F-doping of nanostructured ZnO: a way to modify structural, electronic, and surface properties.

Polycrystalline ZnO is a material often used in heterogeneous catalysis. Its properties can be altered by the addition of dopants. We used gaseous fluorine (F2(g)) as direct way to incorporate fluoride in ZnO as anionic dopants. Here, the consequences of this treatment on the structural and electronic properties, as well as on the acidic/basic sites of the surface, are investigated. It is shown that the amount of F incorporation into the structure can be controlled by the synthesis parameters (t, T, p). While the surface of ZnO was altered as shown by, e.g., IR spectroscopy, XPS, and STEM/EDX measurements, the F2 treatment also influenced the electronic properties (optical band gap, conductivity) of ZnO. Furthermore, the Lewis acidity/basicity of the surface was affected which is evidenced by using, e.g., different probe molecules (CO2, NH3). In situ investigations of the fluorination process offer valuable insights on the fluorination process itself.


Introduction
ZnO is an extensively studied metal oxide with multiple applications in catalysis or within optoelectronic devices [1][2][3][4][5][6] . It is commonly employed in heterogeneous catalysis as support or co-catalyst, for example in Cu/ZnO based systems applied in methanol synthesis 7 or the reverse water-gas shift reaction 8 .
ZnO is an intrinsic n-type semiconductor, containing a high concentration of defects, including oxygen vacancies 9,10 .
Doping provides a versatile route to engineer the properties of the material. Doping of the cation lattice was shown to affect the structural, electronic, and catalytic properties of ZnO as part of catalytic systems [11][12][13] . Anionic N-doping of ZnO is applied to create a p-type semiconductor 14 . Here, we present the effects of using fluoride as an anionic dopant for ZnO to increase its n-type character.
In the literature, fluorine doping of oxides has been employed for the manipulation of electronic and optical properties, for example in case of fluorine modified TiO 2 15,16 . Fluorine doped ZnO has mainly been studied for the application as a transparent conducting oxide and synthesised by methods such as reactive magnetron sputtering 17,18 or chemical vapour deposition 19 . Previous studies have thus focused on the investigation of the F-doped ZnO materials with respect to their optical and electronic properties. 20,21 Further applications of F-doped ZnO thin films include the use as a gas sensing material. 22,23 While ZnO for the application in optoelectronics is required to form thin solid layers, in heterogeneous catalysis high surface area and thus the generation of large amounts of polycrystalline powder is required. Such polycrystalline F-doped ZnO materials have been previously studied with applications in, e.g., photocatalysis. 24 The influence of fluorine treatment on the catalytic properties of the Cu/ZnO system has been subject of a recent study 25 , however, the role of fluorine remains elusive.
In this contribution, we show the impact of direct F 2(g) treatment on ZnO with respect to structural and electronic properties. To control the amount of fluoride in ZnO, different F 2(g) partial pressures were applied with the aim of providing a synthetic pathway to a F-modified nanostructured ZnO. We employ various techniques (XRD, XPS, NEXAFS, SEM-EDX, STEM-EDX, 19 F NMR, EPR and in situ FT-IR) to probe the effect of fluorine doping on the structural and electronic properties of the material. In light of the applicability of this material as a support in heterogeneous catalysis, the surface properties are thoroughly analysed as well (FT-IR, micro-calorimetry and TDS).
follows: 357 g of zinc nitrate hexahydrate were dissolved in 500 mL deionised water with 15 mL nitric acid and filled up to give 600 mL solution. A second solution of 339.16 g sodium carbonate in deionised water to give 2 L was prepared. In an automated reactor (LabMax, Mettler-Toledo), 400 mL deionised water were placed, and 20 g/min of the nitrate solution were added for 30 min while the carbonate solution was added to stabilise the pH at 6.5. The mixture was aged for 60 min before the product was filtered and washed twice with 600 mL water until the conductivity of the filtrate was below 0.5 mS/cm. The resulting white powder (zinc hydroxycarbonate) was then dried at 353 K over night before calcining it in a rotating furnace at 603 K (ramp rate 2 K/min, dwell time 3 h) to yield the white zinc oxide powder (called "ZnO" hereafter). Two samples were fluorinated by application of fluorine gas of differing partial pressures. The needed amount of F 2 was calculated based on the stoichiometry Zn in ZnO for nominal 1 at% and 10 at% of fluorination. The partial pressure of F 2 (g) was adjusted with a calibrated volume. The fluorination was conducted at RT, 1 bar (mixed with N 2 ) and for one hour time of reaction. The sample treated with the lower F 2 partial pressure is termed "ZnO_F1" in the following, and the sample resulting from the treatment using a higher partial pressure "ZnO_F10", respectively. Commercial zinc fluoride (anhydrous 99 % min, alfa aesar ZnF 2 11521, Lot U01C031) was used as a reference.

X-Ray Diffraction
The samples were analysed using X-Ray diffraction (XRD) on a Bruker AXS D8 Advance II theta/theta diffractometer in Bragg-Brentano geometry using Ni filtered Cu α1+2 radiation and a position sensitive energy dispersive LynxEye silicon strip detector. Subsequently, the diffractograms were analysed by whole powder pattern fitting, employing the Rietvelt method as implemented in the TOPAS software (Bruker AXS).

UV-vis Spectroscopy
Diffuse Reflectance UV-vis spectroscopy was performed on a Cary5000 photospectrometer (Agilent), employing a Praying Mantis Diffuse Reflection Accessory and Reaction Chamber (Harrick) using Spectralon as a white standard. The optical band gap of each sample was determined from the appropriate Tauc plots 26 by linear extrapolation of the absorption edge to zero intensity.

Electron Microscopy
Scanning Electron Microscopy (SEM) was performed using a Hitachi S 4800 SEM equipped with a Genesis 4000 Energy Dispersive X-ray (EDX) detector. The microscope was operated at an acceleration voltage of 1.5 kV. EDX spectra were obtained using an acceleration voltage of 5 kV. For the TEM measurements, a Thermo Fisher Scientific Talos F200x S/TEM was used with an acceleration voltage of 200 kV. STEM measurements were performed using an HAADF detector, and the ChemiSTEM super-X EDX system using four SDD detectors was employed to generate Energy Dispersive X-Ray Spectroscopy (EDX) maps.

Electron Paramagnetic Resonance
Electron paramagnetic resonance (EPR) measurements of the fluorinated ZnO samples were conducted in quartz tubes (2.9 mm outer diameter). Continuous wave (CW) EPR spectra at X-band frequencies were obtained at room temperature using a Bruker B-ER420 spectrometer upgraded with a Bruker ECS 041XG microwave bridge and a lock-in amplifier (Bruker ER023M). The samples were measured in a Bruker SHQ or TM 110 resonator applying a modulation amplitude of 5 G and a modulation frequency of 100 kHz. All spectra were background corrected by subtracting a spectrum of an empty quartz tube and the signals were normalised to the sample mass.

X-Ray Photoelectron Spectroscopy
X-ray photoelectron spectroscopy was performed on pressed pellets of the samples using non-monochromatised Al Ka excitation (1486.6 eV) and a hemispherical analyser (Phoibos 150, SPECS). The standard Au4f (7/2) and Cu2p (3/2) procedure was employed to calibrate the binding energy scale, and no correction for charging effects was applied. Elemental compositions were calculated using theoretical cross sections from Yeh and Lindau. 28

Conductivity Measurements
The contact-free and in situ Microwave Cavity Perturbation Technique (MCPT) 29 was employed to measure the conductivity of the (fluorinated) zinc oxide samples at 503 K in inert gas (N 2 ) as well as hydrogen-(5% H 2 /N 2 ) and oxygencontaining (5% O 2 /N 2 ) conditions. The gas flow was held constant at 10 mL/min. The resulting data were fitted and corrected to account for the powder samples' filling fractions using an established formalism. 30,31 2.9 NEXAFS All NEXAFS measurements were performed at the UE56-2 PGM-2 beamline at BESSY II/HZB (Berlin, Germany). All measurements were carried out in a stainless steel NAP-XPS chamber described in detail elsewhere 32 . The samples were pelletised and mounted on a sapphire sample holder between a stainless steel back-plate and a stainless steel lid with a 6 mm hole. The NEXAFS spectra of these samples were obtained in the total electron yield mode by recording a sample current. The photon-energy resolution was set to 700 meV at the Zn L 3edge (≈ 1020 eV), 400 meV at the F K-edge (≈ 680 eV) and 250 meV at the O K-edge (≈ 520 eV) The photon energy was Please do not adjust margins Please do not adjust margins calibrated using the Au 4f line. All NEXAFS spectra were normalised to the incident photon flux.

IR spectroscopy
The fluorine treated ZnO samples were characterised by FTIR and compared to ZnO and ZnF 2 . The measurements were carried out using a Varian 670 FTIR spectrometer equipped with a MCT detector. The spectra were recorded at a resolution of 2 cm -1 accumulating 128 scans. Self-supported wafers (area weight of 20-30 mg cm -2 ) were transferred into an IR cell that is connected to a vacuum line and a gas delivery system. Before the spectroscopic measurements, surface impurities on the samples were removed by heating in the IR cell to 240 °C (200 °C in the case of ZnO_F10) in 50 mbar O 2 until the IR spectra did not show any further changes. The presence of oxygen is necessary in order to avoid the formation of donor centres which may obscure changes of the vibrational bands as a result of their intense and broad absorption in the IR range. After cooling to 25 °C the cell was evacuated and 67 mbar of NH 3 were added. All spectra shown are recorded after evacuation (to remove gas phase NH 3 ) and are referenced against a spectrum taken immediately before NH 3 addition. For the desorption experiments the cell was heated at a rate of 1 K/min, with spectra taken every 10 K. In situ spectroscopy was performed by a Vertex 70 FT-IR spectrometer from Bruker® equipped with a MCT detector. The spectra were recorded at a resolution of 0.5 cm -1 accumulating 100 scans. Self-supporting wafers of 13 mm diameter were placed in an IR cell 33 . The sample was heated to 240 °C in vacuum for 10 h in order to remove surface impurities and subsequently treated with O 2 (120 mbar) for 1 h. 43 mbar 10 % F 2 /Ar mixture was added to the cell, while simultaneously taking IR spectra in order to monitor the changes to the sample.

Microcalorimetry
The microcalorimetric experiments were performed in a Calvet calorimeter (SETARAM MS70) equipped with a customdesigned high vacuum and gas dosing apparatus, which allows for the dosage of probe molecules within a range of 0.02 mmol and the calculation of the amount of adsorbed molecules. The samples were heated to 240 °C at a rate of 5 K/min in UHV for 1 h dwell time. After degassing, oxidation was performed in 5 % O 2 for 30 min two times. Thereafter, the cell was cooled to room temperature and pumped to UHV (10 -7 mbar final pressure), placed inside the calorimeter without contact to air, and connected to the microcalorimetric gas adsorption system. CO 2 was then dosed stepwise at 30 °C, while recording pressure, adsorption temperature, and heat signals during all dosing steps.

Thermal Desorption Spectroscopy
(TDS) was applied for the temperature programmed desorption of oxygen. Therefore, a self-constructed setup which enables the testing of powder samples was used. The setup is equipped with mass flow controllers, an IR-light furnace (BEHR IRF 10) and a mass spectrometer (PFEIFFER VACUUM QME 200). The powder sample was placed on a small quartz-glass boat which is placed in a quartz tube (inner diameter of 14 mm, outer diameter of 20 mm, length of 450 mm) located inside the furnace and connected to the system using Ultra Torr vacuum fittings. Afterwards the system was stepwise brought to 9·10-7 mbar and directly connected to the mass spectrometer. The surface was treated at 503 K for 2.5 h under reductive atmospheres (5 % H2 in Ar). Subsequently, at 348 K the surface was saturated with double labelled C18O2 at 1 bar. The desorption experiment was conducted at 10-6 mbar with a heating rate of 25 K/min.

Sample designation
The co-precipitated and calcined polycrystalline ZnO powder was directly treated with F 2 (g). By employing two different p(F 2 ), fluoride concentrations within the ZnO sample of 1 and 10 at.%, respectively, were targeted. Accordingly, the samples are named as ZnO_F1 and ZnO_F10. For details about the preparation method, please see the experimental section. As reference samples, the pristine ZnO and commercial ZnF 2 were investigated. Figure 1: X-ray diffraction patterns of fluorine treated and untreated ZnO, and commercial ZnF2 as a reference. Indices for ZnO and ZnF2 are given from literature references 34,35 . In order to study the effects of F 2(g) treatment on the crystal structure of ZnO, and to reveal the presence of resulting crystalline by-phases in the samples, X-ray diffraction (XRD) was employed. In Figure 1, we present the X-ray diffraction patterns of ZnO_F1 and ZnO_F10, as well as ZnO and ZnF 2 as references. The XRD pattern is indicative of a hexagonal ZnO phase. The highly fluorinated ZnO_F10 sample contains in addition ZnF 2 as a by-phase, which is absent in ZnO_F1.

Crystal structure
The whole powder pattern of the fluorinated ZnO samples were fitted (Table 1) to reveal possible changes of the lattice parameters and for the quantification of the ZnF 2 -phase. The fluorination had no impact on the lattice parameters a and c, whereas the oxygen position z(O) increased with increasing fluorine doping. From the Rietveld refinement, the crystalline ZnF 2 by-phase observed for the ZnO_F10 sample amount to 8 wt.% (or ~6 at.%). The slightly increasing domain size of ZnO, in particular for Zn_F10, is interpreted as preferential reactivity of F 2 (g) with small crystallites to form, e.g., ZnF 2 moieties. Table 1. Refined XRD parameters, errors given as 3SD.

Quantification of F content.
The fluorine content of the fluorine treated samples was quantified using Energy Dispersive X-Ray Spectroscopy (EDX) in the SEM at an information depth of approx. 100 nm. The F/Zn signal ratio was about 10 times lower for the ZnO_F1 sample than for ZnO_F10 (F/Zn approx. 0.012 for ZnO_F1, and ca. 0.12 for ZnO_F10), which matches the expectation based on the applied fluorine pressure and is interpreted as a complete consumption of the F 2(g) in the reaction with ZnO. According to XRD the fluorine bound in ZnF 2 is calculated to 6 at.%, which means an estimated amount of ~6 at.% of fluorine is left in ZnO_F10 as possible anionic dopant. Additionally, the surface-near molar element ratios of the samples were investigated using XPS (Table 2). While the total oxygen content measured through the O/Zn ratio remains constant upon fluorine treatment, fitting of the O1s line shapes reveal a decrease of lattice oxygen content with increasing F-content in the sample. The latter can be interpreted as a substitutional fluorine doping of ZnO upon treatment with gaseous F 2 . Fitting the O1s spectra (SI 1) reveals an increase of spectral components with higher binding energy, which points to the fact that the sample with higher F-content contains more absorbed oxygen in the form of OH groups or adsorbed water. This might be a sign for an increased surface Lewis acidity. However, estimating the OHconcentration on the ZnO surface, based on the information depth and calculated quantities (Table 2), the coverage is <1 mono layer (ML) without significant influence on the wurtzite lattice. The F 1s spectra show, as expected, a larger F/Zn ratio for ZnO_F10 than for ZnO_F1 (SI 4). The slight increase for ZnO in the F 1s region is explained by minor Zn Auger contributions. The stoichiometric disbalance of the O/Zn ratio (<1) is a typical sign for nanostructured ZnO with an oxygen deficient character, originated in the synthesis history of the precursor. Interestingly, for the ZnO_F1 sample the F/Zn ratio near the surface (XPS information depth, here: ≈ 2 nm) is larger than the bulk F/Zn ratio (determined from EDX/SEM, here: ≈ 100 nm). This finding indicates that the concentration of F might be larger near the sample surface.

Morphology and local structure
Complimentary to the integral structural analysis by XRD, the local structure and morphology of the fluorine treated samples was investigated using Scanning Electron Microscopy (SEM) and Transmission Electron Microscopy (TEM), which are not limited to the crystalline part of the samples. Figure 2  In both regions indicated in Figure 4(b) and (c), ZnO was identified as the main structure, but ZnF 2 was found as a minor phase, which agrees well with the results obtained from EDX ( Figure 3 (c-d)). Since not all reflections present in Figure 4(c) correspond to only one of the two phases ZnO and ZnF 2 , along with the results from EDX mapping it can be concluded that this region contains a higher amount of ZnF 2 than region (b).
The results of the electron microscopy investigations are in agreement with the elemental quantifications and structural analysis already discussed. High p(F 2 ) was shown to form a heterogeneous system as it leads to the formation of a ZnF 2 by-phase, while the low p(F 2 ) results in a phase pure system. The spatially heterogeneous distribution observed for the system exposed to high fluorine pressure is readily explained by a ZnF 2 particle adjacent to an F-doped ZnO one.

Nuclear Magnetic Resonance spectroscopy
The environment of 19 F nuclei within the structure of the fluorine containing samples was investigated by solid state 19 F Nuclear Magnetic Resonance (MAS NMR) spectroscopy which is able to identify local environment of 19 F species both in crystalline and also potentially X-ray amorphous parts. A comparison of the spectra for all F-containing samples can be found in Figure 5. In the spectrum of ZnF 2 a 19 F signal at -205 pm is visible (with corresponding spinning sidebands at -88, -150, -258, and -312 ppm (SI 13)).
The spectrum of the ZnO_F10 sample shows a signal at 205 ppm indicating the presence of ZnF 2 in line with the X-ray diffraction results. The variation of the spinning frequency reveals that the signals at -36 ppm, -90 ppm, -255 ppm and -   310 ppm are due to spinning side bands. In contrast, the wellresolved signals at -148 ppm and -173 ppm indicating a rather uniform coordination environment of the F-nuclei are not found for crystalline ZnF 2 . The signals are in the typical region for a fluorine doped oxide (or "fluoro-oxide"), as could be expected for ZnO:F 38 . The spectrum of ZnO_F1 exhibits a much reduced signal intensity in line with the lower F content, nevertheless the presence of fluorine-related centres is still detectable. The spectrum is characterized by a broad asymmetric line ranging from -150 to -220 ppm. The large width together with the asymmetric line shape gives evidence for a broad distribution of local environments for the fluorine species in this sample. This means, despite a homogeneous spatial distribution of the fluorine within the sample (confirmed by STEM-EDX-maps in Figure 3(a)), their local arrangements is rather non-uniform (different, non-preferential coordination environments). The spectrum shows narrow signal at -14.8 ppm, which does not disappear in a spin-echo test and is neither found for ZnF2 nor for the ZnO_F10 sample. It is a typical solid state signal (line width ≈ 1.5 kHz) and it appears at the position found for fluorine in BaF 2 , however, the elemental analysis did not show any trace of Ba 2+ . Therefore, we cannot assign this signal unambiguously.
For all three samples, the spin lattice relaxation behaviour of 19

Infrared spectroscopy
The effect of F 2 treatment on the surface of ZnO was studied by IR spectroscopy. The spectra of the untreated ZnO and the two samples treated with different amounts of fluorine gas can be found in Figure 6, focusing on the bands between 4000 cm -1 and 3000 cm -1 (OH stretching region). Overall, an excellent agreement between the spectrum of ZnO to reported values 39 can be noted. In contrast, the spectra of the two fluorinated samples differ from ZnO. The sharp peak assigned to OH on the 0001 surface of ZnO (3620 cm -1 ) shifts to lower energy with increasing degree of fluorination (ZnO_F1, 3616 cm -1 ; ZnO_F10 3606 cm -1 ), which indicates a weakening of the OH bond, and thus, a higher acidity. The bands above 3670 cm -1 are assigned to OH stretching modes associated with adsorbed water on the ZnO 1010 facet. These bands are absent for the highly fluorinated sample ZnO_F10. For ZnO_F1, the complex line shape becomes more resolved with two dedicated signals at 3686 cm -1 and 3670 cm -1 , respectively. In addition, the relative intensity of the band assigned to OH groups on the ZnO surface not coupled to coadsorbed water on this facet (3640 cm -1 ) increases significantly with the degree of fluorination. Therefore, it becomes apparent that the fluorine treatment reduces and finally removes adsorption sites of H 2 O on these facets. Furthermore, broad defect bands at 3450 cm -1 and 3560 cm -1 are affected. To create a deeper understanding on the ongoing processes of fluorination and fluoride incorporation, in-situ reference experiments were conducted. To this end, nanostructured ZnO is treated with F 2(g) inside the spectrometer cell, and changes relative to the pristine ZnO were monitored as a function of time. Figure 7 and Figure 8 show the changes for the OH-and the carbonate-region regions of the IR spectrum. The reaction starts instantaneously after the addition of fluorine gas, the spectrum continuously evolves for several hours. The most significant changes occur in the OH stretching region as

Please do not adjust margins
Please do not adjust margins already described above ( Figure 6). Upon exposure of the sample to F 2 , the bands at 3690 cm -1 and 3670 cm -1 , which are assigned to H 2 O and OH with co-adsorbed water, respectively, both on the ZnO 1010 surface, decline. In contrast, the band at ~3640 cm -1 increases as a result of the removal of the coadsorbed water and the possible formation of new OH-groups on former defect sites. This band shifts to 3634 cm -1 at longer reaction times and with higher degrees of fluorination. as well as an overall decline of the intensity of the carbonate bands is observed (Figure 8). Carbonates might be present as trace residuals of the synthesis process or formed on the surface by contact to the ambient 41 . The formation of CO 2 by the replacement of carbonate groups at the ZnO surface by fluoride can be expected to occur according to the following equation: An increasing amount of gas phase CO 2 during fluorination is observed in the spectrum around 2350 cm -1 . In addition, the band at 1650 cm -1 , which gains intensity over time, can be assigned to the bending vibration of H 2 O(g), which is in line with a release of co-adsorbed water upon fluorination deduced from the spectra in the OH region, discussed above. In summary, the IR spectroscopic investigations unravel a couple of changes on the surface of ZnO upon fluorine treatment. The quality and quantity of the OH-groups is significantly altered. OH groups related to co-adsorbed water are gradually decreasing and new OH groups are formed. Besides, some existing OH-groups are shifted to lower energy as sign for an increased Lewis acidity. No preferential surface termination for fluorination is identified, which might be explained by the overall high reactivity of F 2(g) . However, the hydroxyl-and carbonate-related functions seem to react faster than the defect related sites, in particular upon the formation of new (extrinsic or F-induced) ones.

Optical band gaps
The optical band gaps of the fluorine treated and untreated ZnO samples were determined using DR-UV-vis (diffuse reflectance ultraviolet-visible) spectroscopy, and extrapolating the band gap (BG) energies from the appropriate Tauc plots (see SI 14). Figure 9 shows that ZnO and ZnO_F1 have a similar optical BG, whereas the more fluorinated sample ZnO_F10 has a larger optical band gap. Since ZnF 2 is an insulating material, its possible contribution in ZnO_F10 is negligible. Interpreting these results in light of the 19

Conductivity of the samples in varying gas feeds
Using MCPT, the conductivity of the non-fluorinated and fluorine treated ZnO sample in varying gas feeds (100 % N 2 , 5 % H 2 /N 2 and 5 % O 2 /N 2 ) at 503 K was measured. The conductivity of the samples in these gas feeds gives an indication for the interaction of the samples with the gases H 2 and O 2 , and for the reversibility of these processes. Figure 10 presents the conductivity of ZnO, ZnO_F1, and ZnO_F10 in these different gas feeds. The x-axis also denotes the sequence in which the gas feeds were applied. Firstly, the fluorinated systems exhibit an enhanced conductivity (x 3) as compared to the pure ZnO sample. The application of reductive conditions (5 % H 2 /N 2 ) results in an increase of the conductivity as expected for n-type semiconductors. While the conductivity of the ZnO_F1 sample increases by more than one order of magnitude, the other two samples show an increase by about a factor of 2 only, which clearly indicates a significant difference in the electronic properties of the two fluorinated systems. Upon switching back to the inert feed the pristine ZnO exhibits a reversible reduction of the conductivity to the initial value. In contrast to that, the conductivity of the fluorinated system remains significantly enhanced as compared to the initial conditions. The extent of the hysteresis is significantly larger for the ZnO_F1 sample. Applying oxidizing conditions (5 % O 2 /N 2 ) has little effect on the conductivity of the untreated ZnO sample, whereas the conductivity of the fluorinated samples is reduced. The ZnO_F1 system shows the largest reduction and has a comparable conductivity as the untreated ZnO sample under these conditions. Please note that this is significantly smaller than the one observed under inert conditions. The highly fluorinated sample behaves qualitatively similar to the ZnO_F1 sample, however, the extent of the effect is smaller. Upon switching back to inert conditions the fluorinated systems show an increase of the conductivity which becomes similar to their initial values, while the ZnO sample shows little changes. This indicates a qualitative difference between the untreated and F 2 treated samples. While the conductivity of the untreated ZnO can be increased by reductive conditions the effect is reversible already upon switching to inert conditions whereas oxidative treatment has basically no effect on the conductivity. In contrast to that, the increase of the conductivity upon reductive treatment is significantly larger for both of the fluorinated samples, indicating that the fluorination has altered the electronic properties of the materials. Since the optical band gap of ZnO_F1 was not affected by the fluorination, this increase in conductivity is rather explained by surface effects. This means that the activation and chemisorption of H 2 , providing/ donating the electrons, is significantly enhanced. The changes induced by reduction are only partially reversible under inert conditions in contrast to the pristine ZnO. This indicates that the reduced state of the material is stabilized in case of the fluorinated systems, and that the activated H 2 is bound rather weakly. Furthermore, it is interesting to note that an oxidative treatment, which has little effect on the pristine ZnO, reduced the conductivity of the fluorinated systems. This is, however, in line with the weak bonding of hydrogen on ZnO and thus already removed under inert conditions.

Electron paramagnetic resonance
The fluorination of ZnO leads to an enhanced conductivity of the system and significant alterations towards the response to different environments. Thus, it is interesting to gain more insight into the nature of the states that are involved in these changes. Substitutional doping a system with an aliovalent anion will result in states in the band gap. Several of these are paramagnetic and hence, EPR spectroscopy can be used to investigate these species. ZnO was shown to exhibit two major classes of paramagnetic defects observed at g = 1.958 and g = 2.007, which are assigned to shallow donor states and Zn vacancies, respectively 42 . The cw EPR spectra of the untreated and fluorine treated ZnO samples are presented in Figure 11. Overall, the spectra of ZnO, ZnO_F1 and ZnO_F10 are very similar. All samples show a strong, typical ZnO-defect signal around g = 1.958 (359.7 mT). The area of the double integral increases by 1/3 upon low fluorination (ZnO_F1), but decreases for ZnO_F10 (also by 1/3), in comparison to ZnO as reference (see inset Figure 11). The second signal found in ZnO samples (g = 2.007, 351 mT) is almost absent in the spectra of the fluorinated samples. The increase of the defect concentration for the low fluorination level and the subsequent reduction for higher fluorination levels is in line with the NMR observations, which show a very heterogeneous F-environment, hence, an inhomogeneous perturbation of the lattice for the low fluorination level, while a much more homogenous situation is observed for the higher fluorinated sample.
It is interesting to look for a spatial proximity of the defect centers and the F-anions in the lattice. The latter is possible by scrutinizing the coupling between the paramagnetic states to nuclear spins in its surrounding. To this end pulsed EPR measurements of ZnO_F1 were performed, which showed no direct coupling of the ZnO defect at g = 1.958 to fluorine nuclei (see SI 15). This excludes a simple substitutional F-doping of the oxygen sublattice and an associated trapping of the hole center close by.
While the XPS data provide evidence for a significant degree of hydroxylation of the ZnO surfaces the pulsed EPR spectra show, no evidence for a coupling of the paramagnetic states to hydrogen atoms, which renders the corresponding species not to be located in the bulk of the sample. However, due to the limited impact of bulk F-doping (indicated by optical BG measurements), the ZnO-defect signal at g=1.958 might rather be surface related and further explain the conductivity measurements, which significantly increased for ZnO_F1.
Besides the higher quantity of defect sites for ZnO_F1, another criterion like surface acidity might influence the H 2 activation (and conductivity) and is probed later by NH 3 -IR investigations.

NEXAFS
NEXAFS (Near Edge X-ray Absorption Fine Structure) spectroscopy was used in order to elucidate the density of unoccupied states of the samples ZnO, ZnO_F1, and ZnO_F10 near the surface. In ZnO the Zn and O centers are tetrahedrally coordinated. Formally, Zn 2+ in ZnO has 10 d electrons, which completely occupy the 3-t 2 and 2-e orbitals and an empty 4s orbital. Studies on various Zn-salts like nitrates or carbonates, show that this picture is oversimplifying. Instead, a hybridization between Zn and O states is observed, which reflects a certain degree of covalence in these systems. The latter is even more pronounced for ZnO 43,44 . Following the dipole-transition selection rule 45 , the features in the O K-edge spectra (Figure 12  thus explained by a less covalent bonding character upon fluorination and a decrease in electrons transfer from O 2p into Zn 4d states. The main features in the Zn L 3 -edge ( Figure  12(b)) are related to the excitation from Zn 2p electrons to 4s and 4d states. The increase in Zn L 3 -edge resonances represents also an increase in UDOS related to the fluorine incorporation on the expense of oxygen. The high electronegativity of fluorine leads to an increased ionic bonding character. This is further supported by an increase of the 2p to 4s transition visible at 1023.5 eV (more unoccupied 4s states) and into hybridized states (O 2p and likely Zn 4d states) at 1033eV 43,44 . Furthermore, the correlating trends within the O K-edge and Zn L 3 -edge intensities, caused by fluorination, follow the opposite trends observed for N-doped ZnO (p-type doping), where an increase in O K-edge and decrease in Zn L 3 -edge is attributed to an increased covalence by nitrogen incorporation 14 . The peak ration change in the O Kedge (534 eV / 537.5 eV) and in the Zn L 3 -edge (1023.5 eV / 1033 eV) is F substitution at O sites and/or doping into interstitial sites 43 . Generally, the changes of the electronic structure upon fluorination seems to be mainly attributed to ZnO:F, since the major impact is already observed for ZnO_F1. The F K-Edge spectra are given in the supporting information and show the presence of ZnF 2 in the spectrum for ZnO_F10 46 .

IR spectroscopic analysis of the surface acidity probed by NH 3 adsorption
The NEXAFS data indicate an increased ionicity of the system. In turn, this may change the Lewis acidity of the cationic sites on the surface. The latter can be probed by the changes in the vibrational spectrum of appropriate adsorbates. NH 3 was shown to be such a probe for Lewis acidity of surface sites. Figure 13(a) shows the IR spectra of the three samples (ZnO, ZnO_F1 and ZnO_F10) after NH 3 adsorption, all relative to the respective pristine samples. With increasing fluorine content in the sample, the symmetric NH 3 bending vibration, as sensitive probe for the coordination environment, shifts from 1209 cm -1 in ZnO to 1221 cm -1 in ZnO_F1 and finally to 1225 cm -1 in ZnO_F10. This shift is explained by an increase in the electron deficient character of Zn cations in the samples [47][48][49] . In our case, the increase of the Lewis acidity in Zn cations is caused by the fluorination of the samples 50 . Furthermore, ZnO_F10 shows an additional band at 1268 cm -1 , which is already present as a weak feature in the spectrum of ZnO_F1 (tiny signal for ZnO). The range in which this band appears is typical for NH 3 adsorbed on a very electron poor adsorption site in other transition metal oxides, (e.g., V 2 O 5 at 1270 cm - 1 49,51 and MoO 3 at 1260 cm - 1 49 ). From a comparison with the IR spectrum of ZnF 2 after NH 3 adsorption (SI 17), the band cannot be assigned to adsorption sites on ZnF 2 particles. Thus, it is likely that this band is also associated with an electron poor adsorption site, which is already present on ZnO in low concentrations. The concentration of such adsorption sites (defects) in fluorinated samples increases and possibly represents a Zn cation in a particularly fluoride enriched (likely ordered) environment. Temperature programmed IR spectra (Figure 13(b)-(d)) were measured to elucidate the adsorption strength of NH 3 on the different adsorption sites. The stability of the weakly perturbed sites in ZnO_F1 and ZnO_F10 (1221 cm -1 and 1230 cm -1 ) is significantly enhanced (compared to ZnO at 1213 cm -1 ).
The bands at 1260-1269 cm -1 , related to Zn cations likely in a fluoride rich and ordered environment, decline more rapidly than the bands at 1221 cm -1 (ZnO_F1) and 1230 cm -1 (ZnO_F10), respectively. This might be an indication that an additional stabilisation (e.g., by neighbouring OH groups) of the adsorbed molecule on the highly Lewis acidic and fluorine enriched sites is absent (or present for the sites at 1209-1230 cm -1 ). However, the latter bands have a gradually increased stability with the degree of fluorination and Lewis acidity, also visible at the residual intensity at 473 K for, e.g., ZnO_F10. This might further explain the higher conductivity values under hydrogen atmosphere, since Lewis acid sites are known to be relevant for H 2 activation (and here electron donation) 52 .

Analysis of the surface basicity probed by CO 2
The Lewis basicity of adsorption sites was studied using the adsorption enthalpy of CO 2 as probed by microcalorimetry 53 . In Figure 14, the amount of adsorbed CO 2 on ZnO and ZnO_F1 are shown. The total amount of adsorbed CO 2 is reduced by about 1/3 when comparing ZnO and ZnO_F1. In particular, the amount of irreversibly adsorbed CO 2 is reduced by a factor of 3.1, while the amount of reversibly adsorbed CO 2 is increasing for the ZnO_F1 sample as compared to pure ZnO. It becomes evident that fluorination reduces the adsorption strength of CO 2 on the sample surface from 80 kJ/mol to 60 kJ/mol (initial Please do not adjust margins Please do not adjust margins differential heat of adsorption, Figure SI18 (a)). In fact, no significant adsorption of this gas could be observed for ZnO_F10 or ZnF 2 (see Figure SI 18(b)). The relative amount of irreversibly adsorbed CO 2 on ZnO and ZnO_F1 was further investigated by thermal desorption spectroscopy (TDS). The m/z = 48 desorption signal is shown in Figure SI 19. The desorption spectra of both samples are qualitatively almost identical and exhibit several desorption features separated by significant temperature. This indicates that fluorine treatment does not alter the nature of the adsorption sites for CO 2 significantly. Quantitatively, the number of adsorption sites on the fluorinated sample is reduced by a factor of 2.7, which compared favourably with the microcalorimetry (factor of 3.1). The loss in basic adsorption sites is in line with the expectation based on the increased Lewis acidity of the surface (s.a.) as both properties are generally found to be inversely correlated.

Conclusion and Final Discussion
In this contribution, we have presented the direct treatment of ZnO with F 2(g) as a convenient way to modify both the electronic properties and the surface structure of ZnO. Hereby, the degree of fluorine incorporation into the sample can be controlled by variation of the synthesis parameters (partial pressure of F 2 , residence time of the fluorine gas on the sample). The time resolved fluorination indicates a kinetically controlled reaction of F 2(g) with ZnO. In-situ IR spectroscopy obtained during this reaction showed that fluorine removes carbonates on the ZnO surface and displaces adsorbed water, possibly by changing the interaction strength of the surface with H 2 O through replacement of O by F. No preferred ZnO surface termination with respect to their fluorine interaction was identified. The OH groups, also related to defect sites, were found to be altered in the spectrum as well. The treatment resulted in two types of fluorine modified sample: ZnO_F1 shows a spatially homogeneous distribution of fluorine throughout the sample, as evidenced by STEM/EDX mapping and its crystal structure shows only peaks that can be assigned to ZnO (in line with a possible F-doping of this material). In contrast, ZnO_F10, resulting from treatment of ZnO with a higher partial pressure of F 2 , resulted in "over-fluorination" as indicated by the appearance of a ZnF 2 by-phase. This is demonstrated by NMR and X-ray diffraction and underlined by the agglomeration of F near the particle surface in the EDX map presented. In addition, the 19 F-NMR experiments reveal as function of the fluorine content two different kinds of ZnO:F doping: In the low F-doping regime (1-2 at-%) a rather heterogeneous coordination environment (see broad 19 F-NMR signals) and a well ordered coordination with higher degree of doping (until 6 at-%). A combination of XRD, NMR, EPR and insitu IR studies identifies the fluorination as intriguing process where F is very likely locates on lattice and interstitial sites. The substitution of lattice oxygen seems to be the primary process, since F-induced defect sites are detectable only delayed, as possible consecutive adaption of the structure.
Comparing the F/Zn ratios of surface-near and bulk sensitive methods, a homogeneous surface enrichment of fluorine for ZnO_F1 and a heterogeneous for ZnO_F10 is indicated. This behaviour is to be expected for samples resulting from gassolid-reactions and is currently under investigation for ZnO single crystals as part of a separate study. In both samples, the F 2 treatment affected the electronic properties and the interaction of the sample with various gaseous probe molecules. Adsorption of NH 3 and CO 2 as Lewis -acidic and -basic probes, respectively, revealed, in the IR spectra of the samples and in microcalorimetry investigation, that the Lewis basic sites in ZnO were quenched, while the Lewis acidity of the surface increased incrementally with the Fcontent. As a consequence, the interaction of the samples with CO 2 was greatly reduced, as confirmed by microcalorimetry and TDS. Hereby, the fluorinated samples differed significantly in their behaviour, with ZnO_F10 showing an interaction with CO 2 close to ZnF 2 (and close to zero). Furthermore, using conductivity as an indicator, we showed that the less fluorinated ZnO_F1 showed a drastic increase in conductivity (and consequently in H 2 activation) as compared to untreated ZnO and the sample with a higher F-content. Generally, with the degree of fluorination the surface Lewis acidity (on the expense of basicity) is gradually increased as monitored by IR and various probe molecules. The electronic properties, however, are mainly affected by the F-doping, evidenced by DRUVS, NEXAFS and conductivity measurements. This is explained by a combination of more defect sites near the surface (EPR) and a higher Lewis acidity (NH 3 -IR). The increase of the Lewis acidity alone seems not to be sufficient to increase the conductivity significantly (see ZnO_F10). Within an attempt to understand the complex processes upon fluorination, the 19 F MAS NMR experiments serve as important bridge between these parts, identifying that nature and consequence of the occurring doping might also be different (at different levels of F-incorporation). Based on the numerous applied methods, the complexity of the nanostructured ZnO and relevant processes upon fluorination are identified and in parts unravelled. The impact of different local arrangements on, e.g., the surface properties has to be investigated within a separated DFT study.
Since ZnO often finds application as part of a CO 2 reduction catalyst, these findings suggest that the catalytic performance of catalysts derived from fluorine treated ZnO may be influenced significantly 25 . Hereby, an "over-fluorination" might result in a rather inactive catalyst, while a sample with a lower F-content (i.e. moderately doped ZnO:F) may improve the catalytic performance due to the interaction of this material with H 2 and its dual site character of the surface (acid/basic surface sites). Further work on this sample system will be conducted in order to investigate the limits of F-doping without the formation of the ZnF 2 by-phase, as well as the catalytic performance in CO 2 reduction of ZnO:F adding a metal component.

Conflicts of Interest
There are no conflicts to declare.