Novel insight on the local surface properties of ZnO nanowires

Novel insight on the local surface properties of ZnO nanowires (NW) deposited by the evaporation-condensation method on Ag-covered Si substrates is proposed, based on the results of comparative studies by using the scanning electron microscopy (SEM), x-ray photoemission spectroscopy (XPS) and thermal desorption spectroscopy (TDS) methods, respectively. SEM studies showed that ZnO nanowires (nanoribbons) are mostly isolated and irregular, having the average length μm and the average at the level of tens nm, respectively. Our XPS studies confirmed their evident surface non-stoichiometry, combined with strong C surface contaminations, which was related to the existence of oxygen-deficient regions. Additionally, TDS studies showed that undesired surface contaminations (including C species and hydroxyl groups) on the surface of ZnO NWs can be removed almost completely, leading to an increase of the final non-stoichiometry. Both effects are of great importance when using ZnO NWs for the detection of oxidizing gases, because the undesired C contaminations (including C–OH species) play the role of undesired barriers for the gas adsorption, especially at the low working temperature, additionally affecting the uncontrolled sensor ageing effect.


Introduction
Zinc oxide (ZnO) is one of the most popular transparent conductive oxides (TCO), having unique optical properties (wide band gap of 3.37 eV, and large exciton binding energy (60 meV)) as well as electronic properties (high electronic mobility up to 2 cm 2 /V · s). It has a widespread application, mainly in optoelectronics and photovoltaics [1][2][3][4]. In addition, due to the above mentioned high mobility of conduction electrons and good chemical and thermal stability, recently Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
ZnO became one of the most interesting and very promising gas sensing material [5,6].
It should be noted that already in 90s Sberveglieri et al [7], as well as other researchers, have shown that ZnO thin films are very useful for the detection of oxygen. However, even after many years of exploring fundamental sensing parameters, including the response and recovery times, the undesired ageing effect of oxide semiconductors, including ZnO, remains mostly unknown. This is related to the fact that the gas sensing effect appears within the surface space-charge region of sensor material, i.e. at the depth related to Debye length, after its exposure to gas, in this region, the surface state density of sensor material is balanced by acceptors and donors, creating a surface band bending [8].
What is crucial, during the gas sensing effects, is that surface adsorption/chemisorption processes of gaseous species take place mainly at the surface of crystalline ZnO nanowires. These processes are responsible for the charges redistribution within the surface space-charge region of sensor material and the appearance of surface bend bending effect [42]. For the ZnO nanowires, the charge carrier concentration reaches the value of~10 18 [cm −3 ], and a surface depletion region (upward band bending) is usually observed [23], which is directly responsible for the specific surface conduction mechanism. In some cases, the transverse dimension of ZnO nanowires (NWs) can be even smaller than 10 nm. This dimension is comparable to the Debye length, therefore in such a case one can expect the highest detection efficiency of gas molecules adsorbed at their surface.
Having all above in mind, it is evident that the detailed information on ZnO NWs surface chemistry is absolutely indispensable for the deep understanding of gas sensing mechanism, allowing the design and construction of novel gas sensor devices based on this material. X-ray photoemission spectroscopy (XPS) is considered one of the best technique for monitoring the surface chemistry of ZnO NWs, including stoichiometry and undesired surface contaminations. In one of our [43], preliminary XPS studies on ZnO NWs exhibited an evident nonstoichiometry with oxygen deficiencies, even before thermal annealing under UHV conditions. In turn, in our subsequent [44], XPS method was applied for the monitoring of the surface chemistry of ZnO nanowires gas sensors functionalized by chemical modification, allowing an enhancement of NO 2 gas sensor response. This technique was also applied in our recent study for the control of the surface chemistry of nanostructured ZnO thin films deposited by the direct current (DC) reactive magnetron sputtering (DCRMS) method [45].
In this paper, we surface properties of ZnO nanowires deposited the Ag-covered Si substrate obtained by using three comparative techniques, i.e. SEM method for the observation of their surface morphology, XPS method for the determination of their local surface nonstoichiometry and contaminations, TDS method for the control of surface active gases adsorbed at the surface of ZnO NW.

Deposition of ZnO nanowires
ZnO nanowires (NW) were deposited on Si(100) substrates at SENSOR Lab, Department of Information Engineering, Brescia University, Italy, using an evaporation-condensation process in a custom furnace. ZnO powder was placed in the middle of furnace at high temperature (1370 • C) and condensed in a colder region (≈ 700 • C) where the substrates were placed. An inert gas flow (Ar, 100 sccm) was used to move the evaporated material toward the colder region of the furnace. Pressure was fixed at 100 mbar, and it was controlled finely by barometer (MKS, Germany). In order to reach a good uniformity of the ZnO NWs lateral dimensions, a catalyzed-assisted growth has been performed by Ag nanoparticles deposited onto the Si(100) substrates by DC magnetron sputtering (MS) method (Kenotec, Italy).

Surface characterization of ZnO nanowires
For the characterization surface properties of the ZnO NWs, a combination of SEM, XPS and TDS experiments were performed.
The surface morphology studies of ZnO NWs were carried out at SENSOR Lab, Brescia University, Italy, using a Field Emission Scanning Electron Microscope (FE-SEM, Gemini, Leo 1525) operated in the 3-5 keV range.
The surface stoichiometry of ZnO NWs, together with the presence of surface contaminations, was determined by using the XPS method, before and after the subsequent TDS experiments, respectively. For XPS analysis we used a commercial XPS spectrometer (SPECS, Germany) equipped with a x-ray lamp (AlK α 1486.6 eV; XR-50 model), and a concentric hemispherical analyzer (PHOIBOS-100 Model). The XPS experiment have been performed at the background working pressure of~10 -7 Pa. The registered XPS spectra have been calibrated with respect to reference binding energies (BE) using both the XPS Au4f peak at 84.5 eV of an Au foil located on the sample holder, as well as the XPS C1s peak at 284.5 eV of residual C contamination always present on the surface of investigated ZnO nanowires.
The specific active surface gases desorbed from the surface of ZnO NWs were detected by the TDS method during a linear growth of the sample temperature (temperature programmable desorption-TPD), using the in line mass spectrometry (MS). TDS measurements have been performed in a sample preparation chamber equipped with a resistive type sample heating unit, combined with a temperature programmable control unit (OmniVac-Dual Regulated Power Supply PSReg120 model) and mass spectrometer-residual gas analyzer (Stanford RGA200 Model). During every TPD cycle in the temperature range of 50 ÷ 300 • C, the typical gas sensor working conditions, the TDS spectra of selected gases like H 2 , H 2 O, O 2 , and CO 2 have been registered.
Both XPS and TDS experiments have been performed at the Silesian University of Technology, Gliwice, Poland. Other experimental details have been described elsewhere [46][47][48].

Results and discussion
At the beginning of our studies we have focused on the surface morphology of ZnO NWs deposited on top of Ag-covered Si substrate and their SEM images at two different magnification levels and lateral resolutions are illustrated on figure 1.
It is clearly visible that the Ag-covered Si(100) substrate is covered completely by ZnO nanowires (nanoribbons) originating from the Ag metallic nanograins, contributing to the ZnO NWs nucleation growth process. The morphology of the nanowires is more visible in the SEM image at the higher magnification, which confirm that nanowires are mostly isolated and irregular, having various length exceeding one µm, as well as a various diameter (in the range of 10 ÷ 50 nm). This is crucial, having in mind that the gas sensor effect appears within the surface space-charge region of ZnO related to Debye length, which is in the range of several nm.
Afterwards, in to the morphological information above, we have focused on the surface chemistry of ZnO NWs deposited on Ag-covered Si substrate, with a special emphasis on the determination of its variation during the complementary thermal desorption experiments.
The survey spectra for ZnO NWs deposited at Ag-covered Si substrate before and after TPD process in the full binding energy (BE) range 0-1200 eV are presented in figure 2  Considering the relative intensity (height) of Zn2p 3/2 , O1s, and C1s XPS core level lines corrected by their respective atomic sensitivity factors (ASF), the relative concentration of the basic elements like O, Zn and C with respect to all atoms in the subsurface region can be calculated, according to the commonly used analytical procedure [49,50]. However, due to the high undesired background in the observed XPS survey spectra in the full binding energy range (1200 eV) and the contribution of Auger electron emission lines, the relative concentrations of the main elements of ZnO NWs have been calculated on the base of relative intensity of XPS core level lines in a limited range (600 eV) additionally corrected by the transmission function T(E) of CHA PHOIBOS 100 energy analyzer. The obtained data are summarized of table 1.
These respective values summarized in table 1 directly correspond to a total relative concentration (in %) of basic elements i.e. O, Zn and C in subsurface region of ZnO NWs. On the base of above summarized information, that before TPD process the contribution of main elements in the ZnO NWs surface/subsurface region is quite similar (~0.33). This is quite expected and a crucial information for their potential use as gas sensors because it is well known that hydrocarbon contamination is common on the air-exposed samples during sample handling or preparation [50]. Although the ZnO NWs exhibit almost stoichiometric form, after their exposure to air the undesired surface C contaminations can play a specific barrier for the gas interaction with expected surface sites waiting for the gas adsorption. In turn, after TPD process the contribution of main elements in the surface/subsurface region of ZnO NWs drastically changed. A first information is that after TPD process the contribution of undesired C contaminations drastically decreased, even reaching the noise level. In parallel, the relative O concentration with respect to all atoms in the subsurface region also drastically changed, reaching a value very far from the natural stoichiometry. Probably, this is related to the existence of specific forms of surface bondings for the main elements in the subsurface region of the analysed samples. Therefore the deeper analysis of the evolution of XPS Zn2p, O1s, and C1s lines related to main surface elements at ZnO NWs surface before and after TPD process has been performed.
The XPS Zn2p lines reveal two broadly resolved features at the binding energy of about 1045 eV and 1022 eV, which can be ascribed to the Zn2p 1/2 and Zn2p 3/2 for both samples, respectively as visible in figure 3. This is related to the spin orbit splitting of~23.0 eV, what is in a good agreement with available literature [48,51,52]. XPS spectra confirm that Zn element exists mainly in the form of Zn 2+ on the ZnO NWs surfaces, corresponding to Zn atoms in the ZnO lattice [48,53].
However, it is clearly visible that XPS Zn2p 3/2 line is slightly asymmetrical. A precise deconvolution procedure has been performed for these lines the obtained results are shown in figure 4. The deconvolution of Zn2p 3/2 lines exhibits two peaks located at binding energy~1022.0 eV and at~1024.0 eV, respectively, with very high line fitting (RMS = 0.995). The first one can be ascribed to the Zn atoms in the ZnO lattice [50]. In turn, the second one can be ascribed to the same zinc hydroxide Zn(OH) 2 species also observed for the ZnO nanoparticles by Guo et al [54], and for the ZnO thin films by Armelao et al [55]. The existence of specific Zn(OH) 2 on the surface of ZnO NWs before and after TPD process was also observed in the XPS O1s spectral lines what is demonstrated on figure 5. Unlike XPS Zn2p 3/2 lines, the XPS O1s lines exhibit an evident asymmetry with two main parts, which are probably related to the existence of different forms of oxygen bondings at their surface. For their detailed verification, the deconvolution of XPS O1s lines was performed using the Gauss fitting. The obtained results are shown in figure 6.
In the case of ZnO NWs before TPD process, after the deconvolution procedure the XPS O1s line exhibits four components clearly visible at the binding energy of 531.2 eV, 532.8 eV, 533.9 eV and 534.6 eV, respectively.
The first one, which can be treated as the main peak located at 531.2 eV, can only be attributed to the presence of partially reduced ZnO such as ZnO x direcly related to the existence of O 2 ions caused by oxygen deficiencies within the ZnO matrix. This is extremely important information because its binding energy is about 1 eV higher with respect to the commonly observed O 2 ions in the ZnO wurzite structure of the hexagonal Zn ion array (~530 eV) [49]. A similar peak located at 531.2 eV was observed, among others, by Chen et al [56], Kaneva et al [57] and Li et al [58] for various ZnO thin films, as well as by Chen et al [59] for the flower-like and tube-like ZnO nanomaterials. The existence of oxygen-deficient regions at the surface of ZnO NWs is of great importance because they can be responsible for the intrinsic sensing characteristic of ZnO, originating from the native defects (i.e. nonstoichiometry) such as oxygen vacancies and zinc interstitials. These defects are mainly located in the surface region and determine the electrical conductance, in particular in the quasi-mono dimensional structure of ZnO NWs with a large surface-to-volume ratio, having a great impact on the gas sensing characteristics of gas sensor systems and devices.
In turn, a second component of XPS O1s line located at 532.8 eV, evidently lower in intensity, can be attributed to the presence of loosely bounded oxygen on the surface of ZnO belonging to specific species, such as adsorbed H 2 O, as well as adsorbed hydroxyl groups. A similar component was recently observed for ZnO thin films by Armelao et al [55], Hsieh et al [60], Lupan et al [61], Stambolova et al [62], as well as for low dimensional ZnO nanostructures by Guo et al [54], Lee et al [63], Bai et al [64], Kim et al [65] and Kicir et al [66].
Subsequently, a third component of XPS O1s line located at 533.9 eV, evidently higher with respect to the previous one, can be attributed to the adventitious carbon dioxide O-C = O surface bondings existed at the surface of ZnO NWs. Finally, the fourth wide component of XPS O1s line located at 534.6 eV, having the smallest amplitude, can be attributed to the carboxyl acid COOH groups adsorbed at the surface ZnO nanostructures. The similar component(s) were observed for ZnO thin films by Lupan et al [61], and for the porous ZnO single crystals by Yao et al [67].
Only three among the above mentioned XPS O1s line components are clearly visible after the TPD process. Apart from the main XPS O1s component located at 531.2 eV attributed to the presence of partially reduced ZnO as ZnO x and directly related to the existence of O 2 ions in the oxygen-deficient regions, there are two XPS O1s components related to existence of surface contaminations at the surface of ZnO NWs. As before, a second component of XPS O1s line located at 532.8 eV, evidently lower in intensity, can be attributed to the presence of adsorbed hydroxyl groups. A third wide component of XPS O1s line located at 534.6 eV, having the smallest amplitude, is also observed, which can be attributed to the carboxyl acid COOH groups adsorbed at the surface ZnO nanostructures. What is crucial after TPD process the XPS O1s line component at the binding energy of 533.9 eV, attributed to the adventitious carbon dioxide O-C = O surface bondings is not observed. This is probably related to the fact that during the TPD process there is an evident desorption of C contaminations in form of CO 2 , causing an increase of the contribution from loosely bounded oxygen in adsorbed hydroxyl groups. These information on the behaviour of XPS O1s components are in good correlation with the information related to the evolution of XPS C1s peaks and their potential components presented in figure 7.
As was already mentioned, after the TPD process the contribution of undesired C contaminations drastically decreased is confirmed by the fact that the signal-to-noise (S/N) ratio of XPS C1s line was at the level of 3.
Moreover, for the ZnO NWs before TPD process, contrary to XPS O1s lines, the XPS C1s line exhibits only a weak asymmetry. Nevertheless, it is surely related to the existence of chosen forms of carbon bondings at surface, which have been recognized after the deconvolution of XPS C1s lines using Gauss fitting. The obtained results are shown in figure 8. Thanks to the deconvolution procedure, with high line fitting parameter (RMS~0.98), it is evident that the XPS C1s line exhibits three components clearly visible at binding energy of 286.1 eV, 287.8 eV, and 288.8 eV, respectively.
The first one, which can be treated as the main peak located at 286.1 eV, can only be attributed to C-O type surface bondings such as C-OH, commonly observed at the surface of various form of oxides, in the well-known reference book [49]. This is in a good correlation with the fact that adsorbed hydroxyl groups at surface of ZnO NWs have been recognized also in the XPS O1s line(s), as mentioned. Other components (of evidently lower amplitude), located at the binding energy 287.  Because, as was mentioned above, after TPD process the signal-to-noise ratio was 3, the deconvolution procedure would be doubtful. Nevertheless, it is evident that only the main peak located at 286.1 eV can be recognized, attributed to the hydroxyl surface bondings such as C-OH. The existence of these C-OH surface bondings can strongly affect the variation of local surface chemistry of our ZnO NWs for their potential sensor application.
In addition to the above analysis, the adsorbed adventitious carbon dioxide CO 2 combined with carboxyl acid COOH groups adsorbed at the surface of our ZnO NWs have been recognized from the respective XPS O1s line at the binding energy~534 eV, as for ZnO films deposited by atomic layer deposition (ALD) [68].
All the obtained XPS information on the local surface chemistry (stoichiometry/nonstoichiometry combined with undesired surface contaminations) of ZnO NWs before and after the TPD process are in good correlation with the respective information obtained by our complementary studies performer by TDS methods. The TDS spectra of main residual gases desorbed from the investigated ZnO NWs after their exposure to air atmosphere are displayed in figure 9 which for the clarity have been additionally corrected by the ionization probability of respected gases detected in our experiments.
Based on results of our TDS studies shown in figure 9 it is clear that mostly the molecular hydrogen (H 2 ) was desorbed during TPD process with the highest relative partial pressure (in the range of 10 -6 ÷ 10 -7 mbar), starting already even below 100 • C, with an almost flat trend in the temperature range 150 ÷ 350 • C. This is probably related to the high degree of crystalline form of ZnO NWs, because such forms of the molecular hydrogen, in general, can penetrate their subsurface region, especially at lower temperature. To the best of our knowledge, this has never been reported for ZnO NWs in the available literature.
In turn, rather only a very small amount of the molecular oxygen (O 2 ) desorbs from our ZnO NWs at the relative  Figure 9. TDS spectra of main residual gases desorbed from the ZnO NWs exposed to air.
partial pressure close to 10 -9 mbar, also starting even below 100 • C and having evident maximum at about 165 • C. It means that probably the residual oxygen from air atmosphere is only physically adsorbed at the surface of ZnO NWs. Nevertheless, the relative [O]/[Zn] concentration after the TDS process only slightly decreased, as evidenced from our XPS measurements, which will be discussed later on. At this moment it should be underlined, what is commonly known, that water vapour (H 2 O), and various carbon oxides (CO x ) are the main residual gases, playing an extremely important and uncontrolled role in the gas sensing mechanism [5,6]. This is why in our TDS studies a special attention was given on the detection and analysis of these last two residual gas components.
From our TDS spectra shown in figure 9, it is clear that H 2 O desorbs from our ZnO NWs at the partial pressure in the range of 10 -6 ÷ 10 -7 mbar, also starting below 100 • C, with a rather oscillating tendency in temperature range 150 ÷ 350 • C. These oscillations correspond to the three local maxima observed at the TDS spectrum at~170 • C, 250 • C, and 320 • C. It means that there are different forms of the hydroxyl groups at the surface of crystalline ZnO NWs.
In turn, the carbon oxides (CO x ), in our TDS studies only carbon dioxide CO 2 was detected, as shown in the respective TDS spectrum. The relative partial pressure also changed in the similar range of 10 -6 ÷ 10 -7 mbar, as for the case of H 2 O desorption, starting already below 100 • C with the almost constant tendency in the temperature range of 150 ÷ 350 • C, having the maximum at about 170 • C.
In relation to the above TDS information it is clear that the effect of desorption of H 2 O and CO 2 residual gases was in a quite good correlation with the variation of relative concentration of the main elements at the surface of ZnO NWs, leading to an almost complete vanishing of the various forms of carbon C contaminations, as determined in our XPS experiments.
At this moment it should be additionally underlined that thanks to the removal of C contaminations, the surface of ZnO NWs became more nonstoichiometric, what is of great importance when using our samples for the detection of oxidizing gases [5,6]. It directly confirms the possibility to completely remove the C contaminations during heating processes from the surface of ZnO NWs, which can not only allows their better sensitivity, but also the shorter response/ recovery time(s), being still one of the most critical limitation of the conductometric metal oxides gas sensors even based on nanostructures, including ultrathin films and nanolayers [5,6].

Conclusions
Thanks to comparative studies, the local surface properties of ZnO nanowires deposited by evaporation-condensation method on Ag-covered Si substrate were investigated. It was shown from our SEM studies that ZnO NWs are mostly isolated and irregular, having an average length of µm and diameters at the level of tens nm. In turn, from our XPS studies it was confirmed that ZnO NWs are slightly nonstoichiometric, additionally covered by the undesired various forms of C contaminations, including C-OH species, that can be almost completely removed during the subsequent TPD process. This information is of great importance because the undesired C contaminations play the role of undesired barriers for the gas adsorption, and can strongly affect the uncontrolled sensor ageing effect, what still remains unclear and requires additional further studies.