation and naturally reduces the detector saturation effect.

Fig. 7 Real time optical images of droplet evaporation and particle distributions on an Si/SiO₂ ring structure (a) after the droplet was dispensed, (b) after 45 s of evaporation, (c) after 75 s of evaporation and (d) a zoomed-in image after complete evaporation.

Conclusions

Micropatterned surfaces composed of concentric circular arrays of SiO₂ and Si layers have the potential to produce uniform particle distributions and reduce the coffee stain effect from a colloidal or suspended solution droplet after natural evaporation by self assembly. The concentric circular array with alternating 3 μm wide SiO₂/Si layers and a 400 nm deep surface is nonhomogeneous due to the presence of alternating hydrophilic (SiO₂)–hydrophobic (Si) layers as well as the variation in surface roughness. Such a nonhomogeneous surface leads to a faster evaporation rate and a faster reduction in contact angle and droplet diameter. The combined effects of depinning, capillary force and Marangoni convection lead to an even distribution of particles on the top and side walls of the SiO₂ ring. Thus, problems associated with the “coffee-stain” effect are minimized and particles dispensed from a solution are better distributed. Consequently, the structure allows the uniform distribution of signal and avoids detector saturation when a fluorescent sample is inspected by conventional fluorescent analyzers.

Acknowledgements

The authors wish to deeply thank Dr Madanagopal V Kunnavakkam, ST Microelectronics Singapore for very helpful comments and discussions.

References

1. P. Montagne, P. Varcin, M. L. Cuilliere and J. Duheille, Bioconjugate Chem., 1992, 2, 187.
2. R. Bellisario, R. J. Colinas and K. A. Pass, Early Hum. Dev., 2001, 64, 21.
3. Y. Kohara, H. Noda, K. Okano and H. Kambaba, Nucleic Acids Res., 2002, 30, e87.
4. A. S. Brodsky and P. A. Silver, Mol. Cell. Proteomics, 2002, 1, 922.
5. H. Y. Zhu, J. D. Suter, I. M. White and F. D. Fan, Sensors, 2006, 6, 785.
6. M. A. Iannone, T. G. Consler, K. H. Pearce, J. B. Stimmel, D. J. Parks and J. G. Gray, Cytometry, 2001, 44, 326.
7. Q. Ma, T.-Y. Song, P. Yuan, C. Wang and X.-G. Su, Colloids Surf., 2008, B 64, 248.
8. J. Lee, O. Kim, J. Jung, K. Na, P. Heo and J. Hyun, Colloids Surf., B, 2009, 72, 173.
9. http://www.illumina.com/technology/beadarray-technology/direct-hybridization-assay.ilmn.
10. F. Fan and K. J. Stebe, Langmuir, 2004, 20, 3062.
11. P. Roach, N. J. Shirtcliffe and M. I. Newton, Soft Matter, 2008, 4, 224–240.
12. A. L. Dubov, J. Teisseire and E. Barthel, EPL, 2012, 97, 26003.
13. J. Drelich, J. L. Wilbur, J. D. Miller and G. M. Whitesides, Langmuir, 1996, 12, 1913–1922.
14. R. D. Deegan, O. Bakajin, T. F. Dupont, G. Huber, S. R. Nagel and T. A. Witten, Nature, 1997, 389, 827.
15. W. Kusnezow, A. Jacob, A. Walijew, F. Diehl and J. D. Hoheisel, Proteomics, 2003, 3, 254–264.
16. E. W. Olle, J. Messamore, M. P. Deogracias, S. D. McClintock, T. D. Anderson and K. J. Johnson, Exp. Mol. Pathol., 2005, 79, 206–209.
17. C. Wingren, J. Ingvarsson, L. Dexlin, D. Szul and C. A. K. Borrebaeck, Proteomics, 2007, 7, 3055–3065.
18. H. Li, R. F. Leulmi and D. Juncker, Lab Chip, 2011, 11, 528–534.
19. H. Hu and R. G. Larson, J. Phys. Chem. B, 2002, 106, 1334.
20. M. K. Chaudhury and G. M. Whitesides, Science, 1992, 256, 1539.
21. X. Xu and J. Bin, Proceeding of CIST, 2008, p. 186.

---