Vapor‐Phase Infiltrated Organic–Inorganic Positive‐Tone Hybrid Photoresist for Extreme UV Lithography

Continuing extreme downscaling of semiconductor devices, essential for high performance and energy efficiency of future microelectronics, hinges on extreme ultraviolet lithography (EUVL) and addressing associated challenges. One of such challenges is a need for improved EUV photoresists featuring simultaneously high sensitivity, resolution, and etch selectivity. Here, a new, positive‐tone, organic–inorganic hybrid EUV photoresist is demonstrated that delivers a high‐resolution EUVL and electron‐beam lithography (EBL) patterning capability combined with high sensitivity and etch resistance. The new resist, poly(methyl methacrylate) infiltrated with indium oxide (PMMA‐InOx), is synthesized via vapor‐phase infiltration (VPI), a material hybridization technique derived from atomic layer deposition. The weak binding of the gaseous indium precursor, trimethylindium, to the carbonyl group in PMMA allows the synthesis of hybrids with inorganic content distributed uniformly in the resist, enabling high EUVL and EBL sensitivities (18 mJ cm−2 and 300 µC cm−2, respectively) and high‐resolution positive‐tone EUVL patterning (e.g., 40 nm half‐pitch line‐space and 50 nm diameter contact hole patterns) with high Si etch selectivity (>30–40). The low exposure doses required to pattern the PMMA‐InOx hybrid resist, high etch resistance, and processing strategies, which are developed, can pave the way for using infiltration‐synthesized hybrid thin films as reliable positive‐tone EUV photoresists for future semiconductor patterning.


Introduction
The advent of the digital era in the 21st century has resulted in an unprecedented increase in the demand for high-performance electronic products, such as smart phones, computers, sensors for automobiles and medical equipment, data storage devices, and other smart devices for new and upcoming technologies.These technologies use various electronic component structures at micro-or nanoscale, such as, fieldeffect transistors, single-electron transistors, molecular sensors, etc., requiring careful, precise, and defect-free patterning and fabrication.In order to continue to enhance their performance, the size of these components keeps shrinking down, and as of today, the required feature size or critical dimension (CD) is in the sub-10 nm range.To meet this demand, the semiconductor industry has relied upon the continuous advancement in the lithography process to pattern smaller features and its successfuladoption into high-volume manufacturing (HVM).Over the past few decades, patterning features with smaller CD has mainly been achieved by reducing the wavelength () of the light source, considering CD ∝  as per Rayleigh's resolution law (i.e., diffraction limit).Today, the industry has adopted extreme ultraviolet (EUV) light source that has a wavelength of 13.5 nm to pattern features with CD of ≈10 nm and below.Unlike the preceding 193 nm ArF immersion lithography, incorporating EUV lithography (EUVL) into HVM, however, has not been an easy transition. [1]ne of the critical challenges that must be addressed for the EUVL process is a need for suitable photoresist that enables highperformance patternability at EUV wavelength.Traditional photoresists like well-known chemically amplified resists (CARs), which are predominantly made of carbon and oxygen as elements, exhibit poor intrinsic EUV sensitivity, resulting in a lower throughput.The resist thickness required for EUVL is also lower (less than ≈30 nm) compared to the previous lithography technologies since the high-resolution patterns otherwise undergo pattern collapse during typical wet development.Subsequently, these thin, patterned organic resists also do not withstand the etching conditions required for pattern transfer to the underlying substrate, resulting in a resolution-etch trade-off. [2]The planned implementation of high numerical aperture (NA) EUVL demands even thinner resists, thus exacerbating these issues. [3]oreover, CARs exhibit challenges posed by inhomogeneous distribution of resist components and acid diffusion, which limit the resolution of the patterned features and increase line edge/width roughness.Overall, this is referred to as RSL trade-off, with R, S, and L denoting resolution, sensitivity, and line edge/width roughness, respectively.A suitable photoresist material for EUVL is desired to have R < 10 nm, S < 20 mJ cm −2 , and L < 15% for highthroughput manufacturing. [4]o meet these stringent requirements with high reproducibility and low defectivity, both the academia and industry have been pursuing the development of new photoresist materials, and among the various materials investigated, organic-inorganic hybrid materials have emerged as one of the most promising systems.In the hybrid resist, the inorganic components help achieve higher sensitivity because of their higher absorption at the EUV wavelength. [5]The hybrids also display improved mechanical strength, making them less susceptible to pattern collapse.Moreover, the etch resistance of hybrids is comparatively higher, allowing a deeper etch pattern transfer for fabricating high-aspect-ratio structures using thin photoresists. [2]everal hybrid photoresists have been investigated so far for electron beam lithography (EBL) and EUVL as summarized in great detail in recent review articles. [6]Despite significant progresses made, currently available hybrid EUV photoresists have a number of limitations, preventing their use for HVM using EUVL.For instance, most hybrids are chemically synthesized and require complex and slow development and processing steps for production or modification of resist properties, while suffering from a short shelf life.Additionally, negative-tone, crosslinking resists have been the focus of most investigated systems, which are capable of patterning line gratings or pillars but require multiple exposures or complex processing for contact-hole patterning in memory devices.Positive-tone CARs are still the drivers for contact-hole patterning, which, however, are near their performance limit.These limitations necessitate the need to further pursue alternative solutions and identify suitable inorganiccontaining photoresist for EUVL, more specifically ones with positive-tone patternability, but such positive-tone hybrid EUV resists have been rarely reported.
In the recent years, chain-scission based positive-tone EUV resists have started to emerge.One approach called chemically amplified backbone scission (CABS) resist has been to incorporate acid labile molecule as a part of the polymer backbone, such that the photoacid generated on exposure leads to backbone scission. [7]In similar line of thought, polyphthalaldehyde (PPA)based depolymerizable resist systems have also been reported for EUVL, with the help of unbound [8] or polymer-bound [9] photoacid generators.However, their high-resolution EUV patternability has not yet been reported.Zeon chemicals has recently introduced EUV-resist counterpart of their conventional ZEPseries EBL resist. [10]Similarly, Oji Holding Corp. has also introduced biomass-derived chain-scission EUV lithography resist. [11]hile both of these commercial-resist platforms have shown EUVL patterning resolution down to ≈16 nm halfpitch, apart from being costly proprietary compositions, their dose-to-size ≈75 mJ cm −2 is still quite high to achieve required HVM throughput.
In this work, we report a new positive-tone hybrid resistindium oxide (InO x )-infiltrated PMMA (PMMA-InO x )-and its high-resolution EBL and EUVL patterning characteristics and performances.Our proposed solution here is to modify conventional organic photoresists using vapor-phase infiltration (VPI), [12] an ex situ material hybridization method derived from atomic layer deposition (ALD), [13] which can be readily implemented without requiring complex chemical processing by using a standard ALD tool, an already existing industry infrastructure.The new PMMA-InO x hybrid resist detailed in this study not only leverages high EUV-absorbing and etch-resistant inorganic species, viz., InO x , but also exploits the weak interaction of organometallic indium precursor with PMMA matrix for uniform distribution of infiltrated InO x in PMMA during VPI, thus ensuring suitable patterning performance.We find that the hybrid resist has high EUVL and EBL sensitivities with critical dose as low as 18 mJ cm −2 and 300 μC cm −2 , respectively, and can perform high-resolution, positive-tone patterning, demonstrating, for example, 40 nm half-pitch line-space and 50 nm diameter contact hole patterns by EUVL.The hybrid resists also feature high Si etch selectivity (>30-40) due to the increased etch resistance stemming from the infiltrated inorganic InO x , enabling fabrication of a high-aspect-ratio Si nanostructures via a plasma-etchingbased pattern transfer.The results represent the first successful experimental demonstration of high-resolution EUVL patterning characteristics and performance based on the new class of infiltrated hybrid photoresists.The delivered positive-tone, highresolution EUVL patterning, and high etch resistance, along with detailed processing strategy described in this study, should help further advance the performance of infiltrated hybrid EUVL photoresists, contributing to the continuing extreme downscaling of future-node semiconductor devices by EUVL.

Results and Discussion
13a,14] Briefly summarizing the differences, the ALD process is typically implemented on a solid substrate by alternatively exposing a vapor-phase organometallic precursor (e.g., trimethylaluminum (TMA)) and an oxidizing agent (e.g., water vapor) in a cyclic manner, which results in the growth of a surface monolayer of target inorganic material (e.g., aluminum oxide (AlO x )) per cycle.In contrast, during the VPI process, the sample is typically a polymer film (i.e., soft or porous material) into which the precursors are cyclically infiltrated, relying on the sorption of gaseous precursor molecules in the polymer matrix.This yields the growth of molecular-scale target inorganic material within the bulk (free volume) of the polymer film via the reaction of the diffused organometallic precursors (Lewis acid) with various functional groups (Lewis base) available in the polymer matrix. [15][14][15][16] This study unveiled the full tunability of the resist EBL performance in terms of high-resolution patterning, critical dose, patterning contrast, and Si etch selectivity.However, the study identified that the high reactivity of infiltrating organometallic precursors could lead to a non-uniform infiltration of inorganic components along the thickness of a starting polymer film as well as a strong bonding with the infused organic matrix, causing non-ideal resist performance such as an increased exposure critical dose and difficulty of development.These issues could be remedied in the current InO x -infiltrated PMMA hybrid resist.
The VPI synthesis of PMMA-InO x hybrid resist is described in Figure 1.During the synthesis, a PMMA thin film spin-coated on a Si substrate (≈30 nm; Figure 1a) is subjected to InO x infiltration at 85 °C in a commercial ALD system.The number of infiltration cycles controls the amount of InO x infused into PMMA (referred to here onward as "n-cycle hybrid").Each cycle consists of a first trimethylindium (TMIn) precursor exposure for 60 s for its diffusion into the PMMA matrix and subsequent binding reaction with carbonyl groups in the PMMA, which is followed by a 15 s N 2 purge step (Figure 1b).Subsequently, the water exposure halfcycle is implemented for 60 s, leading to the conversion of methyl groups of TMln into hydroxyl groups that provide reaction sites for the next infiltration cycle, and is followed by an N 2 purge for 120 s to complete one infiltration cycle (Figure 1c).
The lithographic patterning performance of InO x -infiltrated PMMA hybrid resist was evaluated by EBL first.The atomic force microscopy (AFM) height map of the EBL dose-exposure matrix for pure PMMA and 4-cycle hybrid is depicted in Figure 2a,b, respectively.Careful observation of the AFM maps reveals that while the highest exposure dose (D MAX ) of 1500 μC cm −2 for pure PMMA exhibits a depth of ≈30 nm, the hybrid resist exhibits a depth less than that.The exposure dose was increased from bottom to top and from right to left.The normalized remaining height extracted from the AFM is plotted in Figure 2c against their corresponding exposure doses, along with their sigmoidal fit.The sensitivity (D 50 ) for pure PMMA was estimated to be ≈200 μC cm −2 , which increased to ≈300 μC cm −2 for the 4-cycle hybrid.The contrast (  M = 1/log 10 (D 0 /D MAX ), with D 0 being the onset dose), however, reduced from 0.78 to 0.68 when the PMMA was infiltrated with 4-cycles of InO x .It should be noted that EBL exposure characteristics do not precisely represent EUVL characteristics because of the difference in exposure and absorption mechanisms. [2,17]Interestingly, data revealed that the residue left behind by pure PMMA was ≈0.7% of the initial resist thickness, whereas it was ≈18% for the 4-cycle hybrid (Figure 2c) at D MAX .
The dose-exposure data extracted from the EUVL patterned samples (Figure 2d) revealed the critical exposure dose (D 50 ; the dose at 50% of the resist height) for 4-cycle PMMA-InO x to be ≈26.2mJ cm −2 , increased by a small amount from D 50 for pure PMMA of ≈15.8 mJ cm −2 .The resist's estimated  M decreased from ≈1.01 for pure PMMA to ≈0.69 for 4-cycle PMMA-InO x (red curve), as evidenced in Figure 2d, becoming shallower compared to the pure PMMA dose curve (blue curve).Similar to the EBL data, ≈10% remaining resist height points toward the leftover, post-development residue, albeit slightly smaller than that from EBL.The attempted EUV high-resolution patterning of concentric ring patterns exposed at ≈80 mJ cm −2 also showed that 4cycle PMMA-InO x leaves behind residue as imaged under scanning electron microscopy (SEM), as opposed to the clean development of pure PMMA (Figure 2e,f).
The surface residue after resist patterning can cause issues during the following pattern transfer and, therefore, it must be addressed.We investigated the cause of post-development residue and developed a mitigation strategy utilizing a resist underlayer.Surface characterization performed within the EBLpatterned square region exposed at D MAX revealed that for pure PMMA, no visible residue was seen in the developed region under SEM while AFM characterization showed root-mean square roughness (R RMS ) of ≈0.11 nm (Figure 2g).In contrast, the 4cycle hybrid clearly showed excessive residue in the developed regions, as shown in Figure 2h with the R RMS comparatively higher ≈1.61 nm.Similarly, the residue was also observed to be appearing for 500 nm line-space patterns generated from the infiltrated hybrid, which was not noticeable in pure PMMA (Figure 2i,j).Separate, additional surface characterization also showed that conventional oxygen plasma descum was incapable of removing this residue (data not shown).The increased D 50 , reduced  M , and the chemical nature of the residue left behind all point toward the residue most likely being consisting of inorganic InO x that did not readily dissolve in the organic solvent developer due to the poor efficacy of the methyl iso-butyl ketone (MIBK)-isopropyl alcohol (IPA) developer in removing inorganic content.
To make further sense of the origin of this residue, it is important to take a closer look at the infiltration mechanism.As mentioned before, VPI is a process that relies on both the diffusion (i.e., infiltration) of a Lewis-acidic organometallic precursor into the polymer matrix and its reaction with various Lewisbasic functional groups within the polymer's bulk.However, depending on the strength of the interaction between the precursor and the polymer, the distribution of the organometallic precursor through the polymer thickness can vary.When a very strong Lewis-acidic precursor (e.g., TMA for AlO x infiltration) [18] is infiltrating into PMMA, which has Lewis-basic carbonyl groups, [19] the precursor quickly binds to the carbonyl functional groups as it diffuses into the polymer from the top surface.As the precur-sor diffuses through the thickness of the polymer, less and less precursor molecules become available to diffuse deeper because they are being consumed.Once a full VPI cycle is completed by water vapor exposure, this results in a decreasing content of AlO x at a deeper depth of PMMA (Figure 3a-left).Moreover, with increasing VPI cycles, the molecular porosity within the polymer is reduced as the space becomes occupied by AlO x .Additional VPI cycles thus will not increase the depth of infiltration because the diffusion pathways are now choked up by the pre-infiltrated AlO x , rather leading to the formation of a thin AlO x layer on the top surface of the polymer, as shown in the Figure 3a-left.Additionally, there is a possibility that small amounts of TMA, during the initial few infiltration cycles, can diffuse all the way to the bottom of PMMA layer and at the PMMA/Si substrate interface (when the exposure time is long enough) and react with the highly Lewis-basic ─OH group present on the native Si oxide-a much stronger Lewis base compared to the carbonyl group in PMMA-which exothermically reacts with the methyl group of TMA.This ultimately yields the growth of a thin AlO x layer at the bottom PMMA/Si substrate interface along with the AlO x infiltration within the bulk of the polymer after VPI cycles (Figure 3aleft).In contrast, during InO x infiltration, the precursor TMIn is known to be a much weaker Lewis acid compared to TMA and interact very weakly with the moderately Lewis-basic carbonyl group within PMMA. [20]This allows TMIn to diffuse much more easily and deeper within the PMMA matrix unlike TMA and readily reach near the bottom PMMA/Si interface in quantity, where it can react to the strongly Lewis-basic ─OH group of the surface native oxide of Si substrate (Figure 3a-right).During the subsequent purge step, the TMIn reacted to the surface ─OH group stays intact, but the TMln infiltrated within PMMA can be easily removed from the bulk matrix because of its very weak binding or lack of it with carbonyl groups in PMMA.The TMIn removal is especially prominent near the top surface since the deeper infiltrated TMIn can be kinetically trapped (i.e., not sufficient time to diffuse out).The completion of a VPI cycle with following water exposure then would result in the InO x infiltration concentrated at the bottom portion of the PMMA layer along with the formation of a InO x layer at the PMMA/Si interface, creating an increasing InO x content at a deeper depth of PMMA, which is opposite of the AlO x VPI case (Figure 3a-left).Significantly, the bottom InO x layer at the PMMA/Si interface will grow thicker during the following VPI cycles and is expected to be responsible for the observed leftover residue after the post-patterning resist development.
Indeed, cross-sectional transmission electron microscopy (TEM) on the PMMA-InO x hybrid resist thin films after 4cycles of InO x infiltration revealed a dense, InO x -rich layer at the Si/PMMA interface as seen in the bright-field TEM and high-angle annular dark-field (HAADF) scanning TEM image (Figure 3b).The corresponding energy dispersive spectroscopy (EDS) elemental map for indium reveals that its distribution within the resist is most dense near the bottom interface (i.e., Si/PMMA interface) and decreases as approaching the top surface of the resist.The elemental map for Si is also shown as a reference for the substrate.This observed gradient of indium distribution seen within the PMMA matrix strongly supports the suggested InO x VPI mechanism in a PMMA film on a Si substrate and the role of the bottom InO x layer in causing the postdevelopment residue.
In order to reduce the residue, the next logical step is to prevent the interaction that occurs between the infiltrated TMIn molecules and the ─OH group of the native oxide on the surface of the Si substrate.We find that when the Si substrate is passivated with hexamethyldisilazane (HMDS) prior to spin-coating the PMMA resist, the InO x growth at the Si/PMMA interface becomes reduced.HMDS is a widely used adhesion promoter for a resist film on a Si substrate, with its monolayer terminated with non-polar methyl groups that would not react with infiltrating TMIn.The cross-sectional TEM on the hybrid resist thin films after 4 cycles of InO x infiltration using the HMDS-passivated Si substrate indeed revealed that the dense, InO x -rich layer at the Si surface was absent because of the HMDS layer between Si and PMMA (Figure 3c).The corresponding elemental map for indium reveals that its distribution within the resist is still most dense near the bottom portion of PMMA layer and decreases as approaching the top surface of PMMA.
In a standard Si wafer, the dangling Si bonds react with moisture and form SiO x /SiOH native oxide on its surface.This makes the substrate hydrophilic and prevents the proper spin-coating of photoresist (typically hydrophobic), resulting in defects in the coated resist film due to poor adhesion.The HMDS passivation can render Si substrates hydrophobic and promote the resist adhesion by replacing the surface ─OH group with methyl groups, following the nominal reaction mechanism on a Si surface schematically illustrated in Figure 3d.In our approach, the passivation of the Si substrate with HMDS served dual purposes, first to improve the adhesion of PMMA onto the substrate and, second, to block the reaction that occurs between the infiltrating TMIn precursor and the strongly Lewis-basic ─OH group on the Si substrate.
The suppressed formation of InO x layer at the PMMA/Si interface by the application of HMDS was further corroborated by measuring the mass gain during InO x VPI: The in situ quartz crystal microbalance (QCM) mass gain measurements showed that when PMMA was infiltrated with InO x with the Si substrate coated with HMDS, there was a significant decrease (by ≈39% on average) in mass gain per InO x VPI cycle, compared to the same InO x VPI in PMMA on a Si substrate without any HMDS passivation (Figure 3e,f).The mass gain recorded for the PMMA on a non-passivated Si substrate originates from both the InO x infiltrated in the bulk matrix of PMMA itself and the InO x layer formation at the PMMA/Si interface.The reduced mass gain upon the application of HMDS passivation on the Si substrate thus reconfirms the suppressed SiOH-InO x formation at the PMMA/Si interface because the Si surface is now terminated with methyl groups of HMDS that do not react with infiltrating TMIn.
As intended, the HMDS passivation of the Si substrate was found to reduce the formation of surface residues after patterning the PMMA-InO x hybrid resist.We reinvestigated the patterning characteristics of the pure PMMA and InO x hybrid resists on HMDS-passivated Si substrates using EBL (Figure 4a-c) as well as EUVL (Figure 4d).This revealed the EBL D 50 for pure PMMA and the 4-cycle InO x hybrid to be ≈250 and ≈325 μC cm −2 , respectively, slightly higher than those of the resists on nonpassivated Si substrates (i.e., ≈200 μC cm −2 for pure PMMA and ≈300 μC cm −2 for 4-cycle InO x hybrid).Meanwhile, the corresponding EBL contrast  M reduced from 0.73 for pure PMMA to 0.69 for the 4-cycle hybrid (as opposed to 1.01 and 0.69 of the pure PMMA and 4-cycle hybrid resists on non-passivated Si substrates) showing that the HMDS passivation did not negatively impact the EBL contrast of PMMA-InO x hybrid resist.
Most significantly, the dose characterization revealed that the residue left behind by pure PMMA was ≈0.3% of the initial resist thickness and ≈10% for the 4-cycle hybrid (Figure 4c) at D MAX of 1500 μC cm −2 , indicating that the remaining residues were roughly 44% lower compared to those on the non-passivated Si substrate.The dose-exposure data extracted from the EUVL patterns (Figure 4d) revealed small drop in the critical exposure dose for both 4-cycle PMMA-InO x (≈17.6 mJ cm −2 ) and pure PMMA of (≈10.2 mJ cm −2 ) compared to the resists coated on non-passivated substrates.
Surface characterization was performed on the dose square exposed at D MAX (EBL) with SEM and AFM.For pure PMMA, no visible residue was seen in the developed region by SEM (Figure 4e), but AFM characterization revealed minor residue with a surface R RMS of ≈0.10 nm (Figure 4e).While the 4-cycle hybrid clearly showed residue in the developed regions, as shown in Figure 4f (R RMS measured to be ≈0.89nm), it was considerably lower compared to the 4-cycle hybrid on the non-passivated Si substrate (Figure 2h; R RMS ≈ 1.61 nm).The results indicate that the HMDS passivation of Si substrates did help reduce the amount of residue but could not eliminate it completely.This is most likely due to the InO x infiltrated within PMMA that can still be left behind on the substrate surface after development.
Considering the residue mostly consists of inorganic InO x , an acid or base rinse was required to remove it.InO x is an amphoteric oxide that can dissolve in both acids and bases.However, the residue also has organic components from the PMMA matrix, which can be easily removed using a short oxygen plasma descum.The post-development, 20 s acid/base rinse along with oxygen plasma descum (20 s, 20 W, 100 mTorr), was implemented on EBL-patterned samples, and the surface characterization was performed on the dose square exposed at D MAX .SEM revealed no visible residue for the 4-cycle hybrid in the developed region (Figure 4g,h) while AFM did a minor residue with a surface R RMS of ≈0.21 nm when rinsed by acetic acid (Figure 4g) and ≈0.17 nm by MF312, a standard developer containing tetramethylammonium hydroxide (TMAH) (Figure 4h).These R RMS are comparable to those of pure PMMA, thus confirming the efficacy of the devised residue cleaning procedure combining acid/base rinse and oxygen plasma descum.
This optimized post-development residue removal strategy was implemented on the high-resolution patterns generated by EUVL. Figure 5 shows various 1:1 line-space patterns of pure PMMA (Figure 5a,b) and 4-cycle hybrid (Figure 5c,d) at pitches 60, 80, and 100 nm, which were subjected to 60 and 70 mJ cm −2 EUV exposure.The patterns printed in pure PMMA showed a clear sign of overexposure with a wiggly line of poor fidelity-an indication of pattern collapse.4-cycle infiltrated PMMA-InO x hybrid at the same exposure conditions, however, showed much improved pattern fidelity.Higher-magnification SEM images from the 40 and 50 nm half-pitch lines are shown in Figure 5e-h, Analysis of the high magnification images revealed that at 60 mJ cm −2 average linewidths of 33.5 nm (with linewidth roughness (LWR) ≈17.4 nm) and 46.3 nm (LWR ≈22.2 nm) for 40 and 50 nm halfpitch, respectively.Interestingly, when the exposure dose was increased to 70 mJ cm −2 , linewidths broadened to an average of 55 and 64.3 nm, respectively, with a reduction of LWR to ≈13.8 and ≈15.3 nm, respectively.18b] A rectangular array of high-density contact hole patterns remains a challenging feat even for state-of-the-art resist compositions.We also tested the EUVL patterning performance of contact hole arrays at center-to-center pitches of 60, 75, and 90 nm, and the as-developed SEM images are shown in Figure 6.After measuring the diameters of several holes for each condition, their distribution is illustrated in Figure 6g-i.All the images showed considerable pattern broadening under the development and descum conditions used.The increase in average diameter was clearly observed for all three pitches when the exposure dose was increased from 80 to 100 mJ cm −2 (from 49.6 to 53.3 nm for pitch 60 nm, from 53.9 to 57.1 nm for pitch 75 nm, and from 55.9 to 57.9 nm for pitch 90 nm).Interestingly, a slight improvement in the CD uniformity was also seen with increased exposure dose (standard deviation decreased from 4.4 to 3.4 nm, 2.8 to 2.4 nm, and 5.8 to 2.4 nm for each pitch, respectively).Compensation of the photon-shot-noise under higher doses might be the cause for such an improvement.
12a] Low-magnification 45°-tilted SEM images of the as-etched patterns are shown in Figure 7a,b, depicting uniform etching across various pattern dimensions.Additionally, higher-magnification SEM images (Figure 7c-e) depict ≈182.5 nm deep pattern transfer, leading to the fabrication of nanostructures with aspect ratios up to ≈3-4.Thirty nanometers thick and pure PMMA with an etch rate ≈10 nm s −1 would have been completely removed within ≈3 s, etching merely ≈30 nm deep into the Si substrate (i.e., etch selectivity ≈1).Whereas, the 4-cycle infiltrated PMMA-InO x hybrid was able to sustain 20 s of cryo-Si etching without consuming its thickness.This exercise clearly demonstrates that despite the weak interaction between the InO x and PMMA networks of the hybrid composition, it was able to exhibit a vast improvement in the etch-selectivity (>30-40) compared to pure PMMA.
In a nutshell, our PMMA-InO x hybrid resist is able to mitigate some of the shortcomings of the previously reported PMMA-AlO x .It not only mitigates the sensitivity loss, while also exhibiting compositional uniformity throughout the resist thickness.12a] The dose-to-size values of ≈60-70 mJ cm −2 for 1:1 line-space patterns and ≈80-100 mJ cm −2 for 1:1 contact-hole patterns for PMMA-InO x resist, are comparable to the recently reported EUVL results of commercial chain-scission positive-tone resists. [10,11]Nonetheless, further optimization of the EUVL patterning performance is necessary to estimate the resolution limits of the infiltrated resist platform.A notable challenge associated with infiltrated resists is their high LWR due to the inherent crystalline tendency of InO x .It may be possible to suppress the excessive grain growth by co-infiltrating hetero-element oxides such as TiO x , ZnO x , along with InO x into the PMMA matrix, as a number of mixed metal oxides exhibit an amorphous structure, circumventing the LWR issue.Another key challenge faced by infiltrated hybrid resists is that the use of conventional solvent developers, prohibits dissolving inorganic species in the exposed region, leaving behind significant residue.Therefore, achieving optimal patterning performance at a lower exposure dose using infiltrated resists is incumbent upon finding an effective developer scheme.

Conclusion
In summary, we have reported a new infiltration-synthesized PMMA-InO x hybrid photoresist capable of high-resolution, positive-tone EUVL patterning, delivering the first successful experimental demonstration of such EUVL performances from the new class of organic-inorganic hybrid photoresists generated by VPI.The PMMA-InO x hybrid resist featured high EUVL and EBL sensitivities with critical doses as low as 18 mJ cm −2 and 300 μC cm −2 , respectively.Under the optimized pre-and post-patterning resist process strategies comprising HMDS underlayer and post-development descum procedures for removing InO x residues, which are devised based on the experimentally elucidated VPI mechanism for the weakly reactive indium precursor in PMMA, the hybrid resist could perform highresolution, positive-tone EUVL patterning, demonstrating 40 nm half-pitch line-space and ≈50 nm diameter contact hole patterns with reasonable CD uniformity.Additionally, the high Si etch selectivity (>30-40) of the hybrid resist enabled the generation of high-aspect-ratio Si nanostructures via plasma etching pattern transfer to the underlying Si substrate.The low exposure doses required to perform high-resolution patterning in the PMMA-InO x hybrid resist, along with the high etch resistance and the developed resist processing strategy and understanding of the infiltration mechanism, can pave the way for using infiltrationsynthesized hybrid thin films as reliable EUVL photoresists that can support the development of next-generation semiconductor devices and microelectronics.

Experimental Section
VPI Synthesis of PMMA-InO x Hybrid Resist: During the synthesis of hybrid nanocomposite resists, PMMA (Polymer Source, molecular weight of 175 kDa) dissolved in anisole (1 wt.%) was spun on Si substrates to attain the thickness of ≈30 nm (Figure 1a).The substrates were then subjected to InO x infiltration at 85 °C in a commercial ALD system (Veeco Savannah S200).The number of infiltration cycles varied between 1 and 10 in order to infiltrate different amounts of InO x into PMMA.Each cycle was composed of first trimethylindium (TMIn) precursor exposure for 60 s by pulsing TMIn for 800 msec and then purging the chamber for 15 s using 100 sccm N 2 (Figure 1b).Subsequently, the water exposure half-cycle was implemented for 60 s by pulsing the water valve for 300 msec and then purged by 100 sccm N 2 for 120 s (Figure 1c).Mass gain measurements were performed using a built-in QCM setup.A quartz QCM crystal, which typically has a gold coating, was sputter coated with 5 nm of SiO 2 at 100 W radio frequency (RF) power under a 10 mTorr Ar atmosphere (Lesker PVD 75) in order to mimic the substrate-resist interface as that of the native SiO 2 on a Si substrate.
EBL and EUVL Patterning: The EBL patterning of the PMMA-InO x hybrid photoresists was performed using a JEOL JBX-6300FS EBL system operating at 100 kV.An exposure dose matrix comprising a 5 × 5 array of 5 μm 2 squares was exposed with 1 nA current with a shot spacing of 8 nm, with exposure dose varying from 50 to 1500 μC cm −2 .The development was carried out using MIBK-IPA solution in a volume ratio of 1:3 for 30 s, followed by 30 s IPA rinse.The nanopatterning of 1:1 line-space patterns with linewidth varying between 100 and 500 nm was carried out using the same current and shot-pitch settings.The EUVL patterning was performed using the Micro-field Exposure Tool 5 (MET5), a synchrotron-based dedicated EUVL tool at the Advanced Light Source of the Lawrence Berkeley National Laboratory.The EUV dose test was performed by exposing a 200 μm × 30 μm full-field area to 30 different exposure doses between 3.5 and 203 mJ cm −2 .High-resolution EUVL patterning was also conducted at doses varying between 32 and 128 mJ cm −2 using a reflective mask containing line space, holes, and concentric ring patterns in the 200 μm × 30 μm field area.The same development conditions as those for EBL were used for the EUVL samples as well.
Resist Characterization: The as-infiltrated PMMA-InO x hybrids were characterized using 200 kV TEM using FEI Talos F200X equipped with EDS elemental mapping capability.The cross-sectional TEM samples were prepared by a standard in situ lift-out procedure using Ga ion milling in a focused ion beam system (FEI Helios 600 Nanolab).The post-development, remaining resist height for the dose matrix regions, and the surface topography of the highest exposure dose region were characterized by AFM using a Park NX20 AFM system with PPP-NCHR tips.The dose matrix and high-resolution patterns were imaged using SEM (Hitachi S-4800).
Surface Passivation with HMDS: For selective samples, the Si substrate was primed with HMDS before applying PMMA.The Si substrate was first cleaned with oxygen plasma (100 W, 200 mTorr, 60 s) and immediately loaded into an HMDS coater oven (YES HMDS Vapor Prime) maintained at 150 °C.The HMDS deposition was then carried out by first baking the substrate in vacuum for 120 s followed by subjecting it to HMDS vapors for an exposure time of 300 s.
Cryogenic Si Etching: An inductively coupled plasma reactive ion etching (ICP-RIE) tool (Oxford PlasmaLab 100) was used to transfer elbow patterns into Si substrates at −100 °C and chamber pressure of 15 mTorr with a SF 6 :O 2 gas ratio of 40 sccm:15 sccm under RF power of 15 W and ICP power of 800 W.

Figure 1 .
Figure 1.Schematic (bottom row) showing VPI of InO x into PMMA.a) PMMA was spin-coated on a cleaned Si substrate; b) first half cycle infiltrating TMIn molecules into PMMA.; c) The second half cycle infiltrating water molecules into the TMIn-infiltrated PMMA to generate a molecular network of InO x imbedded within PMMA.The chemical reaction (top row) shows the potential binding reaction between InO x and the carbonyl group within PMMA during each half-cycle for the 1st VPI cycle.Steps (b) and (c) are repeated "n" times sequentially to incorporate specific amounts of inorganics into the polymer matrix.

Figure 2 .
Figure 2. a,b) Post-development height profile map across EBL-exposure dose matrix as measured by AFM for pure PMMA and 4-cycle PMMA-InO x hybrid resists; c) Post-development normalized remaining height variation with EBL exposure dose and fitted to a sigmoidal curve for pure PMMA (blue) and 4-cycle PMMA-InO x hybrid (red) resists; d) Post-development normalized remaining height variation with EUVL exposure dose fitted to a sigmoidal curve for pure PMMA (blue) and 4-cycle PMMA-InO x (red) resists; e,f) SEM images showing the high-resolution patterns of concentric rings for pure PMMA and 4-cycle PMMA-InO x hybrid resist.g,h) Low-magnification SEM image of EBL D MAX dose square [marked with red dotted box in (a) and (b)], and top-view and three-dimensional (3D)-view AFM scans for pure PMMA and 4-cycle PMMA-InO x hybrid, respectively.i,j) SEM image of 500 nm EBL line-space patterns (white area showing developed space) from pure PMMA and 4-cycle PMMA-InO x hybrid, respectively.All SEM scale bars are 200 nm.

Figure 3 .
Figure 3. a) Schematic illustration of AlO x infiltration within PMMA using TMA (left) and InO x infiltration within PMMA using TMIn (right).b,c) Crosssectional bright-field TEM image, HAADF image, and EDS elemental map for 4-cycle InO x infiltrated PMMA without HMDS and with HMDS on Si substrates.d) Schematic representation of the chemical reaction occurring on a Si substrate during HMDS treatment.e,f) Mass gain characteristics and mass gain per InO x infiltration cycle of PMMA thin film during InO x infiltration on HMDS passivated Si substrate (red) compared to the non-passivated Si substrate (grey).All TEM scalebars denote 10 nm.

Figure 4 .
Figure 4. Investigation of PMMA-InO x hybrid resist patterning performance with the use of HMDS-primed Si substrates.a,b) Post-development height profile map across EBL-exposure dose matrix as measured by AFM for pure PMMA and 4-cycle PMMA-InO x hybrid; c) post-development normalized remaining height variation with EBL exposure dose and fitted to a sigmoidal curve for pure PMMA (blue) and 4-cycle PMMA-InOx hybrid (red) resist.d) Post-development normalized remaining height variation with EUVL exposure dose fitted to a sigmoidal curve for pure PMMA (blue) and 4-cycle PMMA-InO x (red).Low-magnification SEM image of D MAX dose square [marked with red dotted box in (a) and (b)], top-view and 3D-view AFM scans, and SEM image of 500 nm EBL line-space patterns (white area showing developed space) from (e) as-developed pure PMMA, f) as-developed 4-cycle PMMA-InO x hybrid, g) 4-cycle PMMA-InO x hybrid after post-development 20 s acetic acid dip and 20 s oxygen plasma descum.h) 4-cycle PMMA-InO x hybrid after post-development 20 s MF312 dip and 20 s oxygen plasma descum, respectively.All SEM scale bars are 200 nm.

Figure 5 .
Figure 5. Post-development/descum SEM images of 1:1 line-space patterns with −10 nm bias generated by EUV exposure dose of a,c) 60 and b,d) 70 mJ cm −2 in pure PMMA resist (a,b) and 4 cycle InO x infiltrated hybrid resist (c,d).Zoomed-in SEM images of (e,g) 40 and (f,h) 50 nm half-pitch lines from (c) and (d), respectively.Scale bars for (a-d) are 1 μm in length while scale bars in (e-h) are 200 nm.

Figure 6 .
Figure 6.SEM images of EUV exposed post-development/descum contact-hole patterns with different pitch dimensions -a-c) exposed at 80 mJ cm −2 while d-f) exposed at 100 mJ cm −2 .g-i) Distribution of measured contact-hole diameters for pitch 60, 75, and 90 nm, respectively.All scale bars are 200 nm in length.

Figure 7 .
Figure 7. SEM images of EUV exposed post-development/descum contact-hole patterns with different pitch dimensions -a-c) exposed at 80 mJ cm −2 while d-f) exposed at 100 mJ cm −2 .g-i) Distribution of measured contact-hole diameters for Pitch 60, 75, and 90 nm, respectively.Scale bars in (a,b) are 1 μm and in (c-e) are 200 nm in length.