Research Article www.acsami.org
Ultrahigh Resolution Titanium Deep Reactive Ion Etching Bryan W. K. Woo,†,# Shannon C. Gott,†,#,¶ Ryan A. Peck,† Dong Yan,‡ Mathias W. Rommelfanger,§ and Masaru P. Rao*,†,∥,⊥ Department of Mechanical Engineering, ‡Center for Nanoscale Science and Engineering, §Central Facility for Advanced Microscopy and Microanalysis, ∥Department of Bioengineering, and ⊥Materials Science and Engineering Program, University of California, Riverside, Riverside, California 92521, United States
ACS Appl. Mater. Interfaces 2017.9:20161-20168. Downloaded from pubs.acs.org by UNIV OF LOUISIANA AT LAFAYETTE on 01/22/19. For personal use only.
†
ABSTRACT: Titanium (Ti) represents a promising new material for microelectromechanical systems (MEMS), due to its unique combination of advantageous physical and mechanical properties. Recently, realization of this promise has been made possible with the advent of processes that enable deep reactive ion etching (DRIE) of high-aspect-ratio (HAR) structures in bulk Ti substrates. However, to date, these processes have been limited to minimum feature sizes (MFS) ≥750 nm. Although this is sufficient for many applications, MFS reduction to the deep submicrometer range opens potential for further device miniaturization and an opportunity for endowing devices with unique functionalities that are derived from precisely defined structures within this length scale regime. Herein, we report results from studies seeking to create means for realizing such opportunities through extension of Ti DRIE to the deep submicrometer scale. The effects of key process parameters on etch performance were investigated, and the understanding gained from these studies informed the development of a new ultrahigh resolution (UHR) Ti DRIE process. Using this process, we demonstrate, for the first time, fabrication of HAR structures in bulk Ti substrates with 150 nm MFS, smooth vertical sidewalls (88°), good etch rate (587 nm/min), and mask selectivity (11.1). This represents a fivefold or greater improvement in MFS relative to our previously reported processes and a 29-fold or greater improvement over more recent processes reported by others. As such, the UHR Ti DRIE process extends the state-of-the-art considerably, and it opens important new opportunities for Ti MEMS, particularly in the implantable medical device realm. KEYWORDS: titanium, deep reactive ion etching, high-aspect-ratio, nanopatterning, nanofabrication
1. INTRODUCTION Although the materials originally used for microelectromechanical systems (MEMS) were derived from those used in the integrated circuit industry, breakthroughs over the past several decades have enabled the broadening of this selection considerably.1 Titanium (Ti), in particular, has become a promising candidate for MEMS, due in large part to its high fracture toughness, which reduces the risk of catastrophic fracture-based failure compared with more conventional MEMS materials, such as silicon, and other similarly brittle compounds based thereon. The high modulus, strength, solvent resistance, and thermal stability of Ti also provide an opportunity for enhanced performance relative to common polymeric MEMS materials [e.g., polydimethylsiloxane (PDMS) and poly(methyl methacrylate) (PMMA)], whereas its excellent corrosion and fatigue resistance make it more favorable for harsh environment applications than common metallic MEMS materials (e.g., Cu, Ni, Fe, & Al). Finally, its relatively low cost and excellent biocompatibility make Ti an ideal material for MEMS devices intended for use in vivo. With the recent advent of novel micromachining techniques that enable deep reactive ion etching (DRIE) of Ti, opportunity now exists for exploring the potential embodied in this material for MEMS applications. The first of these techniques was © 2017 American Chemical Society
reported by Aimi et al., who developed a cyclic Cl2/O2 etch/ passivation scheme known as the metal anisotropic reactive ion etching with oxidation (MARIO) process.2 Although this process allowed fabrication of high-aspect-ratio (HAR) structures with vertical sidewalls in bulk Ti substrates for the first time, its broader utility was limited by low etch rates (0.5 μm/min) and the relatively isotropic nature of the etch cycle, which limited the minimum feature sizes (MFS) to 1 μm or greater. Soon after, Parker et al. reported an improved process that employed a continuous Cl2 etch scheme known as the Ti inductively coupled plasma deep etch (TIDE) process. This process enabled fabrication of HAR structures with MFS as small as 750 nm in bulk Ti substrates, with smoother sidewalls, higher etch rate (2 μm/min), and comparable mask selectivity (40:1 Ti/TiO2).3 Since then, a number of other processes for Ti DRIE have been reported by others.4−6 Collectively, these techniques have begun to enable the exploration of Ti MEMS in a wide variety of applications, including microneedles for transdermal and ocular drug delivery,7,8 vascular stents,9 neural prosthetic interfaces,10,11 resonant sensors,12 and microfluidic Received: December 22, 2016 Accepted: May 23, 2017 Published: May 23, 2017 20161
DOI: 10.1021/acsami.6b16518 ACS Appl. Mater. Interfaces 2017, 9, 20161−20168
Research Article
ACS Applied Materials & Interfaces devices for dielectrophoretic particle manipulation,13 biomolecular network studies,14,15 thermal ground planes,16 and photocatalytic microreactors.17 However, although the capabilities of the Ti DRIE processes reported to date have been sufficient for exploration of numerous applications, none of these processes are able to achieve MFS less than 750 nm (e.g., Figure 1 illustrates results
resolution (UHR) Ti DRIE process are made with previously reported processes, and the implications of these new capabilities for implantable medical device design are discussed.
2. EXPERIMENTAL SECTION Polished, 200 μm thick, commercially pure (CP) grade 1 Ti bulk polycrystalline substrates were used throughout the study (Tokyo Stainless Grinding Co.). All substrates were patterned with an array of sixteen 5 mm × 5 mm gratings, each with equal linewidth and spacing (484 mm2 total substrate area). A fixed linewidth was used for each grating, and linewidths were varied from 150 nm to 50 μm across the array. Gratings with 200 nm linewidth were used for the etch parameter study, whereas 150 nm linewidth gratings were used for demonstration of the UHR Ti DRIE process capability. All gratings were fabricated using the process described below. The substrates were first cleaned by ultrasonic agitation in acetone and isopropanol baths, followed by deionized water rinse and N2 drying. A 200 nm thick SiO2 etch mask was then deposited using inductively coupled plasma, plasma-enhanced chemical vapor deposition (VLR ICP-PECVD, Unaxis; process conditions: 2 Pa chamber pressure, 400 W ICP source power, 5 W substrate power, 5.9 sccm SiH4, 20 sccm Ar, 10 sccm O2, and 100 °C lower electrode temperature). The substrates were then dehydrated on a hot plate at 115 °C for 5 min, followed by O2 plasma ashing (PEII-A, Technics; process conditions: 100 W substrate power and 40 Pa chamber pressure). A thermoplastic resist (mr-I 7020, micro resist technology) was then applied to the substrates, and the grating patterns were transferred into the resist using nanoimprint lithography (NX2000, Nanonex; process conditions: 140 °C and 500 psi imprint temperature and pressure) and a silicon master that was previously fabricated using deep ultraviolet (DUV) stepper-based photolithographic patterning and fluorine-based anisotropic dry etching. An oxygen-based dry etch was then used to remove the residual resist layer at the bottom of the trenches (E626I, Panasonic Factory Solutions; process conditions: 0.5 Pa chamber pressure, 75 W ICP source power, 65 W substrate power, and 49 sccm O2). This was followed by a fluorine-based dry etch to transfer the grating patterns into the SiO2 mask (E626I, Panasonic Factory Solutions: process conditions: 0.5 Pa chamber pressure, 900 W ICP source power, 50 W substrate power, 20 sccm CHF3, and 20 sccm CF4). Once the etch masks were defined, the grating patterns were transferred into the substrate under varying Ti DRIE process conditions (E626I, Panasonic Factory Solutions). On the basis of our earlier experience in developing the original TIDE process,3 the chamber pressure, the chlorine flowrate, and the oxygen flowrate were targeted as the most relevant process parameters for the study, and a preliminary baseline condition was established with the following parameters to reduce the aggressiveness of the etch and enhance sidewall passivation: 0.5 Pa chamber pressure, 400 W ICP source power, 100 W substrate power, 40 sccm Cl2, 4 sccm O2, and 10 °C lower electrode temperature. During each etch, one parameter was varied about the baseline value, while all others were held constant. The etch time was also held constant at 1 min, across all conditions studied, to minimize the influence of depth-dependent etching effects, for example, aspect ratio-dependent etching (ARDE) resulting from transport limitations within deep narrow trenches.20 All substrates were attached to 150 mm Si carrier wafers using diffusion pump oil (5P Ultra, Santovac). The ductility of Ti precluded the cross-sectioning of the deep etched gratings by substrate cleavage. Consequently, etch characterization was performed using focused ion beam (FIB) milling (Quanta 3D 200i, FEI). Immediately before cross-sectioning, a carbon-based protective film was selectively deposited on a small portion of the gratings using ion beam-assisted deposition in the FIB. Scanning electron microscopy (SEM) was used to image the cross-sectioned gratings (Supra 55, Leo). It was also used to image the floors of 100 μm wide trenches located immediately adjacent to the gratings, which were defined concurrently during the fabrication process. Measurements of the Ti etch depth and the remaining SiO2 mask thickness
Figure 1. SEM micrograph of 200 nm linewidth grating (400 nm pitch) etched using the TIDE process. The bright suspended features in the foreground are SiO2 etch mask fragments. Severe undercutting and feature loss are evident and indicative of overly aggressive conditions for defining HAR deep submicrometer scale features.
from the etching of 200 nm linewidth gratings using the TIDE process). This represents an important limitation because the continuing drive for increased performance and reduced cost necessitates further device miniaturization and hence further reduction in MFS. Moreover, within the context of implantable medical devices and minimally invasive surgical instruments based upon Ti MEMS, motivation for additional miniaturization lies in the opportunity this affords for further reducing invasiveness and hence reducing pain, procedural complexity, and recovery time. Finally, motivation for additional Ti DRIE MFS reduction also lies in the opportunity this may provide for endowing implantable medical devices with unique and highly compelling functionalities. For example, as we have reported earlier, endothelial cell responses are enhanced considerably when Ti substrates are patterned with precisely defined, grating-based topographies, particularly as the MFS is reduced just below the micrometer range (i.e., MFS = 750 nm).18 Because endothelial cells are key contributors to the healing response after implantation of vascular devices (e.g., coronary stents), this suggests potential for rationally designing the microscale surface topography of such devices to mitigate adverse physiological responses associated with delayed healing that are commonly encountered after implantation (e.g., restenosis & late-stent thrombosis).19 As such, this motivates the pursuit of further Ti DRIE MFS reduction deep into the submicrometer realm to explore the ultimate limits of the improvements that can be achieved. Herein, we report results from studies seeking to extend the Ti DRIE process capability to the deep submicrometer scale. Selected process parameters, including chamber pressure, chlorine flowrate, and oxygen flowrate, were varied individually to determine their effects on the Ti DRIE process. The resulting trends were then used to develop an optimized DRIE process that enables, for the first time, fabrication of HAR deep submicrometer scale structures with smooth vertical sidewalls in bulk Ti substrates. Comparisons of this new ultrahigh 20162
DOI: 10.1021/acsami.6b16518 ACS Appl. Mater. Interfaces 2017, 9, 20161−20168
Research Article
ACS Applied Materials & Interfaces
Figure 2. SEM micrographs of 200 nm linewidth grating cross-sections (top row) and floors of adjacent wide trenches (bottom row), showing the effect of chamber pressure at (a) 0.50; (b) 1.00; and (c) 1.50 Pa. All etches were 1 min, and the remaining Ti DRIE parameters were held constant at the baseline conditions (400 W source, 100 W substrate, 40 sccm Cl2, 4 sccm O2, and 10 °C lower electrode). The bright layer atop each grating is a FIB-deposited carbon film that was used to mitigate sputtering-induced faceting during milling. The inset in the first image identifies the SiO2 etch mask and the carbon film that is deposited on the sidewalls of the deep etched Ti trenches during the FIB carbon deposition process. The vertically oriented contrast variations beneath the trench floors are “curtain effect” artifacts produced by differential sputtering during FIB milling.
Figure 3. Effect of chamber pressure on (a) Ti and SiO2 mask etch rates within the 200 nm linewidth gratings; (b) rms floor roughness in the wide trenches adjacent to the gratings; and (c) extent of grating feature sidewall bowing, assessed using a semiquantitative scoring scheme ranging from 0 to 5, where 0 represents no bowing and 5 represents full undercutting and feature loss arising from severe bowing. The remaining Ti DRIE parameters were held constant at the baseline conditions (400 W source, 100 W substrate, 40 sccm Cl2, 4 sccm O2, and 10 °C lower electrode). Data = mean ± standard deviation (n ≥ 5). The error bars are not visible in some cases because of the ordinate axis scale. were made at five different locations across each cross-sectional image and averaged (ImageJ, NIH). These measurements were then used to calculate the Ti and SiO2 mask etch rates for each etch. In addition, the extent of grating feature sidewall bowing was assessed semiquantitatively from the cross-sectional images to facilitate comparisons across various etch conditions. A five-point scale was employed, with 0 representing no bowing and 5 representing bowing sufficiently severe to cause complete undercutting and feature loss. Finally, root mean square (rms) floor roughness was measured within the 100 μm wide trenches adjacent to the gratings using a surface profilometer (Dektak 8, Veeco). These measurements were made using 300 μm line scans that were collected at 13 different locations across each sample and then averaged. Efforts were made to avoid measurements in areas with obvious micromasking; however, in some cases, this was impossible because of the density of micromasking in those specific areas. Measurement of the floor roughness within the 200 nm grating trenches themselves was not possible because of their diminutive dimensions and HARs.
the wide trench floors appear to show a trend of increasing roughness within the grains with increasing pressure. Figure 3a quantifies the variation of the Ti etch rate with chamber pressure within the gratings, which is observed to increase with pressure from 809 nm/min at 0.50 Pa to 1215 nm/min at 1.50 Pa, thus corroborating the visual assessment from the SEM micrographs (Figure 2). Figure 3a also shows that the SiO2 mask etch rate is largely constant at ∼50 nm/min across the pressure range studied. Collectively, this results in an increasing trend of Ti/SiO2 mask selectivity with pressure, ranging from 15.2 at 0.50 Pa to 23.3 at 1.50 Pa. Figure 3b quantifies the variation of floor roughness within the adjacent wide trenches. Roughness is observed to generally increase with pressure, ranging from 14 nm rms at 0.50 Pa to 33 nm rms at 1.50 Pa, thus corroborating the visual assessment from the SEM micrographs (Figure 2). Because the roughness measurements were based on line scans that typically traversed many grains, the reported values encompass contributions from roughness within the grains themselves and roughness associated with slight preferential etching of the grain boundaries (as seen in Figure 2). Figure 3c shows semiquantitative assessments of the grating feature sidewall bowing that increases significantly with increasing pressure, nearing full undercutting at 1.50 Pa. Over the pressure range studied herein, the observed trends are consistent with those reported in the earlier TIDE study.3 For example, both show that the Ti etch rate increases with
3. RESULTS 3.1. Effects of Chamber Pressure. Figure 2 shows the SEM micrographs of the grating cross-sections and the adjacent wide trench floors at the low-, mid-, and high-end of the chamber pressure range (i.e., 0.50, 1.00, and 1.50 Pa, respectively). Etch depths within the gratings are observed to significantly increase with pressure. However, this comes at the expense of increasing sidewall bowing, with many features nearing complete undercutting at 1.50 Pa. The micrographs of 20163
DOI: 10.1021/acsami.6b16518 ACS Appl. Mater. Interfaces 2017, 9, 20161−20168
Research Article
ACS Applied Materials & Interfaces
Figure 4. SEM micrographs of 200 nm linewidth grating cross-sections (top row) and floors of adjacent wide trenches (bottom row), showing the effect of Cl2 flowrate at (a) 20; (b) 40; and (c) 60 sccm. All etches were 1 min, and the remaining Ti DRIE parameters were held constant at the baseline conditions (0.5 Pa chamber pressure, 400 W source, 100 W substrate, 4 sccm O2, and 10 °C lower electrode).
Figure 5. Effect of Cl2 flowrate on (a) Ti and SiO2 mask etch rates within the 200 nm linewidth gratings; (b) rms floor roughness in the wide trenches adjacent to the gratings; and (c) extent of grating feature sidewall bowing. The remaining Ti DRIE parameters were held constant at the baseline conditions (0.5 Pa chamber pressure, 400 W source, 100 W substrate, 4 sccm O2, and 10 °C lower electrode). Data = mean ± 1 standard deviation (n ≥ 5). The error bars are not visible in some cases because of the ordinate axis scale.
pressure, which may be indicative of increased chemical etching resulting from increased reactive species density and/or scattering.21 This is corroborated by the increase in sidewall etching with increasing pressure, as seen in both studies. Moreover, the mask etch rate is observed to be only minimally affected by pressure in both studies, as would be expected, because the etching of the mask materials used in both studies (i.e., SiO2 in the current study and TiO2 in the former) is expected to be more dependent upon physical rather than chemical processes. Finally, although the etch rates observed herein are lower than those reported in the TIDE study, this too is unsurprising because the current study utilizes far lower levels of Cl2, thus reducing the overall aggressiveness of the etch. The Ti etch rate in the current study could also be further suppressed by ARDE effects because of diminutive trench dimensions relative to the earlier TIDE study. 3.2. Effects of Chlorine Flowrate. Figure 4 shows the SEM micrographs of the grating cross-sections and the adjacent wide trench floors at the low-, mid-, and high-end of the Cl2 flowrate range (i.e., 20, 40, and 60 sccm, respectively). Etch depths within the gratings are observed to increase initially with the flowrate, albeit to a lesser extent than seen in the pressure study. However, unlike the pressure study, there is little, if any, apparent difference in the etch depth between the mid-range and highest flowrates. Also apparent is the reduced effect of Cl2 flowrate on sidewall bowing and floor roughening compared with pressure. Finally, slight micromasking is observed at the lowest flowrate, most of which, but not all, is localized at the grain boundaries. Figure 5a quantifies the variation of the Ti etch rate within the gratings, which is observed to initially increase from 650
nm/min at 20 sccm Cl2 to 810 nm/min at 40 sccm, and then remains relatively unchanged at higher flowrates. This corroborates the earlier visual assessment of reduced etch depths relative to the pressure study and the minimal change in the etch depth beyond 40 sccm Cl2. Figure 5a also quantifies the effect of Cl2 flowrate on the SiO2 mask etch rate within the gratings, which is observed to increase from 50 to 67 nm/min as the flowrate increases from 20 to 60 sccm. This trend contrasts that seen in the pressure study, where the mask etch rate was largely unaffected. Collectively, this results in a Ti/ SiO2 mask selectivity trend that first rises from 12.9 at 20 sccm Cl2 to 15.2 at 40 sccm, followed thereafter by decline to 12.7 at 60 sccm. Figure 5b quantifies the variation of floor roughness within the adjacent wide trenches, which shows a slight decrease initially with increasing Cl2 flowrate from 16 nm rms at 20 sccm to 15 nm rms at 30 sccm, after which it remains relatively constant. This corroborates the earlier observation of slight micromasking at the lowest flowrate and lower overall roughness than the pressure study. Figure 5c shows the semiquantitative assessments of the grating feature sidewall bowing, which increases slightly with the Cl2 flowrate, but generally falls below the levels seen in the pressure study. Over the Cl2 flowrate range studied herein, the observed trend for the Ti etch rate is consistent with the earlier TIDE study. Specifically, both studies show that the Ti etch rate increases with the Cl2 flowrate, which is indicative of increased chemical etching arising from increasing reactive species density. This is corroborated by the increase in sidewall etching seen in both studies, albeit to a lesser extent than that seen with increasing pressure. The mask etch rate trend 20164
DOI: 10.1021/acsami.6b16518 ACS Appl. Mater. Interfaces 2017, 9, 20161−20168
Research Article
ACS Applied Materials & Interfaces
Figure 6. SEM micrographs of 200 nm linewidth grating cross-sections (top row) and floors of adjacent wide trenches (bottom row), showing the effect of O2 flowrate at (a) 2; (b) 4; and (c) 6 sccm. All etches were 1 min, and the remaining Ti DRIE parameters were held constant at the baseline conditions (0.5 Pa chamber pressure, 400 W source, 100 W substrate, 40 sccm Cl2, and 10 °C lower electrode).
Figure 7. Effect of O2 flowrate on (a) Ti and SiO2 mask etch rates within the 200 nm linewidth gratings; (b) rms floor roughness in the wide trenches adjacent to the gratings; and (c) extent of grating feature sidewall bowing. The remaining Ti DRIE parameters were held constant at the baseline conditions (0.5 Pa chamber pressure, 400 W source, 100 W substrate, and 40 sccm Cl2, and 10 °C lower electrode). Data = mean ± standard deviation (n ≥ 5). The error bars are not visible in some cases because of the ordinate axis scale.
nm/min at higher flowrates. This contrasts the trends seen in both the pressure and Cl2 studies. Collectively, this results in a Ti/SiO2 mask selectivity trend that first rises from 7.9 at 2 sccm O2 to 18.7 at 5 sccm, followed thereafter by a slight decline to 17.6 at 6 sccm. Figure 7b quantifies the variation of floor roughness with O2 flowrate in the adjacent wide trenches. Roughness remains relatively constant at 14 nm rms across most of the flowrate range, and then increases markedly to 24 nm rms at the highest flowrate, thus quantifying the effect of micromasking observed in the earlier visual assessment (Figure 6). Figure 7c shows the semiquantitative assessments of the grating feature sidewall bowing, which quickly decreases with increasing O2 flowrate.
observed herein, however, differs from the earlier TIDE study. Although the mask etch rate in the current study increases with the Cl2 flowrate, it is largely unaffected in the TIDE study. This therefore suggests that the SiO2 masks used in the current study are more susceptible to chemical attack by Cl2 than the TiO2 masks used in the former study. Finally, as discussed earlier, the lower etch rates seen in the current study can be attributed to the less aggressive etch conditions overall and ARDE effects. 3.3. Effects of Oxygen Flowrate. Figure 6 shows the SEM micrographs of the grating cross-sections and the adjacent wide trench floors at the low-, mid-, and high-end of the O2 flowrate range (i.e., 2, 4, and 6 sccm, respectively). Similar to the pressure and Cl2 flowrate studies, the etch depths within the gratings are observed to increase initially with the O2 flowrate. However, unlike the pressure and Cl2 studies, sidewall bowing is reduced with increasing O2 flowrate, and a transition to positive sidewall tapering becomes apparent at the highest flowrate. Finally, unlike the Cl2 study, micromasking is observed at the highest flowrate, with greater extent and a more uniform distribution. However, floor roughness appears to be otherwise minimally affected by the O2 flowrate. Figure 7a quantifies the variation of the Ti etch rate with the O2 flowrate within the gratings, which is observed to first increase from 655 nm/min at 2 sccm to 877 nm/min at 5 sccm and then slightly decrease to 825 nm/min at the highest flowrate. This corroborates the earlier visual assessment of initial increase in the etch depth, followed by tapering-induced etch depth limitation. Figure 7a also quantifies the effect of O2 flowrate on the SiO2 mask etch rate within the gratings, which is observed to decrease from 83 nm/min at 2 sccm to 53 nm/ min at 4 sccm, followed thereafter by a further decline to 47
4. DISCUSSION As discussed in the original TIDE report,3 Cl2-based dry etching of Ti is generally presumed to proceed by ionization and/or dissociation of molecular Cl2 to atomic Cl, followed by reaction with the Ti substrate to form volatile etch products such as TiCl4.22,23 Furthermore, etch directionality is typically ascribed to the minimization of ion scattering, which minimizes sidewall bombardment and thus lateral etching.24 However, although the elucidation of the mechanisms underlying Ti DRIE is beyond the scope of the current effort, it is unlikely that the highly anisotropic etching observed herein can be attributed to ion directionality alone. We speculate that an ioninhibitor DRIE mechanism, similar in nature to that reported for deep etching of Si in continuous F/O2-based plasmas,25 may be operative as well. Specifically, we suspect that lateral etching is minimized through the formation of a passivating film on the trench sidewalls, whereas vertical etching is facilitated by 20165
DOI: 10.1021/acsami.6b16518 ACS Appl. Mater. Interfaces 2017, 9, 20161−20168
Research Article
ACS Applied Materials & Interfaces preferential ion bombardment on the trench floors, which may (a) remove the passivation film or inhibit its formation; (b) enhance the chemical etching of the Ti substrate by damaging the surface and facilitating reactant species adsorption and dissociation;23,26 and/or (c) facilitate the removal of minority reaction products with otherwise limited volatility under the given etch conditions.26 Drawing further analogy to F/O2-based Si DRIE processes, we speculate that the sidewall passivation film could form by any number of means, including (a) oxidation of the exposed Ti; (b) oxidation and deposition of nonvolatile Ti etch reaction products; or (c) erosion or redeposition of the mask material or species derived therefrom.25 It is also conceivable that sidewall passivation may arise from the deposition and oxidation of reaction products produced by the concurrent etching of the Si carrier wafer. Evidence for this supposition can be found in the study by Tillocher et al.,5 which showed significant micromasking in Cl2-based Ti DRIE of chipscale samples mounted on larger Si carrier wafers. The authors suggested that this may have been caused by the incomplete removal of a SiOCl-based passivation film produced by the oxidation of SiClx reaction products deposited on the trench floors. Although they did not discuss the potential for concurrent deposition of this passivation film on the trench sidewalls, it is reasonable to speculate that such processes could be driving the highly anisotropic etching observed in the current study. Our future studies will focus on confirming the ion-inhibitor Ti DRIE mechanism and developing deeper understanding of the sidewall passivation processes operative therein. However, the trends observed in the current study begin to provide preliminary evidence in this regard. For example, the significant increase in sidewall bowing observed with increasing chamber pressure could be explained by passivation degradation due to scattering-induced bombardment of the sidewalls. Similarly, the increased bowing observed with increasing Cl2 flowrate could result from increasing reactant species density and thus greater chemical etching of the passivation layer. However, the lesser magnitude of bowing increase observed in the chlorine study compared with the pressure study suggests that the sidewall passivation etch mechanism may be more physical in nature. Finally, the observation of reduced sidewall bowing with increasing O2 flowrate coupled with the transition to a positive tapering regime at the highest flowrates indicates that oxygen is a critical contributor to the sidewall passivation process. This is further supported by the appearance of moderate micromasking in the adjacent wide trench floors at the highest O2 flowrates, which suggests the beginnings of an overpassivating etch regime where ion bombardment of the floors is unable to fully remove or prevent passivation film formation on the trench floors. 4.1. UHR Ti DRIE. The observed trends were used to identify Ti DRIE process conditions that maximize etch directionality while maintaining good Ti etch rate and mask selectivity. To this end, the chamber pressure was minimized to reduce ion scattering and sidewall bombardment. Furthermore, the Cl2 flowrate was minimized to limit chemical etching of the sidewalls. Finally, a midrange O2 flowrate was selected to maximize sidewall protection without transitioning to an overpassivating positive sidewall tapering mode. Figure 8 shows results from the application of this optimized process toward the fabrication of 150 nm linewidth gratings. As can be seen, UHR patterning is achieved with HARs (4.0),
Figure 8. SEM micrograph of 150 nm linewidth grating cross-section (300 nm pitch) etched using the UHR Ti DRIE process parameters (0.5 Pa chamber pressure, 400 W source power, 100 W substrate power, 20 sccm Cl2, 4 sccm O2, and 10 °C lower electrode). The etch time was 1 min.
smooth vertical sidewalls (88°), good etch rate (587 nm/min), and mask selectivity (11.1). The measured average linewidth and spacing of the Ti gratings are 148 ± 3.5 and 146.5 ± 4.9 nm, respectively. This is in good agreement with measurements of the imprinted resist masks before oxide etching (144.8 ± 4.9 and 160 ± 5.1 nm), thus demonstrating the high fidelity of the pattern transfer by the UHR Ti DRIE process. Although beyond the scope of the current study, it is easily conceivable that even greater performance could be achieved with the current UHR Ti DRIE process or with minor modifications thereof. For example, Figure 8 shows that significant SiO2 mask thickness remains (∼150 nm) after the 1 min etch. As such, when coupled with the observed selectivity, this suggests potential for achieving even greater etch depths with a similar profile control and thus even higher aspect ratios. However, the ultimate extent of the improvement that can be achieved will be limited by various phenomena (e.g., scattering-induced degradation of the sidewall passivation), unless the process parameters are modulated to enhance passivation with time. Furthermore, while it is also conceivable that further MFS reduction could be achieved, this would require use of a higher resolution patterning technique (e.g., electron beam lithography) because the current 150 nm features represent the resolution limit of the DUV stepper that was used for this study. Finally, it is important to emphasize that the process developed herein would likely need to be further optimized to achieve a similar profile control in other pattern geometries because phenomena such as scattering and ARDE can be strongly pattern-dependent. 4.2. Comparison with Other Ti DRIE Processes. As illustrated in Figure 8, the UHR Ti DRIE process now enables fabrication of HAR features with smooth vertical sidewalls in bulk Ti substrates with MFS as small as 150 nm, thus surpassing the resolution capabilities of other Ti DRIE processes reported to date, most by a considerable margin. For example, this represents a fivefold improvement in MFS relative to the TIDE process (MFS = 750 nm)3 and nearly a sevenfold improvement compared with the MARIO process (MFS = 1 μm).2 Furthermore, this represents a 29-fold or greater reduction in the demonstrated MFS capability compared with Ti DRIE processes more recently reported by others,4−6 although high-resolution etching was not the explicit focus of those efforts. Finally, although an RIE-based process for deep submicrometer scale patterning of bulk Ti substrates 20166
DOI: 10.1021/acsami.6b16518 ACS Appl. Mater. Interfaces 2017, 9, 20161−20168
Research Article
ACS Applied Materials & Interfaces has been recently reported by others,27 significant positive sidewall tapering was observed. This limited the etch depth and precluded the definition of HAR structures with vertical sidewalls, both of which are key requisites for many applications (e.g., electrostatic transducers). 4.3. Opportunities Afforded by UHR Ti DRIE. With the development of the UHR Ti DRIE process, we now have the capability for patterning HAR structures in bulk Ti substrates on the deep submicrometer scale, and hence, we now have the opportunity to extend our understanding of cellular response to surface patterning on length scales well below the 750 nm MFS limit imposed by our earlier Ti DRIE processes. To this end, we have recently shown continued favorable trending of endothelial cell response in vitro with further grating feature size reduction on the submicrometer scale, including enhanced adhesion and proliferation, morphology more reminiscent of the native endothelium, and greater expression of atheroprotective factors.28,29 Taken together, this suggests potential for submicrometer scale surface patterning serving as a new means for enhancing endothelialization in vivo and promoting more normative endothelial function, both of which are crucial for long-term success in coronary stenting. Similarly, we have shown that the grating features within this same length scale regime can enhance the adhesion and morphological response of bone marrow stromal cells in vitro,30 which suggests potential for improving osseointegration and orthopedic implant fixation in vivo. Finally, we have recently shown that submicrometer scale surface patterning can also drive macrophages toward an antiinflammatory, pro-healing phenotype in vitro,31 which has important implications for controlling wound healing and tissue repair response to implanted devices. Collectively, these studies provide compelling evidence supporting the potential embodied in submicrometer scale surface patterning for favorably modulating the cellular responses to implantable medical devices and thus improving the clinical outcomes. It is important to emphasize that these studies have been performed using both materials and manufacturing techniques with direct relevance to medical devices (e.g., submicrometer scale surface patterned stents, whose realization was fundamentally enabled by Ti DRIE).9 Furthermore, it is important to emphasize that the Ti DRIE process is performed using conventional dry etch tools, and the rest of the device fabrication process is typically reliant upon standard semiconductor processing equipment as well (e.g., lithographic patterning and thin-film deposition). Finally, although the use of Ti as a micromechanical material is still emerging, Ti is already used widely for other purposes in traditional MEMS processes (e.g., as an adhesion layer for metallizations), thus indicating that there should be minimal material compatibility constraints. Collectively, this provides a clear path for the eventual translation of such concepts to the clinic and commercial marketplace.
passivation. Using this understanding, we have developed a UHR Ti DRIE process that now enables definition of HAR structures in bulk Ti substrates with MFS as small as 150 nm. This represents a 5-fold improvement relative to our earlier Ti DRIE processes and a 29-fold or greater improvement compared with more recent processes reported by others. Hence, this represents a considerable extension of the state-ofthe-art in Ti DRIE, and it opens important new opportunities, including further miniaturization of Ti MEMS and the potential for enhancing the performance of implantable medical devices through rational design of their surface topographies on the submicrometer scale.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. ORCID
Bryan W. K. Woo: 0000-0002-2162-6200 Present Address ¶
PiMEMS, Inc., Santa Barbara, CA 93101, USA (S.C.G.).
Author Contributions #
Equal contributors (B.W.K.W. and S.C.G.).
Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors acknowledge the following for their contributions to this work: the UC Santa Barbara Nanotech Facility; the UC Riverside Center for Nanoscale Science and Engineering Nanofabrication Facility; and the UC Riverside Central Facility for Advanced Microscopy and Microanalysis. Support for this work was provided by a National Science Foundation CAREER Award (CMMI-1254999) and a UC Riverside Office of Research and Economic Development Collaborative Seed Grant. S.C.G. was supported by a National Science Foundation Graduate Research Fellowship.
■
ABBREVIATIONS MEMS, microelectromechanical systems; Ti, titanium; PDMS, polydimethylsiloxane; PMMA, polymethylmethacrylate; Cu, copper; Ni, nickel; Fe, iron; Al, aluminum; DRIE, deep reactive ion etching; MARIO, metal anisotropic reactive ion etching with oxidation; HAR, high-aspect-ratio; MFS, minimum feature size; TIDE, titanium inductively coupled plasma deep etch; UHR, ultrahigh resolution; CP, commercially pure; ICPPECVD, inductively coupled plasma, plasma enhanced chemical vapor deposition; sccm, standard cubic centimeters per minute; ICP, inductively coupled plasma; DUV, deep ultraviolet; ARDE, aspect ratio dependent etching; FIB, focused ion beam; SEM, scanning electron microscopy; rms, root mean square
■
5. CONCLUSIONS We have reported the development of a new Ti DRIE process that seeks to reduce MFS capability deep into the submicrometer realm for the first time. Exploring the effects of several process variables, we have shown that the chamber pressure and the oxygen flowrate are the most critical to ensuring definition of HAR structures with smooth vertical sidewalls within this diminutive length scale regime, presumably due to their respective contributions to maintaining sidewall
REFERENCES
(1) Spearing, S. M. Materials Issues in Microelectromechanical Systems (MEMS). Acta Mater. 2000, 48, 179−196. (2) Aimi, M. F.; Rao, M. P.; Macdonald, N. C.; Zuruzi, A. S.; Bothman, D. P. High-Aspect-Ratio Bulk Micromachining of Titanium. Nat. Mater. 2004, 3, 103−105. (3) Parker, E. R.; Thibeault, B. J.; Aimi, M. F.; Rao, M. P.; MacDonald, N. C. Inductively Coupled Plasma Etching of Bulk Titanium for MEMS Applications. J. Electrochem. Soc. 2005, 152, C675−C683.
20167
DOI: 10.1021/acsami.6b16518 ACS Appl. Mater. Interfaces 2017, 9, 20161−20168
Research Article
ACS Applied Materials & Interfaces
as a Roadmap to Next Generation Equipment. J. Micromech. Microeng. 2009, 19, 033001. (26) O’Brien, W. L.; Rhodin, T. N. Investigations of the Altered Surface Formed During the Ion-Assisted Etching of Titanium. J. Vac. Sci. Technol., B: Microelectron. Process. Phenom. 1989, 7, 1244−1251. (27) Domanski, M.; Luttge, R.; Lamers, E.; Walboomers, X. F.; Winnubst, L.; Jansen, J. A.; Gardeniers, J. G. E. Submicron-Patterning of Bulk Titanium by Nanoimprint Lithography and Reactive Ion Etching. Nanotechnology 2012, 23, 065306. (28) Vandrangi, P.; Gott, S.; Kozaka, R.; Rao, M. P.; Rodgers, V. G. Endothelial Behavior and Signaling on Micro and Submicro- Patterned Titanium. Circulation 2013, 128, http://circ.ahajournals.org/content/ 128/Suppl_22/A18077. (29) Vandrangi, P.; Gott, S. C.; Kozaka, R.; Rodgers, V. G. J.; Rao, M. P. Comparative Endothelial Cell Response on Topographically Patterned Titanium and Silicon Substrates with Micrometer to SubMicrometer Feature Sizes. PLoS One 2014, 9, No. e111465. (30) Cipriano, A. F.; De Howitt, N.; Gott, S. C.; Miller, C.; Rao, M. P.; Liu, H. Bone Marrow Stromal Cell Adhesion and Morphology on Micro- and Sub-Micropatterned Titanium. J. Biomed. Nanotechnol. 2014, 10, 660−668. (31) Luu, T. U.; Gott, S. C.; Woo, B. W. K.; Rao, M. P.; Liu, W. F. Micro- and Nanopatterned Topographical Cues for Regulating Macrophage Cell Shape and Phenotype. ACS Appl. Mater. Interfaces 2015, 7, 28665−28672.
(4) Zhao, G.; Shu, Q.; Tian, Y.; Zhang, Y.; Chen, J. Wafer Level Bulk Titanium ICP Etching Using SU8 as an Etching Mask. J. Micromech. Microeng. 2009, 19, 095006. (5) Tillocher, T.; Lefaucheux, P.; Boutaud, B.; Dussart, R. Alternated Process for the Deep Etching of Titanium. J. Micromech. Microeng. 2014, 24, 075021. (6) Yamada, S.; Minami, Y.; Sohgawa, M.; Abe, T. Thermal Reactive Ion Etching Technique Involving Use of Self-Heated Cathode. Rev. Sci. Instrum. 2015, 86, 045001. (7) Parker, E. R.; Rao, M. P.; Turner, K. L.; Meinhart, C. D.; MacDonald, N. C. Bulk Micromachined Titanium Microneedles. J. Microelectromech. Syst. 2007, 16, 289−295. (8) Khandan, O.; Kahook, M. Y.; Rao, M. P. Fenestrated Microneedles for Ocular Drug Delivery. Sens. Actuators, B 2016, 223, 15−23. (9) Gott, S. C.; Jabola, B. A.; Rao, M. P. Vascular Stents with Submicrometer-Scale Surface Patterning Realized Via Titanium Deep Reactive Ion Etching. J. Micromech. Microeng. 2015, 25, 085016. (10) McCarthy, P. T.; Otto, K. J.; Rao, M. P. Robust Penetrating Microelectrodes for Neural Interfaces Realized by Titanium Micromachining. Biomed. Microdevices 2011, 13, 503−515. (11) McCarthy, P. T.; Rao, M. P.; Otto, K. J. Simultaneous Recording of Rat Auditory Cortex and Thalamus Via a Titanium-Based, Microfabricated, Microelectrode Device. J. Neural Eng. 2011, 8, 046007. (12) Zhang, Y.; Li, N.; Yan, B.; Feng, X.; Hu, J.; He, S.; Hao, Y.; Chen, J. Fabrication of Laterally Driven Bulk Titanium Devices on Titanium-on-Glass Wafers. J. Micromech. Microeng. 2013, 23, 075026. (13) Zhang, Y. T.; Bottausci, F.; Rao, M. P.; Parker, E. R.; Mezic, I.; MacDonald, N. C. Titanium-Based Dielectrophoresis Devices for Microfluidic Applications. Biomed. Microdevices 2008, 10, 509−517. (14) Hirst, L. S.; Parker, E. R.; Abu-Samah, Z.; Li, Y.; Pynn, R.; MacDonald, N. C.; Safinya, C. R. Microchannel Systems in Titanium and Silicon for Structural and Mechanical Studies of Aligned Protein Self-Assemblies. Langmuir 2005, 21, 3910−3914. (15) Hesse, H. C.; Beck, R.; Ding, C.; Jones, J. B.; Deek, J.; MacDonald, N. C.; Li, Y.; Safinya, C. R. Direct Imaging of Aligned Neurofilament Networks Assembled Using In Situ Dialysis in Microchannels. Langmuir 2008, 24, 8397−8401. (16) Ding, C.; Soni, G.; Bozorgi, P.; Piorek, B. D.; Meinhart, C. D.; MacDonald, N. C. A Flat Heat Pipe Architecture Based on Nanostructured Titania. J. Microelectromech. Syst. 2010, 19, 878−884. (17) Khandan, O.; Stark, D.; Chang, A.; Rao, M. P. Wafer-Scale Titanium Anodic Bonding for Microfluidic Applications. Sens. Actuators, B 2014, 205, 244−248. (18) Lu, J.; Rao, M. P.; MacDonald, N. C.; Khang, D.; Webster, T. J. Improved Endothelial Cell Adhesion and Proliferation on Patterned Titanium Surfaces with Rationally Designed, Micrometer to Nanometer Features. Acta Biomater. 2008, 4, 192−201. (19) Garg, S.; Serruys, P. W. Coronary Stents: Current Status. J. Am. Coll. Cardiol. 2010, 56, S1−S42. (20) Rao, M. P.; Aimi, M. F.; MacDonald, N. C. Single-Mask, ThreeDimensional Microfabrication of High-Aspect-Ratio Structures in Bulk Silicon Using Reactive Ion Etching Lag and Sacrificial Oxidation. Appl. Phys. Lett. 2004, 85, 6281−6283. (21) Madou, M. J. Fundamentals of Microfabrication; CRC Press: Boca Raton, FL, 2002. (22) Blumenstock, K.; Stephani, D. Anisotropic Reactive Ion Etching of Titanium. J. Vac. Sci. Technol., B: Microelectron. Process. Phenom. 1989, 7, 627−632. (23) d’Agostino, R.; Fracassi, F.; Pacifico, C. Dry Etching of Ti in Chlorine Containing Feeds. J. Appl. Phys. 1992, 72, 4351−4357. (24) Campbell, S. A. The Science and Engineering of Microelectronic Fabrication; Oxford University Press: New York, 2001. (25) Jansen, H. V.; de Boer, M. J.; Unnikrishnan, S.; Louwerse, M. C.; Elwenspoek, M. C. Black Silicon Method: X. A Review on High Speed and Selective Plasma Etching of Silicon with Profile Control: An In-Depth Comparison Between Bosch and Cryostat DRIE Processes
■
NOTE ADDED AFTER ISSUE PUBLICATION Due to a production error, this paper was published in print issue 23 with an incorrect Acknowledgments section and several missing requested corrections. The revised version was published to the Web on June 29, 2017.
20168
DOI: 10.1021/acsami.6b16518 ACS Appl. Mater. Interfaces 2017, 9, 20161−20168
