pubs.acs.org/Langmuir © 2010 American Chemical Society
Adsorption and Bonding of First Layer and Bilayer Terephthalic Acid on the Cu(100) Surface by High-Resolution Electron Energy Loss Spectroscopy† Yan Ge,‡ Hilmar Adler,‡ Arjun Theertham,‡ Larry L. Kesmodel,§ and Steven L. Tait*,‡ ‡
Department of Chemistry and §Department of Physics, Indiana University, Bloomington, Indiana 47405 Received April 20, 2010. Revised Manuscript Received May 28, 2010
The self-assembled and highly ordered first layer of terephthalic acid on Cu(100) as well as its bilayer on the same surface are studied here using high-resolution electron energy loss spectroscopy (HREELS), Auger electron spectroscopy, and low energy electron diffraction. These experiments show completion of the first layer before growth of the second layer. HREELS measurements show that the first layer of the acid deprotonates, which is seen in the absence of the OH stretching mode for the acid groups. However, this mode is present in the bilayer structure, confirming that the deprotonation is due to a reaction with the Cu surface and suggesting that there is little mixing of the layers. It has been suggested previously that the TPA monolayer structure is stabilized by an intermolecular hydrogen bonding interaction, but we are not able to resolve any distortion of the CH stretching mode for such an interaction, but instead see evidence for direct bonding to the Cu surface.
1. Introduction Terephthalic acid (TPA, C6H4(COOH)2) growth on surfaces is of high interest for two key fields. It has been shown that TPA is a highly effective ligand for metal-organic frameworks (MOFs), both in a traditional sense1 and for creating two-dimensional MOFs at surfaces for highly ordered surface nanostructures.2 TPA is also an important component in organic thin films research, as it presents a route to create a surface with a chemical functionalization, especially on surfaces such as TiO2, where upright TPA molecules present a termination of carboxyl groups.3 For these reasons, the self-assembly and ordering of TPA has been studied on a wide variety of single crystal surfaces, including Pd,4 Pt,5 Au,6,7 Ag,8 Cu,9-12 Al2O3,13 TiO2,3,14,15 and Si,8 using density functional theory, scanning tunneling microcopy (STM), X-ray photoelectron spectroscopy (XPS), and other surface analysis techniques. † Part of the Molecular Surface Chemistry and Its Applications special issue. *To whom correspondence should be addressed. E-mail:
[email protected].
(1) Yaghi, O. M.; O’Keeffe, M.; Ockwig, N. W.; Chae, H. K.; Eddaoudi, M.; Kim, J. Nature 2003, 423, 705. (2) Lingenfelder, M. A.; Spillmann, H.; Dmitriev, A.; Stepanow, S.; Lin, N.; Barth, J. V.; Kern, K. Chem.;Eur. J. 2004, 10, 1913–1919. (3) Rahe, P.; Nimmrich, M.; Nefedov, A.; Naboka, M.; W€oll, C.; K€uhnle, A. J. Phys. Chem. C 2009, 113, 17471–17478. (4) Canas-Ventura, M. E.; Klappenberger, F.; Clair, S.; Pons, S.; Kern, K.; Brune, H.; Strunskus, T.; Woll, C.; Fasel, R.; Barth, J. V. J. Chem. Phys. 2006, 125, 184710. (5) Kim, Y. G.; Yau, S. L.; Itaya, K. Langmuir 1999, 15, 7810–7815. (6) Clair, S.; Pons, S.; Seitsonen, A. P.; Brune, H.; Kern, K.; Barth, J. V. J. Phys. Chem. B 2004, 108, 14585–14590. (7) Ikezawa, Y.; Masuda, R. Electrochim. Acta 2008, 53, 5456–5463. (8) Suzuki, T.; Lutz, T.; Payer, D.; Lin, N.; Tait, S. L.; Costantini, G.; Kern, K. Phys. Chem. Chem. Phys. 2009, 11, 6498–6504. (9) Stepanow, S.; Strunskus, T.; Lingenfelder, M.; Dmitriev, A.; Spillmann, H.; Lin, N.; Barth, J. V.; Woll, C.; Kern, K. J. Phys. Chem. B 2004, 108, 19392–19397. (10) Tait, S. L.; Wang, Y.; Costantini, G.; Lin, N.; Baraldi, A.; Esch, F.; Petaccia, L.; Lizzit, S.; Kern, K. J. Am. Chem. Soc. 2008, 130, 2108–2113. (11) Martin, D. S.; Cole, R. J.; Haq, S. Phys. Rev. B 2002, 66, 155427. (12) Atodiresei, N.; Caciuc, V.; Schroeder, K.; Blugel, S. Phys. Rev. B 2007, 76, 115433. (13) Ogawa, H. J. Phys. Org. Chem. 1991, 4, 346–352. (14) Prauzner-Bechcicki, J. S.; Godlewski, S.; Tekiel, A.; Cyganik, P.; Budzioch, J.; Szymonski, M. J. Phys. Chem. C 2009, 113, 9309–9315. (15) Tekiel, A.; Prauzner-Bechcicki, J. S.; Godlewski, S.; Budzioch, J.; Szymonski, M. J. Phys. Chem. C 2008, 112, 12606–12609.
Langmuir 2010, 26(21), 16325–16329
Surface functionalization by self-assembly of organic ligands is a method of growing interest for the development of functional, periodic nanometer-scale structures.16,17 TPA can form well ordered, self-organized structures on metal surfaces, such as copper,9-12 gold,6,7 and silver.8 Recent studies have demonstrated the use of TPA to efficiently pattern well ordered nanometer-scale architectures on the Cu(100) surface, which have been studied with STM, low energy electron diffraction (LEED), and XPS.2,9,10 A molecular model of TPA and a schematic illustrating the submonolayer structure on Cu(100), as understood from those studies, are given in Figure 1. Other ligands and metals have also been used with TPA to modify and tailor the resulting nanoscale pattern.2,10,18 High-resolution electron energy loss spectroscopy (HREELS) has seen wide applications in the vibrational spectroscopy of adsorbates at surfaces, especially for organic adsorbates.19 HREELS offers high surface sensitivity as well as the capability to detect both dipole scattering modes (with strong selection rules) as well as impact scattering modes (weak dependence on orientation) and therefore provides insight into molecular orientation.20 Here, we present the first high-resolution electron energy loss spectroscopy studies of this important ligand on a surface. These results on Cu(100) confirm results from the prior studies and also provide further insight into the interaction with the surface and within the TPA layer.
2. Experimental Section The experiments were performed in a two-chamber ultrahigh vacuum system. Terephthalic acid (99.9%, purchased from Sigma Aldrich) was vapor deposited from a home-built Knudsen cell evaporator onto a room temperature Cu(100) substrate at a rate of about 0.1 monolayer per minute. We quantify the surface (16) Barth, J. V. Annu. Rev. Phys. Chem. 2007, 58, 375–407. (17) Barth, J. V.; Costantini, G.; Kern, K. Nature 2005, 437, 671–679. (18) Langner, A.; Tait, S. L.; Lin, N.; Chandrasekar, R.; Ruben, M.; Kern, K. Chem. Commun. 2009, 2502–2504. (19) Kesmodel, L. L. High-resolution Electron Energy Loss Spectroscopy. In Encyclopedia of Surface and Colloid Science, 2nd ed.; Somasundaren, P., Ed.; Taylor and Francis: New York, 2006. (20) Ibach, H. Physics of Surfaces and Interfaces; Springer: Berlin, 2006.
Published on Web 06/15/2010
DOI: 10.1021/la101582k
16325
Article
Ge et al.
Figure 1. Molecular model of terephthalic acid (left) and schematic diagram of TPA submonolayer adsorption on Cu(100) based on LEED measurements in the present study and prior results by STM.2 The unit cell of the TPA (3 3) structure is highlighted by the blue square. coverage of TPA in units of monolayers (ML), where the full monolayer (1.0 ML) is defined by the (3 3) TPA layer on Cu(100), reported previously and discussed below, which has a surface density of 1.71 1014 molecules cm-2. In our studies of this structure, we have used a surface coverage of about 0.6 ML which consists of molecules in the first layer (submonolayer). We also report experiments at approximately 2.0 ML TPA, which we refer to as the “bilayer” to emphasize that these data will contain spectral features from both the first and second layers of TPA. The deposition chamber has a base pressure of 7 10-10 Torr. The electron energy loss spectroscopy (LK Technologies, model LK2000), Auger electron spectroscopy (AES), and LEED analyses were carried out in a separate but connected ultrahigh vacuum (UHV) analysis chamber with a base pressure of 2 10-10 Torr. The Cu(100) single crystal sample was cleaned by repeating cycles of Arþ sputtering at room temperature with a beam energy of 1.5 keV and annealing at 800 K. The Cu(100) surface quality was assessed by the presence of a sharp (1 1) LEED pattern (Figure 2a) with low diffuse background intensity and high copper to carbon signal ratio in Auger spectra (C coverage < 0.06 ML). The HREELS measurements were made with a double-pass 127 cylindrical deflection electron spectrometer operated at 7.5 meV (60 cm-1) resolution with elastic beam rates of about 105 counts per second (CPS) for TPA adsorbed on Cu(100) surface.21 Spectra are normalized to the background intensity above 2500 cm-1.
3. Results and Discussion 3.1. Growth by Auger Electron Spectroscopy and LEED. Figure 2a shows a sharp (1 1) Cu(100) LEED pattern prior to the deposition of TPA. A sharp (3 3) LEED pattern is observed (Figure 2b) after TPA deposition (0.6 ML) onto a clean Cu(100) surface and annealing at 310 K for 10 min, indicating a wellordered structure of the TPA adsorbate first layer. Thus, our TPA structures as shown by LEED are consistent with previous studies by LEED10 and with the TPA monolayer structure observed by STM.2 AES was used to monitor the cleanliness of the surface as well as to observe growth of the TPA monolayer and second layer during step-by-step deposition. The ratio of the carbon to copper AES intensity increases linearly with TPA deposition time but changes slope upon completion of the first layer. The deposition rate was calibrated by observing this change in slope of AES C/Cu ratio versus deposition time. These results were also correlated with the sharpness of the LEED (33) pattern and found to be consistent in determining the completion of the monolayer. We (21) Kesmodel, L. L. J. Vac. Sci. Technol., A 1983, 1, 1456–1460.
16326 DOI: 10.1021/la101582k
expect that the coverages stated here are accurate to better than 20% using these calibration methods. 3.2. HREELS of First Layer and Bilayer TPA on Cu(100). In Figure 3, we present HREEL spectra recorded from first layer (bottom) and bilayer (top) TPA molecules on Cu(100). For the first layer molecules, the most intense bands are at 261, 336, 752, and 879 cm-1. Weaker features can also be observed at 1104, 1342, 1542, and 3032 cm-1. Peak assignments are given in Table 1. Some of the spectral features observed may be an overlap of several modes. In making the mode assignments, we note that there is, to our knowledge, no direct analogue to the TPA/Cu(100) system. Benzoic acid on Cu(110) has been studied by HREELS,22 and it should exhibit some similar vibrational features, especially for the carboxylate-Cu modes. The bonding geometry in that case is quite different from the present case, since the benzoic acid was found to be upright on Cu(110), whereas TPA lies flat on Cu(100) and so we do not expect close agreement for all modes. Some CH bending, CH stretch, and ring modes of a parallel bonded TPA should be similar to those of flat lying benzene adsorbed on a metal surface, so we also list the relevant modes for benzene on Pd(100).23 Finally, we also show IR results of solid phase TPA in Table 1.24 The spectra features at 261 and 336 cm-1 are assigned to ν(Cu-C) and ν(Cu-O), respectively. The former is based on comparison to studies of benzene adsorption on Pd(111) with a feature at 280 cm-1.23 The latter is quite close to the methoxy-Cu stretch observed on Cu(100) surfaces at 290 cm-1 observed previously.25,26 O adsorption on the Cu(100) surface26 leads to a ν(Cu-O) stretch feature at 350 cm-1, and benzoic acid on Cu(110) has a ν(Cu-O) mode at 490 cm-1.22 The difference in the TPA system studied here compared to these studies is likely due to the very different adsorption geometries as well as surface differences. The bands at 1342 and 1542 cm-1 can be assigned to O-C-O symmetric and asymmetric stretches, but there is likely overlap with ring stretch and deformation modes as noted in Table 1. We have also presented HREELS data for a sample of approximately two layers of TPA on Cu(100) in Figure 3. We note that this spectrum is expected to include the features of the first layer spectrum, albeit at a reduced intensity, in addition to new features arising from the second layer of TPA molecules. The intensity of Cu-O drops dramatically and shifts slightly to higher frequency. The Cu-O and Cu-C vibrational modes for the second layer are somewhat different from the features of the first layer TPA adsorbate. As the bilayer forms, there is some screening of these features due to interaction of the incident or scattered electrons with the second layer. The assignment of the OH stretching region is straightforward and indicates the deprotonation of the carboxylate groups in the first layer TPA, which has been reported previously.9,10 We also note that the bilayer exhibits a strong mode at 1752 cm-1 associated with the CdO stretch, a mode at 531 cm-1 due to deformation modes involving the COOH group, and an OH stretch mode at 3571 cm-1, none of which are present in the monolayer. Generally, IR spectra of carboxylic acids, which are usually dimerized, give rise to a CdO stretch band in the region (22) Frederick, B. G.; Ashton, M. R.; Richardson, N. V.; Jones, T. S. Surf. Sci. 1993, 292, 33–46. (23) Waddill, G. D.; Kesmodel, L. L. Phys. Rev. B 1985, 31, 4940–4946. (24) Tellez, C. A.; Hollauer, E.; Mondragon, M. A.; Castano, V. M. Spectrochim. Acta, Part A 2001, 57, 993–1007. (25) Sexton, B. A. Surf. Sci. 1979, 88, 299–318. (26) Ellis, T. H.; Wang, H. Langmuir 1994, 10, 4083–4088.
Langmuir 2010, 26(21), 16325–16329
Ge et al.
Article
Figure 2. Low energy electron diffraction (LEED) of (a) clean Cu(100), showing a sharp (1 1) pattern, and (b) submonolayer coverage (∼0.6 ML) of TPA on Cu(100), showing a (3 3) overlayer. LEED beam energies were (a) 144 eV and (b) 96 eV.
Figure 3. Comparison of the high-resolution electron energy loss spectra of first layer (0.6 ML) and bilayer TPA (2 ML) adsorbed on Cu(100) surfaces.
1725-1700 cm-1.27 However, in very dilute solution or in the vapor phase, when the acid exists as a monomer, the CdO stretch is observed at 1760 cm-1,27 in very good agreement with our peak at 1752 for the bilayer, which we conclude has effectively a monomer character, most likely due to a standing geometry of the second layer of TPA on the surface. 3.3. ν(CH) in HREELS of First Layer and Bilayer TPA on Cu(100). The weak but key band located at 3032 cm-1 in Figure 3 is due to the impact-excited C-H stretch. The frequency of this mode indicates that the aromatic character of the ring is preserved upon adsorption.23,28-30 Similarly, the strong peak at 752 cm-1 is consistent with the out-of-plane CH bending vibration associated with a parallel adsorption geometry. It has been suggested in prior work on the TPA/Cu(100) system that the well-ordered, self-assembled TPA islands on the surface are stabilized by intermolecular hydrogen bonding between the deprotonated carboxylate groups and the CH of a neighboring (27) Socrates, G. Infrared Characteristic Group Frequencies, 2nd ed.; John Wiley & Sons: New York, 1994. (28) Raval, R.; Parker, S. F.; Chesters, M. A. J. Chem. Soc., Faraday Trans. 1996, 92, 2611–2614. (29) Lomas, J. R.; Baddeley, C. J.; Tikhov, M. S.; Lambert, R. M. Chem. Phys. Lett. 1996, 263, 591–596. (30) Xi, M.; Yang, M. X.; Jo, S. K.; Bent, B. E.; Stevens, P. J. Chem. Phys. 1994, 101, 9122–9131.
Langmuir 2010, 26(21), 16325–16329
molecule.9 CH 3 3 3 O hydrogen bonds are usually categorized as “weak,” but have been observed in other systems,31,32 and shown to determine crystal structure in some cases.33 Previous IR work with CH 3 3 3 O hydrogen bonding systems have shown a weak peak shift accompanied by an increase in peak intensity and bandwidth upon hydrogen bonding.31,32,34 However, in this study, we do not find evidence for the first layer TPA CH 3 3 3 O hydrogen bonding model. Our analysis of the CH (and CD) stretching modes does not find any distortion or shift of those modes, which would be expected if they are perturbed by a direct interaction in a hydrogen bond. (Note that there are other carboxylic acid/surface systems where the acid groups do not deprotonate and do dimerize by hydrogen bonding.35) We note that the density of 3 3 TPA layer precludes the possibility of any metal coordination with Cu adatoms in the plane of the molecules above the surface, which is reported for other molecules on Cu(100).36 We do see a clear peak at 323 cm-1 which is very consistent with results of benzoic acid on Cu(110), where the feature at 490 cm-1 was attributed to direct bonding of the O atom to the Cu surface atom. Therefore, we conclude that the bonding of the carboxylate O to the Cu surface may be a more significant interaction than any intermolecular bonding. We suggest that the strong propensity of TPA to form well-ordered islands on Cu(100) may not be due to a direct intermolecular attraction, but rather a substrate mediated process, such as surface layer strain relief, as has been observed in other systems.37 3.4. HREELS of Deuterated TPA. We also conducted a series of measurements of first layer deuterium-containing terephthalic acid, which we compare to the undeuterated TPA in Figure 4 and Table 2. The deuterium substitution is at the side of the aromatic ring of the molecule and not in the acid groups. It is generally expected that the effect of this isotopic substitution will (31) Steiner, T. Angew. Chem., Int. Ed. 2002, 41, 48–76. (32) Desiraju, G. R.; Steiner, T. The Weak Hydrogen Bond in Structural Chemistry and Biology; Oxford University Press: Oxford, 1999. (33) Desiraju, G. R.; Sharma, C. V. K. M. Chem. Commun. 1991, 1239–1241. (34) Lutz, B. T. G.; Jacob, J.; van der Maas, J. H. Vib. Spectrosc. 1996, 12, 197– 206. (35) Barth, J. V.; Weckesser, J.; Lin, N.; Dmitriev, A.; Kern, K. Appl. Phys. A: Mater. Sci. Process. 2003, 76, 645–652. (36) Lin, N.; Dmitriev, A.; Weckesser, J.; Barth, J. V.; Kern, K. Angew. Chem., Int. Ed. 2002, 41, 4779–4783. (37) Tseng, T.-C.; Urban, C.; Wang, Y.; Otero, R.; Tait, S. L.; Alcami, M.; Ecija, D.; Trelka, M.; Gallego, J. M.; Lin, N.; Konuma, M.; Starke, U.; Nefedov, A.; Woll, C.; Herranz, M. A.; Martin, F.; Martin, N.; Kern, K.; Miranda, R. Nat. Chem. 2010, 2, 374.
DOI: 10.1021/la101582k
16327
Article
Ge et al.
Table 1. Comparison of HREELS Features of First Layer and Bilayer TPA on Cu(100) with Benzene on Pd(100) and Benzoic Acid on Cu(110), TPA IR, and Other Data from the Literature modea
first layer TPA on Cu(100) (cm-1)
ν(Cu-C),ν(Pd-C) ν(Cu-O) νas(CdCOOH) þ δ(CdC = O, OdCdO) δ(O-C-O) γ(CH) F(OH) þ F(CH) γ(CH) F(CH) o.o.p. δ(CH), ν(CO) þ δ(C-O-H) β(C-H) ν(CC)
261 336
bilayer TPA on Cu(100) (cm-1)
C6H6 on Pd(100)23(cm-1)
benzoic acid on Cu(110)22 cm-1)
TPA IR24 (cm-1)
280 360
other 290 ν(Cu-OCH3)25,26, 350 ν(Cu-O)26
490
531
525
752
665 758
720
700
879
870
870
1104 1168 (weak)
1093 1162
1115
784 840 1170
944, 832 1113 1168
1320
1342 1371 1420 νS(O-C-O) ν(CdC) þ δ(CdC-C) 1424 δ(CC) 1425 1542 1499 1550b νas(O-C-O) β(C-C-H) þ ν(CdC) 1576, 1509 ν(C-C) 1600 1752 1760 ν(CdO)27 νas(CdO) νas(CdO) þ δ(C-O-H) 1684-1630 ν(C-H) 3032 (weak) 3055 3010 3060 3104, 3064 ν(OH) 3571 a For prior studies, modes are labeled as in the source publication. b This mode was observed for phthalic acid, but not for benzoic acid.22
Table 2. Measured Vibrational Modes for First Layer TPA and Deuterated TPA on Cu(100) mode
TPA on Cu(100) (cm-1)
D-TPA on Cu(100) (cm-1)
ratio
ν(Cu-C) 261 256 1.02 ν(Cu-O) 336 350 0.96 (1.36)a δ(O-C-O) or γ(CH) 752 (553 est.)a γ(CH)/γ(CD) 879 646 1.36 δ(CH)/δ(CD) 1104 802 1.38 ν(CO) 1104 1070 1.03 1342 1357 0.99 νs(O-C-O)/ν(CC) ν(CdC) þ δ(CdC-C) 1436 1542 1543 1.00 νas(O-C-O) ν(C-H)/ν(C-D) 3032 (weak) 2263 1.34 a A peak at 553 cm-1 is not clearly resolved in the D-TPA spectrum, but it may be part of the low energy shoulder on the observed 646 cm-1 peak. Its position is estimated from the ratio for other observed peaks.
Figure 4. HREELS spectra of first layer TPA (C6H4(COOH)2, top) and D-TPA (C6D4(COOH)2, bottom) on the Cu(100) surface. The thin line traces for each spectrum have been scaled by 10.
be that the frequencies of the CH stretching modes, νCH, and rocking modes, FCH, will be higher by a factor of about 1.3-1.4 for the nondeuterated molecule, based on standard harmonic oscillator models using reduced mass arguments. Modes not involving the H/D, such as the CC stretching mode, will be virtually unchanged. Comparison of the deuterated molecule can thus aid in the assignment of some of the spectral features. We see similar deuteration shifts in the current work. In this work, the isotope substitution leads to a significant shift of four modes (approximately 1.36, see Table 2). The C-D stretch mode at 2263 cm-1 not only shifts to lower wave numbers but is substantially stronger than the corresponding C-H stretch mode at 3032 cm-1. The mode at 1104 cm-1 is interpreted to be an overlap of a CH bending mode that shifts to 802 cm-1 in the deuterated TPA and a CO stretching mode that remains unchanged (1070 cm-1) for D-TPA. The CH bending mode at 879 cm-1 corresponds to the CD bend here at 646 cm-1, which correlates very well with benzene on 16328 DOI: 10.1021/la101582k
Pd(100)23 (870 and 675 cm-1, respectively), where the aromatic ring is parallel to the surface as in the present case. The other CH bend for the benzene case is at 720 and 520 cm-1, respectively. We do not clearly resolve a corresponding peak in the deuterated spectrum to relate to our nondeuterated peak at 752 cm-1. However, taking the same peak ratio (1.36), we would expect to find a feature for D-TPA at 553 cm-1, where there seems to be something of a weak shoulder on the 646 cm-1 peak. Other modes exist in the D-TPA spectra at approximately the same frequency as for TPA. The two vibrational modes at 256 cm-1 and 350 cm-1 are observed. They can be assigned to Cu-C and Cu-O stretch respectively. These two modes appear to have a much lower intensity for D-TPA than for H-TPA, but at least part of that difference is due to the background of the elastic scattering peak. The presence of strong losses located at 1357 and 1543 cm-1 is due to OCO stretching modes and at 1436 cm-1 due to ring deformation modes, all of which are virtually unchanged between TPA and D-TPA (Figure 4). 3.5. Off-Specular HREELS Analysis. Representative spectra for specular, 5° off-specular, and 10° off-specular scattering Langmuir 2010, 26(21), 16325–16329
Ge et al.
Article
In the case of the bilayer coverage, we find that most of the major features decrease strongly in intensity as the detector is moved away from the specular beam direction. The exceptions to this (primarily impact scattering modes) include the feature at 3571 cm-1, the OH stretch.
Figure 5. HREELS spectra with detector at specular, 5° off-specular, and 10° off-specular positions for first layer and bilayer TPA on Cu(100) surfaces.
are shown in Figure 5 for first layer and bilayer TPA adsorbed on Cu(100). Measurement of HREELS off-specular angular dependence allows identification of dipole scattering or impact scattering modes, as the latter will be very insensitive to the detector angle. Moving from specular to 5° off-specular scattering decreases the intensities of the losses at 261, 336, and 752 cm-1 by significant factors (1.4-1.8). These modes can be assigned to dipole dominant scattering. The intensities of the losses at 1104, 1342, 1542, and 3032 cm-1 only decrease by small amounts (
