Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Surface Science 643 (2016) 108–116

Contents lists available at ScienceDirect

Surface Science

journal homepage: www.elsevier.com/locate/susc

Stereochemistry and thermal stability of tartaric acid on the intrinsically


chiral Cu{531} surface
Silvia Baldanza a, Jacopo Ardini a,b, Angelo Giglia c, Georg Held a,b,⁎
a
University of Reading, Department of Chemistry, Reading, UK
b
Diamond Light Source Ltd, Oxfordshire, UK
c
CNR-Istituto Officina Materiali, I-34149 Trieste, Italy

a r t i c l e i n f o a b s t r a c t

Available online 28 August 2015 Intrinsically chiral metal surfaces provide enantiospecific reaction environments without the need of co-
adsorbed modifiers. Amongst the intrinsically chiral copper surfaces, Cu{531} has the smallest unit cell and the
Keywords: highest density of chiral sites. XPS, NEXAFS and TPD were employed to investigate the adsorption and decompo-
Tartaric acid sition behaviour of the two chiral enantiomers of tartaric acid on this surface. The results obtained from XPS and
Copper NEXAFS show that at saturation coverage both enantiomers of tartaric acid adsorb in a μ4 configuration through
Photoelectron spectroscopy
the two carboxylic groups, which are rotated with respect to each other by 90° ± ≈ 15° within the surface plane.
Temperature-programmed desorption
NEXAFS
At intermediate coverage the R,R enantiomer adopts a similar configuration, but the S,S enantiomer is different
and shows a high degree of dissociation. Growth of multilayers is observed at high exposures when the sample
is kept at below 370 K. TPD experiments show that multilayers desorb between 390 K and 470 K and decompo-
sition of the chemisorbed layer occurs between 470 K and 600 K. The desorption spectra support a two-step
decomposition mechanism with a O_C_C_O or HO–HC_CH–OH intermediate that leads to production of
CO2 and CO. Enantiomeric differences are observed in the desorption features related to the decomposition of
the chemisorbed layer.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction systems for studying the influence of the substrate geometry on molec-
ular bonding and, hence, on enantioselectivity.
Enantioselective synthesis of organic molecules is one of the great The fcc{531} surface (see Fig. 1) is of particular interest because it
challenges in heterogeneous catalysis, and hence in surface science. It has the smallest unit cell and the highest density of kink atoms offering
requires inducing chirality at a molecular scale onto catalyst surfaces, an ideal template for enantiospecific adsorption of small chiral mole-
which are usually achiral, and maintaining a chiral environment even cules. The structures of Pt{531} [9,10,15,16] and Cu{531} [17,18] have
under reaction conditions. This can be achieved either by modifying been fully characterised experimentally by low-energy electron diffrac-
non-chiral substrates with chiral molecules, e.g., amino acids or tartaric tion (LEED) and scanning tunnelling microscopy (STM) and by density
acid, [1–7] or by using intrinsically chiral surfaces [8–12]. High-Miller- functional theory (DFT) model calculations. The clean Pt{531} surface
index {hkl} surfaces of fcc metals are intrinsically chiral if h ≠ k ≠ l ≠ 0. shows a high degree of roughness whereas Cu{531} appears less
They all have in common kink atoms, where {111}, {100}, and {110} rough. In addition, the nearest neighbour distance of Cu atoms in this
micro-facets intersect. These can be seen as chiral centers and, in analo- surface is 2.55 Å, which offers a particularly good bonding environment
gy to the Cahn–Ingold–Prelog rules, a surface is labelled {hkl}R or {hkl}S if for the oxygen atoms of carboxylate groups, which are at a distance of
the sequence {111} → {100} → {110} of micro-facets around the kink 2.3 Å [19].
atoms describes a clockwise or anti-clockwise rotation, respectively. Adsorption of the small amino acids, glycine, alanine, serine,
Several examples are known, where the adsorption of small chiral mol- homoserine, cysteine, and methionine on this surface was systematical-
ecules induces chiral faceting of otherwise achiral surfaces [13,14]. It is ly investigated in previous work [20–23,12,24]. These molecules show
therefore not implausible that such intrinsically chiral surfaces play a enantiomeric differences depending on the nature of the side chain
part in some cases of chiral modification. They are also good model bound to the chiral center and on coverage. All amino acids investigated
so far adsorb in their anionic state, forming bonds to Cu atoms through
the two O atoms of the carboxylic group and the N atom of the amino
⁎ Corresponding author at: University of Reading, Department of Chemistry,
group. This bonding pattern is also observed on other Cu surfaces,
WhiteKnights, Reading RG6 6AD, UK. such as {111}, {100} and {110} [13,14,25–31,19,32]. Both the {110}
E-mail address: g.held@reading.ac.uk (G. Held). and {311} microfacets of the Cu{531} surface are well suited for this

http://dx.doi.org/10.1016/j.susc.2015.08.021
0039-6028/© 2015 Elsevier B.V. All rights reserved.
S. Baldanza et al. / Surface Science 643 (2016) 108–116 109

diastereomer R,S-TA (also mesotartaric acid), can be synthesised.


TA acts as a chiral modifier of heterogeneous Ni catalysts for the
enantioselective hydrogenation of β-ketoesters [1,3,7,35]. Its adsorption
has been investigated on several Ni [36–40], Cu [41–48], and Pd [49]
single crystal surfaces. In terms of substrate bonding, the carboxylic
end-groups of TA react in a similar way to those of amino acids. Chem-
isorption of TA leads to the formation of monodentate or bidentate
phases after deprotonation of one or both carboxylic groups, depending
on surface coverage. In general, bitartrate adlayers are more stable,
whereby the role of H-bonding interactions is not clear. IR spectroscopy
suggests intermolecular bonds between α-hydroxy groups on
neighbouring molecules [42,36–40] whereas a DFT study by Barbosa
and Sautet finds intra-molecular H-bonds more stable [43]. A second
layer of TA also appears to stabilise the chemisorbed layer through addi-
tional hydrogen bonds [49]. On Ni{110} [36 and 37] the creation of the
bitartrate form occurs spontaneously upon initial adsorption at room
temperature, with no evidence of the formation of ordered supramolec-
ular structures. STM shows that the molecules grow preferentially along
the mirror plane of the surface. On Ni{111} [7,39] formation of the most
stable bitartrate species requires higher temperature than on Ni{110},
reflecting the different reactivities of the two surface terminations.
Growth of TA on Ni{111} occurs along the three crystallographically
Fig. 1. Ball model of the Cu{531} S surface. The unit cell vectors a1 and a2 and the main crys-
equivalent mirror planes of the surface. In contrast to Ni, on Cu{110}
tallographic directions are indicated. The first-layer kink atoms are highlighted by circles. the assembly of TA molecules breaks the mirror symmetry of the surface
thus creating a globally chiral environment [41]. Very recently, Gellman
et al. [48] published the first study of TA adsorbed on intrinsically chiral
bonding arrangement. Amino acids with additional functional groups, single crystal surfaces, Cu{17,5,1} R,S, Cu{651} R,S, and Cu{531} R,S. They
such as–OH or –SH, can form a fourth bond with the surface depending observed a high degree of enantiospecificity in the “explosive” dissocia-
on the steric arrangement of the molecule after adsorption. Thomsen tion behaviour, which was attributed to an autocatalytic decomposition
et al. [23] investigated the adsorption of sulphur-containing cysteine mechanism involving chiral adsorption sites.
and methionine by means of X-ray photoelectron spectroscopy (XPS) Here we present an experimental study investigating the adsorption
and near-edge X-ray absorption fine structure (NEXAFS) spectroscopy of R,R-TA and S,S-TA on Cu{531}. XPS and NEXAFS were used to deter-
and found that the formation of a fourth bond through the sulphur mine the chemical state and the orientation of the molecules after ad-
atom occurs only for cysteine while it is sterically unfavourable for me- sorption, and to identify enantiomeric differences. TPD experiments
thionine. Only one enantiomer was investigated for each amino acid in were performed on layers at different coverages in order to further
this work; therefore it does not allow any conclusion about the characterise the thermal stability and decomposition mechanism of
enantiospecificity of these systems. A more recent study by Eralp et al. the molecules.
[22] investigated adsorption of both enantiomers of serine by means
of XPS, NEXAFS, LEED, TPD and DFT. Serine was found to absorb in a 2. Experiment
μ4 (bonds through the carboxylate, amino and β-OH groups) or μ3
(bonds through the carboxylate and amino groups only) configuration, The experiments were carried out in two different ultrahigh vacuum
respectively, at low coverage (≈ 0.25 ML) and saturation coverage (UHV) systems. XPS and NEXAFS were performed at the BEAR beamline
(0.5 ML). (Note, the nomenclature is adapted from metal–organic com- of Sincrotrone Elettra, Italy. The UHV chamber of the endstation is
plexes, where μ n means that the molecules form n bonds with metal mounted on a goniometer and houses a 66 mm hemispherical
atoms [33].) No difference between the enatiomers was found in electron-energy analyser. Both analyser and manipulator can be
terms of type and number of bonds, but NEXAFS and DFT show very moved independently, allowing a great range of experiment geome-
strong enantiomeric differences at low coverage in terms of adsorption tries. TPD experiments were performed at the University of Reading.
sites and orientation of the molecules, caused by different steric con- For all experiments the base pressure was in the low 10−10 mbar range.
straints of the OH group. Also at a global level different superstructures Standard procedures were used for sample cleaning, including Ar+
were observed in LEED for the two enantiomers and temperature- ion sputtering followed by a final annealing step to 700 K and 1000 K
programmed desorption (TPD) showed a difference in the dissociation in UHV. Cleanliness was checked by either XPS or LEED (even small
temperature between the two enantiomers. In contrast, no such enan- traces of oxygen or carbon lead to long-range faceting, which can be ob-
tiomeric differences were found for homoserine, which differs structur- served in LEED [17]). R,R and S,S enantiomers of TA (Aldrich, purity
ally from serine only by an additional –CH2– group in the side chain N99%) were dosed from a home-built evaporator, which consists of
[34]. Although homoserine assumes a similar μ4 configuration as serine two resistively heated stainless steel crucibles each containing a glass
at low coverage the longer side-chain allows more flexibility and, hence, tube filled with TA powder. K-type thermocouples are attached to the
steric constraints influence the adsorption energy less. Thus, the adsorp- crucibles for accurate and reproducible temperature control. Both enan-
tion site enantioselectivity first increases as the side group increases in tiomers were dosed by heating the crucible to temperatures between
size from –CH3 (alanine) to –CH2OH (serine) and is lost again when it 403 K and 418 K, which led to deposition times of typically 10 min per
is further increased to –CH2CH2OH (homoserine). saturated chemisorbed layer.
Similar to amino acids, tartaric acid (TA) is another important chiral For the experiments performed at ELETTRA, a single crystal with
modifier and probe molecule for enantiospecificity in this context. TA is Cu{531}S termination was used. The sample temperature was measured
a dicarboxylic organic acid with two –OH groups in the α-positions with an IRCON pyrometer (using an emissivity value of 0.05), which
(HOOC–CHOH–CHOH–COOH) and two chiral centres. The naturally oc- was only accurate above 250 °C (523 K). After annealing the crystal
curring form of the acid is R,R-TA (also L-(+) or dextrotartaric acid). Its was allowed to cool for 2 h before TA was deposited. Given the fact
enantiomer, S,S-TA (also D-(−)-TA, or levotartaric acid) and the achiral that no multilayer growth was observed, we estimate that the
110 S. Baldanza et al. / Surface Science 643 (2016) 108–116

temperature was still well above room temperature during dosing (see the½112 direction was parallel to the polarisation vector for an azimuth-
below for details). XP spectra in the C 1s and O 1s regions were recorded al angle of ϕ = 52°. The O KLL Auger signal (kinetic energy 509 eV) was
at normal emission using photon energies of 410 eV and 670 eV, respec- recorded while the photon energy was scanned from 527 eV to 567 eV
tively, a pass energy of 20 eV and energy resolution 0.15 eV. The binding in steps of 0.15 eV. The photon flux, I0, was recorded simultaneously
energy (BE) was calibrated by measuring spectra of the Cu 3d-band, the with the Auger signal using a gold mesh inserted in the beamline. This
low-binding-energy edge of which is known to be at BE 2.0 eV [50 and was used to correct the raw data for spectral features from the beamline.
51]. The intensities of all spectra shown here are normalised with In addition, NEXAFS spectra recorded for the clean surface (for exam-
respect to the low BE background and have a linear background ples see Supporting information) were smoothed and subtracted from
subtracted. Oxygen K-edge NEXAFS spectra were recorded in Auger- those recorded for the TA layers in order to account for features origi-
yield mode with the photon beam perpendicular (i.e., polarisation nating from the substrate itself. Both operations were executed after
vector parallel) to the surface and the analyser at an angle of 40° with normalisation with respect to the pre-edge (low photon energy) back-
respect to the surface normal. Spectra were recorded for different azi- ground. In a final step, the spectra corresponding to the same adsorbate
muthal angles by rotating the sample together with the analyser around coverage were normalised with respect to the post-edge signal (high
the surface normal over a range of 150° or 180°. As indicated in Fig. 1, photon energy). The photon energy was calibrated with respect to the

Fig. 2. XP spectra in the O 1s (top) and C 1s (bottom) regions for different coverages of R,R-TA (left) and S,S-TA (right) dosed at room temperature on Cu{531}S. Coverage increases from
bottom to top; R,R-TA 16% and 35% were prepared by dosing onto the clean surface, R,R-TA 79% was prepared by adding to the 16% layer; S,S-TA 26% and 100% were prepared by dosing
onto the clean surface, and S,S-TA 40% was prepared by adding to the 26% layer. Spectra recorded at normal emission; hν = 670 eV (O 1s) and 410 eV (C 1s).
S. Baldanza et al. / Surface Science 643 (2016) 108–116 111

Table 1
Binding energies of the main peaks in the XP spectra of the two enantiomers of tartaric acid at different coverages on Cu{531} S.

Layer O 1s BE/eV Ratio C 1s BE/eV Ratio

CO(−) COO(−) COH [COO]:[CO(H)] CO(−) COO(−) COH [COO]:[CO(H)]

R,R-TA
16% sat 530.7 531.6 – 0.8:1 285.8 287.6 – 0.7:1
35% sat – 531.5 532.8 2.2:1 – 287.8 286.3 0.9:1
79% sat – 531.6 533.0 1.4:1 – 287.9 286.5 0.8:1

S,S-TA
26% sat 530.2 531.3 – 1.0:1 285.8 287.8 – 0.6:1
40% sat 530.3 531.4 – 2.6:1 285.7 287.6 – 0.9:1
100% sat – 531.4 532.7 2.1:1 – 287.8 286.2 0.9:1

π resonance associated with the carboxylic group using the value for of sample temperature in the desorption profiles. The third set was also
Alanine on Cu{111}, 532.6 eV [32]. Note that the exact energy calibration carried out on the Cu{531}R surface using layers at similar coverages of
is not relevant here as the data analysis is only based on the angular de- the two enantiomers, to investigate possible enantiospecificity in the
pendence of resonances and not on their absolute energies. Due to the temperature behaviour of the system. In all cases, relative masses 2
relatively low flux (10−10 ph/s−1) and large X-ray spot (400 × 50 μm2) (H2), 14 (CH2, double ionised CO), 16 (O), 18 (H2O), 28 (CO) and 44
of the BEAR beamline no indications of beam damage were observed (CO2) were traced as a function of temperature while the sample was
in any of the XPS or NEXAFS experiments. annealed up to 700 K at a constant rate of 1 K s −1. Here we concentrate
For the temperature-programmed desorption (TPD) experiments on the traces of masses 2, 18, 28, and 44.
performed at Reading, the temperature was measured through a ther-
mocouple attached to the sample holder. Three different sets of TPD ex- 3. Results and discussion
periments were performed using an SRS RGA 100 quadrupole mass
spectrometer housed in a differentially pumped tube. The first set of ex- 3.1. XPS
periments was carried out on a Cu{531}S surface with layers at different
coverages dosed at a sample temperature of 300 K, in order to analyse Fig. 2 shows C 1s and O 1s XP spectra recorded after dosing R,R-TA
the effect of the coverage on the chemistry of the adlayer. The second (left) and S,S-TA (right) onto Cu{531}S at an estimated temperature of
set was performed on a Cu{531}R surface dosing at a constant exposure between 380 K and 400 K. At this temperature the uptake stops when
time at 300, 350, 370 and 390 K sample temperature, to study the effect the chemisorbed layer is saturated. Further dosing did not show

Fig. 3. NEXAFS data at the O K-edge obtained from different layers of tartaric acid on Cu{531}S dosed at room temperature. (a) 79% saturation of R,R-TA. (b) 100% saturation of S,S-TA.
(c) 35% saturation of R,R-TA. (d) 40% saturation of S,S-TA. The angles indicated are those of the polarisation vector with respect to the crystal (the ½112 direction corresponds to −52°).
112 S. Baldanza et al. / Surface Science 643 (2016) 108–116

Fig. 4. Experimental data (dots) and fits (lines) of the in plane angular dependence of the π resonance intensity. (a) 79% saturation of R,R-TA. (b) 100% saturation of S,S-TA. (c) 35% sat-
uration of R,R-TA. (d) 40% saturation of S,S-TA. The azimuthal angles indicated are those of the polarisation vector with respect to the crystal (the ½112 direction corresponds to −52°).
Representative fits to the data are included in (a), (b), and (c) with fit angles α1 and α2 (see text for description) indicated in each figure. (e) Illustration of the angles α1 and α2 and ϕ
(azimuth).

additional features in either C 1s or O 1s regions, indicating a growth of compared with the C 1s signal. It must be noted, however, that this ap-
multilayers. Different coverages were achieved varying exposure times proach is still only approximate as the chemical composition and, hence,
and/or dosing onto existing layers. The exact preparation conditions are intensity modulations due to diffraction effects are different for different
stated in the caption of Fig. 2. The binding energies of the main peaks of coverages. We estimate the relative error of the total coverages stated in
all spectra shown in Fig. 2 are listed in Table 1. The coverages were cal- the figures and tables below at around 20%; for individual peaks inten-
ibrated by comparing the integrated O 1s peak areas with that of the sity variations due to photoelectron diffraction in this energy range can
100% saturated S,S enantiomer. We use the O 1s signal as this is gener- be up to 40% [52], however when peaks associated with the same spe-
ally less affected by photoelectron diffraction and/or contamination cies are compared these variations are the same and the peak intensity
is proportional to the coverage of this species.
The O 1s spectrum for low coverage R,R-TA (16% saturation) shows a
Table 2 feature centred at 531 eV which can be fitted with two peaks. These are
Optimum fit parameters (χ2 b χ2min + 20 %) for the angular dependence of the π* reso-
attributed to oxygen atoms of the deprotonated carboxylic group
nances in oxygen K-edge spectra for the two enantiomers of tartaric acid at different cov-
erages on Cu{531}. (COO(−), BE 531.6 eV) and a deprotonated hydroxyl group bound to
the surface (CO(−), 530.8 eV) in accordance with Ref. [22]. Note, we
Layer χ2min α2 − α1 A1 A2 A1 : A2
use the notation COO(−)/CO(−) to indicate that the charge on these
35% sat R,R-TA 0.1909 72°–109° 1.6–2.9 1.6–2.9 0.7–1.3 side groups is unknown as the charge can readily be transferred from
79% sat R,R-TA 0.0622 69°–111° 1.3–2.6 1.3–2.6 0.7–1.3 the metal. The area ratio of these peaks is 0.8:1 ([COO(−)]:[CO(−)])
100% sat S,S-TA 0.0154 75°–102° 1.4–2.1 1.4–2.1 0.8–1.2
whereas one would expect a ratio of 2:1 from the stoichiometry of a
S. Baldanza et al. / Surface Science 643 (2016) 108–116 113

completely deprotonated TA (4 O atoms in the carboxylic groups and 2 molecular orientations on the Cu{531}S surface we get the following
bound to the α-carbons). The explanation for this mismatch cannot be equation:
entirely due to photoelectron diffraction. More likely, a significant frac-
tion of molecules is dissociated at this low coverage with dissociation Iπ ðϕÞ ¼ A1 cos2 ðϕ−α 1 Þ þ A2 cos2 ðϕ−α 2 Þ: ð1Þ
products contributing mostly to the low BE peak. In the C 1s spectrum
for this layer two main peaks are distinguishable at BE 287.6 eV Here α1,2 describes the orientation of the O–C–O triangles in the sur-
(COO(−)) and 285.8 eV (CO(−)) at a ratio of 0.7:1 ([COO(−)]:[CO(−)]), face plane (see Fig. 4(e)), ϕ is the azimuthal angle of the polarization
which also deviates from the stoichiometry ratio of 1:1 but does so vector of the X-ray beam and A1,2 are the relative coverages of carboxylic
less than the oxygen signal. The peak at 284.5 eV originates from a dis- groups with the two orientations. The changes in the intensity of the π-
sociation product. In both the O 1s and C 1s spectra for this layer high BE resonance with φ, Iπ(ϕ), are shown in Fig. 4 for the four layers investi-
features are observed at 534.4 eV and 289.2 eV, respectively, with peak gated. Acceptable fits to Eq. (1) can be obtained for the two layers of
areas of less than 8% and 6% of the total areas. These features are either R,R-TA with 35% and 79% of saturation coverage and the high coverage
satellite peaks or originate from carboxylic groups with a high degree of layer of S,S-TA (100% saturation), whereas no clear angular dependence
protonation [53 and 54], e.g., through formation of H-bonds between is observed for the low coverage layer of S,S-TA (40% saturation). For the
neighbouring molecules. three layers that show an angular dependence compatible with Eq. (1) a
When the coverage of R,R-TA is increased to 35% of saturation cover- range of orientations (α1, α2) leads to similarly good fit results, as long
age a new peak appears in the O 1s spectrum at 532.8 eV, while the low as the difference (α1, α2) is close to 90°. It must be noted that this is
BE peak at 530.8 eV is not observed anymore. The low BE peak in the C not synonymous with orientational disorder, which would lead to no
1s spectrum for this layer is shifted from 285.8 eV in the low coverage angle dependence at all, however more complex arrangements with
spectrum to 286.3 eV. Similar shifts have been observed previously for more than two orientations cannot be excluded completely. A series of
serine on the same surface when the hydroxyl groups loose their coor- constrained fits was performed keeping α1 constant at values between
dination with the surface at higher coverage [22]. Hence, we attribute − 90° and + 90° and mapping the change of χ2 as a function of α2
these peaks to protonated hydroxyl groups (COH) which are not (see Supporting information for more details). The range of acceptable
bound to the surface anymore due to increased competition for surface fit parameters χ2 b χ2min + 20% is listed in Table 2. Examples of the fits
sites from neighbouring molecules. The C atoms bound to these groups are included in Fig. 4. For all layers investigated and most values of α1,
experience a more electronegative environment, and the corre- the optimum difference α2 − α1 is either around 90° + 15° or 90°
sponding C 1s signal is shifted in binding energy by 0.5 eV with −15°. The corresponding values obtained for A1 and A2 are compatible
respect to the surface-bonded groups. The ratios of peak areas are with a ratio of 1:1, suggesting that the carboxylic groups with the two
[COO (−)]:[COH] = 2.2:1 for O 1s and 0.9:1 for C 1s, which is more in different orientations belong to the same molecule. By considering the
line with the stoichiometric ratios. The spectra for the highest coverage geometry of the surface and the XPS results, which indicate that only
of R,R-TA, 79% of saturation coverage, show little difference compared the carboxylate groups form bonds with the surface at these coverages,
with 35% saturation coverage in terms of peak positions. The peak ratios we arrive at an adsorption model where the TA molecules adsorb in a μ4
are [COO(−)]:[COH] = 1.5:1 for O 1s and 1.2:1 for C 1s. configuration through the oxygen atoms of both carboxylic groups ro-
The right-hand side panels in Fig. 2 show the O 1s and C 1s XP spec- tated with respect to each other by 90° ± ≈ 15° within the surface
tra for different coverages of S,S-TA. The most striking difference plane, see Fig. 5.
between these and the spectra of R,R-TA is that no COH peaks (BE
532.8 eV in O 1s and 286.3 eV in C 1s) are observed up to 40% of satura- 3.3. Temperature-programmed desorption
tion coverage. When fitting the main O 1s signal with two peaks these
come out with greater width and lower binding energies than for R,R- For a general overview of the thermal stability and growth behav-
TA, 531.3 eV (COO(−)) and 530.2 eV (CO(−)). The main C 1s peaks are iour of TA on Cu{531}, temperature-programmed desorption (TPD) ex-
very similar in position, 287.7 eV (COO(−)) and 285.7 eV (CO(−)), but periments were performed for different coverages of both enantiomers.
a much stronger intensity is detected for the low BE peak at 284.3 eV as-
sociated with dissociation products (19% of the total intensity for the
26% layer and 18% for the 40% layer). This clearly indicates that the S,S
enantiomer has a higher tendency to dissociate compared with the R,R
enantiomer at low coverage.
The peaks characteristic of the dangling –OH groups are only
observed in the S,S-TA layer at saturation coverage. The peak ratios,
in particular of the O 1s signal, differ from R,R-TA. The ratios are
[COO(−)]:[COH] = 2.1:1 for O 1s and 1.1:1 for C 1s, which is almost a
perfect match of the stoichiometry of intact TA.

3.2. NEXAFS

The corrected and normalised O K-edge NEXAFS spectra for layers at


different coverages of R,R-TA and S,S-TA are shown in Fig. 3. The data
show three distinct resonance peaks, a sharp O 1s → π(COO) resonance
at 532.6 eV originating from the carboxylic group and two broad σ ∗ res-
onances associated with O 1s → σ(C–C) (538.3 eV) and O 1s → σ(C–O)
(540.2 eV) transitions [55,32]. For all layers investigated the intensity of
the π resonance does not go to zero for any angle, which means that at
least two orientations of carboxylic group coexist on the surface [21].
For a layer where all molecules have the same orientation the intensity Fig. 5. Possible μ4 adsorption geometry models for S,S and R,R-tartaric acid on Cu{531}S at
high coverages and for the intermediate coverage layer of the R,R tartaric acid. In these bi-
of the π resonance changes as cos2(ϕ − α), whereby (ϕ − α) is the angle tartrate adsorption geometries the relative orientation of the carboxylate groups is ≈ 90°.
between the polarisation vector and a vector perpendicular to the O–C–O Red and blue circles represent Cu atoms in the first and second layers, respectively.
plane of the carboxylate group [55 and 56]. Assuming two different H atoms are omitted for clarity.
114 S. Baldanza et al. / Surface Science 643 (2016) 108–116

Spectra for relative masses 2 (H2), 18 (H2O), 28 (CO), and 44 (CO2) are multilayer starts to appear. Desorption of mass 28 (CO) is also observed
shown in Fig. 6. They were obtained after dosing S,S-TA at 300 K on in this temperature range. The feature is, however, broader than in the
Cu{531} S. Two main desorption events can be distinguished, between spectra of the other masses and extends up to 470 K, which indicates
390 K and 440 K and between 470 K and about 600 K, respectively. that the production of CO is due to a surface-mediated decomposition
The low-temperature desorption feature is a sharp peak shifting reaction.
from 402 K to 416 K as the initial coverage increases. The sharp peak The high-temperature desorption feature is only observed in the
is observed in the desorption spectra of H2, H2O, and CO2 for exposure spectra for CO and CO2 except for a small peak in the H2 spectrum for
times longer than 10 min. The same masses are detected in the cracking the highest coverage. The CO spectrum shows one peak which shifts
pattern of TA during dosing, which indicates that the peak corresponds from 545 K to 570 K with increasing coverage. The CO2 desorption spec-
to the desorption of intact molecules (note, the range of the mass spec- tra for the lowest coverages, 0.25 Θsat, consists of a single peak at 545 K,
trometer did not cover the actual relative mass of the intact molecule, which is similar in shape to the corresponding CO spectrum. For higher
150). The coverage dependence and sharpness of this peak and the coverages the CO2 desorption signal splits up into two peaks at 500–
fact that it does not saturate suggest that it corresponds to desorption 530 K and 540–550 K. This mass distribution is similar to what was
of multilayers, although the desorption temperature is significantly observed inprevious work on TA [38,57,48] and corresponds to the de-
lower than the range (329 K to 340 K) reported for TA on Pd{111} composition of the molecules from the chemisorbed layer. In all the
after dosing at low temperature [49]. The coverage calibration in Fig. 6 TA adsorption systems investigated so far, decomposition occurs
is based on the area under the desorption signal of mass 44 assuming via cleavage of the C–COO bonds, leading to a detection of H2, H2O
that saturation coverage, Θsat, corresponds to the point where the and CO2 signals together with CO, which was attributed either to

Fig. 6. TPD data for different coverages of S,S-TA after dosing on Cu{531}S at 300 K. Relative masses: (a) 2 (H2), (b) 18 (H2O), (c) 28 (CO), and (d) 44 (CO2); heating rate 1 K s−1.
S. Baldanza et al. / Surface Science 643 (2016) 108–116 115

fragmentation of carbon dioxide in the mass spectrometer [38,57] or to


the decomposition of an intermediate species, the formation of which is
confirmed by RAIRS experiments on Pd{111} [49]. The fact that the CO
and CO2 signals differ in shape and temperature range proves that
both decomposition products are generated on the surface and the CO
signal is not due to cracking of CO2 in the mass spectrometer. This obser-
vation suggests a two-step dissociation process where the carboxylate
groups split off first and desorb as CO2, leaving behind either a HO–
CH_CH–OH (ethene-1,2-diol) or O_C_C_O (ethene-dione) interme-
diate, as suggested for Pd{111} [49]. This intermediate dissociates in the
second step and leads to the desorption of CO. It is unclear, however, at
what point and in which form hydrogen is leaving the surface. XPS indi-
cates that deprotonation of carboxyl and hydroxyl groups occurs
already during adsorption process for low coverages of S,S-TA on
Cu{531} S. In this case the hydrogen atoms can recombine and desorb
from the surface if the sample is at room temperature or above. Never-
theless, one would expect desorption of the remaining hydrogen
accompanying the dissociation of the TA backbone.
The desorption signal in the high-temperature desorption feature
increases even for coverages above Θsat, indicating that the chemisorbed
layer is not saturated yet when the second layer already starts growing.
For higher coverage the maximum of the CO2 desorption signal shifts
downwards in temperature with a significant shoulder at 550 K, where-
as the maximum of the CO signal always occurs at around 560 K. A pos-
sible explanation is that in a more crowded environment the molecules
assume more distorted adsorption geometries which lower the barrier
for dissociation but there is not enough room for both carboxylate
groups to split off the molecule simultaneously, which would then
cause the evolution of CO2 over a wider range of temperatures.
The proposed decomposition is compatible with the mechanism
discussed by Gellman et al. [48], who point out the necessity of empty Fig. 7. TPD data for relative mass 44 (CO2) after dosing different coverages of R,R-TA (bot-
sites for the on-surface decomposition and that these could play a role tom) and S,S-TA (top) on Cu{531} R. Adsorption temperature 300 K; heating rate 1 K s −1.
in enantiospecific differences in the decomposition behaviour. Crucially,
however, our experiments do not show the same explosion-like behav-
iour that was observed in the TPD spectra of their study. They report the temperature was 403–418 K, whereas Gellman et al. used 350–
sharp CO2 desorption peaks at 494 K and 491 K, respectively, for S,S- 370 K. A lower temperature would have led to too long deposition
TA and R,R-TA on Cu{531}, whereas the high-temperature desorption times in our experiments, but this could affect the state of molecules
signal observed in this study is much broader. As a consequence, we in the gas phase (e.g., small clusters vs single atoms) and, hence, the
also observe less distinct enantiomeric effects than in Ref. [48]. Fig. 7 order of molecules on the surface.
shows a comparison of CO2 TPD spectra for different coverages of S,S- The desorption spectra after dosing the S,S-TA on Cu{531}R at differ-
TA and R,R-TA on Cu{531} R. Enantiomeric differences are observed ent temperatures (300 K, 350 K, 370 K, and 390 K), shown in Fig. 3 of
mainly in the relative intensities of the two peaks that make up the the Supporting information, clearly show that multilayer growth is
high-temperature desorption feature. For R,R-TA/Cu{531} R (and S,S- inhibited for temperatures above 370 K. Therefore, the sample temper-
TA/Cu{531} S, see Fig. 6) the ratio of the low-temperature vs high- ature during dosing for the XPS and NEXAFS experiments must have
temperature peak is higher than for S,S-TA/Cu{531} R for comparable been above this value, as multilayer growth was not observed in these
coverages, indicating a higher barrier for the dissociation of the second experiments.
carboxylate group. This agrees with the findings from our XPS experi-
ments where we found a higher degree of dissociation for S,S-TA/
Cu{531} S (note S,S-TA/Cu{531} R should behave like R,R-TA/Cu{531} S). 4. Summary and conclusions
In addition, the desorption spectra of R,R-TA/Cu{531} R and S,S-TA/
Cu{531} S show a plateau-like tail up to 610 K hinting at an extra step Adsorption of the S,S and the R,R enantiomers of tartaric acid on
in the decomposition path or a very stable minority TA species, which Cu{531} was investigated by XPS, NEXAFS and TPD. XPS shows an enan-
is not observed for the opposite enantiomer. tiomeric difference in the degree of dissociation at intermediate cover-
We can only speculate about the reasons for the differences between ages, whereby S,S-TA/Cu{531} S shows a higher degree of dissociation.
the TPD spectra in Ref [48] and our work. Explosion-like decomposition Close to saturation of the chemisorbed layer both enantiomers form
behaviour has also been observed for TA on Cu{110} [45] and on Ni{110} bonds with the substrate predominantly through their de-protonated
[36], however not on Ni{111} [38] or Pd{111} [49], suggesting that the carboxylate groups with the hydroxyl groups not involved in the surface
geometry of the surface plays an important role. Different levels of sur- bond. The angular dependence of π-resonance in the NEXAFS spectra,
face contamination (leading to facetting) or surface roughness could, which originates from the carboxylate groups, indicates that the two
therefore, be responsible for the different results. The influence of carboxylate groups are at an angle of 90° ± ≈ 15° with respect to each
adsorption temperature between 300 K and 390 K is also significant, other, but the absolute orientation could not be resolved unambiguous-
as shown in the Supporting information. The authors of Ref. [48] used ly by NEXAFS, hence potential structural enantiospecificity remains un-
405 K, but in our experiments a higher temperature did not lead to resolved. By combining the above results, we propose that tartaric acid
sharper desorption features; in the contrary, the high-temperature de- at saturation coverage adsorbs on Cu{531} in a μ4 configuration in its bi-
sorption feature even broadens towards lower temperature. Another tartrate form, with the two carboxylic groups forming bonds to Cu
difference is that of the evaporator temperature. In our experiments atoms in the first and second layers rotated by ≈ 90∘ with respect to
116 S. Baldanza et al. / Surface Science 643 (2016) 108–116

each other. This adsorption geometry is similar to that of tartaric acid on [22] T. Eralp, A. Ievin, A. Shavorskiy, S.J. Jenkins, G. Held, J. Am. Chem. Soc. 134 (2012)
9615.
Cu{110} at low coverage [41]. [23] L. Thomsen, M. Wharmby, D.P. Riley, G. Held, M.J. Gladys, Surf. Sci. 603 (2009) 1253.
The chemisorbed layer decomposes between 470 K and 600 K in two [24] M.L. Clegg, L. Morales de la Garza, S. Karakatsani, D.A. King, S.M. Driver, Top. Catal. 54
steps with the production of first CO2 (cleavage of the −COO bond) and (2011) 1429.
[25] S.M. Barlow, R. Raval, Surf. Sci. Rep. 50 (2003) 201.
then CO (cleavage of the OC_CO bond of an ethene-1,2-diol or ethene- [26] V. Humblot, S.M. Barlow, R. Raval, Prog. Surf. Sci. 76 (2004) 43466.
dione intermediate). The growth of multilayers is observed at high [27] G. Jones, L.B. Jones, F. Thibault-Starzyk, E.A. Seddon, R. Raval, S.J. Jenkins, G. Held,
exposures when the sample is kept below 370 K. Multilayer desorption Surf. Sci. 600 (2006) 1924.
[28] C.J. Baddeley, G. Held, Chiral Molecules on Surfaces, Comprehensive Nanoscience
occurs between 390 K and 440 K. and Technology, Elsevier, Amsterdam, 2010 (pg. 105–133).
[29] T. Eralp, A. Shavorskiy, Z.V. Zheleva, G. Held, N. Kalashnyk, Y. Ning, T.R. Linderoth,
Acknowledgement Langmuir 26 (2010) 18841.
[30] T. Eralp, A. Shavorskiy, G. Held, Surf. Sci. 605 (2011) 468.
[31] A. Shavorskiy, F. Aksoy, M.E. Grass, Z. Liu, H. Bluhm, G. Held, J. Am. Chem. Soc. 133
The research leading to these results has received funding from the (2011) 6659.
European Community's Seventh Framework Programme (FP7/2007– [32] S. Baldanza, A. Cornish, R.E. Nicklin, Z.V. Zheleva, G. Held, Surf. Sci. 629 (2014) 114.
2013), CALIPSO under grant agreement no. 312284 and the Marie [33] S.M. Barlow, S. Louafi, D. Le Roux, J. Williams, C. Muryn, S. Haq, R. Raval, Surf. Sci. 590
(2005) 243.
Curie Training Network SMALL funded by the European Community's [34] A.L. Cornish, Investigations into Chiral Adsorption Systems Relevant to Asymmetric
Seventh Framework under grant agreement no. 238804. Heterogeneous Catalysis on Metal Surfaces, University of Reading, Reading, 2011
(PhD Thesis).
[35] C.J. Baddeley, T.E. Jones, A.G. Trant, K.E. Wilson, Top. Catal. 54 (2011) 1348.
Appendix A. Supplementary data [36] V. Humblot, S. Haq, C. Muryn, W.A. Hofer, R. Raval, J. Am. Chem. Soc. 124 (2002) 503.
[37] V. Humblot, S. Haq, C. Muryn, R. Raval, J. Catal. 228 (2004) 130.
Supplementary data to this article can be found online at http://dx. [38] T.E. Jones, C.J. Baddeley, Surf. Sci. 513 (2002) 453.
[39] T.E. Jones, C.J. Baddeley, Surf. Sci. 519 (2002) 237.
doi.org/10.1016/j.susc.2015.08.021. [40] T.E. Jones, C.J. Baddeley, J. Mol. Catal. A 216 (2004) 223.
[41] M. Ortega-Lorenzo, S. Haq, T. Bertrams, P. Marray, R. Raval, Baddeley, J. Phys. Chem.
References B 103 (1999) 10661.
[42] M. Ortega-Lorenzo, C.J. Baddeley, C. Muryn, R. Raval, Nature 404 (2000) 376.
[1] Y. Izumi, Adv. Catal. 32 (1983) 215. [43] L.A.M.M. Barbosa, P. Sautet, J. Am. Chem. Soc. 123 (2001) 6639.
[2] M.A. Keane, Langmuir 10 (1994) 4560. [44] M. Ortega-Lorenzo, V. Humblot, P. Murray, C. Baddeley, S. Haq, R. Raval, J. Catal. 205
[3] M.A. Keane, Langmuir 13 (1997) 41. (2002) 123.
[4] H.U. Blaser, Tetrahedron Asymmetry (1991) 843. [45] V. Humblot, M. Ortega-Lorenzo, S. Haq, R. Raval, J. Am. Chem. Soc. 126 (2004) 6460.
[5] A. Baiker, J. Mol. Catal. A 115 (1997) 473. [46] M. Parschau, T. Kampen, K.-H. Ernst, Chem. Phys. Lett. 407 (2005) 433.
[6] A. Baiker, J. Mol. Catal. A 163 (2000) 205. [47] D.A. Duncan, W. Unterberger, D.C. Jacson, M.K. Knight, E.A. Kröger, K.A. Hogan, C.L.A.
[7] C.J. Baddeley, Top. Catal. 25 (2003) 17. Lamont, T.J. Lerotholi, D.P. Woodruff, Surf. Sci. 606 (2012) 1435.
[8] G.A. Attard, J. Phys. Chem. B 105 (2001) 3158. [48] A.J. Gellman, Y. Huang, X. Feng, V.V. Pushkarev, B. Holsclaw, B.S. Mhatre, J. Am.
[9] A. Ahmadi, G.A. Attard, J. Feliu, A. Rodes, Langmuir 15 (1999) 2420. Chem. Soc. 135 (2013) 19208.
[10] A. Asthagiri, P.J. Feibelman, D.S. Sholl, Top. Catal. 18 (2002) 193. [49] M. Mahapatra, W.T. Tysoe, Surf. Sci. 629 (2014) 132.
[11] G. Held, M.J. Gladys, Top. Catal. 48 (2008) 128. [50] R. Domnick, G. Held, H.-P. Steinrück, Surf. Sci. 516 (2002) 95.
[12] A.J. Gellman, ACS Nano 4 (2010) 41187. [51] M.J. Gladys, O.R. Inderwildi, S. Karakatsani, V. Fiorin, G. Held, J. Phys. Chem. C 112
[13] X. Zhao, R.G. Zhao, W.S. Yang, Surf. Sci. 442 (1999) L995. (2008) 6422.
[14] X. Zhao, J. Am. Chem. Soc. 122 (2000) 12584. [52] D.P. Woodruff, Surf. Sci. Rep. 62 (2007) 13881.
[15] T.D. Power, A. Asthagiri, D.S. Sholl, Langmuir 18 (2002) 3737. [53] A. Valle, V. Humblot, C. Methivier, C. Pradier, Surf. Sci. 602 (2008) 2256.
[16] S.R. Puisto, G. Held, V. Ranea, S.J. Jenkins, E.E. Mola, D.A. King, J. Phys. Chem. B 109 [54] J. Park, J.S. Shumaker-Parry, J. Am. Chem. Soc. 136 (2014) 1907.
(2005) 22456. [55] J. Hasselström, O. Karis, M. Weinelt, N. Wassdahl, A. Nilsson, M. Nyberg, L.G.M.
[17] G. Jones, M. Gladys, J. Ottal, S.J. Jenkins, G. Held, Phys. Rev. B 79 (2009) 165420. Petterson, M.G. Samant, J. Stöhr, Surf. Sci. 407 (1998) 221.
[18] M.L. Clegg, S.M. Driver, M. Blanco-Rey, D.A. King, J. Phys. Chem. C 114 (2010) 4114. [56] J. Stöhr, NEXAFS Spectroscopy, Springer Series in Surface Sciences, Springer, Berlin,
[19] Z.V. Zheleva, T. Eralp, G. Held, J. Phys. Chem. C 116 (2012) 618. 1996.
[20] T. Eralp, Z.V. Zheleva, A. Shavorskiy, V.R. Dhanak, G. Held, Langmuir 26 (2010) [57] B. Behzadi, S. Romer, R. Fasel, K.-H. Ernst, J. Am. Chem. Soc. 126 (2004) 9176.
10918.
[21] M.J. Gladys, A.V. Stevens, N.R. Scott, G. Jones, D. Batchelor, G. Held, J. Phys. Chem. C
111 (2007) 8331.

You might also like