Enhanced sample filling and discretization in thermoplastic 2D microwell arrays using asymmetric contact angles

Abstract

Sample filling and discretization within thermoplastic 2D microwell arrays is investigated toward the development of low cost disposable microfluidics for passive sample discretization. By using a high level of contact angle asymmetry between the filling channel and microwell surfaces, a significant increase in the range of well geometries that can be successfully filled is revealed. The performance of various array designs is characterized numerically and experimentally to assess the impact of contact angle asymmetry and device geometry on sample filling and discretization, resulting in guidelines to ensure robust microwell filling and sample isolation over a wide range of well dimensions. Using the developed design rules, reliable and bubble-free sample filling and discretization is achieved in designs with critical dimensions ranging from 20 μm to 800 μm. The resulting devices are demonstrated for discretized nucleic acid amplification by performing loop-mediated isothermal amplification for the detection of the mecA gene associated with methicillin-resistant Staphylococcus aureus.

FIG. 1.

Operation of a 2D microwell device, with schematic side views of the array (left) and corresponding experimental images (right) at different operating points: (a) air-filled device, (b) and (c) aqueous sample filling, and (d) and (e) sample discretization during oil introduction.

FIG. 2.

Numerical 3D simulation results showing the impact of contact angle asymmetry (upper channel/lower well surfaces) on sample filling. For the given well geometry, complete filling without air trapping is only achieved for the case of asymmetric contact angles of 79° for the upper channel surface and 30° for the microwell surface, while symmetric 30°/30° and 79°/79° conditions result in air trapping.

FIG. 3.

Impact of the channel/well contact angle asymmetry and device geometry on filling efficiency when varying (a) the D/L ratio at a constant channel height (H = 400 μm) and (b) the H/L ratio at a constant well depth (D = 800 μm). Numerical (o) and experimental (×) data shown. All experiments were performed using square microwells with L = 800 μm, with a sample flow rate of 1000 μl/min (n ≥ 3). For the low contact angle experiments, the well surface value was measured to be lower (15°) than the simulated value (30°).

FIG. 4.

Effect of well geometry and capillary number on retention efficiency in square microwells (L, W = 800 μm) for (a) H/L = 0.5, (b) H/L = 0.75, and (c) H/L = 1. As both H/L and D/L increase, the dependence of retention efficiency on the capillary number is reduced. Retention efficiency approaches unity as D/L increases (N ≥ 4, RSD < 2.3%).

FIG. 5.

Effect of well dimensions and orientation on sample retention efficiency (all values in mm). The dominant parameter dictating retention efficiency for both rectangular and circular well openings is the minimum microwell length or width (L,W_min). For the case of circular microwells, the well diameter (Dia) was used for both the length and width parameters. All experiments were performed with a main channel height of H = 200 μm and oil flow rate of 10 μl/min (N ≥ 4).

FIG. 6.

(a) Photograph of a fabricated 1000-element microwell array. (b) Sample (DI water with food dye) filling and (c) discretization using silicone oil containing 0.05% Span80 in a 1000-element array with L,D = 400 μm and channel height H = 200 μm. (d) Histogram of retention efficiency across all 1000 wells (3.3% RSD).

FIG. 7.

(a) Image of a bonded device containing 4900 circular 20 μm diameter microwells. (b) Sample filling and (c) discretization using silicone oil containing 0.05% Span80. Filling of all wells was achieved, with 2.1% RSD variation in volume. The chip design followed the design guidelines, with D/L = 1 and H/L = 10.

FIG. 8.

(a) Images of 29 microwells in a 2D array before and after amplification. (b) LAMP amplification curve on the 2D array platform (average of 29 microwells with 95% confidence intervals), and (c) comparison of time to amplify between a microfluidic 2D array and a conventional thermocycler platform for LAMP (shown with 95% confidence intervals). Assay performance on the 2D array platform and the conventional thermocycler platform is comparable in terms of time to positive (TTP) and the variance in TTP.