Molecular Layer Deposition of a Highly Stable ... - ACS Publications

Research Article Cite This: ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

www.acsami.org

Molecular Layer Deposition of a Highly Stable Silicon Oxycarbide Thin Film Using an Organic Chlorosilane and Water Richard G. Closser,† David S. Bergsman,‡ and Stacey F. Bent*,†,‡ †

Department of Chemistry and ‡Department of Chemical Engineering, Stanford University, Stanford, California 94305, United States

Downloaded via NEW MEXICO STATE UNIV on July 6, 2018 at 13:29:24 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: In this study, molecular layer deposition (MLD) was used to deposit ultrathin films of methylenebridged silicon oxycarbide (SiOC) using bis(trichlorosilyl)methane and water as precursors at room temperature. By utilizing bifunctional trichlorosilane precursors, films of SiOC can be deposited in a layer-by-layer manner, wherein a water co-reactant circumvents the need for plasma, high temperatures, or highly oxidizing precursors. In this manner, films could be grown without the degradation commonly seen in other SiOC deposition methods. Saturation behavior for both precursors was confirmed for the MLD process, and a constant growth rate of 0.5 ± 0.1 Å/cycle was determined. X-ray photoelectron spectroscopy (XPS) and Fourier transform infrared spectroscopy were used to verify the reaction between precursors and to gain insight into the final film composition. Unlike most MLD films, which grow polymers in a linear fashion, XPS analysis indicates that neighboring silanol groups within the films tend to condense, forming a highly cross-linked network structure, whereby, on average, two-thirds of silanol groups undergo a condensation reaction. Further indication of cross-linking is seen by XPS during in situ annealing, which shows exceptional temperature stability of the film up to 600 °C in vacuum, in contrast to linear SiOC films, which are known to degrade below this temperature. The films also exhibit high chemical stability against acids, bases, and solvents. A film density of 1.4 g/cm3 was measured by X-ray reflectivity, while the dielectric constant and refractive index were determined to be 2.6 ± 0.3 and 1.6 ± 0.1, respectively, at a 633 nm wavelength. The low dielectric constant, high ease of deposition, and exceptional thermal and chemical stabilities of this MLD SiOC film suggest that it may have potential applications for electronic devices. KEYWORDS: molecular layer deposition, silicon oxycarbide, SiOC, organosilicon, thin film, atomic layer deposition, low-k, cross-link

■

INTRODUCTION The continued scaling down of feature sizes in semiconductor devices, such as transistors, has brought about a significant demand for new types of materials as well as novel methods for depositing them with a high degree of control.1−3 One material of particular interest is low-dielectric-constant (low-k) films, which are defined as having dielectric constants lower than silicon oxide (k of 3.9).4 These low-k materials are needed as insulation layers in integrated circuits to reduce cross-talk and minimize resistance−capacitance delay between copper interconnect lines, problems that are more pronounced as device architecture sizes decrease.4,5 Interconnects and insulating layers are deposited in the back-end-of-line (BEOL) processing steps, requiring them to have high thermal and mechanical stability as well as the ability to be deposited with thickness and conformality control down to the nanometer and even subnanometer scales.4,6,7 To that end, techniques such as atomic layer deposition (ALD) and molecular layer deposition (MLD) have been developed, which allow for precise and predictable growth of metals,8−11 metal oxides,8,11 metal−organic hybrids,2,12−14 and pure organic films.2,12,15 © XXXX American Chemical Society

Previously studied materials that enable the reduction of dielectric constants include organic films, high-porosity films, fluorine-doped siloxane films, and carbon-doped siloxane films.4,16 Of these, carbon-doped siloxanes, herein referred to as silicon oxycarbides or SiOC, are of particular interest for their tunable dielectric, thermal, mechanical, chemical, and electrical properties.4,5,7 Because of their lower density and decreased polarizability, stemming from the separation of siloxane bonds by carbon, SiOC materials possess a lower dielectric constant than silicon oxide.4−6 SiOC films are also well known for their high chemical and thermal stabilities, which enables their use in high-temperature processing steps for semiconductor-based devices.5,6,17,18 Beyond low-k applications, these films have been used in a variety of other applications including: membrane coatings,19 pore sealings,20 antifouling coatings,19,21 three-dimensional microbatteries,22,23 biomaterial coatings,24,25 enhanced resolution imaging,26 and Received: April 14, 2018 Accepted: June 22, 2018

A

DOI: 10.1021/acsami.8b06057 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces barrier layers,27 and are even shown to be capable of selfregeneration.28 SiOC films can be deposited in solution by sol−gel polymerization29 or from vapor by plasma,20,28,30−37 hot wire,38 initiated39,40 or pyrolytic41 chemical vapor deposition and ALD methods. Unfortunately, many of these deposition techniques result in films with inconsistent and unpredictable structures, due to the highly reactive intermediates that are produced during the synthesis, and are also limited in their ability to produce films with a high degree of conformality.31−33,38 Alternate deposition methods, such as MLD, have previously been used to deposit thin-film silicon oxycarbides, but highly reactive precursors, such as ozone, tended to degrade the carbon within the film upon extended exposure.6 To improve and expand upon existing methods, SiOC films were grown in the present study via MLD by using a bifunctional trichlorosilane precursor and water. Unlike previous thin-film SiOC deposition techniques that require strong oxidants or harsh conditions, we show that by utilizing a nonoxidizing co-reactant along with a trichlorosilane, the SiOC films do not undergo degradation during the deposition and result in predictable and reproducible film growth and properties. Because of the layer-by-layer growth mechanism of MLD, ultrathin films of SiOC can be deposited with angstrom-level thickness control, which can be more difficult by other vapor deposition techniques. The films are grown at room temperature and form a cross-linked network-like structure, leading to extraordinary thermal stability of up to 600 °C under vacuum as well as excellent chemical stability to solvents, acids, and bases. Semiconductor and other potential applications requiring an ultrastable low-k dielectric thin film could benefit from this process owing to the excellent properties of the film as well as the simplicity of its deposition.

groups, releasing HCl or water.42,43,46−48 Previous studies on self-assembled monolayer depositions have shown that the reaction between chlorosilanes and silica requires trace amounts of surface water, which is generally present on silica surfaces unless heated above 150 °C.43,46,47 In the next halfcycle, water is dosed to hydrolyze the remaining chlorosilane groups to hydroxyl groups, releasing HCl in the process.42,43 One of the defining characteristics of MLD is having selflimiting growth, due to the inability of precursors to react with themselves. This self-limiting growth gives rise to two important characteristics: constant growth rate and saturation behavior.12,15 A constant growth rate as a function of precursor dosing cycle is an indication that a reaction is occurring between the surface and exposed vapor. Saturation behavior is observed when longer soaking times of one precursor result in no change to the growth rate, and is an indication of a true MLD process in that each reaction cycle is self-limiting. To characterize the growth rate of the SiOC films using MLD, variable-angle spectroscopic ellipsometry (VASE) was used to measure the film thickness with increasing number of cycles, as shown in Figure 2a. A deposition rate of 0.5 ± 0.1 Å/cycle (n =

■

RESULTS AND DISCUSSION The proposed reaction scheme and the final cross-linked structure for the SiOC film grown by MLD are shown in Figure 1. In this idealized scheme, bis(trichlorosilyl)methane

Figure 1. Proposed MLD reaction scheme between bis(trichlorosilyl)methane and water forming siloxane bonds and the resultant proposed Si2O4C film structure.

(CSM) and water are used as precursors for MLD onto hydroxylated surfaces of SiO2 and Al2O3. Chlorosilanes are known to react on hydrated silica surfaces and hydrolyze upon exposure to water,42−44 with trichlorosilanes being much more reactive than their mono- or bifunctional equivalents.45 The dual-sided functionality of the CSM precursor enables the layer-by-layer growth required of MLD, which is commonly performed using bifunctional precursors.15 In the initial step, CSM adsorbs onto the surface, whereby residual surface water hydrolyzes the chlorosilane into silanol, which can then condense with the hydroxylated surface or neighboring silanol

Figure 2. (a) Growth curve for CSM and H2O cycling. For the growth curve experiments, both CSM and H2O had a soak time of 360 s. (b, c) Saturation plots for CSM and H2O precursors performed by fixing one precursor soak time at 360 s, while varying the other precursor soak time. All dashed curves are weighted best fits and error bars indicate ±1 standard deviation. B


Research Article

ACS Applied Materials & Interfaces 21) was determined at a 95% confidence interval. This deposition rate is slightly higher than that for other MLD and ALD SiOC film growth processes, for which growth rates between 0.2 and 0.3 Å/cycle have been reported.6,20 Saturation for both precursors was tested by measuring the growth rate as a function of precursor soak times, as shown in Figure 2b,c. CSM exhibits saturation at 360 s, while water saturates within 45 s, although a more consistent growth rate was observed when exposure time was increased beyond 45 s. The quicker saturation time for the water precursor is likely due to its higher vapor pressure, ∼19.8 Torr,49 compared to a vapor pressure of ∼6 Torr50 for the CSM precursor, allowing for faster surface saturation. To gain insight into the overall composition and bonding within the SiOC film, X-ray photoelectron spectroscopy (XPS) was performed. Two XPS survey scans are plotted in Figure 3.

Figure 4. XPS fine scan of Si 2p peaks of (top spectrum) representative silicon oxide/silicon substrate, (middle spectrum) SiOC film as deposited onto silicon oxide with component deconvolution, and (bottom spectrum) SiOC film deposited onto aluminum oxide.

has a Si 2p3/2 elemental peak at 99.5 eV (full width at halfmaximum (FWHM) of 1.43 eV) and a Si 2p1/2 peak at 100.1 eV, while the silicon oxide peak is located at 103.4 eV (FWHM of 1.80 eV). The SiOC film deposited on silicon oxide shows a significant reduction of the silicon elemental peak with an associated increase in the oxide peak. This indicates that the elemental silicon has been buried under the MLD film and that the film itself contains silicon oxides. By using the binding energy and FWHM fittings from the clean silicon spectrum, the silicon oxide peak in the SiOC film can be deconvoluted into the peak from the underlying silicon oxide substrate at 103.4 eV and a new peak from the MLD film at 103.0 eV (FWHM of 1.73). This new peak indicates the presence of a C−SiO3/2 bonding type, which has a reported binding energy of 102.7 eV.51 The oxidized Si 2p peak is also the only Si peak detected on the aluminum oxide surface at 103.0 eV. The presence of this peak is further indication that the reaction between precursors has occurred. Silanol compounds are known to readily cross-link on surfaces, forming cyclic and network-type structures.33 As the SiOC film is being deposited, there are three possible scenarios for the silanol groups: they can remain as silanol, they can cross-link with neighboring silanol groups, or they can react with the next precursor dose. In the case where a silanol undergoes no reaction, no condensation occurs. On the other hand, if the silanol groups cross-link together or react with the next precursor dose, a condensation occurs and a molecule of water is released. Knowing this, the extent of reaction within the MLD film can be schematically represented, as shown in Figure 5. With each condensation reaction between silanol groups, the loss of water is expected to alter the Si/O elemental ratio. In the case where the film undergoes no condensation and thus no reactions beyond hydrolysis of the chlorosilane, we would expect the Si/O ratio to be 0.33. In this scenario, the deposition would only be occurring by physisorption of the CSM precursor, without any chemical bonds being formed (aside from the hydrolysis). As the extent of reaction increases, more condensations occur, resulting in fewer oxygen atoms per silicon atom, up to three possible reactions, which would give a Si/O ratio of 0.67. A ratio of 0.67 would indicate that all available silanol groups have either cross-linked with neighboring silanol groups or reacted with the next precursor dosing. If all hydroxyl groups cross-link

Figure 3. (Top spectrum) XPS survey scan of SiOC film deposited onto silicon oxide surface. (Bottom spectrum) SiOC film deposited onto aluminum oxide surface. The Y axis is in counts per second with spectra manually offset.

The top spectrum is from a silicon oxide/silicon substrate coated with 75 cycles (∼47 Å) of SiOC film. To examine the silicon peaks originating from the SiOC, rather than the underlying silicon substrate, 65 cycles of SiOC were also deposited onto a 60 Å thick aluminum oxide-coated silicon oxide surface, as shown in the bottom spectrum (see Figure S1 for an XPS image of the aluminum oxide-coated surface without the MLD film, which shows that the underlying silicon is not detected by XPS). The film as deposited on the silicon oxide/silicon surface shows an oxidized silicon signal at 103 eV and the appearance of a carbon peak at a 285 eV binding energy. A similar spectrum is seen for the film as deposited on aluminum oxide with the addition of aluminum 2p and 2s peaks at 75 and 120 eV binding energies, respectively. Analysis of the SiOC film on aluminum oxide gives an elemental ratio of 1.9 for Si/C, which is similar to the ideal ratio of 2. Determining the oxygen content of the film is complicated by the presence of oxygen in the aluminum oxide substrate, as well as the ability of these films to cross-link, as discussed in detail below. Although the CSM precursor contains six chlorine atoms, less than 1% of chlorine is detected in the films (chlorine 2p binding energy ∼200 eV). This is strong evidence that the CSM precursor is nearly fully hydrolyzed and the reaction between CSM and water has occurred. To further investigate the bonding within the films, XPS fine scans over the Si 2p region were collected on 65 Å thick SiOC films deposited on both silicon oxide/silicon and aluminum oxide substrates. As seen in the top spectrum of Figure 4 (see Figure S2 for deconvolution), clean silicon with native oxide C


Research Article

ACS Applied Materials & Interfaces

Figure 5. Scheme for the reaction between CSM and water with potential cross-linking between adjacent silanol groups. Parentheses represent siloxane bonds formed by dehydration of silanol groups. As the amount of cross-linking increases, so does the ratio of silicon to oxygen. A Si/O ratio between 0.33 and 0.67 is possible.

shows the FTIR spectrum of a 150 cycle MLD film. Multiple Si−O stretches are detected between 1010 and 1090 cm−1, indicating a successful reaction between the CSM precursor and water. This is in good agreement with the XPS analysis, which showed that the film composition contained oxidized silicon. Si−C groups have multiple stretching modes between 1040 and 1180 cm−1, which are seen in the spectra, as well as peaks between 800−810 and 900 cm−1.53,54 The peaks observed at 2845−2860 and 2900−2940 indicate the presence of CH2 symmetric and antisymmetric stretching, respectively, as well as a CH2 deformation at 1405 cm−1, which confirm the presence of methylene bridging within the SiOC film.53 The strong, broad O−H stretch at 3100−3500 cm−1 results from the remaining silanol groups that did not undergo condensation, as well as residual water that may remain on the surface or within the KBr salt. An experiment to remove residual water from the film and KBr substrate was attempted by heating the KBr in a dry nitrogen-purged oven for 24 h at 120 °C before FTIR spectroscopy analysis; however, the large −OH absorption peak was still present, indicating a significant amount of silanol within the film. Although there is no indication of Si−Cl bonding, absorption of these groups ranges from 625 to 420 cm−1, the lower frequency being beyond the range of the analysis.54 A close examination of peaks between 1000 and 1200 cm−1 has previously been used in the literature to identify cyclic versus linear SiOC structures,33 although overlap of the Si−O and Si−C modes in this region makes such identification unreliable in the current spectra.54 SiOC films typically have a density lower than silicon dioxide, which has a density of 2.2−2.3 g/cm3, falling within the range of 0.7−2 g/cm3, depending on the porosity and composition of the film.4,55,56 To determine the density of these films, X-ray reflectivity (XRR) was performed and fit to a sample/SiO2/Si model with the sample model elemental ratios set as CSi2O4, as determined by XPS. With this fit, the density was determined to be 1.4 g/cm3. A comparison of film thicknesses as measured between XRR and VASE was also carried out, and the results showed good agreement. The representative XRR spectrum shown in Figure 7 gave a film thickness of 52 Å, similar to the thickness as measured by VASE of 50 Å. The small discrepancy in the thickness is likely from fitting parameters used for the film and silicon oxide/ silicon substrate in both the XRR and VASE models. Because of the lower density and reduced polarizability of SiOC films compared to SiO2, their dielectric constants tend to be significantly lower.4,16 The density of these films can be tuned by controlling porosity, while the choice of precursor can alter the polarizability of the material.16 This low-k property is of particular interest to the microelectronics industry, which has a need for low-k materials in BEOL

before reacting with the next precursor dose, however, no hydroxyl sites would be available, thus preventing further MLD.52 To explore the extent of reaction within the MLD film, elemental ratios of films thick enough to obscure the substrate signal were examined by XPS. An analysis of six different samples deposited on both aluminum oxide and silicon oxide resulted in an average Si/O ratio of 0.49 ± 0.02 at the 95% confidence interval (Figure S3), which is consistent with an elemental composition of CSi2O4. This composition indicates that the SiOC film undergoes significant condensation during the deposition, whereby, on average, two of the three available silanol groups condense with either a neighboring silanol group or the next CSM precursor dosing, whereas the remaining silanol group undergoes no reaction. It is unlikely that all of the condensations occur from only new CSM monomer layers added to the surface, whereby no cross-links are formed, as this would lead to purely linear polymers containing multiple fourmembered rings. Thus, it is most likely that some of the condensation is the result of cross-linking with neighboring silanol groups, forming a network SiOC film. This extended cross-linking is desirable for many applications in which film stability is critical, since cross-linking has been observed to enhance the temperature and chemical stabilities of SiOC films, discussed in detail below.33 The SiOC MLD film is expected to contain Si−CH2−Si bonding moieties, which originate from the CSM precursor, as well as Si−O and O−H bonds, which only exist if reaction between the CSM precursor and water occurs. To examine the bonding types within the film, Fourier transform infrared (FTIR) spectroscopy was performed using an attenuated total reflectance attachment to increase the signal detected. The MLD film was deposited onto KBr salts, since a silicon oxide/ silicon substrate tended to interfere with the analysis. Figure 6

Figure 6. FTIR spectrum of a 150 cycle MLD SiOC film as deposited onto predried KBr salt powder. D


Research Article


Figure 7. XRR plot of a 100 cycle SiOC film. Analysis of the XRR data gives a density of 1.4 g/cm3 and a root mean square roughness of 1.3 nm. The thickness was determined to be 52 Å, in close agreement to the 50 Å measured by VASE.

applications.4 To investigate the dielectric constant, optical properties of the film were determined by VASE using a Cauchy model for data analysis. The Cauchy model assumes an absorption coefficient of 0,57 which is a good approximation for SiOC-type films, as they have been shown to be mostly transparent at wavelengths above 500 nm.35,58 The Cauchy model has been used in numerous studies to fit VASE data with similar SiOC films.6,22,23,33,35,39 Using six samples of varying thickness, a refractive index of 1.6 ± 0.1 (95% confidence interval) was determined at a 633 nm wavelength.57,59 This value is within the typical range of SiOC-type films, which have been reported to have a refractive index between 1.4 and 1.7.35,58,60 Having an absorption coefficient (k) of 0 allows for the approximation of the dielectric constant (ε) directly from the refractive index (n) from n + ik = √ε, which simplifies to n = √ε resulting in ε = 2.6 ± 0.3, with error propagated.4,16 This estimate is within a reasonable range for similar SiOC materials, which varies from 2 to 3.4,32,36,56 SiOC films typically exhibit very high thermal stability, from 450 to 650 °C in an inert atmosphere, making them suitable for applications that require high-temperature processing steps.4,6,16,61 To test the thermal stability of the MLD films prepared in this study, samples were heated using a heating stage in an in situ XPS chamber under ultrahigh vacuum conditions. For the analysis, the SiOC film was deposited on aluminum oxide to prevent spectral interference between a silicon substrate and the silicon-containing MLD film. Figure 8a shows the percent compositions of each element as the temperature was increased from 25 to 600 °C. For each data point, spectra were taken after slowly increasing the temperature and waiting 5−10 min after the desired temperature was reached for pressure and temperature stabilization. Upon reaching 600 °C, the temperature was maintained for 1 h and another scan was collected (marked by “X” in Figure 8a,b) to compare to the initial 600 °C scan. Although the composition of the film changes slightly throughout the heating steps, no dramatic change occurs at a specific temperature, as is usually the case with other MLD films, which tend to undergo significant degradation within a relatively narrow temperature range.62,63 Throughout the temperature range tested, carbon is the only element that experiences a decrease in composition. During the first heating step to 100 °C, a significant increase in pressure was detected in the ultrahigh vacuum chamber, which is typically a result of outgassing of carbon contaminates.62 This resulted in a decrease of carbon from 16.4 to 13.9%. Above this temperature however, the carbon composition

Figure 8. (a) Elemental compositions of SiOC film during in situ XPS heating showing temperature stability under ultrahigh vacuum conditions. The X’s represent a 1 h 600 °C heat treatment. The dashed lines represent linear fits to the compositional changes. (b) The Si/C ratio changes during the heating experiment. The dashed curve is a fit to an exponential approach function with a horizontal asymptote of 2.

steadily, although only slightly, decreased to 10.2% at 600 °C, ending at 9.4% after 1 h of holding at this temperature. The steady increases in aluminum (9.4−13.3%) and oxygen (56.7− 58.8%) are likely due to the loss of carbon within the film, which slightly increases the escape depth of the ejected electrons. Two possibilities for the observed carbon loss are the removal of adventitious carbon and decomposition of the methylene bridge within the SiOC film. It is expected that if the carbon loss is due only to adventitious carbon removal, then the Si/C elemental ratios would get closer to the ideal value of 2 with no elemental binding energy shifts in XPS. On the other hand, we would expect a decomposition process to result in a change of Si/C elemental ratios away from the idealized film as well as a shift in the elemental binding energies. Namely, if loss of carbon from the film occurred while silicon remained, then the expected Si/C ratio would increase beyond a value of 2. However, as shown in Figure 8b, the Si/C ratio slowly approaches 2 during heating, reaching a value of 1.85 after 1 h of heating at 600 °C. The Si/C increase toward a value of 2 indicates that the carbon loss is a result of removal of excess adventitious carbon and not due to decomposition of the MLD film. This conclusion is further supported by XPS fine scans of silicon and carbon, as shown in Figure 9a,b. If the decomposition process released carbon, leaving silicon behind, then the cleaved silicon atoms would likely form oxides, shifting the observed silicon peak to a higher binding energy. However, there is no indication of a shift in the silicon binding energy throughout the heating experiment. Furthermore, if the loss of carbon was due to decomposition, then any changes in the carbon peak detected at 600 °C should be amplified further after 1 h, but such a trend is not observed. Fine scans of the O 1s peak also show no change during heating (see Figure S4). The absence of decomposition observed by XPS is consistent E


Research Article


studied here provides further evidence that the film forms a network of highly cross-linked siloxane bonds. Because of the structural similarity between SiOC and silica films, SiOCs also tend to be very stable under multiple chemical environments.6,16 To test the chemical stability of the MLD SiOC films, a 6.5 nm thick film was exposed to several solvent, acid, and base etchants. The thickness of the sample was measured before and after each experiment by VASE to determine if the chemical treatment altered the sample thickness. Figure 10 shows that the film thickness remains nearly constant after 6 days of exposure in air, H2O rinsing, and rinsing and sonicating in each ethanol, hydrochloric acid (1 M), and sodium hydroxide (1 M), suggesting stability to a wide range of acids, bases, and solvents. The slight increase in the thickness is likely from trace amounts of moisture or contamination adsorbing into the film. Having a high degree of stability to heat and various chemical etchants enables the SiOC film studied here to be a good candidate for potential electronics applications, for which BEOL processing steps include annealing or etching treatments.

Figure 9. XPS fine scans over (a) Si (2p) and (b) C (1s) peaks during in situ heating with a temperature range of 25−600 °C as well as a 1 h hold at 600 °C.

with previous reports of similar MLD and plasma chemical vapor-deposited SiOC films having temperature stabilities in ambient conditions up to 600 and 650 °C, respectively.6,16,61 Additional temperature stability tests were performed in a nitrogen flow tube furnace. For these experiments, a 6.5 nm thick SiOC film deposited on a silicon oxide/silicon substrate was tested by VASE before and after heating to probe for loss in film thickness, which would indicate decomposition. Annealing experiments (1 h) were performed at 100, 200, 300, and 400 °C. The results of the experiment are shown in Figure 10, with thickness plotted as a percent of the original

■

CONCLUSIONS In this study, a room-temperature MLD process was developed for growing thin methylene-bridged SiOC films using bis(trichlorosilyl)methane and water as precursors. By avoiding highly reactive precursors and harsh deposition conditions, film growth occurs without the degradation typically seen in similar thin-film depositions of SiOC while simultaneously enabling a layer-by-layer growth mechanism. Saturation for both precursors and a constant growth rate of 0.5 ± 0.1 Å/cycle indicate self-limiting MLD. XPS and FTIR spectroscopy analyses were used to confirm the reaction between precursors to form siloxane linkages. The extent of condensation reactions between silanol groups was studied by XPS analysis, showing that an average of two-thirds of the silanol groups condense together, forming a highly cross-linked network-like structure with an elemental composition of Si2O4C. This network architecture enhances the thermal stability of the film, up to 600 °C in vacuum. The Si2O4C structure also displays excellent stability to acid, base, and solvent treatments. XRR analysis indicates that the film density is 1.4 g/cm3, which is similar to other SiOC thin films deposited by various methods. Analysis by VASE showed that the film is a low-k material with a dielectric constant of 2.6 ± 0.3 and a refractive index of 1.6 ± 0.1. Because of the exceptional thermal stability, a lowered dielectric constant compared to SiO2, and angstrom-level deposition control, this SiOC MLD film could play an important role in semiconductor processing without the need of a plasma-enhanced or high-temperature deposition technique.

Figure 10. Film thickness change measured by VASE due to various chemical and thermal treatments on a SiOC film. The enhanced stability of the film is indicative of cross-linking.

thickness. The slight increase in thickness observed after the first heat treatment is attributed to a small amount of contamination, minor film restructuring, or the inherent error of ellipsometry, as even a 3 Å difference in the measurement can lead to a 5% thickness change. After the annealing experiments up to 400 °C, 90.5% of the original film thickness remained on the surface. Previous studies of MLD and CVD SiOC films have also seen a similar 5−10% decrease in thickness after annealing to 400 °C in nitrogen, which is generally attributed to the loss of adventitious carbon.6,33 Previous studies on CVD SiOC-type films have shown temperature stabilities to be strongly dependent on the degree of cross-linking within the film.33 In one study, annealing a SiOC thin film in nitrogen to 400 °C resulted in only a 10% loss of film thickness when cross-linking was present, whereas 30% of the thickness was lost for the film without crosslinking.33 The enhanced temperature stability of the SiOC film

■

EXPERIMENTAL SECTION

Unless otherwise noted, depositions were performed on N-type single-sided polished silicon (100) wafers obtained from WRS materials. The substrates were prepared by cleaning in piranha solution for 15 min, followed by a thorough rinsing in deionized water. (Note: piranha is a potentially explosive solution containing a 7:3 ratio of 98% sulfuric acid and 30% hydrogen peroxide. Proper training and precaution must be taken before preparing and using piranha solution.) The piranha-cleaned substrates were then stored in deionized water until use. Aluminum oxide-coated silicon substrates (60 Å thick) were prepared by ALD using trimethylaluminum and water as precursors at 120 °C. Prior to deposition, silicon and aluminum oxide substrates were dried with compressed air and F


Research Article

ACS Applied Materials & Interfaces exposed to UV−ozone treatment using a Novascan PSD Series Digital UV Ozone System for 15 min. The substrates were then introduced into the MLD chamber. MLD was performed in a reactor described previously64 with a lowvacuum base pressure (∼5 mTorr). Before depositions began, the reactor was purged with nitrogen for 1 h at room temperature after loading the substrates to remove atmospheric water and oxygen from the chamber. Bis(trichlorosilyl)methane (CSM) (97% purity, used as received by Sigma-Aldrich) and deionized water were used as precursors. The precursors and reaction chamber were kept at room temperature throughout the deposition. The precursors, nitrogen flow, and vacuum pump were operated by computer-controlled valves, allowing for timed cycling to be performed. Each cycle consisted of the following steps: (1) an open time in which the precursor valve was opened to the reaction chamber without nitrogen flow or vacuum pumping; (2) a soak time where the precursor valve was closed but vapor remained in the chamber without nitrogen flow or vacuum pumping; and (3) a purging time with nitrogen flow with the vacuum pump open to remove excess precursor, in which the chamber pressure was maintained at ∼1 Torr. Unless otherwise specified, CSM was dosed for a total of 368 s (8 s open time and 360 s soak time) with a 400 s nitrogen purge. Water was dosed for 420 s (20 s open time and 400 s soak time) with an 1800 s nitrogen purge. Film thicknesses were measured by variable-angle spectroscopic ellipsometry (VASE) using a J.A. Woollam Co. α-SE ellipsometer with a spectral range of 380−900 nm at incidence angles of 65, 70, and 75° and the polarizer set to 45°. An optical model consisted of a silicon substrate, a native silicon oxide layer, and a thin film. Similar SiOC films have been shown to be mostly nonabsorbing between 400 and 900 nm,35,58 making the standard Cauchy model appropriate for fitting the deposited film. This Cauchy function has been used extensively for modeling similar SiOC-type films6,23,27,33,35 as well as thin transparent MLD films.65 Fourier transform infrared (FTIR) spectroscopy was performed in attenuated total reflectance mode with a diamond plate using a Nicolet iS50 FTIR spectrometer with a DTGS detector. Spectra were taken using 200 scans at 4 cm−1 resolution ranging from 525 to 4000 cm−1. For the analysis, SiOC films were deposited onto KBr salt powders (>99%, obtained from Sigma-Aldrich Inc.). Prior to SiOC deposition, the KBr salt was dried in an oven for 2 h at 200 °C. Background FTIR scans were collected using KBr salt, dried by the same method. X-ray photoelectron spectroscopy (XPS) was performed using a PHI VersaProbe III scanning X-ray photoelectron spectrometer with Al Kα radiation with 1486 eV energy. The X-ray beam was set to 100 W, 20 kV with a spot size of 200 μm × 200 μm. Survey scans were performed with a 0.8 eV/step resolution and a 224 eV pass energy using three to four scans at 20 ms/step. High-resolution elemental scans were performed with a 0.1 eV/step resolution and a 55 eV pass energy using 6−10 scans at 20 ms/step. An in situ heating stage in the XPS unit was used for temperature stability tests. The stage was slowly heated in a stepwise fashion to prevent overpressuring, with measurements taken every 50−100 °C, after waiting for 5 min for pressure stabilization. X-ray reflectivity was performed using a PANalytical X’pert materials research diffractometer with Cu Kα radiation (λ = 1.54 Å) and a generator power of 45 kV and 40 mA, respectively. A divergence slit of 1/32° and a parallel plate collimator receiving slit were used, and 2θ scans were taken with a step size of 0.005° and a scan time of 0.5 s/step. Data were fit using PANalytical X-ray reflectivity software with a sample/SiO2/Si model, where the sample electron density was converted into a gravimetric density using an elemental stoichiometry, as determined by XPS. Chemical stability of the deposited films was tested by measuring the thickness of film deposited onto silicon substrates by VASE and then immediately placing the substrate containing the film into a 20 mL glass scintillation vial and rinsing or sonicating in water, ethanol, 1 M HCl, or 1 M NaOH for 5 min. The samples were then thoroughly rinsed again with water, dried with compressed air, and tested by VASE to determine the final film thickness.

Temperature stability in nitrogen environment was carried out by testing the initial thickness of a sample by VASE and then placing the sample into a Mellen Series SC13R tube furnace with ∼30 sccm flow of nitrogen gas. The samples were heated at a rate of 10 °C/min using a programmable Omega 7823 process controller until reaching the desired temperature and remained at that temperature for 1 h. After cooling to room temperature, the samples were retested by VASE for thickness comparison.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsami.8b06057.

■

X-ray photoelectron spectroscopy (XPS) images of silicon (100) and silicon with 100 atomic layer deposition cycles of aluminum oxide as a control; deconvoluted XPS images of silicon (100) as a reference for binding energies and full width at half-maximum data; XPS O 1s fine scan showing the temperature stability of the film; table including all of the specific samples used to obtain a stoichiometric ratio of the film (PDF)

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Richard G. Closser: 0000-0002-8971-168X David S. Bergsman: 0000-0002-0141-6417 Stacey F. Bent: 0000-0002-1084-5336 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This study was supported by the National Science Foundation (CHE-1607339). Part of this work was performed at the Stanford Nano Shared Facilities (SNSF).

■

REFERENCES

(1) Mackus, A. J. M.; Bol, A. A.; Kessels, W. M. M. The Use of Atomic Layer Deposition in Advanced Nanopatterning. Nanoscale 2014, 6, 10941. (2) Meng, X. An Overview of Molecular Layer Deposition for Organic and Organic−inorganic Hybrid Materials: Mechanisms, Growth Characteristics, and Promising Applications. J. Mater. Chem. A 2017, 5, 18326−18378. (3) Alf, M. E.; Asatekin, A.; Barr, M. C.; Baxamusa, S. H.; Chelawat, H.; Ozaydin-Ince, G.; Petruczok, C. D.; Sreenivasan, R.; Tenhaeff, W. E.; Trujillo, N. J.; Vaddiraju, S.; Xu, J.; Gleason, K. K. Chemical Vapor Deposition of Conformal, Functional, and Responsive Polymer Films. Adv. Mater. 2010, 22, 1993−2027. (4) Volksen, W.; Miller, R. D.; Dubois, G. Low Dielectric Constant Materials. Chem. Rev. 2010, 110, 56−110. (5) Morgen, M.; Ryan, E. T.; Zhao, J.-H.; Hu, C.; Cho, T.; Ho, P. S. Low Dielectric Constant Materials for ULSI Interconnects. Annu. Rev. Mater. Sci. 2000, 30, 645−680. (6) Zhou, H.; Bent, S. F. Highly Stable Ultrathin Carbosiloxane Films by Molecular Layer Deposition. J. Phys. Chem. C 2013, 117, 19967−19973. (7) Rathore, J. S.; Interrante, L. V.; Dubois, G. Ultra Low-k Films Derived from Hyperbranched Polycarbosilanes (HBPCS). Adv. Funct. Mater. 2008, 18, 4022−4028.

G


Research Article


Polymer for Barrier Applications. J. Appl. Phys. 2012, 111, No. 073516. (28) Zuber, K.; Markanday, J. F. S.; Hall, C.; Evans, D.; Charrault, E.; Murphy, P. Post-Polymerization Surface Segregation in Thin PECVD Siloxane Films Leading to a Self-Regenerative Effect. Plasma Process. Polym. 2017, 14, No. 1600233. (29) Loy, D. A.; Shea, K. J. Bridged Polysilsesquioxanes. Highly Porous Hybrid Organic-Inorganic Materials. Chem. Rev. 1995, 95, 1431−1442. (30) Grill, A.; Neumayer, D. A. Structure of Low Dielectric Constant to Extreme Low Dielectric Constant SiCOH Films: Fourier Transform Infrared Spectroscopy Characterization. J. Appl. Phys. 2003, 94, 6697−6707. (31) Gates, S. M.; Neumayer, D. A.; Sherwood, M. H.; Grill, A.; Wang, X.; Sankarapandian, M. Preparation and Structure of Porous Dielectrics by Plasma Enhanced Chemical Vapor Deposition. J. Appl. Phys. 2007, 101, No. 094103. (32) Grill, A.; Patel, V. Ultralow-k Dielectrics Prepared by PlasmaEnhanced Chemical Vapor Deposition. Appl. Phys. Lett. 2001, 79, 803−805. (33) Burkey, D. D.; Gleason, K. K. Organosilicon Thin Films Deposited from Cyclic and Acyclic Precursors Using Water as an Oxidant. J. Electrochem. Soc. 2004, 151, F105. (34) Schäfer, J.; Fricke, K.; Mika, F.; Pokorná, Z.; Zajíčková, L.; Foest, R. Liquid Assisted Plasma Enhanced Chemical Vapour Deposition with a Non-Thermal Plasma Jet at Atmospheric Pressure. Thin Solid Films 2017, 630, 71−78. (35) Wrobel, A. M.; Uznanski, P.; Walkiewicz-Pietrzykowska, A.; Glebocki, B.; Bryszewska, E. Silicon Oxycarbide Thin Films by Remote Microwave Hydrogen Plasma CVD Using a Tetramethyldisiloxane Precursor. Chem. Vap. Deposition 2015, 21, 88−93. (36) Lee, J.; Jang, W.; Kim, H.; Shin, S.; Kweon, Y.; Lee, K.; Jeon, H. Characteristics of Low-κ SiOC Films Deposited via Atomic Layer Deposition. Thin Solid Films 2018, 645, 334−339. (37) Walkiewicz-Pietrzykowska, A.; Uznanski, P.; Wrobel, A. M. Silicon Carbide, Silicon Carbonitride, and Silicon Oxycarbide Thin Films Formed by Remote Hydrogen Microwave Plasma CVD. Curr. Org. Chem. 2017, 21, 2229−2239. (38) Lau, K. K.; Pryce Lewis, H. G.; Limb, S. J.; Kwan, M. C.; Gleason, K. K. Hot-Wire Chemical Vapor Deposition (HWCVD) of Fluorocarbon and Organosilicon Thin Films. Thin Solid Films 2001, 395, 288−291. (39) Trujillo, N. J.; Wu, Q.; Gleason, K. K. Ultralow Dielectric Constant Tetravinyltetramethylcyclotetrasiloxane Films Deposited by Initiated Chemical Vapor Deposition (ICVD). Adv. Funct. Mater. 2010, 20, 607−616. (40) Chen, N.; Reeja-Jayan, B.; Liu, A.; Lau, J.; Dunn, B.; Gleason, K. K. ICVD Cyclic Polysiloxane and Polysilazane as Nanoscale ThinFilm Electrolyte: Synthesis and Properties. Macromol. Rapid Commun. 2016, 37, 446−452. (41) Kwan, M. C.; Gleason, K. K. Pyrolytic CVD of Poly(Organosiloxane) Thin Films. Chem. Vap. Deposition 1997, 3, 299− 301. (42) Tripp, C. P.; Hair, M. L. Chemical Attachment of Chlorosilanes to Silica: A Two-Step Amine-Promoted Reaction. J. Phys. Chem. 1993, 97, 5693−5698. (43) Tripp, C. P.; Hair, M. L. Reaction of Alkylchlorosilanes with Silica at the Solid/Gas and Solid/Liquid Interface. Langmuir 1992, 8, 1961−1967. (44) Angst, D. L.; Simmons, G. W. Moisture Absorption Characteristics of Organosiloxane Self-Assembled Monolayers. Langmuir 1991, 7, 2236−2242. (45) Fadeev, A. Y.; McCarthy, T. J. Self-Assembly Is Not the Only Reaction Possible between Alkyltrichlorosilanes and Surfaces: Monomolecular and Oligomeric Covalently Attached Layers of Dichloro- and Trichloroalkylsilanes on Silicon. Langmuir 2000, 16, 7268−7274.

(8) Johnson, R. W.; Hultqvist, A.; Bent, S. F. A Brief Review of Atomic Layer Deposition: From Fundamentals to Applications. Mater. Today 2014, 17, 236−246. (9) Kim, H. Atomic Layer Deposition of Metal and Nitride Thin Films: Current Research Efforts and Applications for Semiconductor Device Processing. J. Vac. Sci. Technol., B: Microelectron. Nanometer Struct.Process., Meas., Phenom. 2003, 21, 2231. (10) Singh, J. A.; Thissen, N. F. W.; Kim, W.-H.; Johnson, H.; Kessels, W. M. M.; Bol, A. A.; Bent, S. F.; Mackus, A. J. M. AreaSelective Atomic Layer Deposition of Metal Oxides on Noble Metals through Catalytic Oxygen Activation. Chem. Mater. 2018, 30, 663− 670. (11) Hämäläinen, J.; Ritala, M.; Leskelä, M. Atomic Layer Deposition of Noble Metals and Their Oxides. Chem. Mater. 2014, 26, 786−801. (12) Sundberg, P.; Karppinen, M. Organic and Inorganic−organic Thin Film Structures by Molecular Layer Deposition: A Review. Beilstein J. Nanotechnol. 2014, 5, 1104−1136. (13) Liang, X.; Weimer, A. W. An Overview of Highly Porous Oxide Films with Tunable Thickness Prepared by Molecular Layer Deposition. Curr. Opin. Solid State Mater. Sci. 2015, 19, 115−125. (14) George, S. M.; Yoon, B.; Dameron, A. A. Surface Chemistry for Molecular Layer Deposition of Organic and Hybrid OrganicInorganic Polymers. Acc. Chem. Res. 2009, 42, 498−508. (15) Zhou, H.; Bent, S. F. Fabrication of Organic Interfacial Layers by Molecular Layer Deposition: Present Status and Future Opportunities. J. Vac. Sci. Technol., A 2013, 31, No. 040801. (16) Maex, K.; Baklanov, M. R.; Shamiryan, D.; Lacopi, F.; Brongersma, S. H.; Yanovitskaya, Z. S. Low Dielectric Constant Materials for Microelectronics. J. Appl. Phys. 2003, 93, 8793−8841. (17) Wang, M.; Chandra, H.; Lei, X.; Mallikarjunan, A.; Cuthill, K.; Xiao, M. Atomic Layer Deposition of Carbon Doped Silicon Oxide by Precursor Design and Process Tuning. J. Vac. Sci. Technol., A 2018, 36, No. 021509. (18) Zheng, L.; Dai, Y.-W.; Yu, L.-J.; Chen, L.; Sun, Q.-Q.; Zhang, D. W. SiCOH-Based Resistive Random Access Memory for Backend of Line Compatible Nonvolatile Memory Application. Jpn. J. Appl. Phys. 2017, 56, No. 04CE10. (19) Gleason, K. K.; Yang, R. Antifouling and Chlorine-Resistant Ultrathin Coatings on Reverse Osmosis Membranes. US2014/ 0299538A12014. (20) Jiang, Y.-B.; Liu, N.; Gerung, H.; Cecchi, J. L.; Brinker, C. J. Nanometer-Thick Conformal Pore Sealing of Self-Assembled Mesoporous Silica by Plasma-Assisted Atomic Layer Deposition. J. Am. Chem. Soc. 2006, 128, 11018−11019. (21) Lee, J.; Ha, J.-H.; Song, I.-H. Improving the Antifouling Properties of Ceramic Membranes via Chemical Grafting of Organosilanes. Sep. Sci. Technol. 2016, 51, 2420−2428. (22) Reeja-Jayan, B.; Chen, N.; Lau, J.; Kattirtzi, J. A.; Moni, P.; Liu, A.; Miller, I. G.; Kayser, R.; Willard, A. P.; Dunn, B.; Gleason, K. K. A Group of Cyclic Siloxane and Silazane Polymer Films as Nanoscale Electrolytes for Microbattery Architectures. Macromolecules 2015, 48, 5222−5229. (23) Chen, N.; Reeja-Jayan, B.; Lau, J.; Moni, P.; Liu, A.; Dunn, B.; Gleason, K. K. Nanoscale, Conformal Polysiloxane Thin Film Electrolytes for Three-Dimensional Battery Architectures. Mater. Horiz. 2015, 2, 309−314. (24) O’Shaughnessy, W. S.; Gao, M.; Gleason, K. K. Initiated Chemical Vapor Deposition of Trivinyltrimethylcyclotrisiloxane for Biomaterial Coatings. Langmuir 2006, 22, 7021−7026. (25) Lewis, H. G. P.; Edell, D. J.; Gleason, K. K. Pulsed-PECVD Films from Hexamethylcyclotrisiloxane for Use as Insulating Biomaterials. Chem. Mater. 2000, 12, 3488−3494. (26) Jiao, K.; Zhou, C.; Becerra-Mora, N.; Fiske, J.; Kohli, P. VaporEnhanced Covalently Bound Ultra-Thin Films on Oxidized Surfaces for Enhanced Resolution Imaging. J. Mater. Chem. C 2016, 4, 8634− 8647. (27) Coclite, A. M.; Gleason, K. K. Global and Local Planarization of Surface Roughness by Chemical Vapor Deposition of Organosilicon H


Research Article

ACS Applied Materials & Interfaces (46) Brzoska, J. B.; Azouz, I. B.; Rondelez, F. Silanization of Solid Substrates: A Step Toward Reproducibility. Langmuir 1994, 10, 4367−4373. (47) Lowe, R. D.; Pellow, M. A.; Stack, T. D. P.; Chidsey, C. E. D. Deposition of Dense Siloxane Monolayers from Water and Trimethoxyorganosilane Vapor. Langmuir 2011, 27, 9928−9935. (48) Osterholtz, F. D.; Pohl, E. R. Kinetics of the Hydrolysis and Condensation of Organofunctional Alkoxysilanes: A Review. J. Adhes. Sci. Technol. 1992, 6, 127−149. (49) Rumble, J. R., Ed.; CRC Handbook of Chemistry and Physics, 98th ed.; CRC Press, 1977. (50) Calculated Using Advanced Chemistry Development (ACD/ Labs) Software V11.02 (© 1994-2018ACD/Labs). Accessed through Scifinder. 2018. (51) O’Hare, L.-A.; Parbhoo, B.; Leadley, S. R. Development of a Methodology for XPS Curve-Fitting of the Si 2p Core Level of Siloxane Materials. Surf. Interface Anal. 2004, 36, 1427−1434. (52) Kim, A.; Filler, M. A.; Kim, S.; Bent, S. F. Layer-by-Layer Growth on Ge(100) via Spontaneous Urea Coupling Reactions. J. Am. Chem. Soc. 2005, 127, 6123−6132. (53) Socrates, G. Infrared and Raman Characteristic Group Frequencies; John Wiley & Sons, 2004. (54) Lin-Vien, D.; Colthup, N. B.; Fateley, W. G.; Grasselli, J. G. The Handbook of Infrared and Raman Characteristic Frequencies of Organic Molecules; Elsevier, 1991. (55) Dubois, G.; Volksen, W.; Magbitang, T.; Sherwood, M. H.; Miller, R. D.; Gage, D. M.; Dauskardt, R. H. Superior Mechanical Properties of Dense and Porous Organic/Inorganic Hybrid Thin Films. J. Sol−Gel Sci. Technol. 2008, 48, 187−193. (56) Dubois, G.; Volksen, W.; Magbitang, T.; Miller, R. D.; Gage, D. M.; Dauskardt, R. H. Molecular Network Reinforcement of Sol−Gel Glasses. Adv. Mater. 2007, 19, 3989−3994. (57) Woollam, J. A.; Johs, B. D.; Herzinger, C. M.; Hilfiker, J. N.; Synowicki, R. A.; Bungay, C. L. Overview of Variable-Angle Spectroscopic Ellipsometry (VASE): I. Basic Theory and Typical Applications; International Society for Optics and Photonics, 1999; Vol. 10294, p 1029402. (58) Chang, C.-C.; Wen-Chang, C. Synthesis and Optical Properties of Polyimide-Silica Hybrid Thin Films. Chem. Mater. 2002, 14, 4242− 4248. (59) Herzinger, C. M.; Johs, B.; McGahan, W. A.; Woollam, J. A.; Paulson, W. Ellipsometric Determination of Optical Constants for Silicon and Thermally Grown Silicon Dioxide via a Multi-Sample, Multi-Wavelength, Multi-Angle Investigation. J. Appl. Phys. 1998, 83, 3323−3336. (60) Semenov, V. V.; Ladilina, E. Y.; Lapshina, E. V.; Gorshkov, O. N.; Kasatkin, A. P.; Skamnitskii, D. V.; Shenina, M. E.; Kruglov, A. V.; Dashaev, A. A.; Sharapov, A. N. Induced Refractive Index in Thin Films of Polymeric Organosilanes under UV Irradiation. Russ. J. Appl. Chem. 2011, 84, 1809−1812. (61) Furusawa, T.; Ryuzaki, D.; Yoneyama, R.; Homma, Y.; Hinode, K. Heat and Moisture Resistance of Siloxane-Based Low-DielectricConstant Materials. J. Electrochem. Soc. 2001, 148, F175. (62) Lillethorup, M.; Bergsman, D. S.; Sandoval, T. E.; Bent, S. F. Photoactivated Molecular Layer Deposition through Iodo−Ene Coupling Chemistry. Chem. Mater. 2017, 29, 9897−9906. (63) Zhou, H.; Toney, M. F.; Bent, S. F. Cross-Linked Ultrathin Polyurea Films via Molecular Layer Deposition. Macromolecules 2013, 46, 5638−5643. (64) Loscutoff, P. W.; Zhou, H.; Clendenning, S. B.; Bent, S. F. Formation of Organic Nanoscale Laminates and Blends by Molecular Layer Deposition. ACS Nano 2010, 4, 331−341. (65) Bergsman, D. S.; Closser, R. G.; Tassone, C. J.; Clemens, B. M.; Nordlund, D.; Bent, S. F. Effect of Backbone Chemistry on the Structure of Polyurea Films Deposited by Molecular Layer Deposition. Chem. Mater. 2017, 29, 1192−1203.

I


Molecular Layer Deposition of a Highly Stable ... - ACS Publications

Recommend Documents