Formation of Environmentally Persistent Free Radicals (EPFRs) on the Phenol-Dosed α-Fe₂O₃(0001) Surface

Abstract

Environmentally persistent free radicals (EPFRs) are a class of toxic air pollutants that are found to form by the chemisorption of substituted aromatic molecules on the surface of metal oxides. In this study, we employ X-ray photoelectron spectroscopy (XPS) and ultraviolet photoelectron spectroscopy (UPS) to perform a temperature-dependent study of phenol adsorption on α-Fe₂O₃(0001) to probe the radical formation mechanism by monitoring changes in the electronic structure of both the adsorbed phenol and metal oxide substrate. Upon dosing at room temperature, new phenol-derived electronic states have been clearly observed in the UPS spectrum at saturation coverage. However, upon dosing at high temperature (>200 °C), both photoemission techniques have shown distinctive features that strongly suggest electron transfer from adsorbed phenol to Fe₂O₃ surface atoms and consequent formation of a surface radical. Consistent with the experiment, DFT calculations show that phenoxyl adsorption on the iron oxide surface at RT leads to a minor charge transfer to the adsorbed molecule. The experimental findings at high temperatures agree well with the EPFRs’ proposed formation mechanism and can guide future experimental and computational studies.

Graphical Abstract

INTRODUCTION

Environmentally persistent free radicals (EPFRs) are long-lived surface-bound radicals that are formed by the chemisorption of aromatic hydrocarbons to the surface of transition-metal oxides.¹ The lifetime of EPFRs can range from hours to days under ambient conditions.^2,3 Their long lifetime is attributed to their low reactivity with O_2, which makes EPFRs resistant to decomposition in the ambient environment.^1,4 Based on previous studies, the basic formation mechanism of EPFRs involves initially the physisorption of an organic precursor to the surface of a transition-metal oxide, followed by elimination of H₂O or HCl, depending on the type of the organic precursor, and finally thermally activated chemisorption (~>200 °C) to the metal oxide surface.^2,5,6 During chemisorption, partial/electron charge is, in general, transferred from the organic precursor to the metal oxide, thus reducing the metal cation and forming EPFRs.²

Although environmentally remarkably stable with radical lifetimes of days, EPFRs do pose adverse health impacts to humans.^7-14 They are found to be associated with toxic airborne particulate matter (PM), which mostly result from combustion systems.¹ Upon inhalation in PM, EPFRs subsequently produce reactive oxygen species (ROS), which causes oxidative stress that is responsible for respiratory and cardiovascular diseases.^1,7,8 Therefore, obtaining a more fundamental understanding of the EPFRs’ formation mechanism and their remarkable stability is key to understanding their environmental toxicity and to developing strategies for their prevention.

The formation of EPFRs from substituted and unsubstituted benzene precursors on nanoparticle metal oxides and metal oxides supported on silica used as model fly ash has been studied extensively with electron paramagnetic resonance (EPR).^3,6,15-17 It has been reported that EPFRs are thermally activated, and either a phenoxyl-type or semiquinone-type radicals are formed.³ Also, the correlation between transition-metal concentration in soil (clay) samples and EPFRs’ generation has been explored,^18-22 and it has been found that EPFRs form more readily on anthracene-contaminated Fe(III)-montmorillonite.¹⁸

Iron oxides are typically used as heterogeneous catalysts in many environmental and industrial processes.²³ It has been found that Fe(III) is the most abundant transition metal in airborne PM_2.5 since it comprises 10–70% of the bulk iron content in urban atmosphere.³ Yang et al. reported that Fe(III) in Fe₂O₃ exhibits a higher oxidation potential compared to ZnO, CuO, and NiO.²⁴ However, the high catalytic ability of Fe₂O₃ leads to a lower yield of EPFRs.¹⁵ This observation was attributed to the partial decay of the generated EPFRs due to the high reactivity of Fe(III).²²

Previous EPR studies of substituted benzene precursors (hydroquinone, catechol, 2-monochlorophenol, monochlorobenzene, 1,2-dichlorobenzene, and phenol) adsorbed on supported Fe(III)₂O₃/silica³ have indicated in each case EPFR formation upon adsorption at 230 °C. In the case of phenol, the EPR-determined lifetime (t_1/e) of the subsequently formed surface radical was determined to be 3.8 days. Moreover, the EPR spectral characteristics of all of these labgenerated EPFRs^3,25 were very similar to those observed in combustion-generated airborne PM_2.5 samples.^26,27

To better reveal atomistic details of EPFR formation, our group has previously utilized surface science techniques to elucidate the electronic and vibrational properties of monolayer EPFRs adsorbed on the TiO₂(110), Al₂O₃(0001)/NiAl(110), and $\text{ZnO}(10\overline{1}0)$ and ($000\overline{1}$) singlecrystal surfaces at room temperature as well as elevated temperature (e.g. >225 °C).^28-30 Patterson et al. showed a clear evidence of electron charge transfer from phenol HOMO to the unfilled states of TiO₂, and this result is further confirmed by the downward band bending observed in the Ti 3p core-level spectrum obtained for dosed TiO₂ at high temperature.²⁸ Band bending occurs due to the electric field induced by the electron charge transfer between the adsorbate and the surface; hence, valence band features shifts upward or downward depending on the direction of the electron charge transfer.³¹ In contrast, Thibodaux et al. revealed an upward band bending in ZnO 3d states.²⁹ DFT calculations attributed this result to the chemisorption of the phenoxyl group through its oxygen atom, which coordinates to three Zn atoms on the surface.³² Accordingly, partial electron charge transfers to Zn metal cations.²⁹

In this study, we utilize ultraviolet photoelectron spectroscopy (UPS) and X-ray photoelectron spectroscopy (XPS) on phenol adsorbed at room temperature (RT) and at 250 °C on the α-Fe₂O₃(0001) single-crystal surface to gain novel insights into the electronic structure change and to examine the electron-transfer mechanism as a function of adsorption temperature. Additionally, we have performed density functional theory (DFT) calculations that model the dissociative adsorption of phenol on the Fe-terminated α-Fe₂O₃(0001) surface. This surface science approach enables us to study the atomic-scale formation of EPFRs on the α-Fe₂O₃(0001) singlecrystal surface, as well as to develop a comprehensive atomic-scale understanding of the electronic structure of the composite organic/metal oxide system.

METHODS

For nanoparticle EPR studies, α-Fe₂O₃ metal oxide powder (30 nm, CAS 1309-37-1, 20–60 m²/g, 99.5% purity) was purchased from US Research Nanomaterials and used as received; phenol-dosed samples were prepared using the same dosing manifold described in the previous work.^25,33,34 EPR measurements were performed using a Bruker EMXnano spectrometer with an X-band operating at a microwave frequency of 9.64 GHz. The spectra were obtained at RT, and typical operating parameters were microwave power of 0.3 mW, modulation amplitude 1 G, and time constant 40.96 ms.

For single-crystal α-Fe₂O₃(0001) surface investigations, all photoemission experiments were performed in an ultrahigh vacuum (UHV) system with a base pressure below 2 × 10⁻¹⁰ Torr. The photoemission spectra were obtained at room temperature using an Omicron XM1000 monochromatic X-ray source (Al Kα_I) and SPECS microwave UV light source (utilizes He II line ~40.8 eV) in combination with a SPECS PHOIBOS 150 hemispherical analyzer. All XPS spectra were collected at 25 eV pass energy, while UPS spectra were obtained at 10 eV pass energy. The Fermi edge is determined from a gold foil in electrical contact with the sample, and all XPS spectra are referenced to the Au 4f_7/2 peak at 83.95 eV; no charge neutralizer was applied, and the sample was properly grounded. XPS spectra were analyzed using CasaXPS software. C 1s and O 1s core-level peaks were fitted with a Voigt GL(30) line shape, comprising 70% Gaussian and 30% Lorentzian.

A single-crystal sample of α-Fe₂O₃, obtained from SurfaceNet GmbH (Germany), was cut along the (0001) surface with (10 mm × 10 mm × 0.5 mm) dimensions. A stoichiometric (0001) surface was prepared in a UHV chamber by cycles of 500 eV Ne ion bombardment at 5 × 10⁻⁵ Torr for 20 min followed by 30 min annealing at 700 °C in 1 × 10⁻⁶ Torr O₂ and subsequently cooled in O₂ atmosphere. The sample temperature was monitored by a type K thermocouple in contact with the plate beneath the sample. The surface cleanliness before dosing was checked by XPS and showed no adventitious carbon feature on the surface. Also, the binding energy of O 1s of the clean surface is consistent with the previously reported value. Dosing of phenol was accomplished by introducing phenol vapor into the UHV chamber using a standard leak valve. The phenol was purified by several freeze–pump–thaw cycles. For high-temperature dosing, the sample was brought to 250 °C and dosed with phenol at 5 × 10⁻⁷ Torr.

Cluster models for surfaces were generated using structure 0017806³⁵ obtained from the American Mineralogist Crystal Structures Database.³⁶ The CrystalMaker program³⁷ was used to truncate the structure, and the GaussView program³⁸ was used to terminate clusters by replacing Fe atoms with H atoms and to add phenoxyl to complete the cluster model. The surface consisted of 14 Fe, 36 O, and 34 H atoms. Phenoxyl was added to the surface model to create a model for adsorbed phenoxyl. Ab initio molecular orbitals calculations were performed using the NWChem 6.8 program³⁹ to optimize structures, generate molecular orbital energies for use in calculation density of states, and to prepare input files for NBO and Bader charge analysis. The optimizations used density functional theory with the B3LYP functional, the LANL2DZ effective core basis set for Fe atoms, and the cc-pvdz basis set for all other atoms. The use of the B3LYP functional and the LANL2DZ basis/effective core potential is a common approach to studying transition-metal oxides.⁴⁰ The phenoxyl radical, the surface iron atoms, and the three oxygen atoms bound to the central surface iron atom were allowed to move during the geometry optimization. All other atoms were fixed in place. The NBO 6.0 program⁴¹ was used to calculate NBO charges, and the Bader 1.03 program^42-45 was used to calculate Bader charges.

Other choices for the model chemistry are possible. The calculated ionization energies of small organic molecules using B3LYP and PBE functionals with correlation consistent (cc-pvnz; n = D, T, Q) basis sets had an accuracy similar to other functionals in the prediction of ionization energies.⁴⁶ For coinage metals, a variety of functionals including B3LYP and PBE functionals along with the effective core basis set aug-cc-pvdz-pp was used to calculate structure and atomization energies.⁴⁷ For (FeO)_n clusters, the BP91 functional and the 6-311+g* basis set were used to study of clusters up to the size n = 16.⁴⁸ Recognizing that the qualitative results of the calculations should be independent of the choice of basis set and functional, single-point calculations were performed at B3LYP LANL2DZ/cc-pvdz optimized geometries for a combination of functionals and basis sets. Using the LANL2DZ/cc-pvdz basis set, single-point calculations were performed using PBE and BP91 functionals, and using B3LYP functional, single-point calculations were performed using LANL2DZ/cc-pvtz, LANL2DZ/6-311+g*, and 6-311+g* (all-electron basis set on Fe) basis sets. The Bader charges were calculated for all of the single-point calculations.

RESULTS

Consistent with previous studies,^3,25 EPR spectra (see Figure S1 in the Supporting Information) indicate the formation of a phenoxyl-type radical upon phenol adsorption on Fe₂O₃ nanoparticles at 230 °C. Correlated with previous electronic and vibrational studies on nanoparticles,²⁵ the signature of EPFR formation on phenol-dosed Fe₂O_3, which contain (0001) low-energy surfaces, at elevated temperature is conclusive.

In the case of single-crystal work under UHV, Figure 1a shows the UPS spectra of the α-Fe₂O₃(0001) surface dosed with phenol at room temperature and 250 °C. A Tougaard background is subtracted from the spectrum of the clean surface, whereas a combination of Tougaard and Shirley backgrounds are subtracted from each spectrum of the surface dosed at different conditions. In the clean spectrum, photoelectron emissions observed from 4 to 7 eV are mainly due to O 2p states hybridized with Fe 3d states, while the shoulder at 2.5 eV originates from native Fe states.⁴⁹ Upon exposure to 400 Langmuirs (L) of phenol at room temperature, the overall shape of the valence band changes and new electronic features appear in the region from 5.7 to 16 eV. Notably, a 400 L high temperature (250 °C) dose shows the emergence of a clear feature at 0.9 eV. Similar room-temperature phenolic features are observed in the region from 6 to 16 eV in high-temperature doses. The difference spectra depicted in Figure 1b for the surface dosed at different conditions highlight adsorbate features relative to the clean surface. For the dosed surface, it is well known that the intensity of photoelectron emissions from the underlying substrate is exponentially attenuated due to the formation of a saturated layer of adsorbates on the surface. Therefore, assuming an inelastic mean free path length of ~4 Å (at ~30 eV)⁵⁰ and organic overlayer height of ~1.9 Å, difference spectra were generated by scaling down the intensity of the clean spectrum by 0.65 before subtraction.

Figure 1.

(a) UPS spectra of clean α-Fe₂O₃ (black), 400 Langmuirs of phenol dosed at room temperature (green), 400 Langmuirs of phenol dosed at 250 °C (blue), and 800 Langmuirs dosed at 250 °C (red). (b) Difference spectra obtained from data in (a), plotted with the spectrum of multilayer physisorbed phenol on the sample at −173 °C (black).

Figure 1b shows the difference spectra of the dosed surfaces at different dosing conditions along with the “phenol ice” spectrum, which was obtained by cooling the sample to −173 °C and then dosing with phenol until no Fe₂O₃ features were observed in the UPS spectrum; this data agrees well with previous UPS data and calculated molecular orbitals of isolated phenol molecule.²⁸ For the room-temperature dose, the two peaks at 1.6 and 2.6 eV can be assigned to phenol HOMO and, with some speculation, HOMO-1 bands, respectively. The 1 eV energy difference between the two phenolic bands is consistent with the results reported for phenol/ZnO ($10\overline{1}0$) (HOMO and HOMO-1 were observed at 2.5 and 3.5 eV, respectively).²⁹ The peaks at 5.9 and 8.3 eV are mainly due to the electronic levels of oxygen and carbon in the phenyl ring, while the feature at 10.3 eV is attributed to σ_O-H orbital.⁵¹ The prominent features at 12.8 and 16 eV appear at 1 eV lower in binding energy than the peaks in phenol/TiO₂. In comparison with the phenol ice spectrum, these two peaks can be assigned to phenol σ-orbitals.

The high-temperature doses show an intense peak at 0.9 eV that is neither observed for the room-temperature dose nor in the phenol ice spectrum, indicating that this feature is clearly not phenolic. By referring to the UPS spectra obtained for different iron oxides,^49,52 this feature can be assigned to the emergence of reduced Fe states, which are formed due to phenol adsorption at high temperatures. This peak is followed by a shoulder at 1.4 eV for 400 L phenol, which becomes more distinct and intense at 800 L phenol. Conceivably, this feature can be assigned to the phenol HOMO band, which seems to become more populated at 800 L phenol. On the other hand, the peak tentatively assigned to the phenol HOMO-1 band in the room-temperature dose appears to be depopulated and shifts toward higher binding energy. Furthermore, two features attributed to the electronic levels of oxygen in the phenyl ring exhibit different position and intensity values in high-temperature coverages. Noticeably, the feature assigned to σ_O-H orbital is clearly diminished in the 400 L dose at high temperature. Yet, for 800 L, the intense emission at 8.3 eV makes it difficult to verify the existence of this feature. The phenol σ-orbitals are observed for two high-temperature doses, but they appear to be broader and less intense (especially for 400 L phenol) with respect to the room-temperature spectrum.

XPS is an invaluable surface-sensitive tool to study the elemental composition of the surface.^53,54 Figure 2 shows the XPS spectra of the Fe 2p region of clean and phenol-dosed surfaces at different conditions. The goal here is to probe consequent changes in the oxidation state of iron cations. For native ferric oxide, Fe³⁺ oxidation state is characterized by a satellite peak at 719.3 eV, which is clearly observed in the Fe 2p clean spectrum in Figure 2, with the binding energy of Fe 2p_3/2 at 710.9 eV.⁴⁹ Dosing at room temperature produces almost no alteration in the satellite peak intensity, indicating that the surface mostly retains Fe³⁺ cations. However, at elevated-temperature doses, the intensity of the Fe³⁺ satellite is significantly reduced for the 400 L phenol dose, while it is nearly absent for 800 L phenol with concomitant and clear increase in the area ~715 eV, which is the binding energy characteristic of the Fe²⁺ satellite.⁴⁹ For both high-temperature doses, the FWHM of the Fe 2p_3/2 photoelectron peak becomes broader, indicating the presence of both Fe²⁺ and Fe³⁺ oxidation states. This broadness is accompanied by a structure at 708.7 eV ascribed to Fe ²⁺ states (highlighted by the red arrow in Figure 2), observed in both of the high-temperature doses with a higher intensity in 800 L phenol dose.

Figure 2.

XPS spectra of Fe 2p core levels in the clean surface of α-Fe₂O₃ (black), 400 Langmuirs of phenol dosed at room temperature (green), 400 Langmuirs of phenol dosed at 250 °C (blue), and 800 Langmuirs of phenol dosed at 250 °C (red).

To further gain mechanistic insights into electronic structure changes on the adsorbed phenol on the surface, we collected the O 1s and C1s core-level spectra. Figure 3 shows O 1s spectra of the clean as well as the dosed surfaces of α-Fe₂O₃(0001). The main observation in this figure is the shift of near-surface O 1s peak toward higher binding energies in high-temperature doses, with a maximum of 0.3 eV shift for 800 L at 250 °C. Also, the full width at half-maximum (FWHM) increases at elevated-temperature doses due to an emerging structure at the higher binding energy side of the O 1s peak. Therefore, we have quantified O 1s spectra of the 400 L at room temperature and 800 L at high temperature using curve fitting to identify the features evolving due to phenol adsorption at both temperatures. Figure 4 shows O 1s XPS spectra of the phenol-dosed α-Fe₂O₃(0001) surface at 400 L at room temperature (bottom panel) and 800 L at 250 °C (top panel) The O 1s region of the room-temperature dose is deconvoluted into three components. The main component at 530.003 ± 0.03 eV is due to lattice oxygen;^53-55 the second component at 531.12 ± 0.03 eV is ascribed to surface hydroxyl (OH) species.⁵⁵ For the best fit results, a third component is added to the fitting at 532.25 ± 0.07 eV and can be assigned to water or phenolic oxygen.⁵⁵ All three components are constrained to have the same FWHM. Upon dosing 800 L phenol at 250 °C, we observe a broader O 1s peak. The envelope is again deconvoluted into three components. The lattice oxygen component is shifted 0.3 eV toward higher binding energy. An increase in intensity is observed in the surface hydroxyl species component along with a 0.32 eV shift toward higher binding energy. The third component is now observed at 533.11 ± 0.05 eV and is broader and more intense. This component cannot be assigned to water since the sample is here dosed at 250 °C Therefore, we identify it as due to phenolic oxygen.

Figure 3.

XPS spectra of O 1s core levels in the clean surface of α-Fe₂O₃ (black), 400 Langmuirs of phenol dosed at room temperature (green), 400 Langmuirs of phenol dosed at 250 °C (blue), and 800 Langmuirs of phenol dosed at 250 °C (red).

Figure 4.

XPS spectra of O 1s core levels in (a) 400 Langmuirs of phenol dosed at room temperature and (b) 800 Langmuirs of phenol dosed at 250 °C.

Figure 5 shows C 1s XPS spectra of the phenol-dosed α-Fe₂O₃(0001) surface at 400 L at room temperature (bottom panel) and 800 L at 250 °C (top panel). For the room-temperature dose, the main peak was fitted by one component at 284.03 ± 0.01 eV, which is ascribed to unsubstituted carbons on the phenyl ring (C─C), and a shoulder at 285.37 ± 0.07 eV ascribed to (C─OH).^53,54,56 A broader peak is also observed at 290.9 ± 0.1 eV and is attributed to the π–π* shakeup satellite, which is the characteristic of an aromatic ring on the surface.⁵⁷ For the high-temperature dose, the two components of the main photoelectron peak are shifted 0.4 eV toward higher binding energy. Moreover, the (C─O) component is significantly increased in intensity and overlapped with the main (C─C) component. On the other hand, the π–π* shakeup satellite is shifted 0.64 eV toward higher binding energy, and the best fit result requires the fourth peak at 287.80 ± 0.01 eV, which can be tentatively assigned to (C═O).^53-55

Figure 5.

XPS spectra of C 1s core levels in (a) 400 Langmuirs of phenol dosed at room temperature and (b) 800 Langmuirs of phenol dosed at 250 °C.

DFT calculations were performed to predict the magnitude and direction of electron charge transfer upon phenoxyl radical formation due to the dissociative adsorption of phenol on the stoichiometric α-Fe₂O₃(0001) surface. Since the Fe₂O₃ surface is an insulator, restricted DFT calculations were used. Thus, the stoichiometric surface was treated as a singlet, and the adsorbed phenoxyl was treated as a doublet. The optimized structure models for stoichiometric (Fe₁₄O₃₆H₃₄) and phenoxyl adsorbed at RT (Fe₁₄O₃₇H₃₉) on α-Fe₂O₃(0001) are displayed in Figure 6. The coordinates of the two clusters are given in the Supporting Information. The clusters modeled here may be thought of as representative of a Fe₂O₃(0001)–Fe-terminated surface. The calculations show that phenoxyl radical chemisorbs via the C─O─Fe bond with a binding energy of 2.6 eV. The O─Fe bond length was 1.74 Å, and the C─O─Fe angle was 120.1°. NBO and Bader charges were calculated for all atoms of isolated phenoxyl, for the surface, and the surface with added phenoxyl radical (see the Supporting Information). The changes in charges upon the addition of the phenoxyl radical to the surface are summarized in Table 1. Table 2 presents Bader results for the amount of charge transferred from the surface to phenoxyl for different model chemistries used in single-point calculations.

Figure 6.

Left panel: optimized Fe₂O₃(0001) surface. Iron atoms are colored brown, oxygen atoms are colored red, and hydrogen atoms are colored white. Right panel: optimized Fe₂O₃(0001) surface with phenoxyl radical attached. Carbon atoms are colored gray.

Table 1.

Changes in Charges as a Result of Binding a Phenoxyl Radical to the Fe₂O₃(0001) Model Surface

change in charge for phenoxyl bound to the 0001-Fe2O3 surface	NBO	Bader
phenoxyl	−0.16	−0.23
phenoxyl oxygen atom	−0.06	0.04
carbon atom bound to phenoxyl oxygen atom	−0.10	−0.36
central surface Fe atom	0.15	0.28
remaining surface	−0.01	−0.05

Table 2.

Changes in Bader Charges as a Result of Binding a Phenoxyl Radical to the Fe₂O₃(0001) Model Surface Using Single-Point Calculations from B3LYP LANL2DZ/cc-pvdz Calculations^a

change in charge for phenoxyl bound to the 0001- Fe2O3 surface	PBE LANL2DZ/cc-pvdz	BP91 LANL2DZ/cc- pvdz	B3LYP LANL2DZ/6-311+g*	B3LYP LANL2DZ/cc-pvtz	B3LYP 6-311+g*
phenoxyl	−0.19	−0.19	−0.23	−0.23	−0.24
surface	+0.19	+0.19	−0.23	−0.23	−0.24

^aThe basis set used is indicated in the table. The last column is an all-electron calculation using the 6-311+g* basis set on all atoms.

From Table 1, one can observe a qualitative agreement between NBO and Bader charge analyses. Furthermore, the phenoxyl moiety is reduced by roughly 0.2e, whereas the central iron atom is oxidized by roughly 0.15–0.3e. The rest of the surface has a negligible change in charge, resulting from the addition of the phenoxyl radical. There is also negligible change in the charge on the phenoxyl’s oxygen atom compared to this of the carbon atom. Thus, the reduction of the phenoxyl radical is due mainly to a reduction of the carbon atom bound to the phenoxyl’s oxygen atom. Table 2 demonstrates that there is a qualitative agreement among all model chemistries used in this work.

DISCUSSION

By comparing the three difference spectra in Figure 1b, one can see that both room-temperature and high-temperature difference doses share common features like the phenol HOMO band as well as phenol σ-orbitals to some extent. This strongly suggests that the adsorption of phenol at elevated temperatures did not degrade the phenyl ring into smaller hydrocarbon fragments. Another significant feature in Figure 1b is the clear observation of phenol HOMO and HOMO-1 bands at room-temperature as well as at high-temperature doses with ΔE = 1 eV. This finding is consistent with the DFT calculation of electronic states of a single phenol molecule, except that the binding energies of both bands are lower than the predicted values.^28,29 These two bands overlap in the UPS spectrum of phenol ice; however, they clearly split in the room-temperature spectrum with both bands having almost the same intensity. This might indicate that at room-temperature adsorption, both bands are either equally populated or that the degeneracy is lifted due to the breaking of symmetry. Notably, at high-temperature doses, we observe an intense feature on the low-binding-energy side of the HOMO band, attributed to reduced Fe states, presumably strongly hybridized with the HOMO band and resulting in the enhancement of its intensity. On the other hand, we see that the HOMO-1 band shifts toward higher binding energy with its intensity diminished at 800 L dose. Moreover, the dependence of adsorption behavior on temperature is manifested in the σ_O-H orbital, which is observed in the room-temperature spectrum indicating molecular (associative) adsorption, whereas the absence of this feature in 400 L at high temperature implies that phenol chemisorbed dissociatively. The spectral changes seen in the emission from the electronic levels of phenolic oxygen between ~ 6 and 8.5 eV for different conditions might have to do with the nature of the bonding between the phenolic oxygen and the surface and/or the orientation of the ring with respect to the surface.

The Fe 2p core-level spectra confirm the results observed in UPS. The Fe ³⁺ cation satellite at 719.3 eVs was not modified by the room-temperature dose. However, at 400 L phenol at 250 °C, the intensity of the satellite was found to decrease along with a broadening of the main photoelectron peak, indicating a partial reduction of Fe³⁺ cations. On the other hand, this satellite is completely attenuated for the 800 L phenol at 250 °C and significant changes took place over the Fe 2p region. Although the reduction in Fe³⁺ cations is still partial at this coverage, the density of Fe²⁺ states is higher, as verified by its satellite peak area at ~715 eV. That is, both oxidation states of Fe coexist together at elevated temperatures; however, 800 L phenol dosing generates more reduced Fe states than 400 L phenol. Also, the shift toward higher binding energy observed in O 1s core-level spectra further confirms the reduction of iron oxide in higher-temperature doses.^31,58

Quantification of C 1s and O 1s core-level spectra has provided a deeper insight into the formation of species on the surface due to phenol adsorption at different conditions. Observing both the hydroxyl group and the phenolic O in the O 1s spectra may suggest that phenol has chemisorbed on the dosed surface. However, the C 1s spectrum of both doses may support the same conclusion with the exception that the C─O/C─C ratio is higher than 1:5, which is the ratio indicating the phenol molecule. This finding might result from the bonding between C in the ring and a nearby O in the lattice, which may be allowed depending on the orientation of the chemisorbed phenol molecule on the surface. On the other hand, the increase in the ─OH, organic O, and π–π* transition intensities associated with the appearance of, tentatively, C═O (although uncertain, a quinone-like species due to the subsequent radical formation), and a further increase in the C─O/C─C ratio in the 800 L of phenol dosed at high temperature further verify the chemisorption of phenol on iron cations, indicating the formation of both carbon-centered and oxygen-centered phenoxyl-type radicals and the possible partial degradation of the formed phenoxyl radical to catechol. At the same time, these findings support the changes observed in the electronic levels of phenolic oxygen in the UPS spectra of elevated-temperature doses.

Finally, DFT calculations have shown that the dissociative adsorption of phenol on the Fe-terminated α–Fe₂O₃(0001) surface results in the oxidation of the central iron atom and consequently the reduction of phenoxyl by roughly 0.2e, and hence the formation of a carbon-centered phenoxyl radical chemisorbed to the surface. Seemingly, DFT calculations may support our experimental finding, in which phenol chemisorbs to the α–Fe₂O₃(0001) surface at room temperature since no shift has been observed in the O 1s spectrum of 400 L phenol dosed at room temperature compared to this of the clean surface, indicating that there is almost no charge transfer between the organic molecule and the iron oxide surface. However, it is noteworthy that the charge transfer calculated is minor, which suggests that the partial reduction of native Fe states, observed experimentally at high temperature, may not be solely attributed to the partial electron charge transfer from chemisorbed phenol to Fe³⁺ cations on the surface. Schlögl et al. have reported that 1000 L of atomic hydrogen results in a deep reduction of the α–Fe₂O₃(0001) biphase surface.⁵⁹ Thus, we hypothesize that at high-temperature doses, hydroxyl groups form on the surface due to the dissociative adsorption of phenol on the Fe₂O₃ surface. Further adsorption of phenol forms water which desorbs at high temperature and consequently reducing the metal cation.

CONCLUSIONS

This work has explored the formation mechanism of EPFRs by observing the adsorption of phenol on the α-Fe₂O₃(0001) surface as a function of dose and temperature to investigate the general assumption of EPFRs’ formation. Phenol was found to chemisorb on the surface of α-Fe₂O₃(0001) at room-temperature dosing, as evident by observing the σ_O-H orbital in the respective UPS. Moreover, phenol HOMO and HOMO-1 bands for room-temperature doses did not overlap, unlike the case of the multilayers of physisorbed phenol at −173 °C. On the other hand, the signatures of phenol chemisorption on the surface of α-Fe₂O₃(0001) were observed clearly at high-temperature doses. Fe 2p core-level spectra along with UPS spectra have clearly shown the partial reduction of native Fe states. These observations may suggest a partial electron charge transfer from the organic precursor to the α-Fe₂O₃(0001) surface since the observed reduction of native Fe cations may not be only attributed to charge transfer. Also, all features related to phenol have been observed for both high-temperature doses with different intensities, implying that phenol has mainly chemisorbed with an intact phenyl ring. However, it is realized by quantifying and comparing O 1s and C 1s spectra that both carbon- and oxygen-centered phenoxyl radicals coexist on the surface and may be associated with catechol. This may imply that dosing iron oxide with phenol at high temperatures may have activated certain surface sites, which in turn may have partially degraded chemisorbed phenol.