Solution-Phase Photochemical Nanopatterning Enabled by High

Jun 15, 2017 - A high-throughput, solution-based, scanning-probe photochemical nanopatterning approach, which does not require the use of probes with ...
0 downloads 4 Views 9MB Size
Solution-Phase Photochemical Nanopatterning Enabled by High-RefractiveIndex Beam Pen Arrays Zhuang Xie,†,‡,∥ Pavlo Gordiichuk,†,‡,∥ Qing-Yuan Lin,‡,§ Brian Meckes,†,‡ Peng-Cheng Chen,‡,§ Lin Sun,‡,§ Jingshan S. Du,‡,§ Jinghan Zhu,‡,§ Yuan Liu,†,‡ Vinayak P. Dravid,‡,§ and Chad A. Mirkin*,†,‡,§ †

Department of Chemistry and ‡International Institute for Nanotechnology, Northwestern University, 2145 Sheridan Road, Evanston, Illinois 60208, United States § Department of Materials Science and Engineering, Northwestern University, 2220 Campus Drive, Evanston, Illinois 60208, United States S Supporting Information *

ABSTRACT: A high-throughput, solution-based, scanning-probe photochemical nanopatterning approach, which does not require the use of probes with subwavelength apertures, is reported. Specifically, pyramid arrays made from high-refractive-index polymeric materials were constructed and studied as patterning tools in a conventional liquid-phase beam pen lithography experiment. Two versions of the arrays were explored with either metal-coated or metal-free tips. Importantly, light can be channeled through both types of tips and the appropriate solution phase (e.g., H2O or CH3OH) and focused on subwavelength regions of a substrate to effect a photoreaction in solution that results in localized patterning of a self-assembled monolayer (SAM)-coated Au thin film substrate. Arrays with as many as 4500 pyramid-shaped probes were used to simultaneously initiate thousands of localized free-radical photoreactions (decomposition of a lithium acylphosphinate photoinitiator in an aqueous solution) that result in oxidative removal of the SAM. The technique is attractive since it allows one to rapidly generate features less than 200 nm in diameter, and the metal-free tips afford more than 10-fold higher intensity than the tips with nanoapertures over a micrometer propagation length. In principle, this mask-free method can be utilized as a versatile tool for performing a wide variety of photochemistries across multiple scales that may be important in high-throughput combinatorial screening applications related to chemistry, biology, and materials science. KEYWORDS: beam pen lithography, photochemistry, surface patterning, solution phase, self-assembled monolayer, free radical of the patterning tips,9,16,17 making it potentially useful for surface wettability control,18,19 printed electronics,20−22 biochips,23−25 3D printing,26−29 and cell-material interface studies.30,31 A major limitation of BPL pertains to the requirement for arrays of pyramid-shaped probes with nanoscale apertures.10 These apertures are difficult and costly to generate in a uniform manner. In addition, to create subwavelength feature dimensions in the solution phase,10 a subwavelength tip− substrate distance must be maintained during the patterning process, which is extremely challenging and makes it difficult,

S

canning probe lithographies have been developed to varying extents based upon the concept of using cantilever-based structures to deliver energy or materials (or both) to a surface to effect patterning in destructive or constructive modes, respectively.1−6 Beam pen lithography (BPL) is a cantilever-free scanning probe optical printing technique that utilizes arrays of millions of individually addressable pyramid-shaped polymer tips to direct light onto a surface.7−10 With BPL, subdiffraction-limited patterns can be produced over the centimeter length scale using low-cost instrumentation in a direct-write and massively multiplexed fashion.8,11 Moreover, BPL provides capabilities not only for controlling feature size but also chemical composition, thereby opening opportunities for combinatorial nanoscience, biology, and materials discovery.12−15 Finally, the technique allows one to simultaneously deliver materials (inks) and energy from each © 2017 American Chemical Society

Received: May 11, 2017 Accepted: June 8, 2017 Published: June 15, 2017 8231

DOI: 10.1021/acsnano.7b03282 ACS Nano 2017, 11, 8231−8241

Article

www.acsnano.org

Article

ACS Nano

Figure 1. (A) Schematic illustration of solution-phase photochemical patterning using arrays of pyramidal tips made from high-refractiveindex polymer, e.g., SU-8 (n = 1.65), with the green spheres indicating the dissolved reagents, e.g., photoinitiator (PI). Two tip configurations were examined to confine light at the subwavelength scale, metal-coated and metal-free ones, respectively. (B) Schematic illustration of tipdirected removal of a self-assembled monolayer (SAM) on a Au thin film substrate mediated by photogenerated free radicals (R•) in an aqueous photoinitiator solution. The patterned SAM can be further used to generate metal patterns by subsequent Au etching or molecular patterns by exchanging/backfilling with other thiol-containing molecules.

Figure 2. (A) Finite-difference time-domain (FDTD) simulations for light propagation (λ = 400 nm) in an aqueous environment (n = 1.34) through SU-8 tips (n = 1.65) with 100 nm Au coating (left) and without metal coating (right), respectively. The intensity scales are normalized to the intensity at the 500 nm propagation length in each figure so that intensity profiles below the tips are obvious. The top view images of light spots are obtained at the 500 nm propagation length. Scale bars: 500 nm. (B) Simulated light intensities normalized to the incident light as a function of the propagation length from the tip apex. (C) Scheme of the fabrication process for SU-8 tip arrays: (a) spin coating of SU-8 onto a 100 nm Au-coated silicon master followed by UV curing; (b) sandwiching the liquid PDMS precursor between the UVcured SU-8 layer and a glass slide, followed by PDMS curing at room temperature for 72 h; (c) peeling off the whole assembly from the silicon master to obtain the metal-coated tip arrays; (d) removing the Au coating at the top portion of the pyramids by selective Au etching to obtain the metal-free tip arrays. The edge length of the metal-free portions on the pyramidal tips can be tailored in step (d). (D) Scanning electron microscopy (SEM) image of a typical Au-coated SU-8 tip stripped from the Si master. (E) SEM image of a typical metal-free SU-8 tip after metal etching. (F) SEM image of arrays of SU-8 pyramids over large areas, with metal-free tips as shown in E.

8232

DOI: 10.1021/acsnano.7b03282 ACS Nano 2017, 11, 8231−8241

Article

ACS Nano

commercially available polymers, SU-8, an epoxy-based negative tone photoresist, meets the optical requirements (n > 1.65 for λ < 405 nm, transmittance >50% for λ > 350 nm).39 Moreover, SU-8 is chemically and mechanically stable and has been utilized by others to make scanning probes by a straightforward molding process.40 To explore the optical properties of pyramidal tips made of SU-8 in solution, we carried out finite-difference time-domain (FDTD) calculations of light transmission (λ = 400 nm) through the SU-8 tips (n = 1.65) in H2O (n = 1.34). In the modeling, two pyramidal tips with base edge lengths of 1.5 μm were evaluated, one coated with 100 nm Au and the other without the metal coating (Figure 2A). As predicted, in both of the apertureless configurations, light can be directed to the vicinity of the tip and propagates in the liquid medium with little lateral spreading (Figure 2A). Thus, focused light beams with full width at half maxima (fwhm) of ∼300 nm can be formed from the tip apex along a distance greater than the light wavelength. In contrast, conventional PDMS tips with the same tip configurations fail to focus light (Figure S1A). Our simulations also suggest that, in comparison with a thinner Au coating (40 nm), the 100 nm Au coating on the SU-8 tip allows less light leakage from the tip to the surrounding area (Figure S1B), leading to a circular and uniform light spot throughout the entire propagation path. More importantly, when the light intensity transmitted through the tips was examined, the metal-free design was found to exhibit a remarkable intensity enhancement. In Figure 2B, we plotted the transmitted light intensity normalized to the incident light with the propagation length. The metal-free SU-8 tip exhibits >20-fold more intensity than the incident light at the tip apex, and no intensity decay is observed within the initial 0.5 μm of propagation. Traveling beyond 0.5 μm, the light intensity starts to decay by a factor of ∼5 per 1 μm (Figure 2B). As the light passes through the 100 nm Au-coated tip, the light intensity is ∼30% of the incident light and exhibits exponential decay from the tip apex. The light intensity at the tip apex is lower than that through an SU-8 tip with a 400 nm aperture, which is ∼200% of the incident light. Nevertheless, at the propagation length beyond 500 nm, the light intensity is comparable to or even higher than that through the probe with a 400 nm aperture, and the light spot from the apertureless tip is smaller as well (Figure S1B). These optical simulations led us to hypothesize that arrays of such apertureless SU-8 tips are candidates for high-resolution photopatterning in solution without the need for nanoapertures. Furthermore, the metalfree tip architecture could significantly improve the uniformity of light transmission across the entire array due to the minimized beam intensity decay within 1 μm of propagation path, in addition to providing a means for amplifying light intensity over a long propagation length (∼2 μm) beyond the near field (∼400 nm). To test the hypotheses, we fabricated metal-coated tip arrays made of SU-8 by employing a template-stripping method (Figure 2C, steps a−c, and Figure S2A).40 In a typical experiment, a hydrophobic silicon master used for fabricating pyramidal tip arrays41 was first coated with a 100 nm thick Au layer by evaporation. This metal coating served as both a separation layer and an opaque coating on the final tip arrays. Next, an SU-8 layer was spin-coated over the Au-coated master and exposed to UV light for photo-cross-linking. The SU-8 forms the tips that define each array. Then liquid PDMS precursor was added on top of the backs of the tips followed by a glass slide (1 mm thick) support. After the soft PDMS

and in certain cases impossible, to write uniform submicrometer features over large areas.32,33 Furthermore, writing speed must be extremely slow due to the low transmission efficiency and rapid power dissipation at the near field when the light passes through a tiny aperture.10,34 These facts make it especially difficult to effect many solution-phase photoreactions, such as free-radical transformations, which require >100 times the optical energy typically utilized to process commercial photoresists.35,36 In the scanning near-field photolithography community, improvements in patterning speed have been made by using short-wavelength light (λ < 300 nm) or focused laser sources.14 However, these instrumental improvements are difficult to apply to BPL, since the polymer-based pen arrays block deep UV light or undergo degradation after prolonged irradiation. Also, the high-power light sources limit the illumination area and increase the cost and safety risks. Herein, we introduce a materials-based approach that allows the pyramidal tips in a BPL system to focus light in solution at the subwavelength scale without the need for the tedious fabrication and manipulation of nanoscopic apertures. The innovation is made possible by designing pyramid arrays made from high-refractive-index polymer materials. Two versions of the arrays were explored, ones with either metal-coated or metal-free tips. Specifically, we characterized the optical properties of these tips in the solution phase (e.g., H2O and CH3OH) and evaluated their performance as patterning tools in a conventional liquid-phase BPL experiment (Figure 1A). Arrays of such tips were exploited to effect a free-radical photoreaction in solution that results in localized patterning of a self-assembled monolayer (SAM)coated Au thin film substrate by oxidative removal of the SAM (Figure 1B). Finally, this solution-phase SAM patterning approach was applied to the rapid generation of arbitrary patterns over macroscopic areas at submicrometer resolution, which illustrates the potential for using it to generate architectures important to many fields, spanning electronics, materials science, chemistry, and biology.

RESULTS AND DISCUSSION Prior advances in scanning near-field photolithography and BPL in air have explored the use of two different types of apertureless pyramidal tips or tip array architectures: (1) a hollow silicon nitride tip or a polyurethane tip array with, in both cases, the tips fully coated with a metal layer that allows partial transmission of light through the apex of the tip(s),14,37,38 and 2) a polydimethylsiloxane (PDMS) tip array without a metal coating on the tips.9 Both architectures are able to direct light to a surface at a resolution comparable to architectures containing tips with subwavelength apertures.7,34 Importantly, even in the case of the tip without a metal coating, the incident light can be predominately focused at the tip apex by total internal reflection inside the pyramid due to a large difference in the refractive indices (n) between the polymer tip and the surrounding environment (air). To enable a similar type of patterning capability in solution, one must identify materials and solvents with the appropriate difference in refractive indices. Using a critical incident angle of 54.7° derived from the pyramidal shape, we determined by Snell’s law for refraction that a critical refractive index of 1.65 for the tip material would be required to allow total internal reflection in common solvents such as H2O and CH3OH (n < 1.34). Under these conditions, one should be able to focus light to subwavelength dimensions using the pyramidal tips. Among 8233

DOI: 10.1021/acsnano.7b03282 ACS Nano 2017, 11, 8231−8241

Article

ACS Nano

Figure 3. (A) Dark-field optical microscope image of pattern arrays of Au hole features generated by metal-coated SU-8 tips. These tips were used to induce the removal of an 11-mercaptoundecyl tri(ethylene glycol) (EG3) SAM on a Au thin film substrate under 365 nm light (∼0.25 W cm−2) in the presence of 34 mM LAP photoinitiator aqueous solution. The subsequent Au etching resulted in optically observable features. In each array, the relative tip heights (Ztip) were 0, 0.5, and 1 μm from bottom to top, and the exposure times were 2 and 3 min from left to right. (B) Plot of feature diameter with Ztip with the size measured from 3 × 2 dot arrays written by five tips. (C) Atomic force microscopy (AFM) image of an array of Au features with the smallest size of ∼200 nm at Ztip = 1 μm.

backing layer was cured, the whole assembly was peeled from the master. Arrays of pyramids with a smooth metal coating were obtained, as seen from the scanning electron microscopy (SEM) images (Figure 2D−F). Furthermore, to generate metal-free tips (Figure 2C, step d), we removed the metal layer at the top portion of the pyramids following the previously reported etching procedures.9 Using this protocol, we have prepared an apertureless pyramid tip array made of metal-free SU-8 tips with an average tip radius of curvature of 80 ± 20 nm (Figure 2E). The edge length of the metal-free portions on the pyramids can be tuned by the spin coating step prior to metal etching. The SU-8 tip array after removing all the metal coatings was optically transparent, with a measured transmittance of ∼50% at 365 nm and >80% for λ>400 nm (Figure S2B). In order to investigate light propagation through the asfabricated SU-8 tip arrays in solution, we used a series of monolayer-coated Au thin film substrates as a model platform. Such monolayers, consisting of 16-mercaptohexadecanoic acid

(MHA), 1-dodecanethiol (DT), or (11-mercaptoundecyl)tri(ethylene glycol) (EG3), can be destructively patterned with light through a solution-phase free-radical process involving lithium phenyl(2,4,6-trimethylbenzoyl)phosphinate as the photoinitiator (hereafter designated LAP for lithium acylphosphinate). In the presence of photogenerated free radicals, the surface-bound thiolate groups can be oxidatively desorbed.14,42 The short lifetime of the free radicals (ns scale)43 results in a diffusion length of less than 50 nm, allowing one to effect highly localized surface reactions. As an initial test, we performed conventional photolithography through a photomask on a SAM-modified Au substrate, which was covered with an aqueous solution containing the LAP photoinitiator (1 wt %, 34 mM) followed by Au etching with a solution consisting of 20 mM thiourea, 30 mM iron nitrate, 20 mM hydrochloric acid, and 2 mM octanol in water. Regardless of the choice of SAM, patterns of etched holes on Au were produced with different dosages of light (Figure S3A). Among them, EG3 required the lowest dosage (∼10 and ∼45 J cm−2 under 365 and 405 nm 8234

DOI: 10.1021/acsnano.7b03282 ACS Nano 2017, 11, 8231−8241

Article

ACS Nano

Figure 4. (A) Optical microscope image of Au pattern arrays produced by metal-free SU-8 tip arrays. In each array generated by a single tip, the exposure time was 20, 40, 60, and 80 s in each row from left to right, and the relative tip height (Ztip) was 0 and 1 μm for the bottom and top row, respectively. (B) AFM images of a typical array written by a metal-free tip with a 1.5-μm edge length, in which a subwavelength feature was generated at 20 s of exposure and Ztip = 1 μm. (C) AFM image of a 190 nm feature produced by the metal-free tip at Ztip = 1 μm. (D) Plot of feature diameter as a function of exposure time. The statistics were obtained from arrays produced by 9 metal-free tips with an average edge length of ∼1.5 μm. (E) Feature diameter growth curves for dot arrays generated by metal-free tips with different edge lengths ranging from 1.7 to 5.6 μm in a single experiment. All the patterning experiments were conducted with 34 mM LAP and 365 nm light (∼0.1 W cm−2).

light, respectively) followed by DT (∼50 J cm−2 under 405 nm light) and MHA (∼100 J cm−2 under 405 nm light). Without photoinitiators in solution, no features were generated, as evidenced by the Au etching procedure even at a 15 times higher dosage. X-ray photoelectron spectroscopy (XPS) data showed that the Au etching begins before the SAM is completely removed, as evidenced by the presence of the S 2p signature at 162 eV after irradiation (Figure S3B). To determine the potential of the apertureless tip arrays for effecting free-radical photochemistry at the nanoscale, we first studied the metal-coated tip arrays (∼2 × 2 mm2, ∼ 400 pyramidal tips, each with a 40 × 40 μm base, 100 μm pitch), and our ability to use them to pattern EG3 SAM-modified Au substrates. The tip array was mounted onto a scanning probe system (XE-150, Park Systems) and optically leveled with respect to the substrate.41 Then the LAP aqueous solution (34 mM) was injected between the tip array and the substrate. Next, the pen array was programmed to write dot arrays with exposure times of 1, 2, and 3 min and tip heights (Ztip) ranging from 1.5 to 0 μm (three 0.5 μm steps) relative to the tip contact point. During patterning, the backside of the tip array was illuminated with UV light (365 nm, ∼0.25 W cm−2) to initiate the photoreactions. As a control, we also wrote dot arrays in the dark to see whether the tip contact resulted in the mechanical removal of the SAM. The control patterns never exhibited Au etching, except for some marks due to leveling when the tips were compressed with >2 μm z extensions beyond tip contact. For the experiments carried out under UV exposure, etched Au features with submicrometer sizes were observed only after 2 min of exposure and when the tip− substrate gap was less than 1 μm (Figure 3A). We found that

around 1/3 of the tips were able to generate 3 × 2 dot arrays with sizes of 760 ± 70, 500 ± 80, and 310 ± 30 nm at Ztip = 0, 0.5, and 1 μm, respectively (Figure 3B). The smallest feature size of ∼200 nm could be patterned at Ztip = 1 μm (Figure 3C). In the rest of the patterning area, dot patterns were not written completely at heights Ztip = 1 μm or Ztip = 0.5 μm due to greater separation distances between tip and substrate across the pen array caused by tip height variation and imperfect optical alignment with the substrate, which generally result in a variation of 1−2 μm in height. The whole patterning area was ∼1 mm2, since only a portion of the array was close enough to the substrate to initiate surface reactions, a problem due to poor alignment of the array. Compared with photolithography through a mask (40 s of exposure under ∼0.25 W cm−2), patterning with the metal-coated tips required an exposure time ∼3 times longer, which can be attributed to the optical energy loss during the light transmission through the tip array and the Au coating. In addition, we also explored a methanolic photoinitiator solution containing 34 mM phenylbis(2,4,6trimethylbenzoyl)phosphine oxide. With this solution, an exposure time of 30 s was required to generate dot features of 650 ± 100 nm with the same irradiation power (Figure S4) due to the higher absorption efficiency of this photoinitiator, compared to that of LAP. These experiments not only demonstrate the light focusing effect of the metal-coated SU8 tips in appropriate solutions, but also confirm that the freeradical reactions are spatially confined with little loss of resolution. Similar to the tips with nanoapertures, the metalcoated tips exhibit near-field light transmission behavior in accordance with the simulation results (Figure 2A,B), which makes alignment for the array with the substrate over large 8235

DOI: 10.1021/acsnano.7b03282 ACS Nano 2017, 11, 8231−8241

Article

ACS Nano

incident light. To evaluate generality, we also carried out patterning experiments with 405 nm light (∼0.18 W cm−2) and determined that the shortest exposure time required to obtain dot features at Ztip = 1 μm was 30 s. Again, a ∼8-fold enhancement in light intensity at the focal point was observed (250 s of exposure if using the incident light). The experimental results roughly fit with the simulation that predicts a 10-fold intensity enhancement at the 1-μm propagation length for the 1.5-μm metal-free tip (Figure 2A,B). The difference may be due to the light absorption by the SU-8 material and the photoinitiator solution, as well as the reflection from the Au surface. But overall the transmitted light is significantly amplified through the metal-free SU-8 tip, and the beam intensity can be at least 1 order of magnitude higher compared with the metal-coated tip or the tip with subwavelength apertures. Next, we investigated the relationship between exposure time and resulting feature size by plotting the diameter of each feature as a function of exposure time (Figure 4D, 1.5 μm tips). When the tip was relatively far from the substrate (Ztip = 1 μm), the average diameter of the feature slowly increased from 280 ± 95 to 500 ± 120 nm between 20 and 40 s but increased more rapidly at longer exposure times, with the largest size being 1620 ± 210 nm at 80 s. The slow size increase at the beginning suggests that thiols are removed gradually within the region of the focused light at early stages. After the reaction was initiated, the rate of feature growth increased because of increased chemical accessibility to the damaged monolayer. In addition, the resulting dot features grow with diameters exceeding a micrometer, a consequence of optical edge effects (see simulation, Figure 2A). As the tip was brought closer to the substrate (Ztip = 0 μm), in which the propagated light intensity was higher, the feature size rapidly increased as a function of exposure time. For example, over the span of 60 s, the feature diameter increased from 420 ± 70 to 1580 ± 240 nm, also a consequence of such effects. However, after 60 s, growth in feature size begins to slow down, a result of the decrease in photoinitiator concentration in the local illuminated areas, as well as the attenuation of light intensity in the lateral propagation directions. In addition, the focused light intensity positively correlates with the edge length of the metal-free portions, which leads to variable feature growth from tip to tip (each has a different edge length). We also investigated patterning under 365 nm light (∼0.1 W cm−2) where metal-free tips ranging from 1.7 to 5.6 μm were examined in a single experiment. Over the span of the 60 s exposure, growth of feature size was also nonlinear, consistent with the aforementioned observations, especially for tips with edge lengths greater than 3.5 μm (Figure 4E). With an increasing edge length for the metal-free tips, the required exposure time to obtain features decreased and the growth rate in the maximum growth rate regime increased. For instance, a 1.7 μm tip required 20 s of exposure to produce features with diameters in the 200−300 nm range, while for a neighboring 4.6 μm tip, it only took 10 s to create similar feature sizes (Figure S5C). Apart from the optics, the concentration of the dissolved reagents can be a vital factor that affects photochemical patterning. To study the role of photoinitiator concentration on the patterning process, we explored a solution-exchange strategy to realize patterning under different LAP solutions on one substrate with the same pen arrays. In a typical experiment, we first introduced a drop of 3.4 mM LAP solution on the substrate and generated a dot array using increasing

areas of paramount importance. Indeed, with this architecture, large-area patterning of uniform features is difficult. In addition, the throughput is relatively slow (∼1/3 of the light transmission efficiency through the tip). Nevertheless, this architecture eliminates the steps for fabricating arrays with nanoapertures to generate features with subwavelength dimensions. Next, we evaluated the performance of the metal-free SU-8 tips, which were expected to increase the transmitted light intensity and light propagation length. As a test-case system, a ∼2 × 2 mm2 metal-free tip array with ∼400 pyramids (40 μm base, 100 μm pitch) was fabricated (Figure 2C, a−d, and Figure 2F). Within one array, we varied the edge length of the metalfree portions of the pyramid tips (1.5−5.6 μm) by nonuniform spin coating prior to the etching step used to make the metalfree tips. Notably, when one compares the optical properties of the array of SU-8 metal-free tips with a conventional PDMS polymer pen array, there are some significant differences. With the SU-8 tips, when they are brought close to the surface in water and subsequently irradiated with yellow light (an LED with a yellow filter) from the backside, by optical microscopy one can clearly see well-resolved tip apexes, while in the case of PDMS such tips appear blurry (Figure S5A), a consequence of the superior light focusing properties of the SU-8 in this medium. This observation is quite consequential in that light focusing enables the accurate determination of tip−substrate distance by determination of the tip apexes, which is critical for the optical leveling of BPL arrays in liquid media. With the metal-free SU-8 tip arrays, lithography experiments were conducted on an EG3-modified Au substrate following the procedures described above (34 mM LAP, 365 nm light, ∼0.1 W cm−2). Two rows of dots were produced by each tip, while the tip was within 500 nm of touching the substrate (bottom row, Ztip = 0 μm) and was elevated 1 μm away from the substrate for the top row (Ztip = 1 μm). Within each row, the exposure time was set at 20, 40, 60, and 80 s, left to right, respectively (Figure 4A). After photopatterning, the Au substrate can be etched to more clearly visualize the results of the patterning process and also confirm that >95% tips over the entire area were functional to generate dot features at both Ztip = 0 and 1 μm (Figure 4A). In contrast, metal-free PDMS tips, when used in a similar fashion, produced microring patterns in solution (Figure S5B). Therefore, the SU-8 tips yield significant advantages in terms of resolution, throughput, and feature structure and uniformity due to focused light at the tips. In order to attain a quantitative comparison with our previous simulations (Figure 2A), we characterized several dot arrays produced by the 1.5-μm metal-free tips by AFM (Figure 4B). One can clearly see that exposure time and tip−substrate distance can be used to finely control the feature size, with subwavelength dimension features possible at 20 s exposure times with Ztip = 1 μm (280 ± 95 nm). For comparison purposes, if Ztip = 0 μm, 20 s of exposure results in 420 ± 70 nm diameter features. The smallest features made to date at Ztip = 1 μm are ∼190 nm (also 20 s exposure time) (Figure 4C), but this is difficult to control and more a consequence of tipheight variation within a single array, an observation that points toward the need to produce higher quality, more uniform arrays. Based upon exposure times (20 s with the metal-free tip and 100 s with the incident light in the absence of the tip) needed to pattern the SAM-coated substrate, we estimate that the light transmitted through the SU-8 metal-free tips at Ztip = 1 μm has an intensity at the focal point ∼5 times higher than the 8236

DOI: 10.1021/acsnano.7b03282 ACS Nano 2017, 11, 8231−8241

Article

ACS Nano

Figure 5. (A) SEM image of dot features generated with different exposure times and LAP concentrations of 3.4, 10, and 34 mM, respectively. In each row, the dots were patterned with increasing exposure time from left to right, as indicated by the arrow. The light intensity was ∼0.25 W cm−2 at 365 nm. (B) Plot of feature diameter with exposure time under different LAP concentrations. The statistics were obtained from five arrays.

Figure 6. Large-area patterning with metal-free SU-8 tips. (A) Optical microscope image of arrays of submicrometer Au features over large areas, with exposure times of 5−20 s. Inset: Histogram of the feature sizes at 15 s of exposure across a ∼ 2 × 2 mm2 patterning area. RSD: relative standard deviation. (B) Schematic illustration of generation of DNA arrays by molecular exchange followed by DNA-directed nanoparticle assembly. (C) Optical microscope image of large-area patterns of the assembled Au nanoparticles with dot and line features. (D) Typical SEM image of the patterned lines consisting of Au nanoparticles. The left line was written with faster scanning speed (0.4 μm s−1) compared with the right (0.2 μm s−1), leading to a lower density of nanoparticles. (E) Typical SEM image of the assembled nanoparticles within dot features, with the smallest feature size of ∼500 nm generated at 10 s of exposure. All of the patterning experiments were conducted with 10 mM LAP, 365 nm light, and a light intensity of ∼0.1 W cm−2.

exposure times from 30 to 180 s (365 nm, ∼0.25 W cm−2). We then lifted the pen array, withdrew the first solution, and added the second LAP solution with a higher concentration (10 mM). Patterning was performed again with exposure times between 15 and 120 s, in which the same tip wrote patterns in the regions next to the previous ones. This process was repeated one more time to pattern the SAM in a 34 mM LAP solution

with exposure times between 5 and 60 s. A typical dot array written by a single tip as a function of three different LAP concentrations clearly shows the amount of LAP can be used in conjunction with exposure time to control feature size (Figure 5A). By analyzing the SEM images, one can see that the lower LAP concentration gives rise to smaller feature sizes at constant exposure times (Figure 5B). For example, at 30 s of exposure, 8237

DOI: 10.1021/acsnano.7b03282 ACS Nano 2017, 11, 8231−8241

Article

ACS Nano the feature size decreased from 2490 ± 340 to 350 ± 180 nm as the LAP concentration was reduced from 34 mM to 3.4 mM. Interestingly, for 3.4 and 10 mM LAP cases, as the exposure time was lengthened, the feature size increased initially, but finally decreased, for example, at 120 s exposure for 10 mM and 180 s exposure in 3.4 mM (Figure 5). This phenomenon can be attributed to the depletion of photoinitiators over relatively large distances (10 μm in this case) due to diffusion of photoinitiators from the surrounding space to the tipilluminated regions. Thus, after one area is exposed with light and patterned, there will be insufficient photoinitiators for patterning the next position. The depletion area of photoinitiators correlates with the illumination dosage as well as the photoinitiator concentration; for example, the decrease in feature size for subsequent features generated at higher exposure times (120 and 180 s) disappears in the 10 and 3.4 mM cases with decreasing light intensity (Figure S6A,B), indicating the local depletion distance is less than 10 μm at these dosages. In addition, in the 3.4 mM LAP case, the 120 s exposure time prevented the patterning of the next feature over a distance of 10 μm, but it showed less influence on the next feature as the distance between features increased to 20 μm. However, with 240 s of exposure, the next feature could not be generated at a site 20 μm away from the first feature (Figure S6C). These observations support our hypothesis that the photoinitiator concentration can be dynamically varied at local areas with a radius of tens of micrometers around the tip by photodecomposition and diffusion of photoinitiators, which could contribute to the feature size change during the BPL process, especially in the low concentration and high dosage cases. Because of the continuous writing nature of BPL, in future development, a continuous-flow system might be useful for controlling the reactant concentration as well as the feature size in a dynamic way. Having established the localized free-radical reactions conducted by the SU-8 tips, the massive metal-free tip arrays in combination with the rapid photochemistry can be applied to generate patterns of various materials across multiple length scales and at high throughput. Importantly, high uniformity in feature size can be also enabled over large areas, since the metal-free tips can be fabricated with uniform edge lengths easily (size variation