Surface Science 599 (2005) 41–54 www.elsevier.com/locate/susc
Tertiary amide chemistry at the Ge(1 0 0)-2 · 1 surface Albert J. Keung, Michael A. Filler, David W. Porter, Stacey F. Bent
*
Department of Chemical Engineering, 381 North South Mall, Stanford University, Stanford, CA 94305-5025, United States Received 26 July 2005; accepted for publication 14 September 2005 Available online 27 October 2005
Abstract We have investigated the adsorption of several tertiary amides, including N,N-dimethylformamide, N,N-dimethylformamide-d7, 1-methyl-2-pyrrolidinone, and N-methylcaprolactam, on Ge(1 0 0)-2 · 1 using multiple internal reflection Fourier transform infrared spectroscopy and density functional theory. At 310 K, all four tertiary amides were observed to selectively form a dative bond to the germanium surface through the oxygen atom. While previous work has shown that oxygen dative bonds are unstable near room temperature, tertiary amides exhibit delocalization of electron density from nitrogen to oxygen, which appears to increase the stability of the oxygen dative-bonded state. Partial desorption of these surface adducts on the timescale of minutes indicates weakly bound surface adducts with coverage dependent binding energies. 2005 Elsevier B.V. All rights reserved. Keywords: Density functional calculations; Vibrational spectroscopies; Germanium; Organic molecules; Organo-functionalization of surfaces; Desorption; Amide; Ge(1 0 0)
1. Introduction The past forty years have seen tremendous growth in the fields of microelectronics and biotechnology. The minimum feature size of microelectronic devices has approached 50 nm [1], making many miniaturized and portable devices possible. At the same time, advances in biotechnology have revolutionized the healthcare industry *
Corresponding author. Tel.: +1 650 723 0385; fax: +1 650 723 9780. E-mail address:
[email protected] (S.F. Bent).
through novel materials, drug delivery methods [2], and research tools [3–7]. These two fields have developed to the point where they have great synergistic potential as already evidenced by drug delivering microchips [8] and quantum dots used to image live cells in vivo [9]. Furthermore, biocompatibility and biorecognition at interfaces are key aspects of bioelectronic devices, and there is much experimental and computational work being pursued to understand and address these issues [10– 13]. Since these and other applications rely on organic–inorganic interfaces [14–18], it is important to understand the fundamental chemistry of
0039-6028/$ - see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.susc.2005.09.035
42
A.J. Keung et al. / Surface Science 599 (2005) 41–54
organic compounds with inorganic interfaces such as the group-IV semiconductor surfaces. We focus on the Ge(1 0 0)-2 · 1 surface in the present work. Group-IV semiconductor surfaces are particularly important since the (1 0 0) crystal face of silicon predominates in the microelectronic industry. However, as the limits of traditional device scaling become increasingly apparent, there is renewed interest in semiconductor materials other than silicon. Due to its high electron mobility [19] and similarities to silicon [20], germanium is a potential candidate with which to augment current silicon technology. Moreover, as devices approach molecular dimensions, control of the interface will be imperative. Proper preparation of Ge(1 0 0) under vacuum conditions causes the surface to form a 2 · 1 reconstruction, analogous to that of Si(1 0 0) [20]. While the (1 0 0)-2 · 1 reconstruction is interesting from a fundamental perspective, the rows of surface dimers are also a template with the potential to enable novel molecular-level devices [15] and chemical sensors [17]. The amide linkage forms the backbone of proteins and is prevalent in biological systems. Therefore, studying the chemistry of the amide functional group is a logical first step toward the understanding of larger biological molecules. To this end, we have investigated the adsorption of a series of amides, shown in Fig. 1, on the Ge(1 0 0)-2 · 1 surface. Although amides are relatively unreactive compared to other carbonyl-con-
taining compounds in the solution phase [21], their reactivity may be quite different at the semiconductor surface. Primary and secondary amides possess N–H bonds which are known to dissociate for amines on both the Si(1 0 0)-2 · 1 [22–28] and to a lesser extent Ge(1 0 0)-2 · 1 [29,30] surfaces, forming surface adducts attached to the surface through a semiconductor–nitrogen bond. In addition, another N–H dissociation reaction, similar to the a-CH ‘‘ene’’ dissociation reaction reported for acetone on Ge(1 0 0)-2 · 1 [31], could also occur in amides. However, both of these reaction pathways are expected to be prevented in tertiary amides by the presence of methyl groups on the nitrogen atom. We therefore anticipate that tertiary amides will undergo fewer surface reactions, and thus they were chosen for this study. The resulting spectroscopic analysis and elucidation of tertiary amide reactivity in the present work should also facilitate the study of more complex secondary and primary amides in the future. The tertiary amide functionality is, from a simplistic perspective, a combination of a carbonyl and a tertiary amine group, which several research groups have studied independently on Si(1 0 0) and Ge(1 0 0). Formation of a dative bond, where a lone pair from the impinging molecule is donated to the electrophilic atom of the buckled surface dimer, is reported to be a general bonding motif for both functionalities. However, the strengths of the resulting surface bonds are strongly depen-
Fig. 1. Tertiary amides investigated in this study.
A.J. Keung et al. / Surface Science 599 (2005) 41–54
dent on the identity of the atom on which the lone pair exists as well as on its hybridization. Tertiary amines including trimethylamine, pyridine, and Nmethylpyrrolidine on Si(1 0 0)-2 · 1 [22,23,32–34] and Ge(1 0 0)-2 · 1 [24,35,36] were observed to form stable nitrogen dative-bonded states at room temperature, and their binding energies were calculated to be greater than 19 kcal/mol on both surfaces. In contrast, the dative bonds of carbonyl-containing compounds such as acetone and ethylvinylketone have weak binding energies, calculated to be below 15 kcal/mol on both surfaces [31,37]. These findings can be explained by the greater electronegativity of oxygen compared to nitrogen as well as the sp2 hybridization of the carbonyl group compared to the sp3 hybridization of amines [20]. However, the amide functionality is not simply a combination of carbonyl and amine groups since resonance, depicted in Fig. 2 for N,N-dimethylformamide, plays an important role in the electronic structure of amides. Electron delocalization leads to a pseudo-double bond between the nitrogen and carbonyl carbon, increased electron density on the oxygen, decreased availability for bonding of the nitrogen lone pair, and a 20 kcal/mol barrier to rotation about the main C 0 –N bond [38], where C 0 denotes the carbonyl carbon. The effect of electron delocalization has been observed to influence the reactivity of molecules toward group-IV (1 0 0)-2 · 1 semiconductor surfaces. For example, in their study of cyclic aliphatic and aromatic amines on Si(1 0 0) and Ge(1 0 0), Wang et al. [29] calculated that dative bonding for pyrrole on a Ge2Si7H12 cluster was 6 kcal/mol more favorable through a ring carbon than through the nitrogen lone pair. The authors explain that this increased stability is a direct result of electron delocalization from the nitrogen to the carbon atoms of the aromatic ring. Similarly, tertiary amides exhibit a delocalization of electron density
Fig. 2. Resonance forms for N,N-dimethylformamide depicting the delocalization of electron density from nitrogen to oxygen.
43
from nitrogen to oxygen, possibly increasing the nucleophilicity of the oxygen atom toward the surface. At the same time, tertiary amides are expected to be relatively inert due to the lack of N–H bonds [21]. Thus, tertiary amides should provide an interesting system in which to study the effect of electron delocalization in the N–C 0 @O amide backbone on the relative reactivity of the oxygen atom toward dative bond formation compared to other oxygen-containing functional groups.
2. Experimental and computational details All reactions were performed in an ultrahigh vacuum chamber described previously [30]. Trapezoidally shaped Ge crystals (1 · 14 · 19 mm, 45 beveled edges) were sputtered with Ar+ ions at room temperature (0.5 keV accelerating voltage, 20 mA emission current, 6–8 lA sample current) for 20 min followed by annealing to 900 K for 5 min to prepare the Ge(1 0 0)-2 · 1 surface. Low energy electron diffraction (LEED) confirmed that the proper (1 0 0)-2 · 1 surface reconstruction was achieved, and Auger electron spectroscopy (AES) verified that carbon, oxygen, and nitrogen surface concentrations were undetectable. The back face of the crystal was covered with a tantalum plate to block unwanted adsorption. For multiple internal reflection Fourier transform infrared (MIR-FTIR) spectroscopy experiments, the infrared beam from a Biorad FTS-60A spectrometer passes through two CaF2 windows on the vacuum chamber, limiting the observable infrared spectrum to energies greater than 1100 cm1. The infrared beam is focused on the beveled edge of the Ge crystal and undergoes approximately 20 internal reflections before exiting the chamber, where it is focused onto a narrowband HgCdTe detector. The beam path outside the chamber is purged with air filtered for water vapor and carbon dioxide to reduce the spectral features resulting from these gases. 1000 and 200 scans are co-added at 4 cm1 resolution for chemisorption and multilayer experiments, respectively, and ratioed to a background scan taken immediately prior to dosing each compound. All spectra are corrected for baseline sloping and the presence
44
A.J. Keung et al. / Surface Science 599 (2005) 41–54
of water vapor peaks. The spectral region above 2100 cm1 is not shown since the t(C–H) stretching modes for the tertiary amides in this study are weak and barely observable even in multilayer spectra. N,N-dimethylformamide (Aldrich, 99.9%), N,N-dimethylformamide-d7 (Aldrich, 99.5% D), 1-methyl-2-pyrrolidinone (Aldrich, 99+%), and N-methylcaprolactam (Aldrich, 99%) are clear liquids at room temperature. N,N-dimethylformamide-d7 was transferred under dry air to a sample vial to preserve the isotopic purity of the compound. All compounds were purified by several freeze-pump-thaw cycles prior to dosing. Amides were leaked into the chamber via a variable leak valve with a 0.5 inch diameter stainless steel directed doser positioned 0.5 inch from the sample surface. Room temperature dosing is adequate for N,N-dimethylformamide and N,N-dimethylformamide-d7, whereas 1-methyl-2-pyrrolidinone and N-methylcaprolactam require heating to 70 C via a water bath to achieve 100 mTorr in the manifold. In situ mass spectrometry confirmed the purity and identity of the compounds leaked into the chamber. Sample exposures, given in Langmuirs (1 L = 106 Torr s), are not corrected for ion gauge sensitivity. Single dimer density functional theory (DFT) calculations were also performed to obtain binding energies as well as theoretical spectra of adsorbed species. We employed the Gaussian 03 software package [39] using Becke3 Lee–Yang–Parr (B3LYP) [40–42] three-parameter density functional theory [43]. We modeled the Ge dimer using an unconstrained Ge2Si7H12 cluster with subsurface silicon atoms terminated by hydrogen to maintain the sp3 hybridization present in the bulk crystal diamond lattice. Only the top two atoms were modeled as Ge to keep calculations computationally manageable. Previous results have shown that binding energies found with these hybrid clusters are within 1–2 kcal/mol of those found with Ge9H12 clusters at the B3LYP/6-311++G(2df,pd) level of theory [44]. Geometry optimizations and frequency calculations were performed with the 6-31 G(d) basis set followed by single-point energy calculations performed with the higher 6311++G(2df,pd) basis set. Calculated frequencies
were scaled by a factor of 0.96 [45], and a Gaussian line-shape with a standard deviation of 10 cm1 was used in constructing theoretical spectra. No negative frequencies were found for local minima structures, and reported binding energies have been zero-point corrected. Recent studies of the Si(1 0 0)-2 · 1 and Ge(1 0 0)-2 · 1 surfaces within the cluster approximation have shown that nonlocal charge transfer to adjoining dimers is better modeled by larger, multiple dimer clusters [46,47], and the surface electronic structure can be more accurately realized with quantum capping potentials [48]. Nevertheless, comparisons of binding energies found in this study to those previously reported with similar single dimer clusters are expected to be valid and useful.
3. Results and discussion 3.1. N,N-dimethylformamide We begin this study with the simplest tertiary amide, N,N-dimethylformamide (DMF) (Fig. 1a). As illustrated in Fig. 3, the presence of an oxygen and a nitrogen atom in DMF, both of which possess lone pairs, leads to the possibility of several different reaction pathways on Ge(1 0 0)-2 · 1. After passing through an initial oxygen dativebonded state (a), [2 + 2] C 0 @O cycloaddition (e) and O-dative N–CH3 dissociation (f) products are possible. However, after passing through an initial nitrogen dative-bonded state (b), the potential exists to form both N-dative cis (g) and trans (h) N–CH3 dissociation products. Finally, it is also possible for the nucleophilic Ge dimer atom to attack the carbonyl carbon (not shown) leading to aldehydic C 0 –H dissociation (c) and C 0 –N dissociation (d) products. Infrared experiments and theoretical spectra, shown in Fig. 4 for DMF and DMF-d7, are used to determine which of these structures exist on the surface. Figs. 4a and c show DMF chemisorbed on Ge(1 0 0)-2 · 1 at 310 K and physisorbed at 160 K, respectively. After a 0.005 L saturation exposure of DMF at 310 K, four spectral features are clearly visible (Fig. 4a). The most intense peak at 1640 cm1 is indicative of an intact carbonyl
A.J. Keung et al. / Surface Science 599 (2005) 41–54
45
Fig. 3. Possible surface products for N,N-dimethylformamide, including: (a) O-dative bond, (b) N-dative bond, (c) aldehydic C 0 –H dissociation, (d) C 0 –N dissociation, (e) [2 + 2] C 0 @O cycloaddition, (f) O-dative N–CH3 dissociation, and (g) cis and (h) trans N-dative N–CH3 dissociation.
functionality following adsorption while the weaker modes at 1487, 1429, and 1352 cm1 are likely due to various d(C–H) deformation modes. Similar modes are observed in the physisorbed multilayer of DMF (Fig. 4c). These data point toward the formation of an oxygen dative bond for DMF, a conclusion which we will detail below. However, we will begin by addressing the surface products that are not observed experimentally. A combination of experimental and theoretical evidence as well as the precedent of previous work in the field allows us to rule out each of the products shown in Fig. 3, except the oxygen dativebonded state. Examining the DMF chemisorbed spectrum (Fig. 4a), no t(Ge–H) stretching modes in the 1900–2000 cm1 region are visible, indicating that no C–H dissociation pathways are active. The presence of the [2 + 2] C 0 @O cycloaddition product (Fig. 3e) is more difficult to ascertain since
t(C–O) stretching modes are not easily detected with our experimental setup. However, previous theoretical work has found that [2 + 2] C 0 @O cycloadducts have weak binding energies (near 10–12 kcal/mol) and are not stable at room temperature on Ge(1 0 0)-2 · 1 [31,37]. For DMF, the [2 + 2] C 0 @O cycloaddition product was found to be particularly unstable, calculated at 3.2 kcal/mol above the energy of the reactants, and is therefore unlikely to be observed. This instability is likely due to the involvement of DMF p-electrons in bonding with the surface, resulting in a loss of resonance stabilization. There is another possible cycloaddition product where the oxygen and nitrogen atoms are both bound to the dimer, similar to the 1,3-dipolar cycloaddition products previously reported on Si(1 0 0)-2 · 1 for nitro-containing compounds [49–52]. However, no minima for this species were found theoretically.
46
A.J. Keung et al. / Surface Science 599 (2005) 41–54
Fig. 4. Infrared spectra of N,N-dimethylformamide and N,N-dimethylformamide-d7 on Ge(1 0 0)-2 · 1: (a) saturation exposure of N,N-dimethylformamide at 310 K, (b) theoretical spectrum of N,N-dimethylformamide oxygen dative-bonded, (c) N,N-dimethylformamide multilayers at 160 K (scaled), (d) saturation exposure of N,N-dimethylformamide-d7 at 310 K, (e) N,N-dimethylformamided7 multilayers at 160 K (scaled).
The N–CH3 dissociation products (Fig. 3f–h) can be eliminated on the basis of the following analysis. DMF chemisorbed on Ge(1 0 0)-2 · 1 was observed to undergo time dependent desorption on the order of minutes, a result which will be discussed in detail below. However, previous work on Ge(1 0 0)-2 · 1 indicates that dissociation products with binding energies greater than approximately 20 kcal/mol are not experimentally observed to desorb at room temperature [29]. The O-dative and N-dative N–CH3 dissociation products for DMF are calculated to have binding energies of 36.4 and 40.3 kcal/mol, respectively. It is therefore unlikely that these dissociation products
will desorb on the order of minutes at 310 K, and their presence on the surface can be ruled out. Similar to the [2 + 2] C 0 @O cycloaddition product, adsorption of DMF in the C 0 –N dissociated state (Fig. 3d) leads to a loss of resonance and a calculated binding energy of only 9.2 kcal/mol. This binding energy is more than 25 kcal/mol weaker than the N–CH3 dissociation products discussed above. Although we are not aware of any previously observed surface products with binding energies less than 15 kcal/mol, the potential for desorption on the order of minutes at 310 K cannot be completely eliminated. However, the theoretical spectrum of the C 0 –N dissociation product
A.J. Keung et al. / Surface Science 599 (2005) 41–54
(not shown) does not exhibit any vibrational modes between 1600 and 1700 cm1 and is generally different from the experimentally obtained spectrum. We therefore do not believe the C 0 –N dissociation product is present on the surface at 310 K. We now return to the potential formation of oxygen or nitrogen dative-bonded species (Fig. 3a and b). The fact that the chemisorbed spectrum (Fig. 4a) looks remarkably similar to the multilayer spectrum (Fig. 4c) provides initial evidence of a dative-bonded surface adduct since the structure, and hence normal modes, should not change significantly upon adsorption. A similar observation was made for the case of acetone dative-bonded on Ge(1 0 0)-2 · 1 at low temperature [31]. Furthermore, the theoretical spectrum of DMF dative-bonded through the oxygen to the electrophilic dimer atom of a Ge2Si7H12 cluster is shown in Fig. 4b, and agreement with the experimental spectrum is excellent. In contrast, the theoretical spectrum for the nitrogen dative-bonded product (not shown) exhibits an intense mode at 1760 cm1 which we do not observe experimentally, and no modes are calculated to fall near 1640 cm1. The binding energy calculations also provide insight into whether the oxygen or nitrogen dative-bonded species is more likely. While the binding energy of the oxygen dative-bonded state is found to be 17.8 kcal/mol, the nitrogen dative-bonded structure possesses a binding energy of only 0.1 kcal/mol, a large difference which can be explained by a combination of electronic and steric effects. As illustrated in Fig. 2, delocalization of electron density from the nitrogen to oxygen atom greatly reduces the nucleophilicity of the nitrogen atom from that of an amine while increasing the nucleophilicity of the oxygen atom above that of ketones. In addition, the porbital character of the nitrogen lone pair in amides leads to substantial steric interactions between the nitrogen methyl groups and the surface and reduces the binding energy of the nitrogen dative bond even further. Based on these pieces of evidence, we propose that formation of a dative bond is selective and occurs through the oxygen atom rather than the nitrogen for DMF adsorbed on Ge(1 0 0)-2 · 1 at 310 K.
47
A detailed examination of the vibrational spectrum of chemisorbed DMF (Fig. 4a) yields additional information regarding the normal modes of the oxygen dative-bonded state. Based on our calculations as well as literature sources [53–55], the peak at 1640 cm1 in the chemisorbed spectrum is assigned to a normal mode involving asymmetric tas(N–C 0 @O) stretching coupled with aldehydic dal(C 0 –H) bending. The 25 cm1 shift of this mode from its location in the multilayer (Fig. 4c) is small compared to the 70 cm1 shift observed for the carbonyl stretch in acetone [31]. This can be explained by the presence of intermolecular hydrogen-bonding between the oxygen and aldehydic hydrogen in the DMF multilayer. This interaction redshifts the gas phase frequency of DMF, located at 1715 cm1, to 1665 cm1 in the condensed state, an effect not as strongly observed for acetone (15 cm1) [56]. After accounting for both the 50 cm1 (1715 cm1 to 1665 cm1) hydrogen bonding-related redshift found upon condensing from the gas phase and the 25 cm1 shift observed upon dative-bonding (1665 cm1 to 1640 cm1), the overall redshift is 75 cm1, which is substantially closer to the 85 cm1 redshift observed for acetone. It was seen for low temperature adsorption of acetone on Ge(1 0 0)-2 · 1 [31] that although there is a significant redshift of the carbonyl stretching frequency when the adsorbate dative-bonds through the oxygen, smaller shifts occur for other modes, such as methyl d(CH3) deformations. Similar spectral behavior is seen for DMF in Fig. 4a and c. Therefore, the two modes at 1487 cm1 and 1429 cm1 in the chemisorbed spectrum of DMF are assigned to methyl d(CH3) deformations, while the 1352 cm1 peak is assigned to an aldehydic dal(C 0 –H) bending mode. While these modes are similar to those present in the multilayer spectrum, the asymmetric tas(C–N–C) stretch, observed at 1256 cm1 in the multilayer and predicted by theory to occur at 1228 cm1 in the O-dative-bonded product, is not observed in the chemisorption spectrum. However, this mode is predicted to be approximately 30 times weaker than the tas(N–C 0 @O) + dal(C 0 –H) mode, and we believe the signal-to-noise ratio is not high enough to observe it.
48
A.J. Keung et al. / Surface Science 599 (2005) 41–54
3.2. N,N-dimethylformamide-d7 To help confirm the assignments of hydrogenrelated modes in DMF-d0, we investigated the reaction of N,N-dimethylformamide-d7 (Fig. 1b) with Ge(1 0 0)-2 · 1. Fig. 4d shows a 0.005 L saturation exposure of DMF-d7 chemisorbed at 310 K, and the multilayer spectrum of DMF-d7 is shown in Fig. 4e for comparison. The redshift of the combined asymmetric tas(N–C 0 @O) stretching and aldehydic dal(C 0 –H) bending mode from 1640 to 1613 cm1 for the chemisorbed species upon deuteration is consistent with that observed for DMF-d0 and DMF-d7 in the gas phase (20 cm1) [56]. As expected, the peaks at 1487, 1429, and 1352 cm1 in the chemisorbed spectrum of DMF-d0 corresponding to various d(C–H) deformations are not observed upon deuteration since these deuterated modes fall below the 1100 cm1 cutoff of our CaF2 windows. The appearance of a peak at 1397 cm1 for DMF-d7 was unexpected since the only experimentally observable mode not primarily involving hydrogen in the chemisorbed spectrum of DMFd0 was the tas(N–C 0 @O) + dal(C 0 –H) mode at 1640 cm1. While we do not believe the 1397 cm1 peak in DMF-d7 and the 1352 cm1 peak in DMF-d0 come from similar modes since deuteration normally redshifts frequencies, the 1397 cm1 peak may be a symmetric ts(N–C 0 @O) stretching mode. Our frequency calculations of DMF-d0 adsorbed in the oxygen dative-bonded state do not exhibit a symmetric ts(N–C 0 @O) stretching mode, however, a mode at 1392 cm1 with a clear symmetric ts(N–C 0 @O) stretching motion exists in the theoretical vibrational spectrum for DMF-d7. Thus, we presently assign the absorption feature at 1397 cm1 for DMF-d7 to a symmetric ts(N–C 0 @O) stretch and believe that a similar mode is not observed for DMF-d0 due to strong coupling with d(CH3) and d(C–H) deformations of similar energy.
for these molecules as well as to study the effect of the amide functionality on the a-CH ‘‘ene’’ dissociation reaction observed for acetone [31] and 5hexen-2-one [37]. Both 1-methyl-2-pyrrolidinone (MP) and N-methylcaprolactam (MC) contain an amide functional group embedded in an aliphatic ring (Fig. 1c and d). Unlike DMF, which has no a-CH bonds, these two molecules can potentially undergo an a-CH dissociation reaction following adsorption into an oxygen dative-bonded state, as shown in Fig. 5. Fig. 6 contains the vibrational spectra of MP and MC. The chemisorbed saturation spectrum of 0.1 L MP on Ge(1 0 0)-2 · 1 at 310 K, the theoretical spectrum of MP in the oxygen dativebonded state, and the multilayer spectrum are shown in Fig. 6a–c, respectively. The chemisorbed saturation spectrum of 0.1 L MC at 310 K, the theoretical spectrum of MC in the oxygen dativebonded state, and the multilayer spectrum are shown in Fig. 6d–f, respectively. Interestingly, no t(Ge–H) stretching modes are observed in the
3.3. 1-Methyl-2-pyrrolidinone and N-methylcaprolactam Two other tertiary amides were also studied to explore the generality of oxygen dative-bonding
Fig. 5. a-CH dissociation ‘‘ene’’ reactions for (a) 1-methyl-2pyrrolidinone and (b) N-methylcaprolactam.
A.J. Keung et al. / Surface Science 599 (2005) 41–54
49
Fig. 6. Infrared spectra of 1-methyl-2-pyrrolidinone and N-methylcaprolactam on Ge(1 0 0)-2 · 1: (a) saturation exposure of 1-methyl2-pyrrolidinone at 310 K, (b) theoretical spectrum of 1-methyl-2-pyrrolidinone oxygen dative-bonded, (c) 1-methyl-2-pyrrolidinone multilayers at 160 K (scaled), (d) saturation exposure of N-methylcaprolactam at 310 K, (e) theoretical spectrum of Nmethylcaprolactam oxygen dative-bonded, (f) N-methylcaprolactam multilayers at 160 K (scaled).
chemisorbed spectra for either compound, strongly suggesting that the a-CH dissociation product is not present. This observation may be due to an increase in activation barrier for the a-CH dissociation reaction due to the more rigid geometry of the aliphatic rings. For both compounds, the chemisorbed spectra are similar to their corresponding multilayer spectra with the exception of the redshifted asymmetric tas(N–C 0 @O) stretching modes near 1600 cm1. In addition, the chemisorbed spectra agree well with the theoretically calculated spectra for the oxygen
dative-bonded structures (Fig. 6b and e) suggesting that, like DMF, oxygen dative-bonded structures are the major surface adducts at 310 K. The redshifting of the asymmetric tas(N–C 0 @O) stretching mode at 1614 cm1 for chemisorbed MP (Fig. 6a) to 1582 cm1 for chemisorbed MC (Fig. 6d) is due to the greater ring strain in MP [57]. The same effect is seen in the multilayer spectra. Also, MP and MC do not appreciably hydrogen bond in the condensed state and, as a result, exhibit larger redshifts of their asymmetric tas(N–C 0 @O) stretching modes (60 cm1 and
50
A.J. Keung et al. / Surface Science 599 (2005) 41–54
52 cm1, respectively) upon adsorbing as dative bonds on Ge(1 0 0)-2 · 1. It is also important to note the absence of peaks in the chemisorbed spectrum for both compounds near 1350 cm1, the location of the strong aldehydic dal(C 0 –H) bend feature in the chemisorbed spectrum of DMF (Fig. 4a). This is expected, since both MP and MC lack aldehydic hydrogen atoms, and helps confirm the assignment of the 1352 cm1 mode for DMF. Since MP and MC differ by only two –CH2– groups, the majority of the vibrational features in their chemisorbed spectra are similar. d(CH3)/ d(C–H) deformation modes are assigned to the peaks at 1451/1412 cm1 and 1443/1406 cm1 for MP and MC, respectively. The strong peaks located at 1500 cm1 and 1483 cm1 for MP and MC, respectively, are attributed to more complex normal modes involving coupling of primarily three stretches and one bend: t(C 0 @O) + t(C 0 –C) + t(C 0 –N) + d(C–H). This complex normal mode
assignment is based on our theoretical frequency calculations as well as comparison with the literature [58]. 3.4. Desorption kinetics After adsorption, DMF, DMF-d7, and MP were each observed to desorb on the timescale of minutes at 310 K. Only MC was not found to appreciably desorb with time. As shown in Fig. 7, infrared spectra taken over the span of approximately 1 h for (a) DMF and (b) MP indicate that all peaks attenuate at similar rates for both compounds. In addition, neither compound completely desorbs even at 1 h. This desorption behavior suggests the existence of coverage dependent rate constants, in which molecular desorption occurs at high coverages, with a slowing of desorption once a more stable surface state is attained. Many reasons have been proposed for this type of coverage dependent phenomena on surfaces
Fig. 7. Infrared spectra of time dependent desorption taken periodically for an hour following saturation exposures of (a) N,Ndimethylformamide and (b) 1-methyl-2-pyrrolidinone on Ge(1 0 0)-2 · 1 at 310 K.
A.J. Keung et al. / Surface Science 599 (2005) 41–54
including destabilizing lateral interactions between adsorbed species [59,60] as well as adsorbateinduced changes and reconstructions of the surface [61]. Experimental and theoretical work by Queeney et al. [62] on Si(1 0 0)-2 · 1 has indicated that pre-adsorbed ammonia can alter the charge distribution at the surface and influence the location of subsequent ammonia adsorption. Widjaja and Musgrave [46] investigated the same system theoretically and found significant non-local surface charge effects of adsorbed ammonia where electron density is delocalized to adjacent dimers, thus increasing the stability of NH3 adsorbed in a dative-bonded state. This non-local charge transfer effect was also predicted to ‘‘poison’’ the adsorption of subsequent ammonia molecules on adjacent dimers in the same row [63]. Similarly, we speculate that the coverage dependent desorption observed in the present work could result from weaker binding energies caused by an accumulation of negative surface charge at high coverage. Desorption of a percentage of the dative-bonded species removes some of the charge accumulated at the surface, increases the binding energy of the remaining adsorbates, and ultimately leads to incomplete desorption. In the case of MC, appreciable time dependent desorption is not observed. MC molecules are the heaviest of the tertiary amides studied and possess a greater potential for significant van der Waals interactions. Such lateral interactions between the large aliphatic rings of neighboring absorbates as well as interactions with the surface may lead to higher binding energies at all coverages and prevent time dependent desorption like that observed for DMF, DMF-d7, and MP. The larger ring of MC may also prevent achievement of higher coverages by sterically blocking open surface sites such that no significant surface charge can build up. A more detailed study of the observed time dependence for DMF and MP enables the determination of an approximate binding energy range for the oxygen dative-bonded state. To complete such an analysis, integrated peak areas of the 1640 and 1614 cm1 peaks for DMF and MP, respectively, were selected since they have the greatest initial intensity and greatest separation from neighboring spectral features. Fig. 8 shows the natural log of
51
these integrated intensities (normalized to the initial intensity of each compound) for DMF (d, s) and MP (m, D) plotted as a function of time. The simple molecular desorption of dative-bonded species is expected to follow a first order kinetic rate law and thus have a linear fit in the absence of coverage dependent effects. However, the plots in Fig. 8 do not have constant slopes, implying coverage dependent rate constants for desorption, and therefore coverage dependent Arrenhius parameters [64]. Even with a simple rate expression, k = AeE/RT, extracting binding energies from the data is difficult due to the possible coverage dependence of both variables, the binding energy, E, and pre-exponential factor, A. Often, a constant pre-exponential factor of 1012–1013 s1 is assumed for first order desorption reactions [65]. However, Wang and Seebauer [66] reported in a study of over 45 semiconductor systems that only 10% of the systems exhibit first order desorption pre-exponential
Fig. 8. Plots of the natural log of normalized integrated areas for the 1640 cm1 peak of N,N-dimethylformamide (d/s) and 1614 cm1 peak of 1-methyl-2-pyrrolidinone (m/D) as functions of time. Fits to early (d/m) and late (s/D) data points are shown for N,N-dimethylformamide ( ) and 1-methyl-2-pyrrolidinone (- - -).
52
A.J. Keung et al. / Surface Science 599 (2005) 41–54
factors within that range. Upon inspection of the Ge(1 0 0) systems analyzed by Wang and Seebauer, pre-exponential factors ranged from a minimum of 109 s1 for Na desorption [67] to a maximum of 1014 s1 for HCl and HBr desorption [68,69]. To obtain reasonable limits for the binding energies of the DMF and MP systems as well as to take into account the uncertainty in the pre-exponential factors, we calculated binding energies at early and late times using both the low (109 s1) and high (1014 s1) pre-exponential factors found in the literature. The rate constants or slopes of the plots in Fig. 8 were determined using fits to the equation for first order reactions, lnð//0 Þ ¼ c kt, where k is the rate constant, / is the integrated peak area, /0 is the initial integrated peak area, and t is time. c is included to account for cases where t0 is not equal to 0 (i.e. the latter fits: D, s). Rate constants were found at both early times (m, d) and late times (D, s) of the experiment. We then used the Arrhenius equation, k = AeE/RT, using both limits of the pre-exponential factor, A, to back out the binding energies, E. The calculated binding energies are shown in Table 1 along with the binding energies calculated using density functional theory. Considering the uncertainty and potential coverage dependence of the pre-exponential factor, the experimentally determined binding energy ranges agree with the results of our density functional theory calculations for the oxygen dative bond of each compound. Interestingly, the low limit of the preexponential factor appears to be more consistent with our theoretically determined binding energy. More importantly, the binding energies of DMF and MP in the oxygen dative-bonded state increase
with decreasing surface coverage. As shown in Table 1, we estimate that the binding energies of DMF and MP adsorbed in the oxygen dativebonded state increase by 0.4 and 1.1 kcal/mol, respectively, between the initial and final surface coverage for each compound. This gain in stability illustrates the dramatic effect that changes in the surface environment can have on adsorbates. The above analysis takes into account the coverage dependence of a surface speciesÕ stability as well as a wide range of pre-exponential factors, and it establishes a rough binding energy range for the tertiary amide oxygen dative bond on Ge(1 0 0)-2 · 1. The desorption of DMF and MP over time indicates that molecularly-bonded adsorbates with binding energies between 17 and 25 kcal/mol are just at the cusp of stability in this system.
4. Conclusion The adsorption of a series of tertiary amides has been investigated with MIR-FTIR spectroscopy and DFT calculations. The only observed surface products were oxygen dative-bonded species. Selective dative-bonding through oxygen can be explained by electron delocalization from nitrogen to oxygen as well as the unfavorable steric interactions present at the nitrogen. DFT frequency calculations for the dative-bonded species are in excellent agreement with the experimental spectra. In addition, time dependent desorption was observed for the dative-bonded species on the order of minutes which allowed rough binding energy
Table 1 Binding energies for N,N-dimethylformamide (d/s) and 1-methyl-2-pyrrolidinone (m/D) calculated from experimental data, shown in Fig. 8, at early (d/m) and late (s/D) times fitted to the equation for first order desorption, lnð//0 Þ ¼ c kt k = (109 s1)eE/RT, E (kcal/mol) =
k = (1014 s1)eE/RT, E (kcal/mol) =
DFT, O-dative, E (kcal/mol) =
DMF
d s
17.4 17.8
24.4 24.9
17.8
MP
m D
17.6 18.7
24.7 25.8
17.7
Pre-exponential factors of 109 s1 and 1014 s1 were used to calculate two sets of binding energies. Binding energies calculated with density functional theory are shown for comparison.
A.J. Keung et al. / Surface Science 599 (2005) 41–54
ranges to be experimentally determined. These ranges encompassed the DFT values calculated for oxygen dative-bonded structures. A combination of experimental and theoretical infrared spectra, experimentally determined binding energy ranges, and the observation of time-dependent desorption suggest no dissociation products were formed at 310 K. Amides hint at a method with which to direct reactivity to or away from target sites in a compound. Electron delocalization could be used in the future to selectively increase the nucleophilicity of specific hetero-atoms. In addition, this new understanding of tertiary amide reactivity and the spectral features of their chemisorbed adducts will greatly facilitate future work with secondary and primary amides. Ultimately, developing a comprehensive understanding of amide reactivity will be necessary to investigate the reactivity of larger biological molecules such as polypeptides or nucleic acids.
Acknowledgments This work was supported by a grant from the National Science Foundation (CHE 0245260). A.J.K. would like to acknowledge support from the Stanford PresidentÕs Scholar Program and the Merck Award for Student Research. M.A.F. thanks the National Science Foundation for support in the form of a Graduate Research Fellowship. D.W.P. thanks Stanford University for support in the form of a Stanford Graduate Fellowship. Scientific discussions with Ansoon Kim and James A. Van Deventer were greatly appreciated and enjoyed.
References [1] M. Leong, B. Doris, J. Kedzierski, K. Rim, M. Yang, Science 306 (2004) 2057. [2] R. Langer, N.A. Peppas, AIChE Journal 49 (2003) 2990. [3] M. Chalfie, Y. Tu, G. Euskirchen, W.W. Ward, D.C. Prasher, Science 263 (1994) 802. [4] M. Chalfie, Current Biology 4 (1994) 443. [5] J. Downward, BMJ 328 (2004) 1245. [6] J. Downward, Oncogene 23 (2004) 8334.
53
[7] J. Downward, Oncogene 23 (2004) 8376. [8] J.T. Santini, M.J. Cima, R. Langer, Nature 397 (1998) 335. [9] X. Michalet, F.F. Pinaud, L.A. Bentolila, J.M. Tsay, S. Doose, J.J. Li, G. Sundaresan, A.M. Wu, S.S. Gambhir, S. Weiss, Science 307 (2005) 538. [10] A. Sadana, Bioseparation 3 (1992) 297. [11] V. Hlady, J. Buijs, Current Opinion in Biotechnology 7 (1996) 72. [12] N. Nath, J. Hyun, H. Ma, A. Chilkoti, Surface Science 570 (2004) 98. [13] V.P. Raut, M.A. Agashe, S.J. Stuart, R.A. Latour, Langmuir 21 (2005) 1629. [14] R.J. Hamers, S.K. Coulter, M.D. Ellison, J.S. Hovis, D.F. Padowitz, M.P. Schwartz, C.M. Greenlief, J.N. Russell, Accounts of Chemical Research 33 (2000) 617. [15] M.C. Hersam, N.P. Guisinger, J.W. Lyding, Nanotechnology 11 (2000) 70. [16] G.P. Lopinski, D.D.M. Wayner, R.A. Wolkow, Nature 406 (2000) 48. [17] T. Strother, R.J. Hamers, L.M. Smith, Nucleic Acids Research 28 (2000) 3535. [18] S.F. Bent, Surface Science 500 (2002) 879. [19] R.F. Pierret, Semiconductor Device Fundamentals, Prentice Hall, 1995. [20] M.A. Filler, S.F. Bent, Progress in Surface Science 73 (2003) 1. [21] K.P.C. Vollhardt, N.E. Schore, Organic Chemistry, Structure and Function, W.H. Freeman and Company, New York, 2003. [22] X.P. Cao, R.J. Hamers, Journal of the American Chemical Society 123 (2001) 10988. [23] C. Mui, G.T. Wang, S.F. Bent, C.B. Musgrave, Journal of Chemical Physics 114 (2001) 10170. [24] C. Mui, J.H. Han, G.T. Wang, C.B. Musgrave, S.F. Bent, Journal of the American Chemical Society 124 (2002) 4027. [25] A.J. Carman, L.H. Zhang, J.L. Liswood, S.M. Casey, Journal of Physical Chemistry B 107 (2003) 5491. [26] J.H. Cho, L. Kleinman, Physical Review B 67 (2003) 201301. [27] J.-H. Cho, L. Kleinman, Physical Review B 68 (2003) 245314. [28] J.B. Wu, Y.W. Yang, Y.F. Lin, H.T. Chiu, Journal of Physical Chemistry B 108 (2004) 1677. [29] G.T. Wang, C. Mui, J.F. Tannaci, M.A. Filler, C.B. Musgrave, S.F. Bent, Journal of Physical Chemistry B 107 (2003) 4982. [30] A. Kim, M.A. Filler, S. Kim, S.F. Bent, Journal of the American Chemical Society 127 (2005) 6123. [31] G.T. Wang, C. Mui, C.B. Musgrave, S.F. Bent, Journal of Physical Chemistry B 105 (2001) 12559. [32] G.T. Wang, C. Mui, C.B. Musgrave, S.F. Bent, Journal of Physical Chemistry B 105 (2001) 3295. [33] M.Z. Hossain, S. Machida, Y. Yamashita, K. Mukai, J. Yoshinobu, Journal of the American Chemical Society 125 (2003) 9252. [34] F. Tao, M.H. Qiao, Z.H. Wang, G.Q. Xu, Journal of Physical Chemistry B 107 (2003) 6384.
54
A.J. Keung et al. / Surface Science 599 (2005) 41–54
[35] Y.E. Cho, J.Y. Maeng, S. Kim, S.Y. Hong, Journal of the American Chemical Society 125 (2003) 7514. [36] S. Hong, Y.E. Cho, J.Y. Maeng, S. Kim, Journal of Physical Chemistry B 108 (2004) 15229. [37] G.T. Wang, C. Mui, C.B. Musgrave, S.F. Bent, Journal of the American Chemical Society 124 (2002) 8990. [38] D.M. Schnur, Y.H. Yuh, D.R. Dalton, Journal of Organic Chemistry 54 (1989) 3779. [39] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, J.A. Montgomery Jr., T. Vreven, K.N. Kudin, J.C. Burant, J.M. Millam, S.S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G.A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J.E. Knox, H.P. Hratchian, J.B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W. Ochterski, P.Y. Ayala, K. Morokuma, G.A. Voth, P. Salvador, J.J. Dannenberg, V.G. Zakrzewski, S. Dapprich, A.D. Daniels, M.C. Strain, O. Farkas, D.K. Malick, A.D. Rabuck, K. Raghavachari, J.B. Foresman, J.V. Ortiz, Q. Cui, A.G. Baboul, S. Clifford, J. Cioslowski, B.B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R.L. Martin, D.J. Fox, T. Keith, M.A. Al-Laham, C.Y. Peng, A. Nanayakkara, M. Challacombe, P.M.W. Gill, B. Johnson, W. Chen, M.W. Wong, C. Gonzalez, J.A. Pople, Gaussian 03, Gaussian, Inc., Wallingford, CT, 2004. [40] S.H. Vosko, L. Wilk, M. Nusair, Canadian Journal of Physics 58 (1980) 1200. [41] C. Lee, W. Yang, R.G. Parr, Physical Review B 37 (1988) 785. [42] A.D. Becke, Journal of Chemical Physics 98 (1993) 5648. [43] W. Kohn, L.J. Sham, Physical Review 140 (1965) 1133. [44] C. Mui, S.F. Bent, C.B. Musgrave, Journal of Physical Chemistry A 104 (2000) 2457. [45] A.P. Scott, L. Radom, Journal of Physical Chemistry 100 (1996) 16502. [46] Y. Widjaja, C.B. Musgrave, Surface Science 469 (2000) 9. [47] C. Mui, C.B. Musgrave, Langmuir 21 (2005) 5230. [48] G.A. DiLabio, R.A. Wolkow, E.R. Johnson, Journal of Chemical Physics 122 (2005) 044708. [49] J.A. Barriocanal, D.J. Doren, Journal of Physical Chemistry B 104 (2000) 12269.
[50] J.A. Barriocanal, D.J. Doren, Journal of Vacuum Science and Technology, Part A: Vacuum Surfaces and Films 18 (2000) 1959. [51] S. Bocharov, A.T. Mathauser, A.V. Teplyakov, Journal of Physical Chemistry B 107 (2003) 7776. [52] S. Bocharov, A.T. Teplyakov, Surface Science 573 (2004) 403. [53] G. Durgaprasad, D.N. Sathyanarayana, C.C. Patel, Bulletin of the Chemical Society of Japan 44 (1971) 316. [54] T.C. Jao, I. Scott, Journal of Molecular Spectroscopy 92 (1982) 1. [55] C.H. Langley, N.L. Allinger, Journal of Physical Chemistry A 106 (2002) 5638. [56] S.E. Stein, in: P.J. Linstrom, W.G. Mallard (Eds.), NIST Chemistry WebBook, NIST Standard Reference Database Number 69, National Institute of Standards and Technology, Gaithersburg, MD, 2003. [57] D.L. Pavia, G.M. Lampman, G.S. Kriz, Introduction to Spectroscopy, Brooks/Cole, Australia, 2001. [58] D.P. McDermott, Journal of Physical Chemistry 90 (1986) 2569. [59] P.K. Johansson, Chemical Physics Letters 65 (1979) 366. [60] A.V. Myshlyavtsev, V.P. Zhdanov, Chemical Physics Letters 162 (1989) 43. [61] A.V. Myshlyavtsev, V.P. Zhdanov, Journal of Chemical Physics 92 (1990) 3909. [62] K.T. Queeney, Y.J. Chabal, K. Raghavachari, Physical Review Letters 86 (2001) 1046. [63] Y. Widjaja, C.B. Musgrave, Journal of Chemical Physics 120 (2004) 1555. [64] E.G. Seebauer, A.C.F. Kong, L.D. Schmidt, Surface Science 193 (1988) 417. [65] R.C. Baetzold, G.A. Somorjai, Journal of Catalysis 45 (1976) 94. [66] Z. Wang, E.G. Seebauer, Applied Surface Science 181 (2001) 111. [67] B. Naydenov, L. Surnev, Surface Science 370 (1997) 155. [68] M.P. Develyn, Y.M.L. Yang, S.M. Cohen, Journal of Chemical Physics 101 (1994) 2463. [69] Y.L. Yang, S.M. Cohen, M.P. DÕEvelyn, Materials Research Society Symposium Proceedings 282 (1993) 421.
