Tetrakis(dimethylamido) - American Chemical Society

J. Phys. Chem. C 2010, 114, 14061–14075

14061

Tetrakis(dimethylamido)hafnium Adsorption and Reaction on Hydrogen Terminated Si(100) Surfaces Kejing Li,† Shenggang Li,‡ Ning Li,† David A. Dixon,‡ and Tonya M. Klein*,† Chemical Engineering Department and Chemistry Department, The UniVersity of Alabama, Tuscaloosa, Alabama 35487 ReceiVed: February 12, 2010; ReVised Manuscript ReceiVed: July 1, 2010

The adsorption and reaction of tetrakis(dimethylamido)hafnium (TDMAH) on hydrogen terminated Si(100) were studied by using in situ attenuated total reflectance Fourier transform infrared spectroscopy (ATRFTIR), transmission IR, and quadrupole mass spectrometry (Q-MS). Surface and gas phase reactions were investigated at temperatures between 25 and 300 °C. Density functional theory (DFT) calculations benchmarked by coupled cluster calculations on small models were performed for gas phase decomposition via intramolecular insertion and β-hydride elimination as well as the adsorption and reaction of TDMAH onto a hydrogen terminated Si(100) surface. N-Si and CH2-Si bonds due to reactions on the Si windows were observed in transmission IR, while N-Ge and CH2-Ge bonds on a Ge internal reflectance element (IRE) were observed by ATR-FTIR at 25 and 100 °C. Also observed were the formation of Hf-H bonds and three-member-ring species on the Si surface; the former was confirmed by a control D2O exchange reaction experiment. Both transmission IR and Q-MS indicated the presence of decomposition products dimethylamine (DMA) and N-methyl methyleneimine (MMI). The calculated bond dissociation energies (BDE) at the CCSD(T)/CBS level roughly follow the order of Hf-O > Hf-N > N-H, C-H, Si-N > Si-H, Si-C > N-C, Hf-H > Hf-Si, and the BDEs of the same chemical bond can vary substantially in different molecules. The interface is predicted by DFT calculations to involve Hf-Si, Hf-N-Si, and/or HfNC-Si bonds. TDMAH decomposition products, such as MMI, can form a C-Si or N-Si bond with the silicon surface. The combined experimental and theoretical results suggest that insertion and β-hydride elimination reactions can occur during bidentate chemisorption on the H-Si(100) surface by forming N-Si bonds. Introduction Hafnium oxide (HfO2) is being used as a gate oxide in lieu of a SiO2 insulating layer in the 45 nm technology node for metal oxide semiconductor field effect transistors (MOSFET), reducing leakage current, heat dissipation, and power consumption, as well as increasing device reliability.1 Among all the high-κ candidates, HfO2 is generally accepted as the best material for this application due to its relatively high κ value (κ ) 20-25), high band gap (Eg ∼ 5.6 eV), and thermodynamic stability on Si.2 Previous work has focused on its implementation aspects such as electrical properties,3,4 chemical composition,5-7 interfacial properties,4,8 deposition processes and growth modes,9-12 and gaseous precursor reactions for chemical vapor deposition (CVD) processing.3,8 Several groups have reported the formation of a potentially undesirable interfacial layer before or after the postdeposition annealing for HfO2 grown on Si.13-15 The interfacial structure and chemistry are of special interest because the channel is formed very near this layer and is affected by impurities, dangling bonds, and trapped charges at the interface. Atomic layer deposition (ALD), initially developed by Suntola and co-workers in 1974, has an advantage over CVD in that it can produce smooth films with more uniform chemical compositions over large areas.11,16,17 It involves exposure of the substrate to alternate cycles of a gaseous organometallic * To whom correspondence should be addressed. E-mail: tklein@ eng.ua.edu. † Chemical Engineering Department. ‡ Chemistry Department.

precursor and an oxidizing or reducing reactant. In principle, each step is self-limiting in that adsorption of the precursor stops after the available surface reactive sites are consumed so that one is able to effectively control the thickness at the atomic scale. Low temperature ALD is of special interest for semiconductor back-end processing,18 thin film deposition on temperature sensitive substrates such as fabrics,19 plastic semiconductor substrates used in nonvolatile memory organic field effect transistors,20 and single-wall carbon nanotube field effect transistors,21 where the process temperatures are required to be below 150 °C. The most popular Hf precursors for ALD at present are hafnium alkylamides,3,8,9,11,22 due to the relatively weak Hf-N bond and their high volatility, which enables efficient deposition at relatively low temperatures. Hafnium alkylamides can be used as the gaseous precursor for both the gate and the oxide. For example, Hf[N(CH3)2]4 [tetrakis(dimethylamido)hafnium or TDMAH] has been widely studied for the ALD of HfNx and HfO2 on Si surfaces.11,13,23,24 Fourier transform infrared spectroscopy (FTIR) studies on ALD of Hf[N(CH3)(C2H5)]4 [tetrakis(ethylmethylamido)hafnium, TEMAH] and Hf[N(C2H5)2]4 [tetrakis(diethylamido)hafnium, TDEAH] on hydrogen terminated Si surfaces with H2O as the oxidant source have been performed.25,26 It was found that the ALD processes were initiated by the metal precursor rather than by water, and only ∼50% of the H atoms on the hydrogen terminated Si surface reacted away after about five cycles.26 Unlike TEMAH and TDEAH, an incubation period during ALD deposition using TDMAH onto H-Si(100) has been observed,13 suggesting the

10.1021/jp101363r  2010 American Chemical Society Published on Web 07/29/2010

14062

J. Phys. Chem. C, Vol. 114, No. 33, 2010

presence of a kinetic barrier for the initial TDMAH adsorption step. The dosing amounts for the three precursors are slightly different, which may account for these different observations in the adsorption initialization. Recently, Chen et al. studied the reaction of TEMAH on a model of the 2 × 1 H-Si(100) surface with electronic structure calculations using density functional theory (DFT) at the B3LYP/LanL2DZ level, and predicted that chemisorption of TEMAH on the H-Si surface has a high reaction barrier of greater than 55 kcal/mol in total.27 Furthermore, they predicted that the above reaction forms a Hf-Si bond with ethylmethylamine [EMA, (CH3CH2)CH3NH] as a byproduct. The potential formation of a bond between Hf and Si may be problematic because, if each Hf atom forms bonds with two Si atoms, a strained interface has the potential to form dangling bonds that can trap charges and introduce states in the band gap.2 In the ALD process, it has been proposed that the initially formed Hf-Si bonds may be oxidized to form the Hf-O-Si bonds after the water dose to relieve the bond strain at the interface.2 Unlike the reaction with the OH terminated Si surface, all reactions with the hydrogen terminated Si surfaces to form Hf-Si were predicted to be endothermic.27 As the reaction involves one gaseous molecule (TEMAH) chemisorbing onto the Si surface with the release of another gaseous molecule (EMA), ∆S will be small and ∆G will be predominantly determined by ∆H. Rodriguez-Reyes and Teplyakov28 have studied the adsorption and dissociation of TDMAT on the bare Si(100) surface. They predicted that the N-C bond cleavage is a more thermodynamically favorable pathway than the Hf-N bond cleavage. Especially for the displacement of hydrogen on the hydrogen terminated Si(100) surface which has a large Si-H bond energy of 88 kcal/mol,29 different reaction pathways could be present than previously considered. The thermal chemistry of tetrakis(dimethylamido)titanium (Ti[N(CH3)2]4, TDMAT) which is a precursor similar to TDMAH, has been extensively studied,30-32 and can provide insight into the TDMAH thermochemistry in this study. Different decomposition mechanisms for TDMAT have been proposed, including (1) β-hydride elimination which produces a hydride species and N-methyl methyleneimine (CH3NdCH2, MMI),33 and (2) an intramolecular insertion metallacycle generation reaction,27 followed by the formation of a MMI dative bond to Ti to produce a three-member metallacycle species.30 The product MMI also is formed in the decomposition of other dimethylamido compounds such as As[NMe2]4 [tris(dimethylamido)arsenide, TDMAA].34 Methane was observed during the gas phase decomposition of TDMAT by FTIR at high decomposition temperatures (>350 °C), and was suggested to be formed by a gaseous radical disproportion reaction.30 In the crystal phase, a dimeric structure of TDMAH was observed by X-ray crystallography.35 The current work is a combined experimental and computational study to elucidate the reactions involved in the adsorption of TDMAH on hydrogen terminated Si surface at low temperatures, followed by a D2O or H2O dose to identify the water reactive surface products from the previous dose. We are especially interested in the nature of the interfacial bonds, the potential surface reactions, and the possible reactions initiated by gas phase decomposition products. In situ attenuated total reflection Fourier transform infrared spectroscopy (ATR-FTIR) is a nondestructive surface sensitive technique for monitoring adsorption and reaction phenomena, and provides a way to observe interfacial bonds. Electronic structure calculations, for example, at the density functional theory (DFT) level, provide

Li et al.

Figure 1. (a) FTIR schematic. (b) Transmission gas cell cross section. (c) Attenuated total reflectance compartment. (b) and (c) are interchangeable in the middle component space of FTIR. (b) and (c) are reprinted with permission from Harrick Scientific.

a means to predict the energetics of individual steps for proposed reaction mechanisms, as it is usually impossible to observe all the elementary steps in a hypothesized mechanism experimentally. The calculations also provide IR spectra to help the interpretation of the experimental spectra. Experimental and Computational Methods Experimental Methods. The IR experimental setup is shown in Figure 1. The base absolute pressure was 10-3 Torr. TDMAH (99.99% purity from Strem Chemical Inc.) with a melting point of ∼23 °C was transferred to a transparent quartz bubbler under N2 and remained a light yellow liquid during experiments, indicating that the precursor source is stable. It was heated for about 1 h at 65 °C in a Precision 810 digital oil bath to reach vapor-liquid equilibrium. Deionized D2O or H2O was kept in a 150 cm3 stainless steel vessel half-filled at room temperature. No carrier gas was used for the precursors. In the substrate temperature range of 25-250 °C, the experimental pressure was about 0.7 Torr during the TDMAH dose and 3.6 Torr for the D2O/H2O dose. Longer preheating of the bubbler resulted in higher maximum pressure upon opening but did not affect the steady state pressure obtained after 1 min (Figure S1 in the Supporting Information). A Thermo Nicolet 4700 FT-IR spectrometer with a deuterated triglycine sulfate (DTGS) detector was used with either one of two exchangeable components from Harrick Scientific Inc. designed to fit inside the sample compartment and modified with vacuum compatible fittings (Figure 1b,c). Spectra using this arrangement have a better signal-to-noise ratio than those taken using IR light diverted out of the spectrometer, through windows on a reactor and back to the detector, likely because the minimal path length results in a larger light intensity at the detector. Transmission infrared spectroscopy of the gas phase was performed in a 17 mL cone cross section temperature-controlled stainless steel gas cell with two 25 × 2 mm Si windows. The Horizon multiple reflection ATR flow-through cell was used to observe adsorption and reactions on a HF last trapezoidal cut 50 × 10 × 2 mm internal reflectance element (IRE) with 45° beveled ends and a 280 mm2 surface area on the vacuum side, which is the volume above the IRE depicted by a grey color in Figure 1c. Before introducing the precursor, the inlet lines were heated to 100 °C for 12 h to remove residual moisture, and remained slightly above 65 °C to keep TDMAH vapor from condensing. Intrinsic Si, Ge, and ZnSe IRE crystals were used in the ATR experiments. The spectral collection window was in the middle IR range from 4000 cm-1 down to 1150 cm-1 for Si and down to 700 cm-1 for Ge and ZnSe. Si and ZnSe have

Adsorption and Reaction of TDMAH on H-Si(100) limited transparency at low wavenumbers around and above 300 °C, while Ge is only useful below 150 °C because of increasingly accessible phonon modes at high temperatures. There were 12 reflections of the IR beam on the adsorption side. ZnSe was simply wiped clean with lens paper, while the Si surface was sequentially treated with acetone, a 1:1:5 NH4OH: H2O2:H2O solution at 80 °C, a 1:1:100 HF:NH4F:H2O buffered oxide etch (BOE) solution with a pH ≈9 for 3 min, a deionized water rinse, and N2 blow drying. Ideally this results in an atomically rough 3 × 1 or 2 × 1 hydrogen terminated Si(100) surface,36,37 or a monohydrogen terminated Si(111) surface, with a monohydrogen Si-H vibration frequency at about 2084 cm-1.38 The Ge IRE was treated only with the BOE solution and a deionized water rinse to remove any native oxide. Spectra of the treated surfaces reprocessed (ratioed) with a background spectrum of thermally oxidized Si(100) surface at 700 °C in air showed that Si-H peaks were stable in our experimental time range in air. No residual fluorine from the BOE treatment was observed in X-ray photoelectron spectroscopy (XPS; Physical Electronics APEX system with Omicron EA 125 hemispherical analyzer) measurements, but a small amount of oxygen (532.7 eV) probably from surface hydroxyl groups was detected. The νO-H peak at 3735 cm-1 from ambient moisture or surface Si-OH groups was observed in the single beam FTIR spectra, consistent with the presence of O in the form of a surface hydroxyl group. The surface morphology of the Si(100) crystal was measured by atomic force microscopy (AFM) using an ultrasharp silicon cantilever NSC14/AIBS of 10 nm tip radius (Mikro Masch, Narva) with a force constant of ∼5.7 N/m and a resonance frequency of about ∼160 kHz on a Digital Instruments Dimension 3100 Scanning Probe Microscope with NanoScope VI controller. It showed root-mean-square (rms) roughnesses of about 5.0 nm at the 200 nm scale and 4.0 nm at the 1 µm scale, suggestive of defects and kinks over the large adsorption area of these IRE cyrstals.2 Two sets of experiments were performed using ATR-FTIR: (1) TDMAH liquid drop and (2) adsorption followed by desorption. For experiment 1 a liquid drop of TDMAH was applied at room temperature on both the Si and Ge ATR crystals under a N2 environment (1 atm.). After placement in the FTIR instrument, the crystal was heated at a rate of 100 °C/min to a preset temperature between 25 and 250 °C, but only results at 25 and 100 °C are reported here. All of the IR spectra were reprocessed using a background spectrum taken before the precursor introduction at each corresponding temperature. For the adsorption/desorption experiments, immediately upon TDMAH vapor dosing, multi-internal reflection IR spectra were collected in the range 4000-400 cm-1 with 32 scans at 4 cm-1 resolution, which took approximately 37 s for each spectrum. After completing 15 spectra, the flow of the precursor was stopped and the adsorbed species were left in a vacuum for 10 min to desorb while IR spectra were continually recorded. The dosing amounts are given in langmuirs (1 L ) 10-6 Torr · s). Adsorption experiments were performed with longer exposure time than the usual ALD pulse by 1 or more orders of magnitude in order to maximize the signal-to-noise ratio for ATR-FTIR and to determine if the adsorption saturates. For the transmission IR experiments, the gas cell temperature was set to a value between 25 and 300 °C by a heating band outside the cell. The source TDMAH was kept at 65 °C except for the IR experiments at room temperature. The Si windows were also treated by the procedures used for Si IRE crystals before each experiment. The total beam path length for the IR experiments was 10 cm, and the path length to volume ratio

J. Phys. Chem. C, Vol. 114, No. 33, 2010 14063

Figure 2. (a) Structure of model of the surface. H14Si9 cluster representing H-Si(100). (b) Structure and highest occupied molecular orbital (HOMO) of TDMAH.

was maximized by the conical-shaped interior design in order to maximize the signal. For each temperature, the gas cell was evacuated and a background spectrum was taken before the precursor was introduced. Upon opening the precursor valve, IR spectra were collected similar to that of the ATR experiments in flow mode. After a desired dose the valves to the pump and the precursor were closed for the closed system mode and pressure was recorded as a function of time. An Inficon 200 Quadrex residual gas analyzer (RGA) was connected to the exit side of the flow-through cell for mass spectrometry analysis and the electron impact energy was set to 102 eV, which gives mass spectrometry profiles comparable to those at lower electron impact energies. (e.g., 70 eV).39 This electron energy was used because of the specific requirements of the mass spectrometer.40 The instrumental default resolution is 0.9 amu at 10% peak height. The flow rate was reduced by a 0.3 mm diameter orifice to an estimated 0.5 sccm into the bypass line to reduce the sampling pressure. The pressure in the mass spectrometer was further reduced to 10-10 Torr by a turbomolecular pump backed by a mechanical pump. All molecular ions and other fragments up to 50 atomic mass units were monitored in real time for crystal temperatures from 25 to 250 °C. Computational Methods. Electronic structure calculations were performed to help interpret the experimental observations. The density functional theory (DFT) calculations were performed with the Gaussian 03 software package.41 The energetics of possible gas phase and surface reactions of TDMAH on a model cluster of hydrogen terminated Si(100) surface were calculated. The 2 × 1 reconstructed H-Si(100) surface was represented by a Si9H14 cluster,27 as shown in Figure 2a. The boundary Si atoms were terminated with H atoms along the cleaved Si-Si directions to avoid unrealistic charge transfer. In previous work, this structure has been shown to provide a good representation for the monohydrogen terminated Si(100) surface.27 All reactants and products were fully optimized. The DFT calculations were performed with the B3LYP exchangecorrelation functional.42,43 Transition states were optimized using the STQN method.44-46 The DFT geometry optimizations and frequency calculations were carried out with the 6-31+G(d,p) basis set for the nonmetal atoms and the LanL2DZ large core relativistic effective core potential (RECP) and associated basis set47 for the Hf atom; this combination of basis sets will be collectively denoted as BS1. Single point DFT energies were also calculated with the aug-cc-pVDZ basis set48 for the nonmetal atoms and the small core RECP plus associated augcc-pVDZ-PP basis set49 for Hf, denoted as BS2. All energies were corrected with zero point energies (ZPEs). In order to estimate the bond dissociation energies (BDEs) for the bonds in the reactant (TDMAH), the Si surface, and

14064


Li et al.

possible reaction intermediates, accurate BDEs were calculated for the C-H, N-H, N-C, Si-H, Si-C, Si-N, Hf-H, Hf-N, Hf-O, and Hf-Si bonds in a number of model compounds representative of the chemistry being studied. The geometries of the model compounds were optimized at the B3LYP/BS2 level. Our previous studies on the transition metal oxides have shown that the above method gives good equilibrium geometries when compared with those calculated with the more expensive computational method discussed below.50 Single point energy calculations at the B3LYP/BS2 geometries were carried out at the coupled cluster CCSD(T) level.51 Because the Hf 5p orbitals lie close in energy to the C and N 2s orbitals, the aug-ccpwCVXZ (X ) D, T, Q) basis set52 was used for the nonmetal atoms, and the aug-cc-pwCVXZ-PP basis set49 was used for Hf (with a small core RECP) in order to include the inner shell orbitals (1s for C and N, 2s 2p for Si, and 5s 5p for Hf) in the correlation treatment. These basis sets will be denoted as awCVXZ. Thus, our CCSD(T) energies include the correlation effects of the core electrons and the effect of relativity on the metal due to the use of the RECP. Any additional relativistic corrections to the bond energies are expected to be small, less than 1 kcal/mol.53 The CCSD(T) energies for X ) D, T, Q were then extrapolated to the complete basis set (CBS) limit using a mixed Gaussian/exponential formula:54

E(n) ) ECBS + Be-(n-1) + Ce-(n-1)

2

(1)

In eq 1, n ) 2, 3, and 4 for X ) D, T, and Q, and the CCSD(T)/ CBS energy ECBS, B, and C are the fitting parameters. The BDEs at 0 K (∆E0 K) were then calculated from the sum of the CBS energy and the ZPE. The CCSD(T) calculations were done with the program system MOLPRO.55 Results and Discussion The optimized structure for TDMAH shown in Figure 2b has the expected tetrahedral structure. The average bond lengths of Hf-N, N-C, and C-H are 2.06, 1.46, and 1.10 Å, respectively. The highest occupied molecular orbitals are the lone pair electrons on N. Electron density from the lone pair delocalizes into the σ* orbital of the C-H bond, which weakens the C-H bond and lowers its frequency.7,56-58 The geometry at the N atoms in TDMAH is nearly planar trigonal, suggesting the possibility of π-bonding between vacant 5d orbitals on Hf and the lone pair “p” orbital on N perpendicular to the Hf-N-C-C plane. As discussed later, the Hf-N bond is quite strong, a further indication of the contribution to its bond strength from the additional π back-bonding. Figure 3 shows the IR spectrum of TDMAH in the gas phase at room temperature together with the calculated harmonic spectrum. Due to the effects of anharmonicity in the C-H stretches, a scaling factor of 0.9408 was applied for comparison to experiment. The scaling factor was calculated as the ratio of the frequency of the most intense C-H band in the gas phase spectrum and its corresponding calculated frequency. The low frequency C-H peak at 2781 cm-1 could be a Bohlmann-type band,59 which corresponds to the C-H stretch antiparallel to the lone pair on N. In the calculated spectrum of TDMAH, this corresponds to the C-H symmetric stretches parallel and antiparallel to the N lone pair (for example, C4-H5 and C4-H6 in Figure 2b). The peak at 2873 cm-1 corresponds to the asymmetric C-H stretches of the above C-H bonds. The peak at 2948 cm-1 is due to the C-H stretches from the third C-H bond (for example, C4-H7 in Figure 2b). Although low in

Figure 3. Top: infrared spectrum of TDAMH in gas phase in flow mode at room temperature. Bottom: Calculated IR spectrum at the DFT/ BS1 level with scaling factor 0.9408 applied to the region of C-H stretches for TDAMH.

intensity, these peaks exist in all experimental spectra. The unscaled calculated spectrum for frequencies 200 °C may be due to faster MMI convection. Another possibility is that, at higher temperatures, MMI undergoes further reactions. Interestingly, the C-H stretch began to resemble the intact molecule spectrum with higher temperatures perhaps due to desorbed TDMAH or partially decomposed TDMAH coated on hot reactor walls. The signal-to-noise ratio is smaller at higher temperatures, especially at the lower wavenumbers. However, some characteristic peaks are still discernible. The peaks at 882 and 838 cm-1 grow with increasing temperature while the peak at 1670 cm-1 diminishes, and they remain even after evacuation of the gas cell and after the D2O pulse (Figures S2 and S3 in the Supporting Information). Peaks of similar frequencies were previously assigned to νas(Si-N) in ammonia-treated H-Si(111) (840, 870, and 1040 cm-1)5,26,64 and in SixNy films (840 cm-1)65 and to δCH3 in (CH3)2GeO (857 cm-1).66 By comparison to the calculated DFT spectra, the two peaks at 882 and 838 cm-1 can be tentatively assigned to monodentate νSi-N and δ(Si)C-H2 in species formed by reactions of TDMAH with hydrogen terminated Si surfaces (2b and 2c, Figure 8) or symmetric and asymmetric νSi-N in the bidentate form which will be discussed later in the modeling part (Table 2). For clarification, it is important to note here that the majority of the IR signal in transmission mode arises from the longer path length through the gas phase, and only a small part of the signal will originate from any condensed phase on the window. This is in contrast to ATR-FTIR, where the majority of the signal arises from the evanescent wave that decays exponentially with distance from the surface and which has a penetration depth on the order of one wavelength. As shown in the above gas phase FTIR analysis, MMI and DMA were detected in the transmission IR gas cell together with the observation of Si-N or Si-C on the Si windows. The main characteristic peaks are νCdN in MMI located at 1670 cm-1, δN-H in DMA at 735 cm-1, and νSi-N or νSi-C at 882 and 838 cm-1. Surface FTIR Analysis of TDMAH and D2O or H2O ALD on H-Si(100). Initial growth of thin films occurred immediately upon opening the precursor inlet valve for TDMAH within the

Adsorption and Reaction of TDMAH on H-Si(100)


Figure 8. Surface reactions of TDMAH on H-Si(100). Reaction energies were calculated at B3LYP/BS2 level, and units are given in kcal/mol, all at 0 K.

time scale of the measurement as monitored by the νC-H region. Figures 9 and 10 show IR data for experiments conducted at 25 and 100 °C using the following sequence: (a) first dose, multilayer adsorption of 3.5 × 108 L TDMAH; (b) desorption by pumping for 10 min; (c) second dose of 5.0 × 107 L TDMAH; (d) first dose of 1.0 × 108 L D2O or H2O; (e) third dose of another 5.0 × 107 L TDMAH; (f) second dose of 5.0 × 107 L D2O or H2O. The dose amounts were determined using the steady state pressure as it varied during dosing (Figure S1 in the Supporting Information). A higher dose than traditional ALD was used because it was necessary in order to observe the small Hf-H feature as discussed below. In addition, the saturation time as measured using the maximum and unchanging absorbance of the 1650 cm-1 peak was 6-10 min. After an initial rapid increase, the intensity of the νC-H region monotonically increased with time even after 30 min of exposure (see discussion below), in contrast to hafnium tert-butoxide adsorption,67 where the νC-H region was observed to saturate (Figure S4 in the Supporting Information). The intensity of the C-H stretches decreased during desorption and was almost completely absent after the D2O or H2O dose, consistent with observations

by other groups.25,26 The assignments for the characteristic peaks for the adsorption experiment are given in Table 2. Figure 9 shows the spectra for the adsorption of TDMAH and D2O (left panel) or H2O (right panel) on H-Si(100) at 25 °C. The right panel is an amplified region for Si-H and 1700-1600 cm-1. The last spectrum was taken after a 10 min desorption under vacuum after the H2O pulse to observe the hydroxyl exchange reaction. A TDMAH dose of 2.0 × 107 L was the least amount needed for the 1650 cm-1 peak to saturate. The result demonstrates that TDMAH can initiate the adsorption onto H-Si(100) as shown by the increase in the intensity for the νC-H peaks and the loss of the peaks for νSi-H. After desorption under vacuum for 10 min, there was only a small amount of residual C-H stretches and the intensity of the peak at 2839 cm-1 and the low frequency C-H band at 2781 cm-1 became very weak. In comparison to prior work,76 the adsorption of DMA on a Ge(100) surface showed a peak at 2834 cm-1 during multilayer adsorption, whereas for monolayer adsorption this peak was not observed. The fact that the low frequency C-H band intensity becomes weaker than the higher frequency C-H stretches suggests that N in the residue remaining after

14068


Li et al.

TABLE 2: Peak Positions (cm-1), IR Intensities (km/mol), and Assignments for the Characteristic Experimental IR Peaks from the Calculated B3LYP/BS1 Values expta 1700 (m) (Figure 9) 1662 (m) (Figure 4) 1670 (Figure 7) 1634 (m) (Figure 4) 1650 (Figure 9) 1600 (m) (Figures 9, 11) 1359 (w) (Figure 4) 1210 (w) (Figures 4, 9) 1220 (w) (Figure 11) 945 (s) (Figures 6, 11) 882 (m) (Figure 7) 839 (m) (Figure 11) 838 (m) (Figure 7) 798 (w) (Figure 11) 735 (s) (Figures 6, 7, 11)

assignmentb,c

calcd freqa

calcd intensity

νNdCH2 in 1e νNdCH2 in MMI

1783 1743

297 18

νHf-H in 4b νHf-H in 1b δNH2 in methylamine δCH2 in 1d ν(Hf)N-C in 1d ν(Hf)N-C in 1c νHf-N, νN-C (s) in TDMAH νSi-N in Si-N(CH3)2 νSi-N in 4b νSi-N in 2b νGe-N in Ge-N(CH3)2 νGe-N in Ge-NHfL3 δCH2 in 2c δCH2/δN-H in MMI-Si δCH2/νC-N in 1c δNH2 in SiCH2NH2 δNH in DMA

1648 1612 1665 1362 1222 1255 963 991 782 (s), 753 (as) 816 960, 554 717 804 801 793 826 765

288 325 19 48 212 200 200 188 265, 99 316 127, 41 371 51 108 19 367 131

vs, very strong; s, strong; m, medium; w, weak. b ν, stretching; δ, bending; γ, scissoring; ω, wagging; F, rocking; as, asymmetric; s, symmetric; ip, in plane; op, out of plane c See labeling in Figures 5, 8, and 13. a

Figure 9. Left panel: IR spectra for TDMAH/D2O adsorption on H-Si(100) at room temperature. In sequence: (a) 3.5 × 108 L TDMAH; (b) desorption under vacuum for 10 min; (c) 5.0 × 107 L dose of TDMAH; (d-1) 1.0 × 108 L D2O dose; (e) 5.0 × 107 L dose of TDMAH; (f-1) 5.0 × 107 L dose of D2O. Right panel: amplified H-Si and H-Hf stretching regions of IR spectra for TDMAH followed by H2O adsorption at room temperature. In sequence: (a) 3.5 × 108 L TDMAH; (b) desorption under vacuum for 10 min; (c) 5.0 × 107 L dose of TDMAH; (d-2) 1.0 × 108 L H2O dose; (e) 5.0 × 107 L dose of TDMAH; (f-2) 5.0 × 107 L dose of H2O; (g-2) desorption under vacuum for 10 min.

desorption may be chemically or datively bonded to the surface, which would affect the availability of the lone pair on N for delocalization to the C-H σ* orbital.76 The continuous increase of the integrated C-H stretch IR absorption during TDMAH adsorption and the remaining C-H stretch as well as the Si-H abstraction after desorption indicates that, after the first dose of TDMAH, both physisorbed and chemisorbed TDMAH were present.

A broad set of peaks in the region 2200-2000 cm-1 centered at 2086 cm-1 is readily assigned to νSi-H. After the first dose of 3.5 × 108 L TDMAH, a maximum of one-third of the surface H atoms were abstracted and there was no change after 10 min of pumping. The full width at half-maximum (fwhm) is about 90 cm-1, due to thermal and inhomogeneous broadening,68 and/ or the existence of kink or steps with the presence of both mono and dihydrogen bonds on the Si surface. The peak at 2054 cm-1


Figure 10. IR spectra for the H-Si and H-Hf stretching regions for TDMAH followed by D2O or H2O adsorption on H-Si(100) at 100 °C, sequentially: (a) 3.5 × 108 L TDMAH; (b) desorption under vacuum for 10 min; (c) 5.0 × 107 L dose of TDMAH; (d-1) 1.0 × 108 L D2O dose; (d-2) 1.0 × 108 L H2O dose. (d-1) is after baseline correction.

is consistent with other IR reports of the first ALD dose of TEMAH or TDEAH onto H-Si surfaces.25,26 The red shift of νH-Si to 2054 cm-1 is likely due to the adsorption effect or interaction of Si-H with neighboring Si sites bonded to Hf. The insertion reaction product 1c (reaction 3, Figure 5) can also induce this shift when it adsorbs onto the H-Si surfaces with negative charges on H. Spectra at higher temperatures have fringes that are not readily analyzed in this region. Three new peaks at 1700, 1650, and 1600 cm-1 appear after the first dose TDMAH adsorption. The peak at 1700 cm-1 was also observed after TDMAH adsorption followed by the D2O or H2O dose at 25 °C, but was not readily observed at 100 °C. It is not associated with the peak at 1600 cm-1 as they do not have the same change in intensity. νC-O was measured to be 1705 cm-1 for TDMAT after CO2 insertion.69 The peak at 1700 cm-1 increased after the third TDMAH dose (Figure 9c), which indicates that there may be a carbamato-type (ca. MO2CN, M ) metal) compound formed by TDMAH reacting with the surface after the H2O dose.69 A second possibility is that the νCdN mode is being blue shifted from MMI. The MMI was calculated to have a high desorption energy from Hf-H (discussed below), which indicates that energy needs to be added to desorb it and can explain why it was not observed in the gas phase spectra of TDMAH at room temperature but was observed in those at 100 and 150 °C. Another possibility is that the νCdN mode is from the product of the 1,2-β-hydride elimination reaction, which releases methane (reaction 5, Figure 5). The peak at 1634 cm-1 in heated liquid TDMAH (Figure 4) appears on the surface adsorption located at 1650 cm-1. With the assistance of the calculations, we tentatively assign this peak to the νHf-H of a surface Hf-H species possibly generated by β-hydride elimination. The presence of a transition metal hydride has been reported by several groups. For example, a silica supported transition metal hydride complex, [Zr]s-H, was synthesized from hydrogenolysis of a tris(neopentyl)zirconium surface complex and its IR peak was centered at 1635 cm-1;70 a similar result is observed for [Ta]s-H.71 The asymmetric Hf-H stretch in HfH4 was measured to be 1678 cm-1 72 and calculated to be 1687 cm-1 at the B3LYP/6-31+G(d,p) level; the Hf-H stretch in HHfO was observed at 1626 cm-1.73 This

J. Phys. Chem. C, Vol. 114, No. 33, 2010 14069 peak was not observed in the gas phase spectra, indicating that it results from a surface species. To provide insight into the identity of this Hf-H species, Figures 9 and 10 show the exchange reaction of Hf-H with D2O or H2O at 25 and 100 °C. The peak at 1650 cm-1 disappeared after a pulse of D2O (Figure 9 (d-1) and (f-1), Figure 10 (d-1)) but it remained for some time after a pulse of H2O (Figure 9 (d-2) and (f-2), Figure 10 (d-2)) in both the 25 and 100 °C spectra. The reaction of a surface Zr-H with KOH-water has been reported.70 It is expected that a reaction of Hf-H with water can release H2, which would explain the diminished peak at 1650 cm-1 after pumping for 10 min (Figure 9 (g-2)) after the water pulse. A possible pathway to form the Hf-H species 4b is given in Figure 8. It is likely that the proton exchange reaction generating Hf-D and the OH substitution reaction releasing H2 (reactions 9 and 10, Figure 8) can occur competitively, and that the exchange reaction is faster than the substitution reaction. The Hf-D stretch is reported73 to be around 1230 cm-1 and is calculated to be at 1154 cm-1 for the Hf-H compound 1b in reaction 2; these peaks overlap the D2O bending mode. After a D2O or H2O dose, another TDMAH dose would generate the new Hf-H species 4c. It is worthwhile to note here that the peak height at 1650 cm-1 saturated after about 6 min similar to the O-H stretch saturation of hafnium tert-butoxide (HTB),67 and in contrast to the continuous increase of the C-H stretch region of multilayer adsorbed molecules as discussed above. This indicates that within this time period a submonolayer formed at the interface where β-hydride elimination took place as discussed later. A new peak at 1600 cm-1 exists during the adsorption doses at both 25 and 100 °C. It is an unknown peak prevalent in the decomposition residuals of TDMAT30,31,63 and of TDEAH25 left on the IR windows. It is not likely to be due to the NdC stretch, whose frequency is usually higher than 1600 cm-1 for pure gas compounds containing this functional group. There are many possibilities for this band: NH2 scissoring in methylamine,74 OH2 scissoring in water for Hf silicates prepared in solution75 and from the water dose control experiment, and the CdC stretch in methoxyethylene containing O-CdC.74 Although H2O has an intense bending peak around 1600 cm-1 on the surface, the presence of this peak with no change in intensity after the TDMAH dose indicates that it should not be assigned to H2O. The remaining possible assignment is δN-H2. To identify this species, adsorption of TDMAH on Ge and ZnSe was performed, as these species have a broader observable window for IR as shown in Figure 11. Due to the weak intensity at 1600 cm-1 on Ge, the otherwise intense peak for ωN-H2 around 780 cm-1 66 was hidden by the broad peak at 735 cm-1. We tentatively assign this peak to δN-H2, but further work is needed for a definitive assignment. The C-H bending peak at 1251 cm-1 was broadened, and a deconvoluted peak at 1210 cm-1 may contribute to the broadening shown in Figure 9, left panel. Since there was no apparent peak associated with the peak at 1359 cm-1 for the four-member ring in 1d as in the bulk spectrum in Figure 4b, this peak at 1210 cm-1 is tentatively assigned to a three-member ring 1c, resulting from an insertion reaction (reaction 3, Figure 5) with DMA as the byproduct. After the D2O dose, this peak disappears. As shown in the above FTIR analysis of TDMAH adsorption on H-Si(100), Hf-H and a three-member-ring species were detected by the observation of peaks at 1650 and 1220 cm-1 on the H-Si(100) surface. The Si-H was abstracted during the initial dose of TDMAH. The peaks at 1600 and 1700 cm-1

14070


Figure 11. IR spectra for TDMAH followed by D2O adsorption on Ge and ZnSe IRE for the lower wavenumber region. (a) 3.5 × 108 L dose of TDMAH on ZnSe at 100 °C; (b) 3.5 × 108 L dose of TDMAH on Ge at room temperature; (c) 3.5 × 108 L dose of TDMAH on Ge at 100 °C; (d) D2O dose after TDMAH dose on Ge at 100 °C.

in room temperature experiments are tentatively assigned to carbamato and NH2 containing species, but need further analysis. Surface Adsorption/Desorption FTIR Analysis on Ge and ZnSe. In order to obtain more information at lower frequency, adsorption of TDMAH on hydrogen terminated Ge and ZnSe surfaces was performed. The C-H stretch region is consistent with the adsorption on H-Si(100). Characteristic peaks at 1456, 1245, 1137, 1060, and 945 cm-1 in Figure 11a-c for TDMAH are similar to the gas phase spectra. However, the peaks at 1650 cm-1 for Hf-H species and 1600 cm-1 for NH2 species on Ge and ZnSe surfaces are not as strong as those on the H-Si(100) surface. With a dose of D2O, most peaks due to TDMAH disappeared, showing an abstraction of the dimethylamido ligands. The broad peak centered at 1200 cm-1 is due to δD-O-D, which may overlap the νHf-D around 1154 cm-1 after the exchange reaction. Compared with the IR spectrum for pure TDMAH, there are several new peaks located at 1220, 839, 758, and 735 cm-1. The shoulder peak at 1220 cm-1 observed in both the Ge and ZnSe experiments is assigned as ν(Hf)N-C in three-member ring species. The peak at 839 cm-1 in Figure 11 on the Ge (Figure 11b) surface which is absent on the ZnSe surface (Figure 11a) can be assigned to νN-Ge in comparison to the peak position for νN-Si as discussed above in the gas phase FTIR analysis, which remained after the D2O pulse. Monolayer adsorption of DMA and trimethylamine on Ge showed strong peaks at 887 and 870 cm-1, which have been assigned to the N-C stretch.76 However, N-C stretches usually appear at frequencies higher than 1000 cm-1.77 Upon further evaluation, it is likely that these peaks should be assigned to the Ge-N stretch due to decomposition on the Ge IRE. BDEs of Ge-N and Si-N bonds were reported to be 55 and 78 kcal/mol,

Li et al. respectively (88.8 kcal/mol in H2NSiH3 from the CCSD(T)/CBS calculations, Table 3),78 so these bonds are reasonably strong. In addition, a stronger Si-N than Ge-N predicts the stability of Si-N if formed during the next water dose. DMA has a strong intensity for δCN-H at 735 cm-1 as shown in Figure 5. The small peak at 798 cm-1 is likely to be due to δCH2 in the chemisorbed TDMAH or MMI (see discussion below). In summary, the adsorption of TDMAH on Ge supported the interfacial bonding assignments in the previous transmission IR analysis in gas cell with Si windows by the observation of Ge-N or Ge-C at 839 cm-1. Residual Gas Analysis. Figure 12 shows the mass spectra from 2 to 50 amu for the gaseous effluents from the Si ATR flow-through cell. The background at room temperature was taken before the TDMAH valve was opened. The ATR crystal was heated from room temperature to 250 °C, and mass spectra were obtained at different temperatures. Results are shown for 25, 150, and 250 °C. The background shows a significant amount of moisture in the cell or tubing between the cell and the MS. After the introduction of TDMAH, the H2O peak intensity drops by ∼15%, indicating that H2O is scavenged by TDMAH in the tube lines. TDMAH is very reactive toward water, and the reported TDMAH hydrolysis reaction,25 Hf(NMe2)4 + H2O f (HO)HfN(Me2)3 + HNMe2, was predicted to be exothermic by -20.3 kcal/mol with a calculated barrier height of only 2.2 kcal/mol at the B3LYP/BS1 level. This reaction is likely to occur during ALD due to the usual residual moisture left inside the chamber and gas lines. However, as the hydrolysis reaction is very fast due to the low reaction barrier, it is likely to readily deposit HfO2(s) on the gas line wall, which is less likely to be transported by the flowing gas. Any surface OH groups on defect sites should be consumed quickly upon exposure to TDMAH. With an electron impact energy of 102 eV, most of the detected ions should be fragments of the effluent molecules, while only a small portion of the spectrum is expected to be due to parent molecular ions. The strong peak at m/e ) 44 can be assigned to the ion N(CH3)2+ from a fragment of TDMAH or DMA. The known peaks for DMA are m/e ) 45, 44, 30 (HNCH3+), and 15 (CH3+). The existence of peaks at m/e ) 28 (N ) CH2+), 27, and 26 is consistent with the presence of MMI. MMI can also be identified by the peak at m/e ) 42 (CH2dN-CH2+), which otherwise should follow a Gaussian distribution for DMA and be less intense than the peak at m/e ) 43. A portion of the peak at m/e ) 15 (CH3+) may also come from the fragmentation of MMI. It is a common product from the decomposition of the dimethylamido-metal precursors.31,34,63 Thus, for low temperature ALD using TDMAH, possible reactions in the gas phase or on the surface leading to MMI cannot be neglected. The ratios of the peak at m/e ) 18 to the peak at m/e ) 16 are 24.8, 23.6, 24.9, and 26.3 in Figure 12, parts a, b, c, and d, respectively. This variation indicates that some methane may have formed, accounting for a portion of the peak at m/e ) 16 as CH4+. However, due to its weak intensity, it may also be formed by recombination of fragmented ions, since methane formation is not as favorable among the decomposition processes at these temperatures. The intensity ratio for the CH3+ peak to H2O+ peak grows above 150 °C, because the precursor starts to fully decompose above this temperature, and is consistent with the gas phase IR observation of CH4 at 300 °C (Figure 7). The spectra at 150 and 250 °C are similar. The MS data show that DMA is a major product, although it is impossible to distinguish between the gas phase decomposition reaction and the surface reaction. Other byprod-



TABLE 3: Calculated Bond Dissociation Energies at 0 K in kcal/mol at the B3LYP and CCSD(T) Levels bond C-H H3C-H NH2CH2-H N-H H2N-H (CH3)2N-H N-C H2N-CH3 CH3NH-CH3 Si-H SiH3-H (SiH3)SiH2-H (SiH3)2SiH-H (SiH3)3Si-H Si-C H3C-SiH3 Si-N H2N-SiH3 Hf-H H3Hf-H (NH2)3Hf-H Hf-N H3Hf-NH2 (NH2)3Hf-NH2 H3Hf-N(CH3)2 (NH2)3Hf-N(CH3)2 Hf-O H3Hf-OH (NH2)3Hf-OH Hf-Si H3Hf-SiH3 (NH2)3Hf-SiH3 a

B3LYP/BS2

CCSD(T)/awCVDZ

CCSD(T)/awCVTZ

CCSD(T)/CBSa

100.4 87.2

98.8 88.3

102.4 90.7

103.3 91.3

102.2 87.1

100.0 87.5

104.3 91.1

106.1 92.4

77.5 72.6

77.2 75.1

80.5 78.1

82.6 79.8

86.9 84.3 82.1 79.9

86.0 83.5 81.5 79.7

89.4 86.6 84.4 82.3

90.6 87.8 (85.6) (83.5)

78.8

81.2

85.8

87.3

88.8

91.2

98.9

101.5

77.1 76.8

76.6 76.6

78.2 78.0

78.7 (78.5)

108.6 97.5 94.8 82.4

110.8 100.8 102.3 91.7

114.6 103.8 105.5 -

116.7 (105.9) (107.6) -

137.5 125.5

140.5 128.8

144.6 131.9

146.9 (134.2)

57.0 60.8

59.4 65.0

61.5 66.3

62.7 (67.5)

Those in parentheses are estimated values. See text.

Figure 12. Quadrupole mass spectrometry (Q-MS) spectra. Background (a) at room temperature. TDMAH effluent from the flow-through cell at (b) room temperature, (c) 150 °C, and (d) 250 °C.

ucts such as MMI and methane could also form through competitive reactions.79 Calculations of Gas Phase Reactions. The BDEs calculated at the B3LYP and CCSD(T) levels are given in Table 3. The recommended experimental value80 for SiH4 is 91.7 ( 0.05, so the calculated value at the CCSD(T)/CBS level is in good agreement with experiment. The BDEs calculated at the CCSD(T)/awCVTZ level are smaller than those calculated at the CCSD(T)/CBS level by up to 3 kcal/mol, and those calculated at the CCSD(T)/awCVDZ level are smaller by up to 11 kcal/mol. For the larger compounds, we estimated the

CCSD(T)/CBS value by using the CCSD(T)/awCVTZ value and their differences for the same bond calculated for smaller molecules. The calculated BDEs at the CCSD(T)/CBS level roughly follow the order Hf-O > Hf-N > N-H, C-H, Si-N > Si-H, Si-C > N-C, Hf-H > Hf-Si, and the BDEs of the same chemical bond can vary substantially in different molecules. For example, the BDE for the Hf-N bond in H3Hf-NH2 is larger than those in (NH2)3Hf-NH2 and H3Hf-N(CH3)2 by ∼10 kcal/mol, and is larger than that in (NH2)3Hf-N(CH3)2 by ∼20 kcal/mol. Compared to the BDEs calculated at the CCSD(T)/CBS level, those calculated at the B3LYP/BS2 level are smaller by less than 2 kcal/mol for Hf-H, 3-5 kcal/mol for C-H, N-H, and Si-H, 5-7 kcal/mol for N-C and Hf-Si, 8-10 kcal/mol for Si-C, Hf-N, and Hf-O, and ∼13 kcal/ mol for Si-N. The calculated reaction energies at the B3LYP level are likely to have errors on the order of 5-10 kcal/mol for forming and breaking different bonds. The BDEs in TDMAH and cluster models calculated at B3LYP/BS1 level are given in the Supporting Information (Table S2). The exposure time during the TDMAH pulse is important for ALD, as is the selection of an appropriate substrate temperature. Ideally, a monolayer surface coverage should be fully achieved to prevent unsaturated bonds from being oxidized in subsequent steps. The self-limiting nature is largely dependent on the chosen precursors. It was reported that an ideal ALD temperature window does not exist for the TDMAT/NH3 ALD process for TiN.18 In such cases, the dose amount becomes important: an overexposure of TDMAH may result in multilayer adsorption, while too short a dose leads to uncovered regions rather than the desired atomic layer control. In addition, the ALD process window has been found to depend strongly on an

14072


adequate purge time between precursor pulses.81 To study possible oligomerization effects that may occur during the first dose, the structure of dimer TDMAH [Hf(NMe2)4]2 (1a, Figure 5) was calculated. The dimerization is endothermic by 12.9 kcal/ mol at the B3LYP/BS1 level. The dimer of TDMAH has been reported to exist in the solid phase by crystallization from toluene, with two bridging N atoms between two Hf atoms forming a nonplanar structure of C2V symmetry.35 The optimized Hf-N bridging bond lengths at the B3LYP/BS1 level are 2.326 and 2.360 Å, in reasonable agreement with the crystal structure values of 2.262 and 2.324 Å.35 Reactions 2-5 (Figure 5) show possible decomposition mechanisms for TDMAH in the gas phase: reaction 2 is a β-hydride elimination reaction, reaction 3 is an intramolecular insertion reaction leading to a three-member ring structure, reaction 4 is another intramolecular insertion reaction leading to a four-member ring structure, and reaction 5 is a 1,2-β-hydride elimination reaction. The reaction energies and barriers for these gas phase reactions calculated at the B3LYP level are shown as ∆H and E‡. The reaction energies were calculated at the B3LYP/BS1 level and are within (1.5 kcal/mol different from those calculated at the B3LYP/BS2 level. Both the β-hydride elimination reaction (2) and the three-member-ring formation reaction (3) are endothermic with high reaction barriers. As a comparison, similar calculations were also done for reaction 3 for TDMAT, which has a reaction barrier of 37.4 kcal/mol and a reaction energy of 24.1 kcal/mol at the B3LYP/BS1 level. This reaction was reported to be observed in similar experiments on TDMAH, and it has been proposed to be followed by the formation of a dative bond from MMI.30 Although the energy of reaction 3 for TDMAT is 6 kcal/mol lower than that for TDMAH, all the above gas phase reactions are predicted to be quite endothermic and would not readily occur in the gas phase under our experimental conditions below 300 °C. In addition, we calculated that the decomposition of three-member ring species to MMI and Hf(NMe)2 is thermodynamically unfavorable, which rules out the generation of MMI from three-member ring species. The thermodynamically feasible decomposition reactions (4) and (5) have high reaction barriers of 84.0 and 69.9 kcal/mol, respectively, which is consistent with the observation of methane only at very high temperatures,31 or a long preheating which may assist such slow reactions. The high endothermicity and/or high energy barriers of the thermal decomposition reactions (2)-(5) indicate that these reactions are not likely to occur in gas phase under low temperature conditions. However, since there are unavoidable surfaces in the system such as the quartz bubbler, tubing walls, hydrogen terminated Si windows, or IRE substrate surfaces, these reactions may be exothermic on surfaces and could occur as discussed below. The calculated vibrational frequencies of gas phase reaction products have characteristic peaks which match the experimental IR spectra as described above. The unscaled frequencies of TDMAH and decomposition compounds 1b, 1c, and 1d are listed in Table 2. Calculations of Surface Reactions. Figure 1a shows the structure of Si9H14 used as a model for the Si(100) surface. A Mulliken charge analysis82 shows that the surface H has a charge of about -0.06 e and the BDE of surface Si-H was calculated to be 80.6 kcal/mol for this cluster at the B3LYP/BS1 level. In comparison, the BDE of Si-H is calculated to be 84.1 kcal/ mol for the Si7H14 cluster representing the H-Si(111) surface and 89.1 kcal/mol in SiH4, in good agreement with experimental value80 and the high level CCSD(T)/CBS value. Thus, silicon

Li et al. radicals due to breaking an Si-H terminal bond are not likely to be present under low temperature ALD conditions. Interfacial bonding states would be of interest as the stability of different bonding arrangements determine interface quality and band offset. Incorporation of C contaminants can influence the resistivity and density of the desired high dielectric materials. The ideal interface should have no Si dangling bonds, but have either Si-Hf direct bonding or Si-O-Hf bonding to avoid charge trapping and band offset shifts.83 Potential surface reactions starting from TDMAH on a model 2 × 1 H-Si(100) surface result in different interfacial bonding structures such as 2a, 2b, and 2c in reactions 6-8 (Figure 8). No physisorption energy was predicted at the B3LYP level with either basis set. This is not surprising as most current DFT functionals including B3LYP are known to have difficulty in predicting weak intermolecular interactions such as would be present here. This is also consistent with the weak adsorption observed in the experiment;26 such an adsorption is likely due to weak dispersion interactions. The interfacial bonding species have BDEs in the decreasing sequence Si-N > Si-C > Si-Hf (Table 3), which is consistent with the reaction energy sequence of reactions 6-8 (Figure 8). The only exothermic reaction is (7) for these three reactions. Thus, the most likely reaction between TDMAH and the Si surface is cleavage of the N-C, Si-H bonds and the formation of the 2b moiety with the release of methane and is consistent with our experimental observations of νSi-N or νGe-N in the IR spectra. However, the formation of Hf-Si 2a as shown in reaction 6 is also possible. The Hf-Si bond length is predicted to be 2.901 Å for the product, comparable to the literature value of 2.743 Å in TDMAH-silyl complex.35 The rate of reaction 6 with a reaction barrier of 19.2 kcal/mol calculated at the DFT/BS1 level was predicted to be ∼0.05 s-1 at 298 K using transition state theory.84 The imaginary frequency of the transition state was 500i cm-1, so we estimated the effect of tunneling using the Wigner expression.84 The tunneling correction will increase the rate by about 25% at room temperature. The estimated average residence time of TDMAH in the flow cell is about 0.3 s assuming a pumping speed of 30 L/min (liters per minute) for the total transport volume. Thus from a kinetic point of view, reaction 6 could play a role at room temperature. Reaction 8 has a reaction energy similar to that of reaction 6, which leads to possible C incorporation, and this type of interfacial bonding is not desired. It can be used to explain the observation of δC-H2 in the experimental IR spectra. The possible formation of different interfacial bonds suggests that carbon contamination may remain during low temperature ALD using TDMAH. Whether a byproduct can be easily removed is an important factor with respect to decreasing the contaminant level. We predict the desorption energy for DMA in reaction 6 to be ∼2 kcal/mol (the difference in energy between the product complex and the dissociated products). Thus under vacuum, the byproduct DMA with a low desorption energy could be pumped away quickly for this first step. In order to explain the formation of the Hf-H and threemember ring species observed in the adsorption IR experiments, thermodynamically possible reactions were surveyed. The overall reaction energy including β-hydride elimination and insertion reaction are expected to be less than 0. Reactions 1 + 2 f 2b f 3f f 4b or 4a show two possible reactions for the generation of Hf-H and three-member ring species. TDMAH can chemisorb onto H-Si(100) through N-Si bonds in a bidentate fashion together with the two self-elimination reactions. The total reaction energy to generate 4b or 4a is


Figure 13. Reactions 11 and 12: silylation reactions from MMI to H-Si surface. ∆Hproducts is the total reaction energy; E‡ is the activation barrier; all at 0 K.

exothermic by -8.2 kcal/mol and endothermic by 4.5 kcal/mol, respectively. It is assumed here that the starting surface is ideally hydrogen terminated Si(100). However, any existing defects or kinks may change the thermodynamics dramatically. This model is still useful in explaining the abstraction of H-Si and the formation of Si-N, Si-C, and Hf-H three-member-ring species on the surface. A more detailed quantitative study of the interfacial bond formation as a function of time may help resolve this issue. Bidentate chemisorption was observed for the precursor HTB in our previous work on hydrogen terminated Si surfaces,67 but more evidence such as the Hf/Si ratio is needed. On all counts, these are complex reactions and further studies are needed. When species containing Hf-H are generated, as observed experimentally, the resulting Hf-H bond can undergo proton exchange reaction with D2O 4b f 5a as shown in reaction 9 (Figure 8) or by the H/OH exchange reaction leading to H2 formation with water 4b f 5b shown as reaction 10 (Figure 8). Reaction 10 was calculated to have a reaction barrier of 3.3 kcal/mol and is essentially thermoneutral, as expected. This is consistent with the reaction observed after the D2O or H2O dose. The byproduct MMI can also react with the surface as shown in Figure 13. Reactions 11 and 12 are two possible exothermic silylation reactions with the H-Si surface. Reaction 11 has a reaction barrier of 38.5 kcal/mol and is exothermic by -11.3 kcal/mol. Reaction 12 has a reaction barrier of 34.2 kcal/mol and is exothermic by -21.8 kcal/mol. These reactions can also account for C residue at the interface. With the second water dose, the interfacial Hf-Si bonds in possible chemisorbed species 2a or 3a after the first TDMAH doses could readily break in addition to hydrolysis of the ligands. The reaction of the product L2Hf-Si 3a with H2O to to form Hf-O(H) 4d and a Si-H bond was calculated to have a small barrier of 1.6 kcal/mol with respect to the adsorption state and to be highly exothermic by -57.0 kcal/mol, which may be one of the reasons that Si-H remains after many ALD cycles.25,26 The Hf-Si bond was outside the observable window, but the formation of a small Si-O-Hf peak was observed during an ALD process using TEMAH and D2O.26 Further reaction with another water molecule leads to a similar reaction. Thus, if water is the oxidant source, the Hf-Si bond will not remain during the water dose. Another possible reaction for L2Hf-Si 3a with water generates a Hf-H 4c and an OH group on Si, with an exothermic reaction energy of -31.0 kcal/mol. However, the chemisorbed species with C-Si or N-Si may occur instead due to stronger interfacial bonds. Conclusions Reactions during initial HfO2 ALD using TDMAH and H2O/ D2O involve complex processes. Some aspects of these pathways

J. Phys. Chem. C, Vol. 114, No. 33, 2010 14073 leading to different interfacial bonding formations including Hf-Si, N-Si, and C-Si have been investigated in detail. During the first dose of TDMAH onto a H-Si surface at temperatures 25 and 100 °C, we observed that multilayer physisorption occurred with a large precursor dose. TDMAH can initiate adsorption by abstracting H-Si bonds to form Si-N and Si-C interfacial bonds. The Hf-H and metallacycle species from β-hydride elimination and insertion reactions were identified by the νHf-H at 1650 cm-1 and ν(Hf)N-C at 1210 cm-1 on the hydrogen terminated Si surface as probed by ATR-FTIR. A variety of byproducts such as MMI and DMA can be produced as shown by the detection of νCdN at 1670 cm-1 in MMI and δN-H at 735 cm-1 in DMA in the gas phase by transmission IR and is supported by Q-MS experiments. The proposed insertion reaction and β-hydride elimination were calculated to be thermodynamically not favorable in the gas phase. Instead, TDMAH can form a bidentate chemisorbed species on the hydrogen terminated Si surface by forming two N-Si bonds through cleavage of N-CH3 on the complex and H-Si on the surface. The MMI can either undergo a silylation reaction leading to the incorporation of contaminants or desorb into the gas phase. Acknowledgment. This work was supported in part by NSF CAREER Award No. 0239213 to T.M.K. This work was supported by the Chemical Sciences, Geosciences and Biosciences Division, Office of Basic Energy Sciences, U.S. Department of Energy (DOE) under Grant DE-FG02-03ER15481 (catalysis center program) and by the National Science Foundation (CTS-0608896), through the NIRT program. D.A.D. also thanks the Robert Ramsay Chair Fund of The University of Alabama for support. The authors greatly appreciate Dr. C. Heath Turner for helpful discussions, Dr. Alan Lane and Dr. Wei Li for help with the mass spectrometry instrumentation, Dr. Joseph Thrasher and Dr. Alfred Waterfeld for providing window materials for the gas cell, the Alabama Supercomputer Center for the computing resources, and. Dr. Myrna Matus for discussions about transition states. Supporting Information Available: Figures: pressure change as a function of bubbler equilibrium time at a bubbler temperature of 65 °C; gas phase spectra and residual spectra on the Si windows after 30 min of pumping at 300 °C; gas phase spectra of TDMAH/D2O at room temperature; comparison between HTB and TDMAH adsorption behavior on H-Si(100) at 100 °C; pressure recorded as a function of time at temperatures from 25 to 200 °C. Tables: harmonic vibrational frequencies and infrared intensities calculated at the B3LYP/BS1 level for TDMAH and DMA; bond dissociation energies at 0 K in kcal/ mol calculated at the B3LYP/BS1 level; zero-point energies (ZPEs) calculated at the B3LYP/BS2 level and electronic energies calculated at the CCSD(T)/awCVXZ (X ) D, T, Q) and CCSD(T)/CBS levels; Cartesian coordinates and energies for the reactants and products calculated at the B3LYP/BS1 Level; Cartesian coordinates for the bond dissociation energies calculated at the B3LYP/BS2 level. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) International Technology Roadmap for Semiconductors 2009 FEP. http://www.itrs.net/Links/2009ITRS/2009Chapters_2009Tables/2009_FEP. pdf. (2) Robertson, J. Rep. Prog. Phys. 2006, 69, 327.

14074


(3) Williams, P. A.; Jones, A. C.; Tobin, N. L.; Chalker, P. R.; Taylor, S.; Marshall, P. A.; Bickley, J. F.; Smith, L. M.; Davies, H. O.; Critchlow, G. W. Chem. Vap. Deposition 2003, 9, 309. (4) Zoolfakar, A. S.; Hashim, H.; Taylor, S. 9th International Conference Solid-State and Integrated Circuit Technology, Beijing, China, 2008; IEEE: Beijing, 2008; p 1268. (5) Wang, Y.; Ho, M. T.; Goncharova, L. V.; Wielunski, L. S.; RivillonAmy, S.; Chabal, Y. J.; Gustafsson, T.; Moumen, N.; Boleslawski, M. Chem. Mater. 2007, 19, 3127. (6) Ganem, J. J.; Trimaille, I.; Vickridge, I. C.; Blin, D.; Martin, F. Nucl. Instrum. Methods Phys. Res., Sect. B: Beam Interact. Mater. At. 2004, 219-220, 856. (7) Milanov, A.; Thomas, R.; Hellwig, M.; Merz, K.; Becker, H.-W.; Ehrhart, P.; Fischer, R. A.; Waser, R.; Devi, A. Surf. Coat. Technol. 2007, 201, 9019. (8) Deshpande, A.; Inman, R.; Jursich, G.; Takoudis, C. Microelectron. Eng. 2006, 83, 547. (9) Kukli, K.; Ritala, M.; Sajavaara, T.; Keinonen, J.; Leskela¨, M. Chem. Vap. Deposition 2002, 8, 199. (10) Zhong, L.; Zhang, Z.; Campbell, S. A.; Gladfelter, W. L. J. Mater. Chem. 2004, 14, 3203. (11) Hausmann, D. M.; Gordon, R. G. J. Cryst. Growth 2003, 249, 251. (12) Green, X. L.; Allen, A. J.; Li, X.; Wang, J.; Ilavsky, J.; Delabie, A.; Puurunen, R. L.; Brijs, B. Appl. Phys. Lett. 2006, 88, 032907. (13) Hackley, J. C.; Demaree, J. D.; Gougousi, T. J. Vac. Sci. Technol., A 2008, 26, 1235. (14) Chen, S. C.; Lou, J. C.; Chien, C. H.; Liu, P. T.; Chang, T. C. Thin Solid Films 2005, 488, 167. (15) Deshpande, A.; Inman, R.; Jursich, G.; Takoudis, C. Microelectron. Eng. 2006, 83, 547. (16) George, S. Chem. ReV. 2010, 110, 111. (17) Puurunen, R. J. Appl. Phys. 2005, 97, 121301. (18) Elam, J. W.; Schuisky, M.; Ferguson, J. D.; George, S. M. Thin Solid Films 2003, 436, 145. (19) Hyde, G. K.; Scarel, G.; Spagnola, J. C.; Peng, Q.; Lee, K.; Gong, B.; Roberts, K. G.; Roth, K. M.; Hanson, C. A.; Devine, C. K.; Stewart, S. M.; Hojo, D.; Na, J.-S.; Jur, J. S.; Parsons, G. N. Langmuir 2010, 26, 2550. (20) Ortiz, R. P.; Facchetti, A.; Marks, T. J. Chem. ReV. 2010, 110, 205. (21) Rinkio¨, M.; Johansson, A.; Paraoanu, G. S.; To¨rma¨, P. Nano Lett. 2009, 9, 644. (22) Kan, B.-C.; Boo, J.-H.; Lee, I.; Zaera, F. J. Phys. Chem. A 2009, 113, 3946. (23) Wang, W.; Nabatame, T.; Shimogaki, Y. Jpn. J. Appl. Phys. 2006, 45, L1183. (24) Biecuk, M. J.; Monsma, D. J.; Marcus, C. M.; Becker, J. S.; Gordon, R. G. Appl. Phys. Lett. 2003, 83, 2405. (25) Kelly, M. J.; Han, J. H.; Musgrave, C. B.; Parsons, G. N. Chem. Mater. 2005, 17, 5305. (26) Ho, M. T.; Wang, Y.; Brewer, R. T.; Wielunski, L. S.; Chabal, Y. J.; Moumen, N.; Boleslawski, M. Appl. Phys. Lett. 2005, 87, 133103. (27) Chen, W.; Sun, Q. Q.; Xu, M.; Ding, S. J.; Zhang, D. W.; Wang, L. K. J. Phys. Chem. C 2007, 111, 6495. (28) Rodriguez-Reyes, J. C. F.; Teplyakov, A. V. J. Phys. Chem. C 2007, 111, 4800. (29) Wu, C. J.; Carter, E. A. Chem. Phys. Lett. 1991, 185, 172. (30) Dubois, L. H.; Zegarski, B. R.; Girolami, G. S. J. Electrochem. Soc. 1992, 139, 3603. (31) Driessen, J. P. A. M.; Schoonman, J.; Jensen, K. F. J. Electrochem. Soc. 2001, 148, G178. (32) Norton, E. T.; Amato-Wierda, C. Chem. Mater. 2001, 13, 4655. (33) Musher, J. N.; Gordon, R. G. J. Mater. Res. 1996, 11, 989. (34) Salim, S.; Lim, C. K.; Jensen, K. F. Chem. Mater. 1995, 7, 507. (35) Cai, H.; Yu, X.; Chen, S.; Qiu, H.; Guzei, I. A.; Xue, Z. L. Inorg. Chem. 2007, 46, 8071. (36) Chabal, Y. J.; Raghavachari, K. Phys. ReV. Lett. 1985, 54, 1055. (37) Chabal, Y. J.; Raghavachari, K. Phys. ReV. Lett. 1984, 53, 282. (38) Higashi, G. S.; Chabal, Y. J.; Trucks, G. W.; Krishnan, R. Appl. Phys. Lett. 1990, 56, 656. (39) http://www.inficongasanalyzers.com/en/pdf/Optimization_of_Electron_Energy.pdf. (40) Inficon 200 Quadrex residual gas analyzer with High Performance and Compact Sensors: 334D%20 TWare%2032%20OM.pdf″>http://www.schoonoverinc.com/PDFs/074-334D%20 TWare%2032%20OM.pdf. (41) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.;

Li et al. Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, revision E.01; Gaussian, Inc.: Wallingford, CT, 2004. (42) Axel, D. B. J. Chem. Phys. 1993, 98, 5648. (43) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785. (44) Peng, C.; Ayala, P. Y.; Schlegel, H. B.; Frisch, M. J. J. Comput. Chem. 1996, 17, 49. (45) Peng, C.; Schlegel, H. B. Isr. J. Chem. 1993, 33, 449. (46) Schlegel, H. B. J. Comput. Chem. 1982, 3, 214. (47) Wadt, W. R.; Hay, P. J. J. Chem. Phys. 1985, 82, 284. (48) (a) Dunning, T. H. J. Chem. Phys. 1989, 90, 1007. (b) Kendall, R. A.; Dunning, T. H.; Harrison, R. J. J. Chem. Phys. 1992, 96, 6796. (c) Woon, D. E.; Dunning, T. H. J. Chem. Phys. 1993, 98, 1358. (49) Figgen, D.; Peterson, K. A.; Dolg, M.; Stoll, H. J. Chem. Phys. 2009, 130, 164108. (50) Li, S.; Dixon, D. A. J. Phys. Chem. A 2007, 111, 11908. (51) (a) Purvis, G. D., III; Bartlett, R. J. J. Chem. Phys. 1982, 76, 1910. (b) Raghavachari, K.; Trucks, G. W.; Pople, J. A.; Head-Gordon, M. Chem. Phys. Lett. 1989, 157, 479. (c) Watts, J. D.; Gauss, J.; Bartlett, R. J. J. Chem. Phys. 1993, 98, 8718. (52) (a) Woon, D. E.; Dunning, T. H., Jr. J. Chem. Phys. 1995, 103, 4572. (b) Peterson, K. A.; Dunning, T. H., Jr. J. Chem. Phys. 2002, 117, 10548. (53) Craciun, R.; Picone, D.; Long, R. T.; Li, S.; Dixon, D. A.; Peterson, K. A.; Christe, K. O. Inorg. Chem. 2010, 49, 1056. Li, S.; Dixon, D. A. J. Phys. Chem. A 2010, 114, 2665. (54) Peterson, K. A.; Woon, D. E.; Dunning, T. H., Jr. J. Chem. Phys. 1994, 100, 7410. (55) Werner, H.-J.; Knowles, P. J.; Lindh, R.; Manby, F. R.; Schutz, M.; Celani, P.; Korona, T.; Mitrushenkov, A.; Rauhut, G.; Adler, T. B.; Amos, R. D.; Bernhardsson, A.; Berning, A.; Cooper, D. L.; Deegan, M. J. O.; Dobbyn, A. J.; Eckert, F.; Goll, E.; Hampel, C.; Hetzer, G.; Hrenar, T.; Knizia, G.; Ko¨ppl, C.; Liu, Y.; Lloyd, A. W.; Mata, R. A.; May, A. J.; McNicholas, S. J.; Meyer, W.; Mura, M. E.; Nicklass, A.; Palmieri, P.; Pflu¨ger, K.; Pitzer, R.; Reiher, M.; Schumann, U.; Stoll, H.; Stone, A. J.; Tarroni, R.; Thorsteinsson, T.; Wang, M.; Wolf, A. MOLPRO, Version 2008.1, a package of ab initio programs; Cardiff University, Cardiff, U.K., 2008. see http://www.molpro.net. (56) Ernstbrunner, E. E.; Hudec, J. J. Mol. Struct. 1973, 17, 249. (57) Thomas, H. D.; Chen, K. H.; Allinger, N. L. J. Am. Chem. Soc. 1994, 116, 5887. (58) McKean, D. C.; Ellis, I. A. J. Mol. Struct. 1975, 29, 81. (59) Bohlmann, F. Angew. Chem. 1957, 69, 641. McKean, D. C.; Ellis, I. A. J. Mol. Struct. 1975, 29, 81. Thomas, H. D.; Chen, K. H.; Allinger, N. L. J. Am. Chem. Soc. 1994, 116, 5887. (60) Soulet, A.; Duquesne, L.; Jursich, G.; Inman, R.; Misra, A.; Blasco, N.; Lachaud, C.; Marot, Y.; Prunier, R.; Vautier, M.; Anderson, S.; Clancy, R.; Havlicek, M. Semicond. Fabtech 2005, FT27. (61) Hausmann, D. M.; Kim, E.; Becker, J.; Gordon, R. G. Chem. Mater. 2002, 14, 4350. (62) Stolkin, I.; Ha, T.-K.; Gu¨nthard, Hs. H. Chem. Phys. 1977, 21, 327. (63) Yun, J. Y.; Park, M. Y.; Rhee, S. W. J. Electrochem. Soc. 1998, 145, 2453. (64) Dai, M.; Wang, Y.; Kwon, J.; Halls, M. D.; Chabal, Y. J. Nat. Mater. 2009, 8, 825. (65) Liu, Y. C.; Furukawa, K.; Gao, D. W.; Nakashima, H.; Uchino, K.; Muraoka, K. Appl. Surf. Sci. 1997, 121/122, 233. (66) Khabashesku, V. N.; Boganov, S. E.; Kudin, K. N.; Margrave, J. L.; Nefedov, O. M. Organometallics 1998, 17, 5041. (67) Li, K.; Dubey, S.; Bhandari, H. B.; Hu, Z.; Turner, C. H.; Klein, T. M. J. Vac. Sci.Technol., A 2007, 25, 1389. (68) Jakob, P.; Chabal, Y. J.; Raghavachari, K.; Christman, S. B. Phys. ReV. B 1993, 47, 6839. (69) Chisholm, M. H.; Extine, M. W. J. Am. Chem. Soc. 1977, 99, 792. (70) Quignard, F.; Lecuyer, C.; Choplin, A.; Basset, J.-M. J. Chem. Soc., Dalton Trans. 1994, 7, 1153. (71) Vidal, V.; The´olier, A.; Thivolle-Cazat, J.; Basset, J.-M. Science 1997, 276, 99. (72) Chertihin, G. V.; Andrews, L. J. Phys. Chem. 1995, 99, 15004. (73) Zhou, M.; Zhang, L.; Dong, J.; Qin, Q. J. Am. Chem. Soc. 2000, 122, 10680. (74) Shimanouchi, T. Tables of Molecular Vibrational Frequencies Consolidated Vol. I; National Bureau of Standards: Washington, DC, 1972; pp 1-160. (75) Neumayer, D. A.; Cartier, E. J. Appl. Phys. 2001, 90, 1801.

Adsorption and Reaction of TDMAH on H-Si(100) (76) Mui, C.; Han, J. H.; Wang, G. T.; Musgrave, C. B.; Bent, S. F. J. Am. Chem. Soc. 2002, 124, 4027. (77) Stuart, B. Infrared Spectroscopy: Fundamentals and Applications; Wiley: New York, 2004. (78) Baldwin, C.; Lappert, M. F.; Pedley, J. B.; Poland, J. S. J. Chem. Soc., Dalton Trans. 1972, 18, 1943. (79) We note that this precursor is likely to coat the quadrupole sensor, so a long exposure time is not recommended. (80) Luo, Y.-R. ComprehensiVe Handbook of Chemical Bond Energies; CRC Press: Boca Raton, FL, 2007.

J. Phys. Chem. C, Vol. 114, No. 33, 2010 14075 (81) Liu, X.; Ramanathan, S.; Longdergan, A.; Srivastava, A.; Lee, E.; Seidel, T. E.; Barton, J. T.; Pang, D.; Gordon, R. G. J. Electrochem. Soc. 2005, 152, G213. (82) Mulliken, R. S. J. Chem. Phys. 1955, 23, 1833. (83) Robertson, J.; Peacock, P. W. Microelectron. Eng. 2004, 72, 112. (84) Steinfeld, J. I.; Francisco, J. S.; Hase, W. L. Chemical Kinetics and Dynamics, 2nd ed.; Prentice Hall: New York, 1999.

JP101363R

Tetrakis(dimethylamido) - American Chemical Society

Recommend Documents