Molecular layer deposition (MLD) processes involving two precursors are commonly employed for the growth of conformal thin films. However, the use of two precursors limits the combinations of material properties that can be accessed during film synthesis. Here, we develop a robust, three-precursor MLD process for a hybrid film that incorporates a desirable acrylate, methyl-methacrylate (MMA), together with aluminum into its repeating structure. We report a film growth rate of 3.5 Å/cycle at 110 °C, constant growth per cycle between 100 and 130 °C, and good stability of the film when exposed to ambient. We propose reaction pathways for the incorporation of MMA into the film, and by using infrared spectroscopy and x-ray photoelectron spectroscopy, we identify the reaction pathway as a non-zwitterionic aminoacrylate reaction. This study offers new insight into the use of more than two precursors in the design of an acrylate-based MLD film and provides a framework that can be adopted for subsequent three-precursor film designs.
Methyl-methacrylate (MMA), like most acrylate-based films, has garnered scientific interest over the years because of its immense industrial applications.1,2 For use as a surface coating in applications such as solar cells,3 tissue adhesives,4 and sensors,5 this α-β unsaturated ester is widely sought after because of its thermal stability, durability, and good adhesive properties.6,7 In addition, the electrochemical stability and high mechanical flexibility of MMA7 make it valuable for applications in batteries and other electrochemical devices where chemical and mechanical instabilities abound. As such, MMA has been used as a solid electrolyte,8 an electrode binder,9 and a battery separator10,11 to mitigate instabilities in lithium-based batteries. Yet, the intrinsic physicochemical properties of MMA fall short of some of the stringent requirements of certain technological systems. For instance, while MMA films assuage safety hazards in batteries, they present low ionic conductivities. As a result, MMA films are typically blended with other conductive materials that compensate for their low ionic conductivity.8,12 This advantage of combining the beneficial properties of MMA with those of other films makes the synthesis of MMA-based hybrid films an exciting area of research.
Most MMA-based metal hybrid films have been demonstrated using traditional techniques like spin coating,13,14 plasma spray,15 and initiated chemical vapor deposition.16 However, owing to the high surface area and high aspect ratio of today’s modern devices, some of these techniques may not be able to provide control over uniformity and conformality required for thin films. Atomic layer deposition (ALD) and molecular layer deposition (MLD) are techniques that provide excellent conformality and precise thickness control of thin films via a series of sequential and surface-limited reactions.17–19 ALD allows the synthesis of inorganic films with metal centers, while MLD enables the synthesis of organic films. By combining both of these techniques in two-precursor schemes, hybrid inorganic-organic films with versatile properties can be deposited, as reported in the literature, some of which include aluminum alkoxides (widely known as alucones)20 and zinc alkoxides.21 These alkoxides possess a fine blend of mechanical flexibility, afforded by their organic precursor (usually ethylene glycol), and high bond strengths, afforded by their inorganic precursor.22,23 However, because typical hybrid schemes use two precursors, there is a limit to the combinatorial space of film properties that can be achieved. In addition, promising precursors like MMA and other acrylates are difficult to incorporate into two-precursor MLD schemes, owing to the challenge of finding precursors that react favorably with MMA in cyclic deposition systems. Three-precursor MLD expands the number of bonds that can be accessed at the surface, making the incorporation of precursors like MMA more likely.
Here, using a three-precursor scheme, we report the synthesis of a new aluminum-methyl-methacrylate (Al-MMA) hybrid film grown using MLD. In the order of their surface reactions, we pulse trimethylaluminum (TMA), a volatile inorganic precursor useful for reacting favorably with surface appended hydroxyl groups;24 ethanolamine (EA), a heterobifunctional organic linker; and MMA. Using EA as a heterobifunctional linker mitigates termination reactions as a result of its distinct —OH and —NH2 appendages and provides flexible connection pathways between the metal center and the third precursor.25 First, by demonstrating the possibility of combining more than two precursors in surface-limited reaction schemes, we form an Al-MMA hybrid film using a strategy that expands the number of films that can be synthesized using MLD. Second, we propose mechanistic pathways for this novel three-precursor synthesis process. Finally, using a suite of characterization techniques—x-ray photoelectron spectroscopy (XPS) and infrared (IR) spectroscopy—we validate one of the proposed reaction schemes. This work serves as a demonstration for motivating the use of more than two components in the synthesis of thin films by MLD.
TMA, EA (≥99.5%), and MMA (99%) were purchased from Sigma-Aldrich. TMA and EA were stored in a nitrogen-purged glovebox, and MMA was stored in a refrigerator until use. For all deposition experiments, thin films were grown on single-sided polished silicon (100) wafers purchased from WRS Materials. Right before use, silicon wafers were cleaned for 15 min in a Novascan PSD series digital UV ozone system to remove residual organic contaminants and then placed into the MLD vacuum chamber for experiments.
Hybrid thin films were deposited using a hot-wall, MLD flow reactor described previously,26,27 pumped by a Leybold Trivac rotary vane pump. The reaction chamber was heated by external heating tapes, and its temperature was maintained at the respective deposition temperatures used in this study. TMA and MMA were held at room temperature, while EA was heated to 50 °C during deposition, owing to its low vapor pressure. Nitrogen (N2) was used both as a carrier and purge gas. The total N2 flow rate was fixed at 20 SCCM, which resulted in a base pressure of 1200 mTorr in the reactor. All temperatures were controlled by variable AC power transformers, and pressure was measured using a K. J. Lesker convection gauge. Precursors and N2 purge gas were introduced into the reactor from source manifold using solenoid valves controlled by labview.
After establishing saturation conditions, the Al-MMA hybrid films were grown at 110 °C with repeated cycles of the following pulsing sequence: the TMA precursor was dosed into the reaction chamber for 5 s after which the reacting system was allowed to soak (nitrogen turned off and pump closed) for 5 s, providing the TMA with sufficient time to react with the surface species. Afterward, the unreacted TMA and by-products of its reaction were purged with N2 for 90 s. The corresponding EA pulse-soak-purge sequence was 5-30-180 s, while the pulse-soak-purge sequence for the third precursor, MMA, was 5-30-180 s. Long purge times were used to assure the removal of excess precursor and reaction by-products. After deposition, samples were removed from the reactor for ex situ analysis.
As-deposited thin film thicknesses were measured by variable-angle spectroscopic ellipsometry (VASE) immediately after the samples were removed from the reaction chamber. VASE was performed using a J.A. Woollam Co, α-SE ellipsometer in the spectral range of 380–900 nm at incidence angles of 65° and 70°. The optical model consisted of a silicon substrate and a 1.65 nm thick native silicon oxide layer. The refractive index values were derived from ellipsometer parameters for this hybrid film deposited on silicon substrate using the Cauchy model.
FTIR spectra were measured using a VERTEX 70 FTIR spectrometer from Bruker with a Harrick VariGATR attachment fitted with a germanium attenuated total reflectance (ATR) plate and a mercury-cadmium-telluride detector. Background signal was collected by measuring IR absorbance from an empty ATR plate. MLD hybrid thin film deposited Si wafer samples were pressed onto a Ge ATR plate using a constant force of approximately 800 N. Liquid samples of EA and MMA were placed directly on the Ge ATR plate and the signal was collected without any applied force. Spectra were recorded at 4 cm−1 resolution with 128 scans.
XPS was performed on a PHI VersaProbe III scanning x-ray photoelectron spectrometer with Al Kα X-ray source radiation of 1486 eV energy. The x-ray beam spot size was performed on 200 × 200 μm with 100 W power and 20 kV for all experiments. Three to four scans of 0.8 eV/step resolution and 20 ms/step at a 224 eV pass energy were used for the survey scans. Higher resolution elemental scans were carried out with a 0.1 eV/step resolution with a 55 eV pass energy for 20 ms/step.
III. RESULTS AND DISCUSSION
A. Film growth properties
Growth properties of MLD films are typically characterized by the film growth per cycle (GPC), the saturation behavior for each precursor, and the growth behavior across a range of temperatures.17,22 In this study, we determine these growth properties for the three-precursor MLD process, with results shown in Figs. 1 and 2. Owing to the stability of the three precursors between 100 and 130 °C,25,28 we use 110 °C for the saturation and GPC tests. Saturation tests are carried out to ascertain if the reaction of gaseous precursors in MLD processes are surface reaction limited. To test for saturation, we systematically vary the pulse lengths of each precursor molecule while supplying the two other precursor molecules in excess (5 s) for 20 MLD cycles with cyclic doses of TMA, EA, and MMA. Figures 1(a)–1(c) plot the GPC as a function of TMA, EA, and MMA pulse length, respectively. The data show that the film growth rate saturates at a GPC value of 3.5 Å/cycle when the pulse length of TMA exceeds 2 s [Fig. 1(a),], that saturation occurs when the EA exposure exceeds 1 s [Fig. 1(b)], and that saturation occurs at an MMA pulse length above 1 s, again with a GPC of 3.5 Å/cycle [Fig. 1(c)]. However, some of the data reveal a nonideal behavior at 0 s pulse times. In Fig. 1(b), we observe that for a 0 s pulse length of EA, the recorded growth rate is almost 1 Å/cycle. The observation of film growth without dosing EA is unexpected since the surface product of the TMA reaction should ideally not react with MMA. We speculate that the anomalous growth at 0 s of EA is a consequence of the presence of trace moisture in the reactor, which can react with TMA, resulting in the formation of Al2O3. This speculation is consistent with the growth rate of 1 Å/cycle for Al2O3 that we achieve using TMA and H2O in our reactor. In addition, evident in Fig. 1(c) is a high growth rate of ∼2.5 Å/cycle when the pulse length of MMA is 0 s, indicating that TMA reacts directly with EA in a cyclic scheme, without the need for a third precursor (MMA), as demonstrated in a previous report.25 Finally, using the established saturation conditions, the growth rate of the hybrid film is maintained at 3.5 Å/cycle over a wide range of cycle numbers [Fig. 1(d)], showing that its growth is consistent, self-limiting, and reproducible. It is noteworthy that the observed growth rate of our hybrid film (3.5 Å/cycle) is different from an end-to-end chain length estimation of all three precursors (15.5 Å/cycle). This difference between real and ideal growth rates suggests that our film adopts nonideal growth configurations because of the flexibility of our organic precursors, similar to other MLD reports.29–31
Probing the temperature-dependence of the MLD process is critical for generalizing deposition conditions and understanding the effect of deposition temperature on film properties. Between 70 and 170 °C, we determine the GPC of the hybrid film after 20 cycles of MLD and observe three distinct regions of growth behavior, regions I, II, and III (Fig. 2). In region I where the temperature is less than 110 °C, GPC is larger than the other regions with values between 4.3 and 7 Å/cycle, possibly because of the condensation of precursors. Within 110 and 120 °C (region II), GPC remains relatively constant at 3.5 Å/cycle. In region III, at temperatures greater than 120 °C, GPC reduces with the temperature, leading to values between 2.5 and 3.1 Å/cycle. The decrease in GPC at higher temperature is tentatively attributed to the desorption of metallic components in the film, as has been previously reported for other similar hybrid MLD processes, including for a three-precursor MLD process.20,32 Region II can be established as the so-called MLD window because GPC remains constant within it.
We also investigate the stability of the films in the ambient. For films grown at 110 °C, thickness does not change significantly over the course of seven days when exposed to air (supplementary Fig. 1).43 This result suggests that the film remains stable in air on the time scale of at least days. Using IR, we also show that the film retains its key features (CH2 and C=O bonds) after one day of air exposure (supplementary Fig. 2),43 also supporting that the film is stable. See the supplementary material43 for evidence of hybrid film stability.
B. Elucidating reaction mechanisms
Because this represents the first report of three-precursor MLD of Al-MMA hybrids, we consider possible reaction mechanisms associated with each step of the cycle. Figure 3(a) shows the three precursors in the order that they are introduced into the reactor, with precursor 1 (TMA) introduced first, precursor 2 (EA) introduced next, and precursor 3 (MMA) introduced last, in each MLD cycle. The first reaction step, shown in Fig. 3(b), involves the reaction of TMA with the terminating hydroxyl groups on the silicon substrate. This reaction has been repeatedly verified in numerous ALD and MLD processes,20,24,25 and it proceeds via the formation of an aluminum—oxygen bond and the evolution of methane. The Al—O compound remains coordinated to the surface sites, and CH4 is purged out.
The second reaction step has also been reported previously.25 EA reacts with the surface species formed in the first reaction step. The nucleophilicity of the oxygen and nitrogen atoms in EA make them the primary points of electron exchange and bond formation between EA and the appended molecule. However, the reaction of the hydroxyl species on EA with the appended molecule has shown to be the dominant pathway,32 through which it forms an Al—O bond and generates CH4 as a by-product [Fig. 3(c)]. Here, the reaction sequence shows one of the benefits of using a heterobifunctional molecule in the form of EA: as demonstrated in a previous report30 by using a precursor with functional groups of different binding probabilities, we reduce the possibility of self-termination by reaction of both ends of the precursor, a pervasive phenomenon in MLD processes.33
The third step in the MLD process involves a reaction between MMA and the amine group of the surface species. No reaction of this type has been documented in the gas phase, and as a result, we consider in more depth the plausible pathways for electron exchange between the appended —NH2 group and the MMA molecule. MMA is an alpha-beta unsaturated ketone, indicating that its double bond creates an electron deficiency within the orbitals of the alpha and beta-carbon atoms, making them electrophilic. One possible reaction pathway is the donation of a lone pair of electrons from the appending —NH2 group to the beta-carbon resulting in the formation of a reaction intermediate that requires resonance stabilization, as shown in Fig. 4(a). This intermediate structure is likely in resonance with the intermediate formed in Fig. 4(b), in which charge is localized atop oxygen. The resonance structure could then be stabilized by a proton transfer induced by the electrophilicity of N and the nucleophilicity of the alpha-carbon atom, in the intermediate molecule [Fig. 4(a)]. This pathway is similar to that proposed by Klemarczyk.34 It would lead to the formation of an aminoacrylate group at the surface (product 1).
A second possible reaction pathway for the third step in the MLD process derives from the first and proceeds with a nucleophilic attack on the beta-carbon. However, the resonance stabilization follows a slightly different pathway in this mechanism: the nucleophilic alpha-carbon further donates electrons to the adjacent carbonyl group, resulting in an electronegativity-induced removal of electrons from the carbonyl group by oxygen [Fig. 4(b)]. A proton transfer follows, resulting in the formation of an enolate (product 2), a tautomer of product 1 [Fig. 4(b)]. This resonance stabilization pathway was proposed by Duffy et al.35 It is noteworthy that product 1 likely dominates the keto-enol equilibrium with product 2.36 Nevertheless, we still consider the formation of product 2 to provide a more complete account of possible products.
The third possible reaction pathway, known as amidation, involves a reaction between the nitrogen atom and the carbon atom in the carbonyl structure. The carbon atom is electrophilic and as a result is expected to serve as an electron acceptor from the nucleophilic nitrogen atom appended to the surface [Fig. 4(c)]. Subsequently, there is a stabilization of resonance structures that involves the removal of the methoxy leaving group, and a proton transfer toward the formation of methanol and an amide (product 3) [Fig. 4(c)]. This pathway has also been documented in primary amine amidation studies.37
All three plausible products possess distinct bonds that, if identified experimentally, could clarify the reaction mechanism present in this three-precursor MLD process. To help determine the mechanism, we utilize structural and chemical characterization tools, namely IR spectroscopy and XPS, to analyze the MLD film.
To facilitate the identification of the chemical bonds in the film, we compare the IR spectrum of the deposited hybrid film with reference spectra of liquid samples of two of the MLD precursors, EA and MMA. Figure 5(a) shows the IR spectra of the deposited hybrid film and the MMA and EA reference samples, while Figs. 5(b) and 5(c) show the enlarged portions of the spectra over the CH, OH, and NH2 stretching range and C=O stretching range, respectively. The MMA and EA spectra shown in Fig. 5 are consistent with those reported in the literature.38,39 The EA spectrum displays its characteristic —CH2 symmetric and antisymmetric stretches at 2860 and 2930 cm−1 [Fig. 5(b)]. The peaks at 3290 and 3360 cm−1 correspond to the two —NH2 stretches of a primary amine, while the peak at 3180 cm−1 corresponds to an —OH stretch [Fig. 5(b)]. In addition, the MMA spectrum displays its characteristic vibrational peaks. At 1730 cm−1, we identify a sharp peak that corresponds to the C=O bond present in MMA [Fig. 5(c)]. Also, the symmetric and antisymmetric stretches of the —CH3 bond at 2950 and 3000 cm−1, respectively, can be seen in the MMA spectrum [Fig. 5(b)].
The spectrum of the hybrid film in Fig. 5(a) exhibits many similarities with the two reference spectra, as expected since most of the bonds in both precursors should remain unchanged in the resulting film. The broad —OH peak between 3200 and 3600 cm−1 in the film spectrum is likely due to water adsorbed by the film after air exposure, as has been observed with similar MLD films.20,32 The presence of —CH2 peaks at 2860 and 2930 cm−1 [Fig. 5(b)] provides support that the EA backbone in the surface species remains intact after MMA reacts with the surface. It is evident that the stretching mode from the C=O bond in MMA at 1730 cm−1 retains its position [Fig. 5(c)] in the hybrid film, indicating that the carbonyl group does not undergo a chemical transformation during the addition of MMA. This finding leaves products 1 and 3 as the only plausible options of the proposed reaction schemes, because product 2 does not retain the carbonyl group. Furthermore, the C=O bond stretching mode in product 3 would be expected to appear near 1650 cm−1 as is typical for amides,40,41 but the observation of the C=O stretch at 1730 cm−1 in the hybrid film indicates that product 3 is not likely. Moreover, we do not observe prominent peaks in the hybrid spectrum from C=C bonds, which would appear at a wavenumber of ∼1600 cm−1 [Fig. 5(b)]. The absence of a C=C stretching peak suggests the absence of products 2 and 3, since both contain alkene groups. Hence, the IR spectra suggest that product 1 dominates.
To further corroborate the product identified by IR, we carry out XPS to analyze the chemical composition of a 10 nm thick MLD film [Fig. 6(a)]. To avoid issues with selective removal of some elements in these sensitive MLD films, XPS data were collected without surface sputtering. The XPS survey spectrum indicates the presence of all the expected component atomic species, Al, C, O, and N, in the percentages indicated in Table I. The expected monolayer repeat unit in the film for product 1 is Al2(C7NO3)3, and the “ideal” Al, C, O, and N percentages for this structure are also included in Table I. It is evident that the measured and ideal percentages differ. The disparity in composition between the observed and predicted structures indicates higher than expected Al and O content and lower C and N content. We postulate that this disparity is a result of TMA reacting with trace moisture during the MLD process or the reaction of unsaturated Al bonds in the film with H2O in the atmosphere, resulting in the formation of Al2O3 domains within the hybrid MLD film. To rationalize the XPS-measured composition of the film, we define a term called the blended composition, which is derived from assuming that the MLD film is composed of subdomains of Al2O3, providing a final film stoichiometry of the form Al2(C7NO3)3:n(Al2O3). By varying n, identified with the relative contribution from Al2O3 subdomains (supplementary Fig. 2),43 we find that a value of n = 2 fits a blended film composition that is more similar to the XPS film composition obtained as shown in Table I. The small difference that still exists between our blended composition and the XPS composition is very likely a result of the numerous sources of nonidealities (unreacted bonds, water absorption, and partial replacement of bonds) during three-precursor MLD processes. Similar discrepancies between expected and observed film compositions were also reported in the other three-precursor MLD reported in the literature.32 See the supplementary material43 for blended composition as a function of number of Al2O3 domains.
|Elements .||XPS composition (%) .||Ideal composition for product 1 (%) .||Blended composition (%) Al2(C7NO3)3:2(Al2O3) .|
|Elements .||XPS composition (%) .||Ideal composition for product 1 (%) .||Blended composition (%) Al2(C7NO3)3:2(Al2O3) .|
To provide further evidence for the formation of product 1, we carry out high-resolution XPS analysis for all the atomic species that constitute the MLD film. The Al 2p XPS spectrum is shown in Fig. 6(b). Although the Al atom in the MLD film is not in a unique bonding environment that could help identify differences between the three plausible products, the measured binding energy of ∼75 eV confirms that it is in an oxidized state [Fig. 6(b)].27 On the other hand, the C atom present in the film has a bonding environment that may facilitate the identification of product 1. From the C 1s high-resolution spectrum shown in Fig. 6(c), a strong shoulder at higher binding is observed on the main carbon peak, characteristic of a carbon atom in a double bond with oxygen, which is typically observed at a binding energy of 289 eV.42 Only products 1 and 3 contain a C=O group, ruling out product 2. Combined with the results of the IR measurements showing that product 3 is unlikely, these XPS results lend credence to the formation of product 1. Other high-resolution elemental peaks for N and O are shown in supplementary Fig. 3.43
The combination of IR and XPS results lead to the assignment of species 1 as the most likely product of the reaction between MMA and the surface bound EA groups. Based on the proposed mechanisms in Fig. 4, this product assignment suggests that the synthesis of the Al-MMA hybrid film involves the following sequence of steps: formation of an aminoacrylate through the transfer of electrons from nucleophilic oxygen atoms to electrophilic aluminum atoms, the transfer of electrons from nucleophilic nitrogen atoms to the beta-carbon of MMA, subsequent rearrangement of the C=C bond, and the transfer of protons from the primary amine structure. We also propose that, after the growth of the first monolayer of the hybrid film, subsequent reactions proceed via the mechanisms shown in Figs. 3 and 4(a), with the only difference being the reaction of TMA with the surface, which occurs via the formation of an Al—N bond as shown in supplementary Fig. 5.43 This mechanism is consistent with the presence of a C=O bond in our film after repeated MLD cycles. See the supplementary material43 for the proposed reaction between TMA and the repeating unit of the hybrid film. Using MLD, we demonstrate a process combining three precursors toward the formation of an Al-MMA hybrid film, opening the prospect of creating robust film properties by expanding precursor sample space.
We have demonstrated the synthesis of an Al-MMA hybrid film using a three-precursor molecular layer deposition process. We report a film growth rate of 3.5 Å per cycle under MLD in which saturation conditions are achieved for all three precursors, TMA, EA, and MMA. In addition, we verify a constant GPC for this film over a wide number of MLD deposition cycles. We also establish a narrow MLD window (constant growth region) of the film to be between approximately 110 and 120 °C. Furthermore, we propose three plausible reaction pathways for the growth of the hybrid film, and using XPS and IR spectroscopy, we conclude that the most plausible pathway involves the formation of an aminoacrylate-derivative as the repeating unit in the film. By extension, chemically similar precursors to those studied here could hold high potential in similar MLD configurations. For example, by using this three-precursor MLD scheme, the first precursor could be a metal organic compound like TMA and diethylzinc (DEZ), the second precursor could be any precursor with hydroxyl and amine groups like ethanolamine and 4-aminophenol, and the third precursor could be any alkyl-acrylate or alkyl-cyanoacrylate that has desirable chemical and physical properties. In all, this work reports a concept that can be used to broaden MLD film properties.
S.T.O., H.A., and S.F.B. conceived the idea and designed the experiments. S.T.O. and H.A carried out experiments. N.E.R. helped with data analysis. S.T.O., H.A., and S.F.B. wrote and edited the manuscript. All authors read and discussed the manuscript. S.T.O and H.A. contributed equally to this work.
S.T.O acknowledges support from the Knight Hennessy Scholarship for graduate studies at the Stanford University. H.A. acknowledges support from the TomKat Center Postdoctoral Fellowship in Sustainable Energy at Stanford. N.E.R. and S.F.B. acknowledge support from the National Science Foundation under Grant Award No. CHE-1904108. Part of this work was performed at the Stanford Nano Shared Facilities (SNSF), supported by the National Science Foundation under Award No. ECCS-1542152. The authors declare no conflict of interest.
The data that support the findings of this study are available from the corresponding author upon reasonable request.