Fabrication and modelling of fractal, biomimetic, micro and nano-topographical surfaces

Natural surface topographies are often self-similar with hierarchical features at the micro and nanoscale, which may be mimicked to overcome modern tissue engineering and biomaterial design limitations. Specifically, a cell’s microenvironment within the human body contains highly optimised, fractal topographical cues, which directs precise cell behaviour. However, recreating biomimetic, fractal topographies in vitro is not a trivial process and a number of fabrication methods have been proposed but often fail to precisely control the spatial resolution of features at different lengths scales and hence, to provide true biomimetic properties. Here, we propose a method of accurately reproducing the self-similar, micro and nanoscale topography of a human biological tissue into a synthetic polymer through an innovative fabrication process. The biological tissue surface was characterised using atomic force microscopy (AFM) to obtain spatial data in X, Y and Z, which was converted into a grayscale ‘digital photomask’. As a result of maskless grayscale optical lithography followed by modified deep reactive ion etching and replica molding, we were able to accurately reproduce the fractal topography of acellular dermal matrix (ADM) into polydimethylsiloxane (PDMS). Characterisation using AFM at three different length scales revealed that the nano and micro-topographical features, in addition to the fractal dimension, of native ADM were reproduced in PDMS. In conclusion, it has been shown that the fractal topography of biological surfaces can be mimicked in synthetic materials using the novel fabrication process outlined, which may be applied to significantly enhance medical device biocompatibility and performance.


Introduction
Bio-inspired, functionalized surfaces, containing hierarchical features at the micro and nano-topographical scale, may significantly enhance device performance [1][2][3]. As advanced fabrication techniques are expanding the boundaries of nanoscale surface design, we are increasingly seeking inspiration from nature to overcome and progress from modern manufacturing Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
limitations [4]. Many natural surfaces contain complex, hierarchical, topographical features, which have evolved over billions of years, to perform a specific function exceptionally efficiently [5]. Applying the concept of bio-mimicry has previously led to the design and fabrication of a diverse variety of advanced devices, such as super-adhesive materials, self-cleaning materials, miniature flying machines, soft-flexible robotics and biomimetic implant surfaces which promote implant biocompatibility [6][7][8][9]. The gecko foot, for example, is covered in millions of 0.2-0.5 μm diameter spatula-shaped projections, which interact with the underlying surface through van der Waals forces, allowing the Gecko to hang upside down on almost any surface due to a remarkable adhesive force of 10 N cm 2 [10,11]. Attempts at mimicking the exceptional adhesive properties of the Gecko foot in polyimide have been attempted, through a fabrication method using E-beam lithography and dry etching with oxygen, with potential applications in robotics, sports clothing, healthcare and the military [12][13][14].
In contrast to binary photolithography which produces planar structures, grayscale photolithography utilises a variable-dose exposure, which controls development depth and allows the fabrication of 3D structures [34]. During grayscale photolithography, a UV-sensitive photoresist is exposed to a controlled, variable dose of UV-light, which is dictated by the grayscale level contained in the pattern. The dose corresponds with penetration depth, volume of crosslinked photoresist and feature depth after development (for positive tone resists) [26]. Grayscale photolithography is typically performed using grey-tone masks (GTM's) and is a high-throughput process primarily used to mass produce a final object of interest, despite disadvantages including substantial mask optimisation and limited flexibility [35].
In contrast, maskless grayscale photolithography, performed using a direct laser-writing system (laser lithography) and a computer generated grayscale 'digital mask' is a more flexible, cost and time effective fabrication process if the aim is to manufacture a template from which the mass production of the desired surface can be made [35].
Devices fabricated to date using maskless 3D grayscale photolithography include spherical microlens arrays [25], cantilevers, miniature bridges [26], pyramids and miniature buildings [36], however, to the best of our knowledge, maskless 3D grayscale lithography has not yet been used to produce hierarchical, 3D, micro and nano-topographical templates for high-throughput duplication of biomimetic topographies through replica molding.
Thus, in this work, an innovative grayscale fabrication technique was developed and optimised, which utilised three dimensional digital data, obtained by atomic force microscopy (AFM) of a biological surface as a grayscale 'digital mask'. Maskless grayscale photolithography was followed by deep reactive ion etching (DRIE), to fabricate a hard template, and replica molding, to produce polydimethylsiloxane (PDMS) topographical surfaces which were replicas of the native biological surface; thereby creating a biomimetic synthetic surface (figure 1).
As a proof of concept, the hierarchical, micro and nano-topographical features of acellular dermal matrix (ADM) are replicated in PDMS. ADM is an allogenic, decellularized, extracellular matrix (ECM) protein construct containing precisely optimised topographical cues which promote wound healing and minimisation of the foreign body reaction when utilised in vivo [37,38]. Replicating the topographical cues contained in ADM into PDMS may significantly enhance implant performance while reducing complications, as cells may be less likely to develop an acute foreign body reaction towards an implant surface it recognises 'as self'. Biomimetic surface topography may significantly enhance silicone biocompatibility through modification of cell attachment, proliferation, differentiation and attenuated foreign body reaction [39][40][41][42].
Lastly, data gathered during quantitative characterisation of ADM was used to model an ADM surface in MATLAB. This further proof of concept project aimed to determine whether it is possible to create a computer generated model of the ADM surface, containing similar roughness values and fractal properties to the native ADM images obtained through AFM, but without needing to perform the AFM measurements. Collecting a sufficient number of high quality AFM images was a significant rate limiting step in the fabrication of the biomimetic surfaces, which could be ameliorated through the application of a modelled surface. The ability to accurately model biological surfaces may facilitate quick and efficient upscaling of the grayscale technology outlined in this paper to fabricate biomimetic implant surface topographies at industrial scale.

Materials and methods
2.1. Fabrication of ADM PDMS surfaces 2.1.1. Characterising ADM using AFM ADM was imaged using a Bruker Dimension Icon® AFM. Samples were imaged using ScanAsyst™ Air probes (silicon nitride, nominal k=0.4 N m −1 , tip radius=2 nm) and conducted in ScanAsyst™ mode. The method of preparing ADM for characterisation can be found in supplementary data. Peak Force Tapping™ (PFT) amplitude was 150-100 nm, and PFT frequency was 1 kHz. Scan rates varied between 0.5 and 1 Hz. Images were taken with 512 samples per line and using a Z-limit of 15 μm. A large, intact area of ADM was imaged through obtaining numerous 90×90 μm 2 AFM scans using offsets in the X and Y direction. Scans were performed in at least three different areas of the ADM sample and on three different patient samples. Further details on optimising the use of AFM for the collection of reliable biological topographical data can be found in supplementary data and Supplementary figures S1 and S2.

Creating a grayscale 'digital mask' of ADM for photolithography
Using the stitching feature within Mountain Maps® 7 imaging software (Digital Surf © , France) numerous, adjacent 90×90 μm 2 AFM images of ADM (figure 2(a)) were stitched together, to create a large area pattern of ADM. The montage was produced using X and Y offsets to correctly align images; thereby forming a large intact area of ADM without leaving stitch lines (figures 2(c) and (d)).
As the minimum resolution of the laser lithography system used was 0.5 μm in X and Y, images were re-scaled prior to exposure, through averaging the heights of surrounding pixels. For example, if exposing a single 90×90 μm 2 grayscale AFM image ( figure 2(b)), the pattern is re-scaled to 180 pixels per line, prior to exposure and in the laser lithography system, pixel size is set at 0.5 μm in X and Y.
A two-dimensional (2D) topographical AFM image of ADM, where colour within the image represents height data, can be converted to a grayscale 'digital mask', where each pixel is assigned a grayscale level which corresponds relatively to a feature height on the biological surface. To achieve this, the ADM montage was converted to an 8 bit grayscale image, consisting of 256 grayscale levels, using the open source scanning probe analysis software Gwyddion (http:/gwyddion. net/) which could then be read by a laser lithography system (figures 2(b) and (c)).

Silicon wafer preparation
The following fabrication protocol was optimised for the particular equipment used and further details can be found in the patent application [43]. All processing was carried out in a class 100 clean room, to ensure the surfaces remained free from airborne contamination. A 2×2 cm 2 plain silicon wafer was sonicated for 5 min each in acetone, deionised water and isopropyl alcohol, dried with a stream of dry nitrogen gas and dehydrated on a hot plate set at 200°C for 10 min Resist adhesion was promoted using hexamethlydisilazane (HMDS) which was spun onto the wafer at 4000 RPM for 60 s.

Lowering photoresist contrast, improving exposure linearity and optimisation
Immediately following the application of HMDS, positive tone photoresist S1813® was spun onto the silicon wafer at 4000 RPM for 60 s using a photoresist spinner, producing a thickness of 1.3 μm.
Next, the soft-bake temperature was optimised at 72°C for 1 min 30 s to reduce photoresist contrast and improve exposure linearity. Further details on lowering photoresist contrast can be found in supplementary data and supplementary figure S3.
Following exposure, the resist was developed in MF-319® developer solution for 30 s with gentle agitation followed by 30 s in deionised water to stop the Figure 2. Creating a grayscale digital mask of acellular dermal matrix (ADM) for maskless grayscale photolithography. First, a 90×90 μm 2 atomic force microscopy (AFM) scan of ADM was obtained (a), which was converted to an 8 bit (256 levels) grayscale image (b), and after stitching of numerous AFM images, formed an ADM grayscale montage for exposure (c). A three dimensional (3D) colour representation of the ADM montage is shown in (d). Scale bar=20 μm. Reproduced from [64]. development and then dried with a gentle stream of nitrogen gas.
A grayscale wedge design was used to optimise exposure dose/development linearity of the grayscale photolithography process (supplementary figures S4 and S5).

Maskless grayscale 3D photolithography
Maskless grayscale photolithography was performed using a laserwriter (Microtech Laserwriter LW405). The prepared grayscale bitmap image was loaded into the instrument and the pixel size was set at 0.5 μm in X and Y. A gallium nitride (GaN) solid state laser, operating at 405 nm wavelength (h-line), was used with a 40× objective having 0.65 NA (table 1). A laser power of 57.7 mW and a filter of 1% were applied. An effective exposure dose of 0 J cm −2 was assigned to black pixels (pixel 0, no exposure) and a dose of 0.11 J cm −2 was assigned to white pixels (pixel 256, maximum exposure), with corresponding doses within this dose range assigned to each pixel according to the grayscale pattern. The grayscale levels are created through different lengths of laser pulses of between 5 and 100 nanoseconds.
2.1.6. Pattern transfer into silicon using modified DRIE to create master template An Oxford Plasmalab 100 ICP65 System (Oxford Instruments, UK) deep reactive ion etcher (DRIE), running a modified Bosch process recipe (discussed in results) was used to permanently transfer the exposed ADM pattern from the photoresist into the silicon wafer, which subsequently acted as a template to produce PDMS stamps through replica molding. Increased isotropic etching and lower etch selectivity were required for the accurate transfer of the ADM pattern from the photoresist into the silicon. Therefore, etch selectivity was optimised through the addition of an oxygen (O 2 ) only step to a recipe containing optimised pressures, gases, flow rates, RF and ICP powers, step times and repeats. The modified DRIE parameters can be found in table 2.

Replica molding to create biomimetic silicone surfaces from master template
The PDMS used to create ADM PDMS surfaces for characterisation was kindly donated by Mentor Corporation (Mentor Corporation, Texas, USA) and was the same medical grade mixed dimethyl dispersion used to manufacture the Company's commercially available implant surfaces. The silicon master was vapour treated with a silanizing agent, Trichloro (1H, 1H, 2H, 2H-perfluorooctyl) silane (FDTS) for 10 min in a desiccator under vacuum. This reduced surface free energy in order to ease the release of PDMS from the silicon master. The PDMS was spun onto the silicon wafer (100 RPM) and de-gassed in a desiccator for 1 h to remove any bubbles and to aid the transfer of features within the silicon template into the PDMS. To cure and crosslink the PDMS, it was baked for 18 h in an oven at 80°C. A 1.5×1.5 cm 2 PDMS stamp containing the fabricated surface topography of ADM was cut out using a scalpel and characterised.

Surface characterisation
The grayscale fabrication process was optimised using an exposed grayscale wedge (supplementary figures S4 and S5) and was characterised using a contact profilometer (Veeco Dektak stylus profiler). Native ADM and ADM PDMS samples were quantitatively characterised using AFM as described previously. All samples were qualitatively characterised using SEM after coating with 2 nm chromium (Cr) and 12 nm of gold (Au). Imaging was performed on a Carl Zeiss ULTRA PLUS SEM system using a working distance of 2.7 mm and EHT voltage level of 7-10 kV.

Modelled ADM surface generated in MATLAB
The significant rate limiting step in the fabrication of the biomimetic surface was obtaining AFM images of the ADM surface. Modelling an ADM surface which has the same topographical, fractal and roughness values as native ADM, would accelerate the fabrication process while also facilitating efficient upscaling and distribution of the technology to the manufacturing industry.
Therefore, as proof of concept, using the quantitative data gathered from the comprehensive characterisation and analysis of ADM, a model ADM surface was generated using MATLAB code derived from Kroese and Botev [44]. The surface was modelled, as a fractional Brownian process, based on the intrinsic embedding method proposed by Stein [45]. A fractal dimension of 2.3 was used to reflect that obtained for the reference ADM surface. Several surfaces were generated to test the method, but only the surface with statistical parameters closest to that obtained from the ADM data shown in this work is presented.

Results
3.1. Exposure of ADM pattern into photoresist through maskless grayscale photolithography A 90×90 μm 2 grayscale AFM image of ADM was exposed, etched and replica moulded to demonstrate the optimised grayscale fabrication process, in a proof of concept approach.
The grayscale ADM pattern to be exposed is illustrated (figure 3(a.i)); alongside a corresponding line profile of the ADM image (figure 3(a.ii)), as indicated by the white dashed line. The Sz value of the ADM pattern is 3.2 μm.
The exposed ADM pattern in photoresist is shown ( figure 3(b.i)); alongside the corresponding line profile ( figure 3(b.ii)). The grayscale ADM pattern was deliberately inverted in photoresist ( figure 3(b)), and is returned to correct profile after replica molding of the template. The grayscale photolithography parameters were optimised so that features present in the native ADM image were scaled down when exposed into photoresist. It was necessary to scale-down the features of ADM in the photoresist, as the feature sizes present in ADM (∼3-6 μm) were greater than the thickness of the photoresist (1.3 μm). Therefore, the Sz value of the exposed ADM pattern in photoresist is 0.35 μm.

DRIE step to transfer ADM pattern from photoresist into silicon
The etch selectivity during DRIE has to be tailored to the feature size present in the original image and to the feature sizes exposed into photoresist. In this example, an etch selectivity of 9.2:1 was required to scale-up ADM features exposed in photoresist to their original size in the native ADM pattern.
The etched ADM pattern in silicon after DRIE is shown ( figure 3(c.i)); alongside the corresponding line profile ( figure 3(c.ii)). The Sz value is 3.4 μm which indicates an etch selectivity of 9.6:1 was achieved using the optimised DRIE recipe. The ADM pattern transferred into silicon had a Sz value of 3.4 μm which is very similar to the native ADM Sz value of 3.

Characterisation of ADM PDMS surfaces and comparison to native ADM
ADM PDMS surfaces were characterised and compared with native ADM, to evaluate the accuracy of the grayscale fabrication technique to reproduce ADM features in PDMS and is summarised in table 3. Further analysis of native ADM and discussion of the fractal properties can be found in supplementary data. The grayscale fabrication technique was able to reproduce the topography of ADM at micro and nano length scales, which can be seen in figure 5.
In comparisons at 90×90 μm 2 , the Sa (arithmetic mean) values of ADM PDMS (figure 5(b)) were within 5 nm of native ADM ( figure 5(a)), while the Sz (maximum peak to valley distance) value was within 655 nm. The excess Sku (kurtosis) and Ssk (skewness) were all ∼0 in ADM PDMS surfaces, which is representative of native ADM, indicating that in addition to the reproduction of ADM roughness, topographical features have also been replicated. The FD (fractal dimension) value of ADM PDMS at 90×90 μm 2 was 2.29, which is the same value as in native ADM and indicates that the fractal properties of the surface at this length scale have also reproduced.
At 10×10 μm 2 , the topography of ADM was also accurately mimicked in PDMS and the Sa value of ADM PDMS ( figure 5(d)) was within 27 nm of native ADM ( figure 5(c)), while the Sz value was within 200 nm. Again, excess Sku and Ssk were approximately 0, as they are in native ADM. The FD value of ADM PDMS at this length scale was 2.27, in comparison to 2.28 in native ADM. Lastly, at 1×1 μm 2 , the Sa value of ADM PDMS (figure 5(f)) was within 1 nm of native ADM (figure 5(e)), while the Sz value was within 2 nm. The excess Sku and Ssk values were all close to 0, as they are in native ADM, and the FD value is 2.25 in ADM PDMS at this length scale, in comparison to 2.29 in native ADM.  Figure 6(a.i) shows a model surface of ADM generated as a fractional Brownian process based on the intrinsic embedding method proposed by Stein [45], adjacent to the native ADM surface on which it was modelled This figure demonstrates that it's possible to accurately model the ADM surface, which possesses similar roughness and topographical characteristics as the native ADM surface. Therefore, the modelled ADM grayscale image could replace the native AFM image of ADM as the grayscale pattern exposed into photoresist, thus removing the need to obtain AFM images of the surface. Utilising the modelled surface could enable a significantly faster fabrication process, in addition to efficient upscaling and distribution of the grayscale technology outlined in this paper to the manufacturing industry.

Discussion
Inspired by the concept of bio-mimicry, ADM topography was reproduced in silicone through an innovative 3D, grayscale fabrication technique. This carefully optimised and characterised process is capable of high throughput reproduction of 3D, fractal, nano and micro-scale biomimetic topographies in synthetic surfaces.
For the first time, X, Y and Z spatial data acquired through AFM was used to expose a biological surface topography into a positive-tone photoresist, using maskless grayscale photolithography; followed by pattern transfer into silicon through a modified DRIE recipe. A master template was created and replica molding was used to produce PDMS stamps containing 3D biomimetic topography ( figure 7).
silicone implant surfaces [64]. The biomimetic surface holds potential as being able to reduce the acute inflammatory cellular response towards it, which requires further in vivo confirmation in the future.
Maskless grayscale photolithography followed by DRIE and replica molding offers a number of advantages for the fabrication of biomimetic micro and nano textured surfaces. The resolution of features produced by maskless photolithography is in theory only limited by the diffraction limit of the wave-length of light as the numerical apertures used when fabricating miniaturised devices is typically >0.5 [65]. Therefore, in general, the smallest feature which can be reproduced is equivalent to, or slightly smaller than, the wavelength of light used [65]. Sub-micron features are possible with precise optimisation of the whole optical lithography process including sample preparation, soft-bake, photoresist contrast and development. Furthermore, the maskless system provides the flexibility to adjust and optimise the grayscale 'digital mask' design, with minimal cost and time, which is a significant advantage over the use of a GTM for fabricating templates. Lastly, maskless grayscale photolithography is a single exposure method which makes processing time quicker than multi-exposure systems or E-beam lithography, which is a high resolution, but slow and expensive patterning process [35].
However, grayscale fabrication processes require a low contrast photoresist which develops linearly with exposure dose, to allow more grey levels to be realised and the fabrication of varying relief structures (supplementary figures S3 and S4). To achieve this, the softbake temperature was reduced from 100°C to 72°C, and soft-bake time from 3 min to 1 min 30 s. Reducing the soft-bake time and temperature results in increased solvent retained in the photoresist. This increases the amount of dark erosion (areas of photoresist which are removed at low dose) during development; thereby lowering the contrast [66]. This allows 3D structures to be exposed in the photoresist as the feature height after development corresponds with exposure dose, which corresponds with grayscale level. (Further discussion on lowering photoresist contrast for the fabrication of 3D features is provided in supplementary data.) Finding a balance between exposure dose and development time is also important to enable the accurate reproduction of micro and nano-topographical features into thin (1.3 μm) resists. The penetration depth of light is greater than resist thickness in thin resists, even at lowest exposure dose, and overdevelopment results in the loss of resolution. Therefore, the development time was correlated with the optimised resist contrast, to achieve the fabrication of high resolution features at the correct scale. In the final optimised process, the development time was shortened to 30 s from 40 s, to account for the increased dark erosion as a result of the cooler/shorter soft-bake.
Features of ADM were scaled-down to accommodate them in resist during photolithography and were subsequently scaled-up to their original size during DRIE. In the demonstrated example, an etch selectivity of 9.2:1 was calculated (etch rate of silicon: etch rate of photoresist) to scale features back up to their original size. However, the Bosch process was designed to fabricate high-aspect ratio features, capable of etch selectivity's greater than 75:1, which is significantly greater than what is required here [67]. Etch selectivity during DRIE has previously been reduced through the addition of an oxygen (O 2) only step which increases the photoresist etch rate [23,24,68]. O 2 plasma reacts with organic material in the photoresist forming CO, CO 2 and H 2 O, which are readily removed from the surface [69]. Furthermore, these reactions are exothermic, which increases the temperature at the sample surface, further increasing reaction rate [23]. Therefore, it was decided to include a 3 s oxygen only step at a flow rate of 30 sccm, between etch and passivation steps, which reduced etch selectivity to 9.6:1. Furthermore, an unacceptable amount of 'scalloping' was observed with the first DRIE recipes, which was diminished through decreasing the silicon etch step time (SF 6 ) from 6 to 3 s, which reduces the amount of silicon etched each step [70][71][72]. The oxygen only step may also have contributed to reduced scalloping [73,74]. The etch rate was also reduced as a result of the shorter etch step time and therefore the number of repeats was increased from 30-40 to 80-100 to achieve the desired feature depth (Etch rate 0.25 μm min -1 ). Further discussion on the application of DRIE to fabricate nanoscale topographies can be found in supplementary data. Replica molding is an established technique used to create replicas from a hard master template in a soft polymer based material and has been shown to replicate features down to 1.5 nm in PDMS [33,75]. Using this technique ADM topography was reliably reproduced with characterisation in PDMS with characterisation demonstrating no loss of features above the lithography resolution limit.
Interestingly, characterisation of ADM revealed a fractal, self-similar surface containing micro and nano-scale hierarchical features. The nature of ADM self-similarity suggested there was potential to model the ADM surface, allowing features to be modified (either enhanced or reduced), depending on the requirements of the fabricated surface. Therefore, as proof of concept, the spatial data gathered from the comprehensive characterisation of ADM, was input into modelling software (MATLAB) and using MATLAB code derived from Kroese and Botev [44], a computer generated replica of ADM topography was produced, which could be used for grayscale fabrication. The ADM surface was accurately modelled in MATLAB and possessed the same fractal dimension (2.3) as the reference surface ( figure 6). This potentially replaces the need to use grayscale AFM images of biological surfaces and facilitates faster fabrication speeds in addition to effortless distribution to industry and large scale manufacturing. Furthermore, the flexibility offered by a modelled surface enhances the ease at which this technology can be implemented and the number of potential applications.
The primary application of this technology is the enhancement of medical device performance, through reproducing biomimetic, fractal, ECM topographical features onto synthetic prosthesis surfaces, which may encourage implant integration while reducing foreign body reaction as has been demonstrated in previous studies [76,77]. Medical implant biocompatibility is of great current interest and significance to the National Health Service, as an ageing population increases the demand for tissue replacements [78]. Thus, the design and manufacture of novel, micro and nano-topographical implant surfaces, which improve tissue integration, reduce complications and enhance outcomes, are of paramount importance in improving long term patient healthcare [79]. However, the techniques described in this paper can be used to fabricate biomimetic, nano and micro-scale topographical surfaces for any application.

Conclusions
For the first time, a biomimetic, 3D, nano and microtextured silicone surface has been fabricated utilising the X, Y and Z spatial data of a biological surface (ADM), obtained through AFM. The rendered 2D topographical grayscale pattern was precisely exposed into an optimised photoresist using maskless grayscale photolithography. The pattern was successfully transferred into silicon using a modified version of the Bosch process and was used to create PDMS stamps through replica moulding. This comprehensively characterised and optimised process can fabricate hard Figure 7. The similarity between native acellular dermal matrix (ADM) topography and the biomimetic ADM topography in polydimethylsiloxane (PDMS), replicated through the innovative grayscale fabrication process outlined in this paper, is highlighted in this figure. The grayscale 90×90 μm 2 atomic force microscopy (AFM) image of native ADM is shown in (a) and after it is has been reproduced in PDMS (b). In combination with the scanning electron microscopy (SEM) image in (c), it is obvious to see the accuracy of the novel grayscale fabrication technique, optimised in this work, to create biomimetic topographies in silicone. Scale bar=20 μm.
templates containing precisely controlled 3D, biomimetic features at a faster rate, with more flexibility and at lower cost, than conventional grayscale photolithography requiring a mask. Potential applications include significantly enhancing microelectronic and medical device performance.