Interaction of a Self-Assembled Ionic Liquid Layer ... - ACS Publications

Dec 29, 2015 - formation of an ordered (2D) crystalline phase, which coexists with a mobile (2D) liquid. DFT-D calculations reveal that adsorbed [BMP]...
0 downloads 3 Views 5MB Size
Letter pubs.acs.org/JPCL

Interaction of a Self-Assembled Ionic Liquid Layer with Graphite(0001): A Combined Experimental and Theoretical Study Florian Buchner,† Katrin Forster-Tonigold,† Maral Bozorgchenani,‡ Axel Gross,†,§ and R. Jürgen Behm*,†,‡ †

Helmholtz-Institute-Ulm (HIU), Electrochemical Energy Storage, Helmholtzstraße 11, D-89081 Ulm, Germany Ulm University, Institute of Surface Chemistry and Catalysis, Albert-Einstein-Allee 47, D-89081 Ulm, Germany § Ulm University, Institute of Theoretical Chemistry, Albert-Einstein-Allee 11, D-89081 Ulm, Germany ‡

S Supporting Information *

ABSTRACT: The interaction between (sub)monolayers of the ionic liquid 1-butyl-1methylpyrrolidinium bis(trifluoromethylsulfonyl)amide [BMP]+[TFSA]− and graphite(0001), which serves as a model for the anode|electolyte interface in Li-ion batteries, was investigated under ultrahigh vacuum conditions in a combined experimental and theoretical approach. High-resolution scanning tunneling microscopy (STM), X-ray photoelectron spectroscopy (XPS), and dispersion-corrected density functional theory (DFT-D) calculations were employed. After vapor deposition at 300 K, XPS indicates molecular adsorbates with a 1:1 ratio of cations/anions. Cool down to ∼100 K leads to the formation of an ordered (2D) crystalline phase, which coexists with a mobile (2D) liquid. DFT-D calculations reveal that adsorbed [BMP]+ and [TFSA]− species are arranged alternately in a row-like adsorption structure (cation−anion−cation−anion) and that adsorption is dominated by dispersion interactions between adlayer and substrate, on the one hand, and electrostatic interactions between the ions in a row, on the other hand. Simulated STM images of that structure closely resemble the experimental molecular resolved STM images and show that the resolved features mostly stem from the cations.

I

electrolyte, the interactions between different electrolyte components, like IL+Li on Cu(111),30 we now moved, in a next step, to a more realistic electrode material. Here we report on the interaction of the battery-relevant IL 1-butyl-1methylpyrrolidinium bis(trifluoromethylsulfonyl)amide ([BMP]+[TFSA]−) with graphite, where the latter is a frequently used but still structurally well-defined anode material, as an essential first step for the fundamental understanding of processes at the graphite electrode|electrolyte interface. We present first results of a combined experimental and theoretical study on the adsorption and structure formation of (sub)monolayers of [BMP]+[TFSA]− with a graphite(0001) model electrode under UHV conditions, employing scanning tunneling microscopy (STM), X-ray photoelectron spectroscopy (XPS), and dispersion-corrected density functional theory (DFT-D) calculations. In general, ex situ microscopy investigations on the interaction of IL (sub)monolayers on graphite(0001) performed under UHV conditions are hardly available. Carstens et al. studied the adsorption of 1-methyl-3-octylimidazolium bis(trifluoromethylsulfonyl)amide ([OMIM]+[TFSA]−) on graphite by STM under UHV conditions at 100 K; however,

onic liquids (ILs), which are generally defined as molten organic salts with a melting point below 100 °C, have found increasing interest in science and technology1−6 and specifically as solvents in battery electrolyte for improved performance and safety of next-generation Li-ion and Li-air batteries.1,2,7 Their high ionic conductivity, electrochemical stability, and low flammability make them ideal candidates for these applications. Despite numerous studies on the interaction between ILs and solid electrode surfaces, which have been conducted either in situ (in an electrochemical environment),8−13 or ex situ (in vacuo),14−19 little is known about one of the most crucial aspects, the reactive interaction at the solid|electrolyte interface and the consequent formation and characteristics of a passivating thin film known as solid|electrolyte interphase (SEI), which is built up during the first potential cycles of battery operation and which plays a decisive role for its function and performance20−22 on the molecular scale. To learn more about the processes at the electrode|electrolyte interface on the molecular level, including the interactions between the molecules and with the surface, their intramolecular conformation, and their chemical state, we have started to investigate the interaction of individual components of electrolytes such as ILs23−27 and carbonates28 with well-defined flat model electrode surfaces in model studies performed under ultrahigh vacuum (UHV) conditions. Following previous studies on the interaction of (sub)monolayers of ILs with noble and transition-metal surfaces23−27,29 and to mimic the © XXXX American Chemical Society

Received: November 3, 2015 Accepted: December 29, 2015

226

DOI: 10.1021/acs.jpclett.5b02449 J. Phys. Chem. Lett. 2016, 7, 226−233

Letter

The Journal of Physical Chemistry Letters

1s spectral range the Cgraphite substrate peak is the dominant signal (284.4 eV) (Figure 1a). The Calkyl peak (284.7 eV) (carbon atoms of the cation, which exhibit only carbon neighbors) is hidden below the substrate peak, Chetereo species (286.7 eV) (carbon atoms of the pyrrolidium ring with a nitrogen neighbor) appear as a shoulder at the high BE side of the Cgraphite peak, and Canion atoms (carbon atoms of the anion with fluorine neighbors) appear at 292.8 eV. A satisfactory fit of the measured spectra was achieved by setting the peak area ratio to 5:4:2, which is the nominal ratio of Calkyl/Chetereo/Canion in [BMP]+[TFSA]−.22,29 In addition, a shake-up satellite of the main peak Cgraphite appears at ∼290.5 eV. The N 1s signature of as-deposited [BMP] +[TFSA]− (sub)monolayers displays two contributions, which are assigned to the one nitrogen atom of the cation in the pyrrolidinium ring (Ncation) at 402.7 eV and to that of the anion (Nanion) at 399.5 eV, respectively (Figure 1b). The Ncation/Nanion intensity ratio of one, in conjunction with the peaks in the C 1s spectral range, verifies that [BMP]+[TFSA]− is molecularly adsorbed on graphite(0001) at room temperature with a balanced cation/ anion ratio. In addition, we found that the XP signals completely disappear upon heating to 450−500 K (see Supporting Information (SI I)); that is, the molecules desorb without decomposition, which is similar to the behavior of [BMP]+[TFSA]− on Ag(111) and Au(111). In contrast, for the adsorption on Cu(111) XPS measurements revealed a decomposition of the IL already slightly above room temperature, leading to decomposition products such as CFx,ad, SOx,ad, and Sad.26 Next, the [BMP]+[TFSA]− (sub)monolayer covered sample was cooled to ∼100 K. The STM image in Figure 2a displays the boundary between a highly ordered (2D) crystalline phase in which the molecular building blocks are arranged in row-like structures and an adjacent (2D) liquid/gas phase. The latter is characterized by noisy features within the phase, which are induced by rapidly diffusing species that are too mobile to be resolved by the scanning STM tip. Such a scenario is typical for a (2D) adsorption−desorption equilibrium between the two phases even at 100 K, as it was reported for [BMP]+[TFSA]− on Ag(111).23 To minimize tip-induced effects we applied very mild tunneling conditions (Ut ≈ − 0.6 to − 1.4 V, It ≈ 10−20 pA, Rt ≈ 30−140 GΩ). Nevertheless, we found massive changes of the phase boundaries from image to image; often the entire (2D) crystalline phase disappeared. Tip-induced effects cannot be ruled out, although we did not find any evidence of them. The contour of the phase boundary in Figure 2a is mostly straight parallel to the rows and strongly structured perpendicular to them, with different rows reaching different lengths. This is indicative of stronger intrarow than inter-row interactions between adsorbed species. This is very different from the frizzy features, which we related to surface mobility at the phase boundary on Ag(111),23 pointing to rather different interactions in the adlayer on the two surfaces. In general, the observations at 100 K indicate weak attractive intermolecular interactions and a very low barrier for surface diffusion. The high-resolution STM images in Figure 2b,c resolve the (2D) crystalline phase. In Figure 2b the long-range ordered (2D) crystalline phase (30 × 30) nm2 is characterized by molecular rows with an inter-row distance of 1.55 nm on average, which fluctuate, however, considerably. (The unusual high fluctuation of the inter-row distances is clearly visible in the STM image in the Supporting Information (SI II).) Upon closer inspection of the images (Figure 2b,c), the adlayer seems

they were not able to resolve individual molecules under these conditions.31 On the contrary, in an in situ STM measurement on graphite(0001) in [OMIM]+[TFSA]− they resolved slightly different ordered arrangements at cathodic potentials of ∼ −2 V and correlated these to adsorbed cations, presumably with the octyl chains aligned along the graphite atomic lattice, but they conceded that the structure could maybe also be a bilayer.31 Very recently, Elbourne et al. studied 1-ethyl-3methylimidazolium bis(trifluoromethylsulfonyl)amide ([EMIM]+[TFSA]−) on graphite(0001) by in situ atomic force microscopy and resolved a height-modulated row-like structure at open-circuit potential.32 On the basis of the changing dimension of the respective rows upon applying different potentials (±0.3 V) they estimated that the adlayer rows could be composed of alternating sequences of anion− cation−cation−anion; however, it was not possible in these in situ experiments to unambiguously identify the individual adsorbed ions/ion pairs. This was possible in previous UHV studies, where we could unambiguously identify anions and cations at the [BMP]+[TFSA]−|Ag and Au interface by highresolution STM imaging.23−25,27 By combining UHV STM and dispersion-corrected density functional theory (DFT-D) calculations we could resolve that the alkyl chain of the cation points toward the vacuum and that the anion is adsorbed in a cis conformation, with the oxygen atoms on the surface and the trifluoromethyl groups toward the vacuum next to the cation.23 A similar identification of the molecules and details of the nature of the interaction at the IL|electrode interface would be the ultimate goal also on more battery-relevant graphite electrodes. The chemical state of a (sub)monolayer of [BMP]+[TFSA]− molecules vapor-deposited on graphite(0001) at room temperature was probed by XPS (Figure 1a,b; spectra at the bottom of each panel). For a better comparison, we also present the more intense multilayer spectra, which were acquired under the same conditions, as reference (Figure 1a,b; spectra at the top of each panel). All spectra were acquired in the surface-sensitive grazing emission mode (80° to the surface normal, information depth ∼1 to 2 nm). For [BMP]+[TFSA]− (sub)monolayers in the C

Figure 1. (a,b) C 1s and N 1s XP spectra of (sub)monolayers of [BMP]+[TFSA]− on a graphite(0001) surface recorded at r.t. (bottom of each panel); multilayer reference spectra are shown in the top part of each panel. A molecular representation of [BMP]+[TFSA]− is inserted above the panels. Atoms are color-coded C (black), N (blue), O (red), S (gold), and F (green). 227

DOI: 10.1021/acs.jpclett.5b02449 J. Phys. Chem. Lett. 2016, 7, 226−233

Letter

The Journal of Physical Chemistry Letters

(marked with yellow ellipses in Figure 2b). We assume that the small height difference of the molecular features is mainly caused by electronic effects, most likely arising from different adsorption sites (different interactions with the substrate). In contrast with the results on Ag(111) and Au(111) we did not observe pronounced circular protrusions, which were the dominant features on those surfaces and which, in combination with findings from the calculations, were assigned to the alkyl chains of the cations pointing toward the vacuum. This indicates already a different adsorption geometry and different substrate−adsorbate interactions of the cation on graphite compared with the adsorption on metal surfaces. We recently demonstrated that in the most stable configuration of an isolated [BMP]+ cation the pyrrolidinium ring adopts an envelope conformation in which the butyl group occupies either the axial position (perpendicular to the ring) or the equatorial position (parallel to the ring).23 The equatorial position was found to be energetically slightly more favorable (by 18 meV) for the isolated cation. Hence, we assume that the cations adopt a conformation with the alkyl chain parallel to the ring on graphite(0001). The orientation of the alkyl chain with respect to the pyrrolidinium ring is expected to strongly affect the molecular pattern on the surface. Indeed, the relatively large distance of 1.55 ± 0.15 nm between the molecular rows is sufficient to allow the cation to be adsorbed flat on the graphite surface, with the alkyl chain in equatorial position. The experimental STM images point to a commensurate, rectangular ⎡⎣ 62 40 ⎤⎦ unit cell of the adlayer with the dimensions 8.58 Å × 14.86 Å, with distances of dinter‑row = 1.49 nm and dintrarow = 0.86 nm between adjacent rows and between ion pairs within a row, respectively. Dispersion-corrected DFT calculations, which were performed using this unit cell, result in an adsorption structure in which [BMP]+ and [TFSA]− are alternately aligned in a row (Figure 3a,b). [TFSA]− adopts a cis orientation with respect to the S−N−S plane and the oxygen atoms point toward the surface. The nitrogen atom of [TFSA]− is ∼5.59 Å above the graphite surface, whereas [BMP]+ is much closer to the surface (distance N-graphite: 4.10 Å). Different configurations or conformations of the ions did not yield more stable adsorption structures: For example, an adsorption structure similar to the one shown in Figure 3a,b but with the trans conformation of [TFSA]− is ∼40 meV less stable; an adsorption structure in which the ions are 4-fold coordinated by counterions is ∼440 meV less stable. Furthermore, it is energetically most favorable when the alkyl chain of [BMP]+ is aligned along the zigzag axes of graphite(0001) (see Figure 3a), and the molecular rows would concomitantly follow the armchair axes; however, other orientations of the alkyl chain are only slightly less favorable by only 15 meV, which might explain the occurrence of some molecular rows where the ion pairs adopt deviating azimuthal angles (see Figure 2b). A specific orientation of the domains is supported by STM images, which reveal different azimuthal orientation by multiples of 120° (Supporting Information (SI III)), reflecting the symmetry of graphite(0001). Hence, these domains are rotationally aligned with respect to the graphite atomic lattice. The relative orientation of [BMP]+ and [TFSA]− in its most stable adsorption geometry can be rationalized when looking at the electrostatic potential mapped onto an isosurface of the charge density of the isolated ions (Figure 4). Such an analysis should provide an indication of the location of reactivity sites of a molecule with respect to the attack or attachment of charged

Figure 2. (a−c) STM images of (sub)monolayers of [BMP]+[TFSA]− on graphite(0001) after vapor deposition on the substrate held at r.t. and subsequent cool down to ∼100 K ((a) U = −1.40 V, I = 10 pA; (b) U = −1.00 V, I = 20 pA; (c) U = −0.78 V, I = 20 pA). White and yellow ellipses indicate individual ion pairs in different adsorption geometries. The STM images in panels (a) and (b,c), respectively, were acquired in two separate experiments after equal preparations, resulting in compareable but slightly different coverages.

to consist of a sequence of alternating rows with slightly different apparent heights (Δh ≈ 0.8 Å). Each of the rows is made up of elliptical protrusions with largely the same azimuthal orientation, exhibiting intrarow distances of 0.80 ± 0.05 nm. This is illustrated by superimposed filled ellipses (white), representative of the elevated rows and by unfilled ellipses, representative of rows with decreased apparent height. We note that the images also contain some rows that are builtup by elliptical protrusions with a different azimuthal angle 228

DOI: 10.1021/acs.jpclett.5b02449 J. Phys. Chem. Lett. 2016, 7, 226−233

Letter

The Journal of Physical Chemistry Letters

Figure 4. Electrostatic potential (in hartree/e) mapped onto an isosurface (isosurface electron density = 4 × 10−4 e bohr−3) of the total charge density of [TFSA]− (upper figures) and [BMP]+ (lower figures). The blue regions reveal the most positive electrostatic potential a positive test charge experiences; that is, they show regions where a negative charge would attack preferentially. In contrast, the red regions reveal the most negative electrostatic potential a positive test charge experiences; that is, they show regions where a positive charge would attack preferentially.

Explaining the rather large inter-row distances is more delicate. Varying the distance between the rows shows that there is a rather flat potential energy curve without a true minimum in the range of the experimental observed inter-row distance (see Supporting Information (SI IV)). In that case we calculated commensurate structures with different inter-row distances, where these distances were stepwise varied by always one lattice constant of the graphite. Here we tacitly assumed preferential adsorption sites of the ion pairs on the graphite lattice, which seems to agree with experimental findings, while the calculations do not provide clear evidence of it. The absence of sizable interactions between the rows of ion pairs is in good agreement with the rather large fluctuations in the inter-row distances observed experimentally. The adsorption energy of [BMP]+[TFSA]− in its most stable adsorption structure on graphite(0001) is −1.49 eV. Adsorption in this structure is only possible due to dispersion interactions; the interaction energy based on pure RPBE values is repulsive by ∼0.12 eV. Thus, dispersive forces play a decisive role in this system. A detailed analysis of the interactions in the systems shows that about half of the total adsorption energy is due to interactions between the ionic liquid and graphite (Eint(graphite‑IL adlayer) = −0.77 eV, with a contribution of −1.32 eV from dispersion interactions and 0.55 eV interactions calculated by pure RPBE). The other half of the adsorption energy is due to the interaction between the ion pairs within the structure (Eint(IL layer) = −0.72 eV, 40% dispersion and 60% electrostatic interactions). Moreover, the interaction between ionic liquid pairs within the rows of the structure is about −0.71 eV, whereas it is only about −5 meV between adjacent rows of ion pairs. Considering the very weak interactions between neighbored rows of ion pairs, the interactions between IL pairs are mainly interactions between neighbored ion pairs along the rows, assuming that these interactions are of comparable order of magnitude in the free IL layer and in the adsorbed layer. On the basis of the much stronger interactions between the ion pairs in a row than between adjacent rows one would expect shorter strings of these rows to be present in the (2D) liquid/ gas phase, which are not observed in the STM images. Considering the little corrugation in the adsorption potential,

Figure 3. (a,b) Top and side views, respectively, of the most stable adsorption geometry of [BMP]+[TFSA]− on graphite(0001). (c,d) Simulated STM images of the structure shown in panels a and b. (U = −0.78 eV, isosurface electron density = 5 × 10−7 e Å−3). (e,f) Experimental STM image for comparison; the lattice constants are a = 1.55 ± 0.15 nm, b = 0.80 ± 0.05 nm, and α ≈ 90° ((e) U = −1.00 V, I = 20 pA; (f) U = −0.78 V, I = 20 pA).

particles, that is, whenever electrostatic interactions are decisive. Regions of the most positive electrostatic potential show where a negative charge prefers to attack; regions of the most negative electrostatic potential show where a positive charge prefers to attack. The regions of [TFSA]− where a positive (partial) charge attaches preferentially (marked in red in Figure 4) are within the S−N−S plane both in front of and behind the N atom. The region of [BMP]+ where a negative (partial) charge attaches preferentially (marked in blue in Figure 4) is located along the line defined by the N atom and the center of the triangle formed by the three C atoms surrounding the N atoms. An arrangement of [BMP]+ and [TFSA]− in the way that the regions of preferred attachment of the counterions of [BMP]+ and [TFSA]− contact each other leads to a structure that is very similar to the most stable adsorption geometry of [BMP]+[TFSA]− on graphite(0001). Indeed, the optimized cation−anion distance (calculated within two ion pairs arranged in a row in the gas phase) is comparable to their distance on the surface. 229

DOI: 10.1021/acs.jpclett.5b02449 J. Phys. Chem. Lett. 2016, 7, 226−233

Letter

The Journal of Physical Chemistry Letters

urgently required as a basis for the systematic improvement of future batteries.

however, they are expected to be rather mobile as well and in that case would not be resolved in STM images. Finally, the most stable adsorption structure is used to simulate STM images (Figure 3c,d). Obviously, the simulated STM images agree fairly well with the experimental STM images (Figures 3e,f). The protrusions observed in the STM image stem mostly from the cation with only slight contributions from the anion. In detail, the brightest spots are located at the positions of atoms of [BMP]+ that are directly above C atoms of the uppermost graphene layer. Although the CF3 groups of the anion are most elevated atoms of the adsorption structure (see Figure 3b), they are hardly visible in the STM image. This is due to the fact that the anion hardly contributes to the local density of states at the Fermi level. In summary, aiming toward a more detailed understanding of the molecular processes at model electrode surfaces during the formation of the electrode|electrolyte interphase, we have investigated the interaction of (sub)monolayers of the ionic liquid [BMP]+[TFSA]− with a graphite(0001) model electrode under UHV conditions in a combined XPS, STM, and theoretical study. On the basis of the results of that study we arrive at the following conclusions: (1) As evident from the XPS measurements, [BMP]+[TFSA]− ions/ion pairs adsorb as molecularly intact anions and cations with a balanced ratio at 300 K. Upon heating to 450−500 K the IL desorbs without decomposition. (2) Weakly attractive adsorbate−adsorbate interactions and a low surface diffusion barrier are concluded from the coexistence of an ordered (2D) crystalline phase, which is determined by molecular rows and a highly mobile (2D) liquid phase in the [BMP]+[TFSA]− adlayer on the graphite(0001) surface at ∼100 K at (sub)monolayer coverages, which were observed in STM images recorded after vapor deposition of (sub)monolayers at room temperature and subsequent cool down to ∼100 K. (3) According to periodic DFT calculations including dispersion corrections, the stability of the adsorbed (2D) crystalline phase is caused by an equal amount of adsorbate− substrate and adsorbate−adsorbate interaction, that is, to the same extent by the interaction between graphite and the ionic liquid and the one between the adsorbed ions. Adsorbate− substrate interactions are almost exclusively of dispersion type. Adsorbate−adsorbate interactions are dominated by electrostatic interactions, where the interactions between the ion pairs within the molecular rows are by a factor of 150 larger than the interactions between the molecular rows. The specific charge distribution within the individual ions on graphite(0001) prompts adsorbed [BMP]+ and [TFSA]− to be arranged alternatingly in the rows (cation−anion−cation−anion), where [TFSA]− is located further away from the surface. There is hardly any electronic (nondispersive) interaction between [TFSA]− and the graphite(0001) surface. (4) On the basis of the calculations it is mainly the cation that leads to the observed elliptical protrusions in STM images. Different from adsorption on noble-metal surfaces, the alkyl chains of [BMP]+ are preferentially aligned parallel to the graphite surface, along the zigzag axes of graphite(0001). To the best of our knowledge, this is the first time that ions could be unambiguously identified at the IL|graphite interface and that the nature of the adsorbate−substrate and adsorbate− adsorbate interaction could be clarified, which is essential for a detailed understanding of the processes at the electrode| electrolyte interface. We believe that model studies like this are



METHODS Experimental Methods. The experiments were carried out in a commercial UHV system (SPECS) with a base pressure of 2 × 10−10 mbar. It consists of two chambers, one containing an Aarhus-type STM/AFM system (SPECS Aarhus SPM150 with Colibri sensor), which is capable of measurements in a temperature range of 90 and 370 K by cooling with LN2 and resistive heating, and the other one is equipped with an X-ray source (SPECS XR50, Al−Kα and Mg−Kα) and a hemispherical analyzer (SPECS, DLSEGD-Phoibos-Has3500) for XPS measurements. The highly oriented pyrolytic graphite(0001) (HOPG) single crystal was purchased from MaTeck (ZYA, mosaic spread 0.4 ± 0.1°). Samples used for the XPS measurements exhibited a cuboid shape with a size of 10 mm × 10 mm ×