Analysis of TiO2 Atomic Layer Deposition Surface Chemistry and

Dec 11, 2018 - Soc. , Just Accepted Manuscript. DOI: 10.1021/jacs.8b10689 .... Lanthanides were once thought to have no role in biology, but that dogm...
0 downloads 0 Views 3MB Size
Subscriber access provided by YORK UNIV

Article 2

Analysis of TiO Atomic Layer Deposition Surface Chemistry and Evidence of Propene Oligomerization using Surface-Enhanced Raman Spectroscopy Ryan A. Hackler, Gyeongwon Kang, George C. Schatz, Peter C. Stair, and Richard P. Van Duyne J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.8b10689 • Publication Date (Web): 11 Dec 2018 Downloaded from http://pubs.acs.org on December 15, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

Analysis of TiO2 Atomic Layer Deposition Surface Chemistry and Evidence of Propene Oligomerization using Surface-Enhanced Raman Spectroscopy Ryan A. Hackler†‡, Gyeongwon Kang†, George C. Schatz†, Peter C. Stair†‡, Richard P. Van Duyne*† Department of Chemistry and Center for Catalysis and Surface Science, Northwestern University, Evanston, Illinois, 60208, United States ABSTRACT: Atomic layer deposition (ALD) of TiO2 was performed in tandem with in-situ surface-enhanced Raman spectroscopy (SERS) to monitor changes in the transient surface species across multiple ALD cycles. A self-assembled monolayer of 3mercaptopropionic acid was used as a capture agent, to ensure that nucleation of the titanium precursor (titanium tetraisopropoxide [TTIP]) occurs. Comparisons between the Raman spectra of the neat precursor and the SER spectra of the 1st ALD cycle of TiO2 reveal typical ligand exchange chemistry taking place, with self-limiting behavior and intact isopropoxide ligands. However subsequent cycles show drastically different chemistry, with no isopropoxide ligands remaining at any point during the 2nd and 3rd cycles. Continuous exposure of either TTIP or isopropanol after the 1st cycle shows unlimited, chemical vapor deposition (CVD)type growth. Comparisons with alternative precursors (aluminum isopropoxide, titanium tert-butoxide, and titanium propoxide) and DFT calculations reveal that for the TTIP precursor, isolated TiO2 sites play a role in the dehydration of off-gassing isopropanol. The resulting propene then undergoes oligomerization into six-carbon olefins, before polymerizing into indistinguishable carbon products that accumulate on the surface. The emergence of the dehydration chemistry is expected to be exclusively the result of these isolated TiO2 sites, and as such is expected to occur on other surfaces where TiO2 ALD is feasible. This work showcases how seemingly innocuous ALD can evolve into a CVD process when the products can participate in various side reactions with newly made surface sites.

INTRODUCTION Atomic layer deposition (ALD) of TiO2 has been done on a variety of substrates for applications ranging from thin film overcoats to stabilize various catalysts and plasmonic substrates,1 to photovoltaic and catalyst fabrication.2-4 The wide-reaching utility of TiO2 in thin film applications stems from its thermal and chemical stability, non-toxicity, semiconductor properties, and relative ease in production.5 Due to the interplay between thin film structure and function, much work has been done to optimize ALD processes for various TiO2 thin film applications. Despite efforts to optimize thin film depositions, there is still ambiguity regarding the reaction mechanisms and intermediates associated with these ALD processes and how they play a role in properties such as film growth, purity, and uniformity.6-9 Cleveland et al.,7 for example, concluded titanium tetraisopropoxide (TTIP) surface reaction intermediates influence TiO2 growth rate by comparing their TTIP/O3 ALD results with TTIP/H2O ALD results. As such, any technique that can yield novel information about the TiO2 ALD process and its intermediates would prove invaluable in developing sophisticated materials. Surface-enhanced Raman spectroscopy (SERS) is one such technique that is capable of probing molecules and intermediates near a surface by exciting the localized surface plasmon resonance (LSPR) of a noble metal (Ag, Au, or Cu) nanostructure.10-12 The electromagnetic (EM) mechanism requires the probe molecule to be near the plasmonic surface (99%) for 24 h.

The temperature was assumed to be 298 K, and each Raman peak was broadened with a Lorentzian function with a fullwidth half-maximum of 20 cm-1.

In-Situ QMS Measurements. In-situ QMS (RGA300 Stanford Research Systems) was used to measure the TiO2 ALD reaction products. TiO2 ALD was performed during QMS measurements in the same manner as the SERS experiments (with a timing sequence of 60s-60s-60s-60s). MS signals for propene (m/z = 41) and isopropanol (m/z = 45) were monitored during the deposition. An acquisition time of 1 s was used for all QMS measurements. In-Situ QCM Measurements. In-situ QCM (Inficon Q-POD + 6 MHz Colorado Crystal Corp.) was used to measure the TiO2 ALD growth rate over several cycles. To emulate the AgFON surface used in SERS experiments, 200 nm of Ag was thermally deposited onto the crystal and incubated in a 2 mM ethanolic solution of MPA (Sigma-Aldrich, >99%) for 24 h. The crystal was exposed to TTIP and water for 180 s each, with purge steps of 60 s.

2

𝑆𝑝 = 45𝛼𝑝′ + 7𝛾′𝑝2 𝛼𝑝′ =

𝛾′𝑝 =

1 2

1 3

∑(𝛼 )

′𝑖𝑖 𝑝

𝑖

∑3(𝛼 ) (𝛼 ) ′𝑖𝑗 𝑝

′𝑖𝑗 𝑝

― (𝛼′𝑖𝑖)𝑝(𝛼′𝑖𝑖)𝑝

𝑖,𝑗

RESULTS AND DISCUSSION SER spectra were taken of the AgFON after incubation in MPA to verify nucleation sites were available for TiO2 ALD (see Figure S2). Most of the vibrational modes agree with previously reported values,27 including the strongest modes at 652 (ν(C-S)G), 734 (ν(C-S)T), and 904 (ν(C-COO)) cm-1. The only ambiguous mode is the weak peak located at 804 cm-1 (either ρ(CH2) or δ(S-H)). If the 804 cm-1 mode is the result of S-H bending, then the MPA is likely not chemisorbed onto the AgFON. This is unlikely, however, due to the affinity for thiol

X-ray Photoelectron Spectroscopy. X-ray photoelectron spectroscopy (XPS) was performed at the Keck-II Center at Northwestern University on a Thermo Scientific ESCALAB 250Xi. A monochromatic Al Kα (1486.74 eV) X-ray source was used with a 400 μm spot size. Computational Methods. All Raman spectra were calculated using the Amsterdam Density Functional (ADF) software package.22 Full geometry optimization, frequency, and polarizability calculations presented in this study were carried out using a triple- polarized type (TZP) Slater basis set and the Becke-Perdew (BP86)23,24 generalized gradient approximation (GGA) exchange-correlational functional. For the surface bound species, a Ag20 cluster was used as a surface. The zeroorder regular approximation (ZORA)25,26 was used to include the relativistic effect for all calculations. Static polarizability derivatives were calculated by two-point numerical differentiation of the polarizability using the RESPONSE module implemented in ADF. The differential Raman scattering cross sections for the pth vibrational mode was calculated according to the following equation: 𝑑𝜎 𝜋2 ℎ 1 = (𝜔 ― 𝜔𝑝)4 (𝑆 ) 𝑑Ω 𝜀20 8𝜋𝑐𝜔𝑝 𝑝 45(1 ― exp ( ― ℎ𝑐𝜔𝑝 𝑘 𝑇)) 𝐵

adsorption onto silver, and the carboxylic acid group appears to be intact to act as the nucleation site. Figure 1. SER difference spectra collected in-situ during the TiO2 ALD process on MPA-functionalized AgFON. Panel (a) shows the

ACS Paragon Plus Environment

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1st TTIP dose (in red) and 1st H2O dose (in blue), with the normal Raman spectrum (NRS) of the TTIP precursor (in black) for comparison. Panel (b) shows the 2nd TTIP dose (in red) and 2nd H2O

Page 4 of 14

dose (in blue), with the TTIP precursor spectrum shown for comparison. All major peaks are annotated, and data are shifted vertically for clarity. Data displayed in arbitrary data units (ADU).

Table 1. Vibrational modes and corresponding assignments from in-situ SER spectra during ALD. Shaded rows represent vibrations associated with olefins. i.p. = in-plane TTIP Precursor

1c TTIP SERS

1c H2O SERS

2c TTIP SERS

2c H2O SERS

Assignment

562

-

-

-

-

νs(Ti-O)

616

-

-

-

-

νs(Ti-O)

-

-

-

684

695

τ(CH2)

-

-

-

831

831

ρ(CH2)

850

834

861

-

-

νs(CCC), ν(Ti-O)

-

-

-

899

-

νs(C-(CO)=C)

923

-

-

-

-

ρr(CH3)

1023

1019

1018

-

-

ν(C-O), νs(Ti-O)

1123

1155*

1167*

1148

1150

νas(CCC)

1181

1155

1167

-

-

ρr(CH3), νs(Ti-O)

-

1267

1232

1269

1268

δ(C-H(vinyl) i.p.)

1331

1347

1347

-

-

δ(C-H)

1445

1462

1474

-

-

δas(CH3)

-

-

-

1492

1471

δ(C-H(vinyl))

-

-

-

1543

1517

ν(C=C)

-

-

-

1587

1550

ν(C=C)

2867

2863

2860

2860

2862

νs(CH3), ν(C-H)

2916

2922*

2920*

2920*

2929*

νs(CH3)

2932

2922

2920

2920

2929

νas(CH3)

2971

2969

2966

2966

2965

νas(CH3)

The in-situ SER difference spectrum taken during dosing of TTIP shows a multitude of peaks that appear before disappearing upon exposure to water in the subsequent spectrum (see Figure 1a). The difference spectra were collected by subtracting the previous background spectrum from the current SER spectrum after precursor dosing. In the low wavenumber region, most of the peaks present in the normal Raman spectrum of the TTIP precursor can be found in the 1st TTIP dose SER difference spectrum. The only exceptions include a weak mode at 923 cm-1 (ρr(CH3)) in the neat TTIP precursor, and the appearance of a mode at 1267 cm-1 in the SER spectrum that cannot be found in the neat TTIP precursor. Likewise, the only mode that uniquely shows up in the 1st H2O dose is at 1232 cm-1. In the 1st TTIP and 1st H2O difference spectra, the asymmetrical peaks found at 1155 and 1167 cm-1, respectively, are likely the agglomeration of two or three peaks, two of which can be seen in the normal Raman spectrum of TTIP (1123 and 1181 cm-1). The modes found in the SER spectra match up well with the modes found in the Raman of the neat liquid,20 suggesting some amount of isopropoxide ligands stay coordinated to the titanium center, while the other ligands undergo typical ligand exchange chemistry at the carboxylic acid site of the thiol SAM. The parity in intensity of the peaks

between the two spectra suggest essentially all surface species generated are removed with water, as expected in ALD-type behavior. The CH region (see Figure S3) shows the same, selflimiting behavior. During the second cycle, however, a different type of surface chemistry emerges (see Figure 1b). Not only do the SER spectra show new peaks indicative of new species, but the spectra also lack any modes associated with the neat TTIP precursor. The most noticeable of these missing peaks is the strong C-O stretching mode located at 1023 cm-1. This would suggest that the second ALD cycle results in unique ligand exchange chemistry, where all four of the isopropoxide ligands are displaced from the titanium center during the first half cycle. Alternatively, a different deposition mechanism, akin to chemical vapor deposition (CVD), takes place during the second cycle and introduces new surface species. The peak seen at 1267 cm-1 in the first cycle shows up more prominently in the second cycle, suggesting the new surface chemistry was mildly present during the first cycle. Multiple spectra were also collected throughout the course of a single dose to evaluate the new surface chemistry taking place and the identity of the new surface species. By monitoring the change in peaks from spectrum to spectrum while a single dose is occurring, one can determine whether the chemistry is self-limiting. If the

ACS Paragon Plus Environment

Page 5 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

chemistry is not self-limiting, one can evaluate the continual generation of specific chemical species. During the first 30 s of

exposure

Figure 2. SER difference spectra taken from 30 s doses of TTIP (in red) and H2O (in blue) during the first (a and b) and second (c and d) cycles. Positive peaks represent new species being generated from the previous spectrum. Negative peaks represent removal of surface species from the previous spectrum. Data displayed in ADU.

to the TTIP precursor (see Figure 2a), multiple modes indicative of TTIP surface species can be seen. These modes match up well with the modes seen in the first ALD cycle in Figure 1. Another 30 s of TTIP exposure shows a small disappearance of the 1155 cm-1 mode, and the appearance of modes associated with the new surface species as previously seen in the 2nd ALD cycle (i.e. the mode at 1267 cm-1). A cumulative exposure of 90 s of TTIP, however, shows the reduction of these unknown modes, leaving only TTIP modes. The subsequent water exposure shows removal of the TTIP modes (see Figure 2b), with 90 s being sufficient to remove the isopropoxide ligands. These spectra are clear indicators of the self-limiting chemistry that’s expected in ALD. The exposures during the second cycle (see Figure 2c, d), however, show a more complex and continuous surface reaction taking place. Throughout the 30 s pulses of TTIP during the second cycle, modes around 1587, 1269, 1148, 899, and 684 cm-1 continue to grow in. Modes around 1492 and 1543 cm-1 fluctuate as either positive or negative peaks during the TTIP exposures. During the first 30 s of water exposure for the second ALD cycle, almost all the peaks observed are positive. It is only

after 60 s of water exposure that removal of these surface species begins to take place. Overall, it appears that extended exposure of water is not sufficient to remove these new surface species being generated from the second cycle TTIP exposure. Furthermore, this trend of incomplete species removal continues onto the third ALD cycle (see Figure S4), where the peak area during TTIP exposure is larger than the peak area during water exposure. These spectra suggest new, strongly adsorbing species are forming from constant exposure of TTIP during a CVD-like process. The peak fluctuations also suggest a multitude of species created from the isopropoxide ligands are present during this process. QCM measurements of the deposition process (see Figure S5) corroborate the behavior observed via SERS and shows the expected sub-monolayer growth rates occur only during the 1st ALD cycle, with multiplelayer growth rates occurring in subsequent cycles. These growth rates range from 1 to 4 Å per cycle and are indicative of non-ALD reaction behavior. The isopropoxide ligands from TTIP are suspected to form either isopropanol, acetone, diisopropyl ether, and/or propene, the most likely of which being isopropanol.9 It is possible that

ACS Paragon Plus Environment

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

the resulting isopropanol then undergoes β-hydride elimination to form propene, based on the work by Johnson and Stair, who saw ALD-type behavior for TTIP on a MoOx substrate (at 100 °C and base pressure 1 x 10-6 Torr) for the first ALD cycle, but upon a second ALD cycle saw more CVD-type behavior where propene is likely to form.28 They concluded the dehydration pathway becomes dominant under lower pressures because the lower flux of TTIP molecules leaves the sub-monolayer of initial TiO2 sites unsaturated and available for dehydration chemistry. These findings agree with the results Bondrachuk et al. 29 found on TiO2(110) from 300 to 450 K on undercoordinated Ti4+ ions. While the system interrogated herein was done under higher pressures than Johnson, the dehydration pathway can still play a role under the reaction conditions listed due to the potential for undercoordinated TiO2 sites. XPS results suggest such sites are on the surface as Ti3+ sites30 after 5 ALD cycles (see Figures S6-S8). QMS results from TiO2 ALD on just the steel walls of the reactor show negligible amounts of propene as an off-gassing species (see Figure S9). This would suggest the isolated TiO2 sites are capable of dehydration, due to being undercoordinated, while normal TiO2 sites cannot perform this chemistry given the high flux of TTIP. If undercoordinated TiO2 sites are playing a role in the formation of propene, then we should see evidence of propene. Furthermore, if we move away from undercoordinated TiO2 sites to fully coordinated TiO2 sites, then we should see a dramatic decrease in propene production near the surface. Use of several ALD cycles appears to be enough to grow fully coordinated TiO2 sites, as the SER signal from the products being generated on the surface decreases significantly (see Figure S8).

Figure 3. SER difference spectra taken from 30 s cumulative doses of isopropanol after one TiO2 cycle on AgFON-MPA had been deposited using TTIP, as well as the sum of all isopropanol exposure spectra. Data displayed in ADU.

Several control experiments were performed to confirm that propene was in-fact forming due to dehydration via the TiO2 sites. Firstly, isopropanol was dosed onto AgFON-MPA after one TiO2 cycle and monitored via SERS (see Figure 3) to determine whether the isopropoxide ligands were participating in the change in chemistry, regardless of whether additional TiO2 sites were being generated at the same time. Secondly, three alternative precursors (aluminum isopropoxide [AlIPO],

Page 6 of 14

titanium tert-butoxide [TTB], and titanium propoxide [TP]) were used and monitored via SERS during the deposition to determine the role TiO2 and ligands play in the new chemistry. AlIPO was used to determine whether the TiO2 sites were responsible for the dehydration chemistry or if Al2O3 could perform a similar role. TTB was chosen because tert-butoxide ligands are not expected to undergo dehydration as easily, and therefore the CVD-type behavior was not expected to occur despite TiO2 being present. Conversely, TP was selected as a comparable precursor to TTIP where dehydration chemistry was expected to occur. SER spectra collected for AlIPO and TTB experiments showed ALD behavior during multiple cycles (see Figure 4 and Figures S10-S12), whereas SER spectra from TP ALD exhibited CVD growth akin to TTIP ALD (see Figures S13-S14). SER spectra of the extended isopropanol exposures onto TiO2/AgFON-MPA (see Figure 3) show several modes that appear to match up with the peaks seen during the 2nd cycle of TiO2 ALD, most noticeably various modes in the 1500 – 1600 cm-1 region, as well as peaks at 1369, 1256, and 1156 cm-1. The peaks seen during the varying amounts of isopropanol exposure are far broader than during TiO2 ALD, however the peaks are still distinguishable up until 300 s of dosing. The lack of noticeable features by 300 s of dosing is expected to be the result of product buildup, where access to the TiO2 sites responsible for dehydration becomes blocked. The sum of all the spectra highlights the nature of the species that builds up on the surface as amorphous carbon products. Without new TiO2 sites being generated on the surface or prolonged access to the initial TiO2 sites, the products evolving from propene eventually cease, leaving only the amorphous carbon. The AlIPO dosing in both the first cycle and subsequent cycles (see Figures S10 and S11) showcases exemplary ALDtype behavior seen through SERS, as well as confirms that the TiO2 plays a role in the dehydration chemistry following the first cycle. All peaks that appear with each AlIPO dose are removed with the subsequent water dose. Furthermore, the peaks seen in later cycles match what was observed in the first cycle, suggesting there is no new chemistry occurring. Most of the peaks seen experimentally also match up with DFTcalculated spectra for suspected surface species; the only differences are a shift in the COO- stretch from MPA and the absence of the lower Al-O vibrational modes. The lack of Al-O vibrations is due to the low Raman scattering cross-section of those moieties. Spectra taken from dosing TTB (see Figure S12) also show ALD behavior with little to no changes in the chemistry upon multiple cycles. Vibrational modes associated with the neat precursor match up with peaks in the SER difference spectra during the ALD process, except for an unknown additional peak at 1092 cm-1. The use of TTB showcases how tert-butanol is incapable of undergoing this dehydration chemistry, and thus no buildup of carbonaceous material occurs. The SER spectra from TP ALD, on the other hand, showed ALD chemistry for only the 1st cycle. Peaks analogous to the olefin species from TTIP were also found in the 2nd and 3rd TP ALD cycles (see Figures S13-S14). The dehydration chemistry is believed to occur via an E2 mechanism as established previously on TiO2,31 and as such is the reason why the primary and secondary alcohols underwent dehydration while the tertiary alcohol did not.

ACS Paragon Plus Environment

Page 7 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society Dehydration of isopropanol catalyzed by the isolated, undercoordinated titania sites leads to the formation of propene, which then undergoes oligomerization to form carbonaceous deposits that remain on the surface (see Scheme 1). As mentioned previously, the undercoordinated titania sites are necessary in order for dehydration to occur at these temperatures (70 °C), as titania sites made on the walls of the reactor and after several ALD cycles do not show any indication of isopropanol dehydration occurring. This is in agreement with the low-temperature pathway that Bondarchuk et al. found on undercoordinated TiO2 (110).29 It remains unclear as to how the propene undergoes oligomerization on the surface. It is possible that an interface between the TiO2 sites and the silver surface promotes oligomerization; work by de Oliveira et al.32 found oligomeric species resulting from propene oxide to remain on the surface of their Ag/TiO2 catalysts for propene epoxidation under H2/O2 exposure. The authors drew similarities with their findings and those typically seen in propene epoxidation using Au/TiO2 catalysts as well.33 It is possible a similar phenomenon is occurring during TiO2 ALD, although work detailing under what conditions oligomerization occurs for the Ag/TiO2 interface has yet to be done. Oligomerization on Au/TiO2 catalysts has been observed as low as 50 °C, albeit under H2/O2 exposure at 170 kPa.34

Finally, DFT calculations of several molecules were performed to determine the exact structure of the surface species being made in the 2nd and 3rd ALD cycles (see Figure 4). Comparisons between the DFT calculations and experimental results show six-carbon olefins as the most likely candidate. Models of 2,4-hexadiene and 1,3,5-hexatriene coordinated to a TiO4-Ag20 cluster yielded all the appropriate vibrational modes within reasonable proximity to the experimental data. Calculations were performed with and without the silver cluster to ensure the emulated silver surface had a negligible impact on the vibrational modes (see Tables S1 and S2). Modes within the 1600 to 1480 cm-1 range are indicative of a conjugated system (likely olefins). The fact that multiple peaks show up in this region experimentally suggests that a few unique olefins are present. The 1350 to 1200 cm-1 region contains one strong and wide band that matches the vinyl C-H bending mode in both olefin models. The other lesser modes match up well as an asymmetric C-C-C stretch (1150 cm1), C-O stretch (1018 cm-1), C-(CO)=C stretch (899 cm-1), CH 2 rocking (831 cm-1), and CH2 twisting (684 cm-1). The 1,3,5hexatriene model shows strong Ti-O vibrational modes that are not expected to show up in the experiment due to the amorphous nature of the resulting TiO2. A similar phenomenon was seen in the case of Al2O3 ALD from an earlier study,14 where the Al-O vibrational modes were not seen via SERS. All in all, the deviation in ALD behavior seen during SERS is the result of an interaction that occurs between the titania sites and the isopropanol byproduct after the 1st ALD cycle. Figure 4. SER difference spectrum (black) of the second cycle of TiO2 ALD using TTIP with DFT-calculated Raman spectra for two potential byproducts (red and blue) for comparison. Simplified structures for the DFT-calculated species are shown on the bottom two panels. Each region of matching vibrational modes is highlighted with the corresponding modes. The lower Ti-O vibrational modes are expected to be absent in the SER spectrum due to the amorphous TiO2 and their low Raman scattering crosssection.

CONCLUSION The spectroscopic evidence provided herein suggests propene is formed from the isopropoxide ligands of TTIP via dehydration chemistry at isolated titania sites made during the previous ALD cycle. Propene then undergoes oligomerization to form distinguishable six-carbon olefins before polymerizing into amorphous carbon, like what was seen in the case of continuous isopropanol exposure. New titania sites can form atop the amorphous carbon with continuous TTIP exposure, resulting in extended six-carbon olefin production done concurrently with TiO2 ALD. Mass spectrometry results show the isolated titania sites responsible for this chemistry do not form on the walls of the reactor, suggesting the MPA nucleation layer allows for isolated, undercoordinated titania sites to form when a generic surface may not. The vibrational modes associated with the six-carbon olefins can even be seen during the 1st ALD cycle, albeit at a much lower intensity than in subsequent cycles. Furthermore, it appears the olefins on the surface during the 1st cycle can be removed with water exposure, unlike the 2nd and 3rd ALD cycles. The observation of analogous precursors under the same conditions show that this behavior is not unique to TTIP, and that the deviation in ALD chemistry can occur so long as there is undercoordinated TiO2 sites and off-gassing alcohols that dehydrate via an E2 mechanism. This work showcases the importance of monitoring the surface during ALD for byproducts that interfere with film growth, introducing impurities and perturbing product function.

ASSOCIATED CONTENT The Supporting Information is available free of charge on the ACS Publications website at acs.org. SER spectra of MPA-functionalized AgFON, SER difference spectra in the CH region and after several cycles during TiO2 ALD using TTIP and TTB, as well as Al2O3 ALD using AlIPO, QMS measurements

ACS Paragon Plus Environment

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

during depositions, and tabulated data for DFT calculations of surface species present during TiO2 ALD.

AUTHOR INFORMATION Corresponding Author *E-mail: [email protected] ORCID Ryan A. Hackler: 0000-0003-4698-6074 Gyeongwon Kang: 0000-0002-8219-2717 George C. Schatz: 0000-0001-5837-4740 Richard P. Van Duyne: 0000-0001-8861-2228 Notes The authors declare no competing financial interest.

ACKNOWLEDGMENT The authors gratefully acknowledge financial support from the Northwestern University Institute for Catalysis in Energy Processes (ICEP). ICEP is funded through the U.S. Department of Energy, Office of Basic Energy Sciences (Award Number DE-FG0203ER15457). G.K., G.C.S., and R.P.V.D. acknowledge support from the Air Force Office of Scientific Research MURI (FA955014-1-0003). This work made use of the Keck-II facility of Northwestern University’s NUANCE Center, which has received support from the Soft and Hybrid Nanotechnology Experimental (SHyNE) Resource (NSF ECCS-1542205); the MRSEC program (NSF DMR-1720139) at the Materials Research Center; the International Institute for Nanotechnology (IIN); the Keck Foundation; and the State of Illinois, through the IIN.

REFERENCES (1) Biener, M. M.; Biener, J.; Wichmann, A.; Wittstock, A.; Baumann, T. F.; Bäumer, M.; Hamza, A. V. ALD Functionalized Nanoporous Gold: Thermal Stability, Mechanical Properties, and Catalytic Activity. Nano Lett. 2011, 11, 3085-3090. (2) Standridge, S. D.; Schatz, G. C.; Hupp, J. T. Toward Plasmonic Solar Cells: Protection of Silver Nanoparticles via Atomic Layer Deposition of TiO2. Langmuir 2009, 25, 2596-2600. (3) Park, K.; Zhang, Q.; Garcia, B. B.; Zhou, X.; Jeong, Y.-H.; Cao, G. Effect of an Ultrathin TiO2 Layer Coated on Submicrometer-Sized ZnO Nanocrystallite Aggregates by Atomic Layer Deposition on the Performance of Dye-Sensitized Solar Cells. Adv. Mater. 2010, 22, 2329-2332. (4) Liang, Y.-C.; Wang, C.-C.; Kei, C.-C.; Hsueh, Y.-C.; Cho, W.H.; Perng, T.-P. Photocatalysis of Ag-Loaded TiO2 Nanotube Arrays Formed by Atomic Layer Deposition. J. Phys. Chem. C 2011, 115, 9498-9502. (5) Ritala, M.; Leskelä, M. Niinisto, L.; Haussalo, P. Titanium Isopropoxide as a Precursor in Atomic Layer Epitaxy of Titanium Dioxide Thin Films. Chem. Mater. 1993, 5, 1174-1181. (6) Aarik, J.; Aidla, A.; Mändar, H.; Uustare, T. Atomic Layer Deposition of Titanium Dioxide from TiCl4 and H2O: Investigation of Growth Mechanism. Appl. Surf. Sci. 2001, 172, 148-158. (7) Cleveland, E. R.; Henn-Lecordier, L.; Rubloff, G. W. Role of Surface Intermediates in Enhanced, Uniform Growth Rates of TiO2 Atomic Layer Deposition Thin Films Using Titanium Tetraisopropoxide and Ozone. J. Vac. Sci. Technol., A 2012, 30, 01A150/1-01A150/6. (8) Hackler, R. A.; McAnally, M. O.; Schatz, G. C.; Stair, P. C.; Van Duyne, R. P. Identification of Dimeric Methylalumina Surface Species during Atomic Layer Deposition Using Operando Surface-Enhanced Raman Spectroscopy. J. Am. Chem. Soc. 2017, 139, 2456-2463. (9) Rahtu, A.; Ritala, M. Reaction Mechanism Studies on Titanium Isopropoxide-Water Atomic Layer Deposition Process. Chem. Vap. Deposition 2002, 8, 21-28.

Page 8 of 14

(10) Willets, K. A.; Van Duyne, R. P. Localized Surface Plasmon Resonance Spectroscopy and Sensing. Annu. Rev. Phys. Chem. 2007, 58, 267−297. (11) Lai, S. C. S.; Kleyn, S. E. F.; Rosca, V.; Koper, M. T. M. Mechanism of the Dissociation and Electrooxidation of Ethanol and Acetaldehyde on Platinum as Studied by SERS. J. Phys. Chem. C 2008, 112, 19080-19087. (12) Wang, A.; Huang, Y.-F.; Sur, U. K.; Wu, D.-Y.; Ren, B.; Rondinini, S.; Amatore, C.; Tian, Z.-Q. In Situ Identification of Intermediates of Benzyl Chloride Reduction at a Silver Electrode by SERS Coupled with DFT Calculations. J. Am. Chem. Soc. 2010, 132, 9534-9536. (13) Schatz, G. C.; Van Duyne, R. P. Electromagnetic Mechanism of Surface-Enhanced Spectroscopy. In Handbook of Vibrational Spectroscopy; Chalmers, J. M., Griffiths, P. R., Eds.; Wiley: New York, 2002; Vol. 1, pp 759-774. (14) Masango, S. S.; Hackler, R. A.; Henry, A.-I.; McAnally, M. O.; Schatz, G. C.; Stair, P. C.; Van Duyne, R. P. Probing the Chemistry of Alumina Atomic Layer Deposition Using Operando Surface-Enhanced Raman Spectroscopy. J. Phys. Chem. C 2016, 120, 3822-3833. (15) Cardinal, M. F.; Vander Ende, E.; Hackler, R. A.; McAnally, M. O.; Stair, P. C.; Schatz, G. C.; Van Duyne, R. P. Expanding Applications of SERS Through Versatile Nanomaterials Engineering. Chem. Soc. Rev. 2017, 46, 3886-3903. (16) Li, M.; Dai, M.; Chabal, Y. J. Atomic Layer Deposition of Aluminum Oxide on Carboxylic Acid-Terminated Self-Assembled Monolayers. Langmuir 2009, 25, 1911-1914. (17) 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. (18) Seo, E. K.; Lee, J. W.; Sung-Suh, H. M.; Sung, M. M. Atomic Layer Deposition of Titanium Oxide on Self-Assembled-MonolayerCoated Gold. Chem. Mater. 2004, 16, 1878-1883. (19) Greeneltch, N. G.; Blaber, M. G.; Henry, A.-I.; Schatz, G. C.; Van Duyne, R. P. Immobilized Nanorod Assemblies: Fabrication and Understanding of Large Area Surface-Enhanced Raman Spectroscopy Substrates. Anal. Chem. 2013, 85, 2297-2303. (20) Moran, P. D.; Bowmaker, G. A.; Cooney, R. P.; Finnie, K. S.; Bartlett, J. R.; Woolfrey, J. L. Vibrational Spectra and Molecular Association of Titanium Tetraisopropoxide. Inorg. Chem. 1998, 37, 2741-2748. (21) Masango, S. S.; Hackler, R. A.; Large, N.; Henry, A.-I.; McAnally, M. O.; Schatz, G. C.; Stair, P. C.; Van Duyne, R. P. HighResolution Distance Dependence Study of Surface-Enhanced Raman Scattering Enabled by Atomic Layer Deposition. Nano Lett. 2016, 16, 4251-4259. (22) Baerends, E. J.; Ziegler, T.; Autschbach, J.; Bashford, D.; Berces, A.; Bickelhaupt, F. M.; Bo, C.; Boerrigter, P. M.; Cavallo, L.; Chong, D. P.; Deng, L.; Dickson, R. M.; Ellis, D. E.; van Faassen, M.; Fan, L.; Fischer, T. H.; Fonseca Guerra, C.; Franchini, M.; Ghysels, A.; Giammona, A.; van Gisbergen, S. J. A.; Götz, A. W.; Groeneveld, J. A.; Gritsenko, O. V.; Grü ning, M.; Gusarov, S.; Harris, S. E.; van den Hoek, P.; Jacob, C. R.; Jacobsen, H.; Jensen, L.; Kaminski, J. W.; van Kessel, G.; Kootstra, F.; Kovalenko, A.; Krykunov, M. V.; van Lenthe, E.; McCormack, D. A.; Michalak, A.; Mitoraj, M.; Morton, S. M.; Neugebauer, J.; Nicu, V. P.; Noodleman, L.; Osinga, V. P.; Patchkovskii, S.; Pavanello, M.; Philipsen, P. H. T.; Post, D.; Pye, C. C.; Ravenek, W.; Rodríguez, J. I.; Ros, P.; Schipper, P. R. T.; van Schoot, H.; Schreckenbach, G.; Seldenthuis, J. S.; Seth, M.; Snijders, J. G.; Sola, M.; Swart, M.; Swerhone, D.; te Velde, G.; Vernooijs, P.; Versluis, L.; Visscher, L.; Visser, O.; Wang, F.; Wesolowski, T. A.; van Wezenbeek, E. M.; Wiesenekker, G.; Wolff, S. K.; Woo, T. K.; Yakovlev, A. L. ADF2017, SCM, Theoretical Chemistry; Vrije Universiteit: Amsterdam, The Netherlands, http://www.scm.com. (23) Becke, A. D. Density-Functional Exchange-Energy Approximation with Correct Asymptotic Behavior. Phys. Rev. A 1988, 38, 3098-3100. (24) Perdew, Y. Density-Functional Approximation for the Correlation Energy of the Inhomogeneous Electron Gas. Phys. Rev. B 1986, 33, 8822-8824.

ACS Paragon Plus Environment

Page 9 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

(25) Van Lenthe, E.; Baerends, E. J.; Snijders, J. G. Relativistic Regular Two-Component Hamiltonians. J. Chem. Phys. 1993, 99, 4597-4610. (26) Van Lenthe, E.; Baerends, E. J.; Snijders, J. G. Relativistic Total Energy Using Regular Approximations. J. Chem. Phys. 1994, 101, 9783.9792. (27) Castro, J. L.; Lopez-Ramirez, M. R.; Arenas, J. F.; Otero, J. C. Surface-Enhanced Raman Scattering of 3-Mercaptopropionic Acid Adsorbed on a Colloidal Silver Surface. J. Raman Spectrosc. 2004, 35, 997-1000. (28) Johnson, A. M.; Stair, P. C. Alternative Low-Pressure Surface Chemistry of Titanium Tetraisopropoxide on Oxidized Molybdenum. J. Phys. Chem. C 2014, 118, 29361-29369. (29) Bondarchuk, O.; Kim, Y. K.; White, J. M.; Kim, J.; Kay, B. D.; Dohnalek, Z. Surface Chemistry of 2-Propanol on TiO2(110): Low- and High-Temperature Dehydration, Isotope Effects, and Influence of Local Surface Structure. J. Phys. Chem. C 2007, 111, 11059-11067.

(30) Jin, C.; Liu, B.; Lei, Z.; Sun, J. Structure and Photoluminescence of the TiO2 Films Grown by Atomic Layer Deposition Using Tetrakis-Dimethylamino Titanium and Ozone. Nanoscale Res. Lett. 2015, 10, 1-9. (31) Carrizosa, I.; Munuera, G. Study of the Interaction of Aliphatic Alcohols with TiO2 II. On the Mechanism of Alcohol Dehydration on Anatase. J. Catal. 1977, 49, 189-200. (32) de Oliveira, A. L.; Wolf, A.; Schuth, F. Highly Selective Propene Epoxidation with Hydrogen/Oxygen Mixtures over TitaniaSupported Silver Catalysts. Catal. Lett. 2001, 73, 157-160. (33) Nijhuis, T. A.; Visser, T.; Weckhuysen, B. M. Mechanistic Study into the Direct Epoxidation of Propene over Gold/Titania Catalysts. J. Phys. Chem. B 2005, 109, 19309-19319. (34) Nijhuis, T. A.; Huizinga, B. J.; Makkee, M.; Moulijn, J. A. Direct Epoxidation of Propene Using Gold Dispersed on TS-1 and Other Titanium-Containing Supports. Ind. Eng. Chem. Res. 1999, 38, 884-891.

Insert Table of Contents artwork here

ACS Paragon Plus Environment

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Scheme 1: Reaction diagram and surface intermediates for the ALD pathway (top) and CVD pathway (bottom) of TiO2 thin film growth using TTIP. The ALD pathway leads to uniform growth with off-gassing isopropanol, whereas the CVD pathway involves dehydration of the isopropanol byproduct, which then becomes incorporated into the film after oligomerization. 238x116mm (300 x 300 DPI)

ACS Paragon Plus Environment

Page 10 of 14

Page 11 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

Figure 1. SER difference spectra collected in-situ during the TiO2 ALD process on MPA-functionalized AgFON. Panel (a) shows the 1st TTIP dose (in red) and 1st H2O dose (in blue), with the normal Raman spectrum of

the TTIP precursor (in black) for comparison. Panel (b) shows the 2nd TTIP dose (in red) and 2nd H2O dose (in blue), with the TTIP precursor spectrum shown for comparison. All major peaks are annotated, and data are shifted vertically for clarity. 254x381mm (300 x 300 DPI)

ACS Paragon Plus Environment

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 2. SER difference spectra taken from 30 s doses of TTIP (in red) and H2O (in blue) during the first (a and b) and second (c and d) cycles. Positive peaks represent new species being generated from the previous spectrum. Negative peaks represent removal of surface species from the previous spectrum. 203x172mm (300 x 300 DPI)

ACS Paragon Plus Environment

Page 12 of 14

Page 13 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

Figure 3. SER difference spectra taken from 30 s cumulative doses of isopropanol after one TiO2 cycle on AgFON-MPA had been deposited using TTIP, as well as the sum of all isopropanol exposure spectra. 271x210mm (300 x 300 DPI)

ACS Paragon Plus Environment

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 4. SER difference spectrum (black) of the second cycle of TiO2 ALD using TTIP with DFT-calculated Raman spectra for two potential byproducts (red and blue) for comparison. Simplified structures for the DFT-calculated species are shown on the bottom two panels. Each region of matching vibrational modes is highlighted with the corresponding modes. The lower Ti-O vibrational modes are expected to be absent in the SER spectrum due to the amorphous TiO2 and their low Raman scattering cross-section. 241x279mm (300 x 300 DPI)

ACS Paragon Plus Environment

Page 14 of 14