Here, we present a novel experimental approach that enables transfer of vertically-aligned Si nanowires from their original Si substrate onto an optically transparent glass substrate while preserving the geometrical features. We demonstrate the potential of these substrates for label-free live-cell phase contrast imaging of dynamic cellular processes such as cell division, morphology change, and migration. This approach will advance our understanding of cellular responses to extracellular bio-physical cues— which in turn is likely to improve the design of future cellular manipulation technologies.
Figure 1 depicts in detail the nanofabrication pipeline for the SiNW array harvesting and subsequent transfer onto transparent glass substrates. First, e-beam lithography was used to directly write a customized lithographical mask array of nanoscale-circles—200 nm diameter, 1 µm pitch—on an electron-resist film coated onto a Si substrate. Second, a 50 nm thick Ni layer was evaporated and the resist layer lifted off, resulting in the formation of Ni disk-like shapes that served as programmable lithographical etching masks. Third, SiNW arrays were generated using DRIE, and the Ni etch-masks removed. This step was followed by deposition of a 30 nm thick Au-layer over the NW array using e-beam direct-angle evaporation. Fourth, a PMMA layer was deposited by spin coating, followed by immersion in an Au-etchant solution. This step is crucial for selectively removing the residual Au-layer on the top and sidewalls of the NWs. The remaining Au layer on the bottom of the SiNW substrate served as an anti-adhesion agent, which later promoted the NW arrays’ selective harvesting and their subsequent transfer from the Si donor substrate onto the acceptor transparent glass substrate. The PMMA resist layer served as a lower embedding layer that could be further removed after the imprint process. Fifth, a silicate-based UV-curable Ormostamp solution was cast on top of the donor substrate NW array; the acceptor glass substrate was then placed precisely above the SiNW arrays (sandwich-type interface), followed by a UV curing step. Sixth, once cured, the Si-donor and glass-acceptor substrates were mechanically separated by placing a razor blade between the samples (only at the corner, without touching the pillar array), while applying a small force.
Figure 1. Steps from the fabrication of programmed vertically configured SiNW arrays, and their post-detachment from the Si substrate and their subsequent printing onto the glass substrate.
(Figure 2a, b) shows zoom-out and zoom-in SEM images SiNW arrays that were generated using DRIE. (Figure 2c, d) shows optical and SEM images of the Si-donor substrate after SiNW array harvesting. The NW array was successfully transferred onto the glass substrate along with its embedded cured-Ormostamp/PMMA/Au layers. The NW array was exposed through immersion in Au-etchant solution followed by removal of the top PMMA layer via immersion in anisole solution. The lower layer of the exposed nanowires remained embedded in the optically-transparent Ormostamp. This process results in accurate, precise, and high-efficiency SiNW array transfer from Si to glass, preserving the original programmable design (Figure 2e-g; optical and SEM images). (Supporting information Figure S2) shows additional Si geometry onto an optically transparent substrate. (Supporting information Figure S3) shows incomplete NW transfer arrays onto the glass substrates, while (Supporting information Figure S4) shows destructive NW transfer on the glass substrates.
Figure 2. SiNW fabrication and their subsequent transfer onto an optically transparent substrate. (a–b) SEM images of the negatively tapered SiNW arrays fabricated via e-beam lithography and DRIE, (a) the zoom-out, and (b) the zoom-in. (c–d) Optical and SEM showing the Si donor substrate post-harvest. (e–g) Optical image showing SiNW transfer onto a glass substrate (e); and tilt SEM images of a zoom-out and zoom-in of the transferred SiNW array onto the glass substrate.
Arrays of SiNWs on glass substrates allows for the dynamic observation of cell behavior as a function of nanowire interaction over long time scales, (Figure 3, Supporting information movies 1-2). Using phase contrast microscopy, cellular features like the nucleus (Fig. 3b,d; blue arrows) and lamellipodia (Figure 3b,d; red arrows) can be observed in living MDA-MB-231 breast cancer cells growing on flat control (Figure 3a) or SiNW substrates (Figure 3c).
Figure 3. Phase contrast images of MDA-MB-231 breast cancer cells cultured for over 3 hours on flat glass (a, b) or transparent SiNW surfaces (c, d). Insets (b, d) illustrate the label-free localization of cellular nuclei (blue arrows) and lamellipodia (red arrows). Inset (e) illustrates the direct visualization of nanowire arrays by means of phase contrast microscopy.
As time lapse microscopy generates hundreds of label-free phase contrast images, the manual segmentation of individual cells is challenging. To address this, we trained a convolutional neural network to automatically identify and track cell outlines (Supporting information movie 3). To leverage the unique advantages of our transparent SiNW arrays, we focused on dynamic cellular characteristics that would be difficult or impossible to observe in fixed cells. While fully spread cells were observed on SiNW substrates (Figure 3d), we found that the presence of SiNWs results in significant decreases in cell area and cell velocity of MDA-MB-231 cells (Figure 4c, d), although the initial cell viability, as measured by the number of attached cells per mm2, was not affected (Figure S5). Interestingly, the number of cells undergoing division events on SiNW surfaces was significantly higher (p=0.014) than on control glass substrates, (Figure 4a).
This increased division frequency, which causes cells to round up, likely also accounts for the slight decrease in cellular aspect ratio on SiNW surfaces (Figure 4b). To date, nanowire structures have not been implicated in regulating the cell cycle, which highlights the unique void these transparent structures fill in the field of nanoscale material–cell interfaces. Future studies will likely leverage this ability to probe these phenomena, as well as other dynamic cell behaviours, like cell spreading, contact guidance, and collective migration. Given that the highly metastatic MDA-MB-231 cells used for this experiment feature a relatively fast doubling time, it is possible that the effect on division frequency may be tempered in other lines that divide less frequently. Perhaps most intriguingly, the ability to visualize the nanowire structures directly via phase contrast microscopy (Figure 3e) hints at the possibility of correlating directional cell behaviour, either at the whole cell scale area with respect to individual protrusions, to the alignment of the long-range nanowire patterns.
Figure 4. Cell behaviour as a function of SiNW substrate. For division frequency (a), each circle corresponds to the proportion of cells dividing in a single field of view. For all other plots (b–d), each circle corresponds to a single cell averaged over 24 h. Error bars represent 95% confidence intervals of the mean. *p<0.05.
This novel experimental approach enables harvesting of vertically configured SiNWs from a Si-donor substrate and their subsequent transfer onto a glass substrate with minimal artifacts. A considerable advantage of this approach, apart from high efficiency harvesting and subsequent transfer onto an optically transparent substrate, is compatibility with live-cell phase contrast imaging. We have demonstrated proof-of-concept of dynamic characterization of cellular morphology, division, and migration, without the need for immunocytochemical markers, on a nanowire-structured substrate. This approach opens a new dimension for the ongoing study of cellular responses to nanoscale extracellular biophysical cues, which is likely to facilitate the development of improved nanoscale cellular manipulation technologies.