The strengths of dual-wavelength vSLA are its speed and ease of use. Existing SLA and micro-VAM systems can fabricate high-resolution microfluidics.12,1812. A. P. Kuo, N. Bhattacharjee, Y.-S. Lee, K. Castro, Y. T. Kim, and A. Folch, Adv. Mater. Technol. 4, 1800395 (2019). https://doi.org/10.1002/admt.20180039518. J. T. Toombs, M. Luitz, C. C. Cook, S. Jenne, C. C. Li, B. E. Rapp, F. Kotz-Helmer, and H. K. Taylor, Science 376, 308–312 (2022). https://doi.org/10.1126/science.abm6459 However, they still take hours to produce useable devices. Our vSLA method can produce finished microfluidic devices in less than 10 min. Additionally, the dual-wavelength printer can be made from an ordinary projector and a UV LED, with no other translating or rotating parts. The layers of the device are formed as they appear in the slices, and all calculations are limited to scalar equations. vSLA and SLA operate identically in the horizontal projection plane, and the novelty of vSLA comes from the ability to control the location of curing in the vertical z direction. Hence, we primarily report results about the z heights of microfluidic features and not x or y dimensions.

A. Characterization of resin photoinitiation and photoinhibition behavior

The vSLA resin (Table I) was comprised of photoinitiator, photoinhibitor, co-initiator, oligomer, reactive diluent, and blue light absorber. The blue light-activated photoinitiator, CQ, and the UV light-activated photoinhibitor, o-Cl-HABI, are required for this dual-wavelength system. The amine co-initiator EDAB is also necessary to improve the initiation rate because CQ alone is an inefficient initiator of free radical polymerization.2727. L. F. J. Schneider, L. M. Cavalcante, S. Consani, and J. L. Ferracane, Dent. Mater. 25, 369–375 (2009). https://doi.org/10.1016/j.dental.2008.08.003 The rest of the resin formulation is flexible, though several factors should be considered when formulating a new resin, including monomer reactivity, cured polymer properties, transparency in blue and UV wavelengths, and viscosity. Monomers must be susceptible to radical polymerization. Here, we used multifunctional acrylates for their high reactivity and mechanical properties. CN991, a high molecular weight oligomer, was used because of its optical transparency and hardness. To reduce the viscosity, HDDA was added as it was found to be more susceptible to photoinhibition by o-Cl-HABI compared to other reactive diluents.2323. M. P. de Beer, “Advances in additive manufacturing and microfabrication,” Ph.D. dissertation (University of Michigan, 2020). Another reactive diluent, TMSPMA, was used as a silane coupling agent2828. S. Ebnesajjad, Surface Treatment of Materials for Adhesive Bonding, 2nd ed. (Elsevier, 2014), pp. 301–329. to improve the adhesion of the polymer to the glass surfaces of the resin vat. Finally, a blue light absorber was added to reduce hblue, which increases vertical resolution by decreasing light penetration depth.2929. N. Bhattacharjee, C. Parra-Cabrera, Y. T. Kim, A. P. Kuo, and A. Folch, Adv. Mater. 30, 1800001 (2018). https://doi.org/10.1002/adma.201800001 The cured polymer was found to be hemocompatible, so microfluidic devices made with this resin could be used for the red blood cell analysis (see Fig. S-3 in the supplementary material for polymer hemocompatibility).For this resin formulation, the resin-specific parameters of the dose model were found experimentally and listed in Table II. UV-vis spectrophotometry was used to obtain hblue, which was calculated from the linear combination of individual resin component absorbances at 458 nm. Initially, hUV was also calculated from resin absorbance data, but a better representation of the experimental dead zone data was produced when hUV was fit with least squares regression to Eq. (3). Thus, the fitted value of hUV was used instead. The difference between the calculated and fitted values of hUV may be due to photochromism observed in o-Cl-HABI.3030. A. Kikuchi, Y. Harada, M. Yagi, T. Ubukata, Y. Yokoyama, and J. Abe, Chem. Commun. 46, 2262 (2010). https://doi.org/10.1039/b919180a Photochromism is more significant during vSLA due to the high intensities of UV light used, skewing the value of hUV calculated from spectrophotometry data. The values of m, Dcrit, and β were also obtained by fitting Eqs. (2) and (3) to cure depth and dead zone data with least squares regression.

TABLE II. Resin-specific dose model parameters from UV-vis spectroscopy, cure depth, and dead zone experiments.

VariableValueUnitshblue450.5μmm0.8910…Dcrit37.75(mW/cm3)mshUV192.6μmβ0.2377…The cure depth and dead zone data that were used to generate the dose model parameters are plotted in Figs. 2(a) and 2(c). The standard deviation error bars associated with cure depth data are noticeably larger than those associated with the dead zone data, which we attribute to variable oxygen concentration in the resin. Dissolved O2 is a powerful free radical scavenger that decreases the polymerization rate.3131. N. B. Cramer, C. P. O’Brien, and C. N. Bowman, Polymer 49, 4756–4761 (2008). https://doi.org/10.1016/j.polymer.2008.08.051 Although resins were sparged with N2, resin was transferred to the resin vat in air, and we suspect different concentrations of dissolved oxygen were introduced at this stage, leading to variability of cure depth data. Figure 2(b) shows the large difference in dose required to cure an equivalent height in air- and N2-sparged resins. Ultimately, N2-sparged resin was used despite the variability because more rapid polymerization allowed microfluidic devices to be fabricated in mere minutes.The dose model parameters were used to predict the location of curing and the properties of the gelled polymer for any exposure setting. Figure 3(a) shows the dose as a function of the exposure time and height into the resin, which is plotted using Table II parameters and constants Iblue and IUV of 3.21 and 8.65 mW/cm2, respectively. When the dose is equal to or greater than Dcrit, the resin becomes gels (orange curing region). For doses less than Dcrit, the resin remains fluid enough to be washed away by post-process rinsing (beige region). If the surface plot of the dose is viewed from above [Fig. 3(b)], we see that as the exposure time increases, the top side of the curing region increases logarithmically in height, while the bottom side approaches a steady-state value. Because Eq. (1) cannot easily be solved for z, Eqs. (2) and (3) were used to approximate the top and bottom boundaries of the curing region for exposure times greater than 15 s.The dose calculated by Eq. (1) can instead be used to understand the degree of polymerization of the cured polymer. Higher dose corresponds to a higher conversion of polymerizable groups, which increases the crosslinks between monomers. At a sufficient crosslink density, the resin gels, and as more crosslinks form, the storage modulus increases.3232. G. I. Peterson, J. J. Schwartz, D. Zhang, B. M. Weiss, M. A. Ganter, D. W. Storti, and A. J. Boydston, ACS Appl. Mater. Interfaces 8, 29037–29043 (2016). https://doi.org/10.1021/acsami.6b09768 While any dose above Dcrit will induce gelation, the degree of polymerization and the green strength of the gelled polymer vary with the height into the resin. For example, in Fig. 3(c), the magnitude of the dose plot suggests that the polymer gelled at a height of 500 μm would be more polymerized than at 150 and 2000 μm. Figure 3(c) also shows that there is no dose almost immediately below the dead zone, while there is non-negligible dose far above the cure depth. This is an important feature of dual-wavelength vSLA because the dose accumulates. Even if two exposures individually do not produce D≥Dcrit, if they are applied consecutively and the cumulative dose is larger than Dcrit, the resin will cure. For vSLA, where multiple exposures are used, we can assume that no dose is added in the dead zone, while the dose accumulated above the cure depth is equal to the blue light-only dose (Ibluehbluee−zhblue)mt.

B. vSLA of microfluidic channels

To design the exposure sequence for microfluidic fabrication, the microfluidic device was divided into one or more layers, corresponding to the locations that new features appear in the z direction. For example, a microfluidic device with a single horizontal channel could be divided into three layers. The first layer spans from the bottom of the resin vat to the bottom of the channel, the second layer is spatially patterned to form a void that becomes the channel, and the third layer spans from the top of the channel to the top of the resin vat. Using Eq. (2), any combination of Iblue and t that yield the dose needed to cure to the ending height of each layer can be used, although high intensity blue light and a short exposure time are preferred. Finally, the corresponding UV intensity that produces the required dead zone for the starting height of each layer is calculated from Eq. (3). For simplicity, we limited fabrication to only three exposures, with constant light intensity. However, more layers and exposures can be used to make smooth continuous features in the z direction by increasing the number of layers or by using grayscale patterning with blue light slices. Table III shows exposures used to make various microfluidic devices.TABLE III. Geometry print parameters per layer and predicted layer starting (zdz) and ending heights (zcd), calculated using Eqs. (2) and (3). Geometries are categorized by the location of the channels, which can exist on a single level or multiple levels. zcd is based on the total blue light dose that includes the accumulation of dose contributions from exposures before or after the current layer. m = 0.8910.GeometryExposureTime (s)Iblue (mW/cm2)IUV (mW/cm2)zdz (μm)zcd (μm)Single levelSingle exposure1300.7200>1000Straight or obstructed13.50.72002972153.2100>100031202.1426.9654>1000Curved13.50.72004652303.2100>100031202.1426.9654>1000MultilevelSerpentine1150.72006592353.218.65136>100032402.1426.9654>1000Crossing1150.72007822303.218.65135>175032402.1426.9654>1750

1. Single blue light exposure fabrication

Microfluidic devices can be made using a single blue light exposure, although this is not true vSLA because there is no control over the bottom curing boundary. The blue exposure is patterned to have black pixels in the shape of the channel [Fig. 4(a)]. Using a single 30 s patterned exposure, we formed channels that adhered to the top of a 1 mm thick confined volume of resin [Fig. 4(b)]. Because the black pixels used for patterning emitted a non-negligible amount of blue light, a thin layer was cured at the bottom of the channels, evidenced by the green tint of the blue dye. However, the low dose led to a low degree of polymerization, and parts of this layer were eroded during resin removal, which is seen in blue regions around certain inlet ports.

2. Dual-wavelength vSLA of single level channels

While the single blue light exposure device remained attached to the resin vat, standalone devices were made by introducing UV light for true vSLA. These standalone microfluidic devices were fabricated using three exposures [Fig. 5(a)] to form geometries as shown in Figs. 5(b)–5(d). In the third exposure, UV light was applied to cure the top layer without curing through the channel underneath, so the device could ultimately be detached from the resin vat. The accuracy of Eqs. (2) and (3) was analyzed using images of channel cross sections [Fig. 5(e)]. The average heights (±standard deviation) of layers 1 and 3 were 242 ± 54 and 321 ± 32 μm, respectively, and the average channel height in layer 2 was 386 ± 55 μm. The layer 1 height of 242 μm corresponds to the layer 1 zcd. Subtracting the layer 3 height from 1 mm gives the layer 3 zdz of 679 μm. These values agree relatively well with the predicted layer 1 zcd of 297 μm and layer 3 zdz of 654 μm from Table III. From the microscope image, we also see that the channel has rounded corners, which is likely the result of incomplete resin removal and over-cure in the channel corners. During flushing, the velocities of IPA at the edges and corners would be lower, allowing resin to remain in those regions. Furthermore, light diffraction is known to increase dose at the intersections of features and cause over-curing, resulting in the smoothing of sharp edges.33,3433. C. Sun, N. Fang, D. M. Wu, and X. Zhang, Sens. Actuators A: Phys. 121, 113–120 (2005). https://doi.org/10.1016/j.sna.2004.12.01134. A. A. Bhanvadia, R. T. Farley, Y. Noh, and T. Nishida, Commun. Mater. 2, 1 (2021), 2021 21. https://doi.org/10.1038/s43246-021-00145-y

3. Dual-wavelength vSLA of multilevel microfluidic channels

vSLA was also used to fabricate multilevel structures, such as serpentine and crossing channels [Fig. 6(a)]. Both geometries are commonly featured in microfluidics. Serpentine channels are used for efficient mixing,3535. T. G. Kang, M. K. Singh, P. D. Anderson, and H. E. H. Meijer, Microfluid. Nanofluidics 7, 783 (2009), 2009 76. https://doi.org/10.1007/s10404-009-0437-2 and crossing channels are used in compact microfluidic devices and for functional structures such as membrane microvalves.3636. W. Zhang, S. Lin, C. Wang, J. Hu, C. Li, Z. Zhuang, Y. Zhou, R. A. Mathies, and C. J. Yang, Lab Chip 9, 3088 (2009). https://doi.org/10.1039/b907254c Unlike single-level devices, these multilevel devices were not standalone as they were fabricated with only three exposures. There was also significant deviation between the predicted channel heights [Table III] and the real channel heights. For exposures used in multilevel devices, complete curing was predicted above 136 μm, which would have cured through the channels at the top of the device. However, open channels are clearly seen in Figs. 6(b) and 6(c). The breakdown of Eq. (2) in the case of the multilevel devices may be due to the unknown concentration of dissolved oxygen in the resin, but further studies are needed to verify this claim.The crossing channel device highlights one of the advantages of vSLA, namely, the ability to form suspended objects. The middle layer of the device or the separator [Fig. 6(c)] was initially a freely suspended layer. From Fig. 4(b), we know the curing region grows both vertically upward and downward over time. Thus, the separator formed suspended in the resin until it grew downward and fused with the layer below it, turning the separator into an overhanging structure. For the separator to form correctly, it was paramount that the resin was stationary. Flowing resin exerts stresses that can deform or displace delicate features. For this reason, overhanging structures are generally printed with supports, and suspended features are impossible to make in conventional SLA. Although vSLA happens layer-wise, the ability to form a suspended feature classifies this system as a volumetric technology.

C. Concentrating dose by tuning absorbance height and light intensity

The separator between the crossing channels was exceptionally fragile and easily damaged during post-processing; however, the green strength could be improved by exposing the separator layer to higher doses, increasing the degree of polymerization. Figure 6(d) shows a successfully post-cured crossing channel device with no mixing of two different dyes, indicating that the channel separator was intact. According to Eq. (1), the dose is increased by (i) increasing the light intensities, (ii) decreasing the absorbance heights of the resin, or (iii) increasing the exposure time. Separately, increasing the light intensities or the exposure time widens the span of the curing region, while decreasing the absorbance height shrinks the curing region. However, if all three are adjusted together, then the dose can be increased while maintaining the span of the curing region. For example, Fig. 7(a) shows the dose profile for original exposure settings used in the separator layer, and Figs. 7(b) and 7(c) show that the dose can be concentrated by scaling the resin absorbance and light intensities. The scaling factors for Iblue, hblue, and hUV were arbitrarily chosen, and IUV was calculated from Eq. (1) to maintain the dead zone location (z=zdz). To maintain the cure depth (not shown in the example), the required exposure time would be calculated from Eq. (1) for z=zcd. Concentrating the dose like this requires high intensity light sources, so the operational limits of the dual-wavelength printer should be considered when designing microfluidic devices and formulating the resin.Figure 7 illustrates an important limitation about the print depth of our vSLA method. Dual-wavelength vSLA is specifically suited to flat, slab-like designs commonly found in conventionally fabricated microfluidic devices. The light absorbing resin, used to prevent excessive light propagation and improve z resolution, prevents the timely accumulation of the critical dose at depths far into the resin. Thus, while our vSLA prints can be as broad as the projector allows, they are no more than a few mm tall.