Surface and bulk properties of sputter deposited undoped and Sb-doped SnO2 thin films

Sensors and Actuators B 139 (2009) 665–672

Contents lists available at ScienceDirect

Sensors and Actuators B: Chemical journal homepage: www.elsevier.com/locate/snb

Surface and bulk properties of sputter deposited undoped and Sb-doped SnO2 thin films Christoph Körber ∗ , Péter Ágoston, Andreas Klein Institute of Materials Science, Darmstadt University of Technology, D-64287 Darmstadt, Germany

a r t i c l e

i n f o

Article history: Received 5 February 2009 Received in revised form 19 February 2009 Accepted 31 March 2009 Available online 8 April 2009 PACS: 82.65.+r 73.30.+y 79.60.-i 82.80.Pv Keywords: Tin oxide Doping Sputter deposition Thin films Surface potentials Photoelectron spectroscopy Electrical transport measurement

a b s t r a c t Electronic surface and bulk properties of sputter deposited polycrystalline intrinsic and Sb-doped SnO2 thin films have been investigated by a combination of in situ photoelectron spectroscopy, electrical fourpoint conductivity, and optical transmission measurements. The work function and ionization potential of the polycrystalline films increase with increasing oxygen content in the sputter gas by ∼1.4 and ∼1 eV, respectively. The changes are explained by the different surface terminations known for single crystalline SnO2 . Comparison of surface and bulk Fermi level positions indicates flat band situation for most cases but the presence of a depletion layer for Sb-doped films deposited under oxidizing conditions. Large changes of electrical conductivity depending on the oxygen content in the sputter gas were observed for undoped SnO2 which can be understood in terms of different concentrations of oxygen vacancies. In contrast, literally no changes occur for SnO2 :Sb, which is attributed to the too high formation energy of compensating defects like oxygen interstitials or Sn vacancies. © 2009 Elsevier B.V. All rights reserved.

1. Introduction Transparent conducting oxides (TCOs) like SnO2 , In2 O3 and ZnO, have a wide range of applications in optoelectronic devices, owing to their unique combination of high electrical conductivity and optical transparency in the visible spectrum of the light [1–3]. The requirements for the use of a TCO as an electrode material in optoelectronic devices in brief are optical transparency, high conductivity (high dopability), partially also smooth surfaces and a possibility to tune the work function. The most popular TCO, used for transparent electrodes, with conductivities up to 104 S/cm, is Sn-doped In2 O3 , also referred to as indium tin oxide (ITO) [2,3]. The drawback of ITO is the high price of In, which fortifies the demand on In-free TCOs. Inexpensive alternative TCOs for optoelectronic applications are Al-doped ZnO (ZnO:Al) and F-doped SnO2 (SnO2 :F). Conductivities > 103 S/cm can be accomplished with SnO2 :F thin films, prepared via chemical vapor deposition (CVD), using HF for doping [2,4]. Very rough sur-

∗ Corresponding author. Tel.: +49 6151169661. E-mail address: [email protected] (C. Körber). 0925-4005/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.snb.2009.03.067

faces are obtained from the CVD process. This limits the application of SnO2 :F for organic light emitting devices, where smooth surfaces are required in order to minimize the electrical field at the TCO/organic interface during operation. Sputter deposition of TCO thin films from ceramic targets provides a method to coat large areas and is therefore highly attractive for commercial applications. Furthermore, this technique often leads to smooth surfaces with a typical rms roughness of ∼1–2 nm [5–7]. The most common dopant used for sputter deposited SnO2 thin films is Sb (SbSn ), since doped ceramic targets can be readily used and smooth surfaces are obtained. However, Sb-doped SnO2 , so far, has been prevailingly of scientific interest, since its conductivities are in general one order of magnitude smaller than those of CVD SnO2 :F [2,4,8]. The smaller conductivities are most often attributed to a segregation of Sb3+ at internal and external interfaces, respectively [9–15]. An important issue for the application of TCOs in optoelectronic devices is the electronic lineup at the TCO/active material interface. For the application in organic light emitting diodes, e.g., the work function of the TCO is crucial for the hole injection into the organic conductor and has been extensively investigated for In2 O3 and ITO (see, e.g., Refs. [16–21]) as well as ZnO and ZnO:Al (see Ref. [22] and

666

C. Körber et al. / Sensors and Actuators B 139 (2009) 665–672

references therein). The work function of a semiconductor can be affected by changes of the Fermi level position and surface dipole, respectively. The latter corresponds to a change in the ionization potential IP , which is defined as the energy difference between the vacuum level and the valence band maximum. Sputter deposition of TCO thin films provides a possibility to affect the Fermi level position and surface dipole, respectively, by adding reactive gases like, e.g., O2 to the process gas, allowing for engineering of TCO properties [16,22–25]. SnO2 is also a very important material used for semiconductor based gas sensors (Taguchi sensors) and has been extensively studied over the past decades [26–33]. The working principle of SnO2 based gas sensors is the change of resistivity in dependence on atmospheric conditions. A well accepted model for the change of resistivity is the modification of band bending in the near-surface region, induced by adsorption/desorption of gas molecules at the oxide surface [29,30,32,34]. For a n-type semiconductor like SnO2 , the adsorption of, e.g., oxygen, which involves the formation of negative ions (e.g., O2 − ), leads to an upward band bending (depletion layer) at the surface and that effect changes the potential barrier of intergrain junctions in Taguchi sensors [29,30,32,34,35]. Hence, the resistivity of the SnO2 gas sensor is increased in the near-surface region. Photoelectron spectroscopy (PES) allows to measure the Fermi level position of a material at the surface [36]. Furthermore, surface potentials such as work function and ionization potential can be directly determined with this technique. In contrast to the highly surface sensitive PES, electrical and optical measurements provide information about the electronic bulk properties of materials. However, the position of characteristic energy levels and free carrier concentrations are most often not directly accessible from electrical and optical data. In this work we combine surface sensitive PES with bulk sensitive electrical and optical measurements in order to determine surface potentials and reveal the defect structure (doping mechanism) of sputter deposited polycrystalline SnO2 and SnO2 :Sb thin films, depending on the deposition conditions. We show, that the Fermi level position and surface dipole, i.e., work function and ionization potential, of SnO2 and SnO2 :Sb can be systematically controlled by changing the oxygen content in the sputter atmosphere. Bulk sensitive electrical conductivity and optical transmission measurements suggest different defect mechanisms for undoped and Sb-doped SnO2 . Furthermore, a depletion layer at the surface of oxidized SnO2 :Sb is revealed in contrast to flat band conditions in the case of undoped SnO2 and reduced SnO2 :Sb.

2. Experimental Thin film deposition and photoelectron spectroscopy were performed at the DArmstadt Integrated SYstem for MATerial research (DAISY-MAT). The system provides a central distribution chamber, which combines a Physical Electronics PHI 5700 multi-technique surface analysis system with several preparation chambers, allowing for rapid sample transfer between preparation and analysis chambers without breaking ultrahigh vacuum (UHV) conditions [36,37]. X-ray photoelectron spectra were recorded using monochromated Al K␣ radiation with an energy resolution of ∼0.4 eV, determined from the Gaussian broadening of the Fermi edge emission of a sputter-cleaned Ag sample, which was also used to calibrate the binding energy scale. UV photoelectron spectra were acquired under normal emission using He I radiation from a gas discharge lamp with an energy resolution of less than 0.2 eV. A sample bias of −1.5 V was applied while recording the UV spectra. SnO2 and Sb-doped SnO2 (3 wt% Sb2 O5 ) thin films were deposited from ceramic targets with diameters of 2 in and purity

of 99.99%, purchased from Kurt J. Lesker (Hastings, England) and MaTeck (Jülich, Germany), respectively. Pure Ar or Ar/O2 mixtures were used as sputter gases. Defined Ar/O2 ratios were established by means of mass flow controllers (MKS, Esslingen, Germany). All films were prepared via radio-frequency (RF) magnetron sputtering on quartz glass slides. A power of 25 W and target-tosubstrate distance of 7.5 cm were employed. The total gas pressure during sputtering in the deposition chamber was 0.5 Pa in all cases. During deposition, substrates were heated using a home-made heater with a commercial halogene lamp and a tantalum reflector shield. Temperature was monitored using two independent thermocouples attached to the sample holder. A substrate temperature of 400 ◦ C was used as a standard for the deposition of the thin films. The base pressure of the deposition chamber was 10−6 Pa. Four-point conductivities were measured at room temperature with linear contact geometry using spring-loaded probes gently pressing on the film surfaces. Currents were applied via a Keithley Precision Current Source (model 6220) and voltages measured with a Keithley Multimeter (model 2700). Different spots on the thin films were probed and various currents applied in order to determine an average conductivity for each specimen. Film thicknesses were determined from optical spectroscopy and range between approx. 300 and 700 nm. Transmission spectra were recorded using a PerkinElmer Lambda 900 UV/VIS/NIR spectrometer, providing a wavelength range from 180 to 3000 nm. Baseline corrections were applied from the transmission spectrum of an uncoated quartz substrate.

3. Photoemission results Fig. 1 shows core level and valence band spectra of undoped and Sb-doped SnO2 thin films, deposited with different oxygen contents in the sputter gas (xO2 ). All films were grown at 400 ◦ C substrate temperature. Wide range survey spectra (not shown), recorded with monochromatic Al K␣ radiation, revealed no contamination of the deposited thin films. The Sn 3d5/2 and O 1s core levels for both undoped and Sb-doped SnO2 are shifted to lower binding energies with increasing xO2 . For the undoped films, the core level emissions show a symmetric line shape. The Sn 3d5/2 emissions of the SnO2 :Sb films are asymmetric for low oxygen contents in the sputter gas, due to an additional component at the high binding energy side. This effect is attributed to a screening effect, measured with photoemission, and has been already described for highly doped SnO2 :Sb ceramics [38,39] and ITO thin films [40]. The Sb 3d5/2 emissions of the SnO2 :Sb films overlap with the O 1s emissions, which complicates the interpretation of those spectra. However, the 3d3/2 emissions of Sb are clearly resolved. Due to the low concentration of Sb in the SnO2 :Sb films, the signal-tonoise ratio is rather small. Yet, the Sb 3d3/2 emission lines appear relatively broad, suggesting that more than one component contribute to the overall line shape. Atomic sensitivity factors, which are required for the calculation of the Sb concentration, are only available for the Sb 3d5/2 emission line. Therefore, the intensity of the 3d5/2 emission was calculated by multiplying the intensity of the 3d3/2 emission with the branching ratio of 1.5, according to the degeneracy of the 3d levels. Correspondingly, the intensity of the Sb 3d5/2 emission was subtracted from the O 1s signal in order to determine the O intensity. With increasing xO2 , a slight increase of O concentration vis-àvis with a decrease of Sn concentration at the surface is observed from quantitative analysis of the XP core level spectra (see Fig. 2). In the case of Sb-doped SnO2 , no variation of Sb concentration, within the experimental limits of quantitative XPS analysis (5–15% relative error), is detected. The Sb concentration is consistent with the

C. Körber et al. / Sensors and Actuators B 139 (2009) 665–672

667

Fig. 1. Core level and valence band spectra of undoped (SnO2 ) and Sb-doped SnO2 (SnO2 :Sb) thin films, recorded with monochromatic Al K␣ radiation. Sb 3d3/2 core level spectra are only displayed for the doped specimens. The oxygen content in the sputter gas during deposition is denoted for each film. The binding energy of the valence band maximum (EVBM ) was determined from the intersection of the linear extrapolations of the leading edge and the background in the valence band spectra (see VB spectrum of SnO2 film with 20% O2 in sputter gas).

doping level of the SnO2 :Sb target, used for thin film deposition. All SnO2 and SnO2 :Sb films appear to be oxygen deficient, even at high oxygen contents in the sputter gas during deposition. However, the formation of SnO instead of SnO2 , that could be in principle possible under reducing conditions, does not occur. The derivation of the observed stoichiometry from the expected 2:1 ratio might, however, also be related to improper sensitivity factors. XP valence band spectra of SnO2 and SnO2 :Sb films (Fig. 1) deposited under reducing conditions (i.e., no or only few oxygen in the sputter gas) show additional emissions in the bandgap with an onset at approx. 2 eV binding energy. With increasing xO2 , their intensity is decreased. Furthermore, the binding energies of the valence band maximum, which is determined from the intersection of the linear extrapolations of the leading edge and the background of the VB spectra, are shifted to lower values. Except for the additional emissions within the bandgap of the material, the line shape of the VB spectra is not affected by the xO2 , as confirmed from VB difference spectra (not shown). The additional

Fig. 2. Surface composition of SnO2 (circles) and SnO2 :Sb (triangles) thin films vs. oxygen content in the sputter gas, determined from quantitative analysis of XP core level spectra.

emissions in the bandgap of SnO2 have been already reported by various authors [38,41–45]. Model experiments and DFT calculations on the SnO2 (1 0 1) surface revealed, that those bandgap states are attributed to Sn2+ at the surface [42,46]. Reducing conditions lead to a removal of the bridging and some of the in-plane oxygen, hence reducing the oxidation state of the Sn surface atoms. Oxidizing conditions, however, produce stoichiometric surfaces with Sn in its 4+ oxidation state, according to the bulk of the material. Corresponding results were also reported for the SnO2 (1 1 0) surface [42,47,48]. With increasing oxygen content in the sputter gas, a more stoichiometric SnO2 surface is prepared and the bandgap emissions are strongly depressed. This also explains the increase of oxygen concentration with increasing xO2 . The data in Fig. 2 indicate, that a fully oxidized surface is obtained for xO2  5%. UP spectra, recorded with monochromatic He I radiation, of SnO2 :Sb thin film specimens are displayed in Fig. 3. The films were deposited with different xO2 , as denoted. With increasing oxygen content, the secondary electron cut-off (see Fig. 3(left)) is shifted to lower binding energies, i.e., the work function of the material is increased. The main peak in the VB emissions (see Fig. 3(center)) is attributed to O 2p derived states [49]. With increasing xO2 , this peak and the valence band maximum binding energy are shifted to lower values. At the same time, the intensity of the emissions within the bandgap of the material decreases, as also observed in the XP valence band spectra. The films, deposited under reducing conditions, reveal a distinct Fermi edge in the UP spectra (see Fig. 3(right)). The intensity of the Fermi edge emissions decreases with increasing oxygen content. Fig. 4 summarizes the effect of variations of xO2 on the surface potentials of the SnO2 and SnO2 :Sb thin films, measured by photoemission. Both, core level and valence band maximum binding energies are shifted parallel by variation of xO2 . Taking a bandgap of 3.54 eV for SnO2 [50] into account, Sb-doped films, prepared under reducing conditions, reveal a Fermi level position above the conduction band minimum and high core level binding energies, corresponding to degenerate doping. Films deposited under oxidizing conditions (approx. 5% and higher xO2 ) reveal low core level binding energies and a Fermi level position in the bandgap. The undoped SnO2 films show similar shifts of binding energies. However, their core level binding energy is always smaller than for the Sb-doped films and the Fermi level is situated within the bandgap. A strong effect of xO2 variations is observed for the work function and ionization potential of both SnO2 and SnO2 :Sb films (see Fig. 4). The work function is increased by ∼1.4 eV by changes of xO2

668

C. Körber et al. / Sensors and Actuators B 139 (2009) 665–672

Fig. 3. Ultraviolet photoelectron spectra, recorded with monochromatic He I excitation, of SnO2 :Sb thin films, deposited with different oxygen contents in the sputter gas, as denoted. The spectra show the secondary electron cut-off (left), valence band (center) and a magnified presentation of the Fermi energy regime (right).

from 0 to 25%. At the same time the ionization potential increases by ∼1 eV. The changes mostly occur for xO2 between 0 and 5%. The same correlation, though the absolute changes are only ∼0.4 eV, is observed for the Fermi level position. This behavior is quite different than observed for, e.g., ITO, where work functions shifts equal Fermi level shifts and no change of the ionization potential is observed with a variation of xO2 during sputter deposition [24,25,40]. 4. Electrical and optical properties Conductivities of sputter deposited undoped and Sb-doped SnO2 thin films vs. oxygen content in the sputter gas are provided in Fig. 5. Undoped SnO2 thin films, prepared under reducing conditions, i.e., no or only few oxygen in the sputter gas, reveal conductivities of ∼0.5 S/cm. Addition of 2% O2 to the sputter gas leads to a drop in conductivity by several orders of magnitude to ∼3 × 10−5 S/cm. Owing to the experimental limitations of the used 4-point conductivity setup, no conductivities could be determined

Fig. 4. Fermi level position with respect to the valence band maximum, work function, and ionization potential of sputter deposited SnO2 and SnO2 :Sb thin films vs. oxygen content in the process gas. The bold lines represent the general trend, rather than a fit of the data points. The position of the conduction band minimum is indicated by the dotted line.

Fig. 5. Conductivities of undoped (circles) and Sb-doped (triangles) SnO2 thin films vs. oxygen content in the sputter gas.

C. Körber et al. / Sensors and Actuators B 139 (2009) 665–672

Fig. 6. Conductivity of undoped (circles) and Sb-doped (triangles) SnO2 thin films vs. Fermi level position with respect to the valence band maximum. The marker size corresponds to the oxygen content in the sputter gas (scale different for SnO2 :Sb and SnO2 ). The line represents calculated theoretical values assuming a constant carrier mobility of 20 cm2 /V s, an effective electron mass of 0.3me , a fundamental bandgap of SnO2 of 3.54 eV, and parabolic conduction band dispersion. The experimental error is indicated by the cross-hair. The position of the conduction band minimum is represented by the dotted line.

for undoped SnO2 films, prepared with more than 2% oxygen in the sputter gas. Their conductivity, however, has to be smaller than ∼10−5 S/cm. Reduced Sb-doped thin films show conductivities of ∼280 S/cm, which is two orders of magnitude larger than in the case of undoped SnO2 . Hence, Sb-doping is promoting the conductivity of the SnO2 thin films. A slight decrease in conductivity of approx. one order of magnitude to ∼20 S/cm is visible for high oxygen contents of 20–25%, whereas for oxygen contents of less than ∼10%, almost no change of conductivity is observed. Yet, even very high oxygen contents in the sputter gas do not depress the conductivity of Sb-doped SnO2 to the value of undoped SnO2 in contrast to the observations for ZnO:Al and ITO [23–25]. Fig. 6 provides conductivities of different SnO2 and SnO2 :Sb thin films plotted vs. Fermi level position with respect to the valence band maximum determined from X-ray photoelectron spectroscopy. Theoretical conductivities were deduced from numerically calculated carrier concentrations and a constant mobility of 20 cm2 /V s, which is an intermediate value for polycrystalline SnO2 [2]. Carrier concentrations in dependence on Fermi level position were calculated according to:



∞

n=

NCB (E)f (E) dE

(1)

ECBM

where ECBM is the effective density of states in the conduction band and f (E) the Fermi function. An effective mass of 0.3me [50], a fundamental bandgap of SnO2 of 3.54 eV [50], and parabolic conduction band dispersion were assumed for the calculations. High Fermi level positions above the conduction band minimum vis-à-vis with high conductivities, are in general visible for reduced SnO2 :Sb specimens. The conductivities of those films agree fairly well with the trend of the calculated values. In the case of SnO2 :Sb, a strong deviation between the calculated and measured conductivities is observed for oxidized specimens. The latter show a Fermi level position below the conduction band minimum. Reduced undoped SnO2 films with a comparable Fermi level position to oxidized SnO2 :Sb films show an almost three orders of magnitude smaller conductivity. The conductivities of undoped SnO2 speci-

669

Fig. 7. Optical transmission spectra of undoped (circles) and Sb-doped (triangles and squares) SnO2 thin films. The composition of the sputter gas is denoted in the figure. Film thicknesses amount to 371 nm (triangles), 368 nm (squares) and 740 nm (circles), respectively.

mens are strongly dependent on Fermi level position according to the theoretical results. Optical transmission spectra of undoped and Sb-doped SnO2 thin films on quartz glass substrates are displayed in Fig. 7. Different sputter gas compositions were used in the case of SnO2 :Sb, as denoted in the figure. Virtually no differences between the transmission spectra of reduced and oxidized SnO2 :Sb are visible. No shift of the leading edge at short wavelengths, representing the onset of the optical band-to-band absorption (absorption edge) is observed. The reduced transmission at long wavelength in the case of SnO2 :Sb is attributed to the high carrier concentration in the material, which leads to excitation of plasmons. Also at this plasmon edge, no differences between oxidized and reduced SnO2 :Sb can be detected. The comparable absorption edge and plasmon energies of SnO2 :Sb deposited with xO2 = 0 and 5% agree with the electrical measurements, which yielded approximately similar conductivities. The absorption edge of undoped SnO2 is located at longer wavelengths compared to SnO2 :Sb. The blue-shift of the absorption edge for the doped materials is attributed the higher free carrier concentration and is well-known as the Burstein–Moss-shift [51,52]. Only transitions from occupied valence to unoccupied conduction band states are possible (renormalization of the fundamental bandgap may be neglected, here). The lower carrier concentration of the undoped specimen is also reflected by the missing plasmon edge. Again, this is in agreement with the much lower conductivities of undoped SnO2 films compared to SnO2 :Sb (see also Fig. 6). 5. Discussion 5.1. Surface potentials A strong increase of work function and ionization potential of the SnO2 and SnO2 :Sb thin films with increasing oxygen content in the sputter gas is observed. The work function of a material can be affected by both variations of the Fermi level position and the surface dipole, whereas changes of the ionization potential are exclusively caused by surface dipole modifications [25]. One effect, contributing to an increase of work function of the presented films, is the decrease of Fermi level position with respect to the valence

670

C. Körber et al. / Sensors and Actuators B 139 (2009) 665–672

band maximum, as observed from binding energy shifts of core level and valence band emissions. The second contribution is due to the oxidation of the SnO2 surface with increasing oxygen content in the sputter gas, which is changing the surface dipole and thus affecting the ionization potential and work function. SnO2 and SnO2 :Sb films, prepared under low oxygen partial pressure, show reduced surfaces with effective Sn2+ oxidation states, causing photoelectron emissions in the bandgap of the material. Owing to the relatively small binding energy shift between Sn2+ and Sn4+ and the too small surface sensitivity of XPS, the two oxidation states cannot be resolved in Sn core level emissions [53]. However, evidence for the oxidation/reduction of the SnO2 and SnO2 :Sb surfaces is provided by the changes of surface composition vs. oxygen partial pressure in the sputter gas, obtained from quantitative XPS analysis (see Fig. 2). This is in good agreement with previous results on polycrystalline SnO2 films [54]. Furthermore, this is also consistent with PES measurements and calculations on SnO2 single crystal surfaces with well defined surface terminations [42,43,45,46,48]. Films, prepared under oxidizing conditions, do not show emissions in the bandgap. All surface Sn atoms are fully oxidized to Sn4+ , leading to an increases of work function and ionization potential by ∼1.4 and ∼1 eV, respectively. A corresponding influence of oxidized and reduced surface terminations on surface potentials, i.e., surface dipole modifications, has also been observed for sputter deposited polycrystalline In2 O3 and ITO surfaces [16,25]. However, in this case, changes of the surface dipole could not be introduced by variations of oxygen content in the sputter gas. Annealing under highly oxidizing conditions was required in order to increase the surface dipole. Absolute values for work functions of ∼5.7 and ∼4.7 eV are available for the oxidized and reduced SnO2 (1 0 1) surfaces, respectively [42,46]. For the (1 1 0) surface, only relative changes in work function between the oxidized and reduced surface termination are reported which range between 0.5 eV [55], 0.6–1.2 eV [56] and 0.7–1.2 eV [43]. These values are in good agreement with out values for polycrystalline films. In addition, changes of surface dipoles with crystallographic orientation of the surface, which are well known for ZnO [22,25,57,58], GaAs [59], and CdS and CdTe [60], have yet not been reported for semiconductors with inversion symmetry as SnO2 . The observed changes in ionization potential are therefore only assigned to changes of surface termination as reported for single crystal SnO2 [42,43,45,46,48].

5.2. Defect mechanism Sb-doping of SnO2 is accomplished by the substitution of Sn sites in the rutile SnO2 lattice with Sb atoms (SbSn ), as was also confirmed by X-ray absorption measurements [12,61,62] and Mössbauer spectroscopy [63]. Sb-doping leads to shallow defect levels close to the conduction band minimum of SnO2 [64,65]. Sb can only act as a donor in SnO2 in its 5+ oxidation state. However, it can also be incorporated into SnO2 as Sb3+ , which results in deep trap levels for conduction electrons [9,11,13]. Due to its much larger size compared with Sn4+ of the host lattice, Sb3+ is preferentially located at grain boundaries or the surface of SnO2 :Sb [9–15]. The dual valency of Sb is also reflected by the broad Sb 3d3/2 emission lines of the SnO2 :Sb thin films, presented in this study. Owing to the weak intensity and poor signal-to-noise ratio of the Sb 3d3/2 emissions, we did not attempt to quantify the Sb5+ /Sb3+ ratio. However, the Sb concentrations, determined from XPS measurements, are in good agreement with the initial Sb concentration of the ceramic SnO2 :Sb target, used for sputter deposition. The used target composition of 3 wt% Sb2 O5 corresponds to a density of Sb atoms of ∼8 × 1020 cm−3 . With a moderate mobility of 20 cm2 /V s [2] a conductivity of 2600 S/cm would result.

In contrast, the highest observed conductivity of the investigated SnO2 :Sb films is approx. one order magnitude lower. However, the observed conductivities agree with conductivities reported for SnO2 :Sb films, prepared with similar techniques [2,4,8]. It is therefore likely that the poor conductivities of the SnO2 :Sb films are associated with the presence of Sb3+ species. Hence, the doping efficiency of Sb is only ∼10% for the investigated specimen. In comparison, much higher conductivities of approx. 5000 S/cm for F-doped SnO2 films, can be achieved by CVD processing with HF as precursor [2]. The conductivity of the presented SnO2 :Sb films is virtually not affected by changes of the oxygen content in the sputter gas (see Fig. 6). However, tremendous changes are observed for undoped SnO2 . The different behavior towards changes in oxygen partial pressure during deposition are attributed to different prevailing defect mechanism in Sb-doped and undoped SnO2 , respectively. In the case of the intrinsic material, n-type conductivity is generated by an inherent oxygen deficiency of SnO2 , leading to doubly ionized oxygen vacancies (VO ) in the crystal lattice [66,67]. With increasing oxygen partial pressure in the sputter atmosphere, SnO2 is oxidized, i.e., the concentration of VO , and thus of free electrons, is decreased. Fonstad and Samson [67] observed corresponding changes of conductivity after annealing of undoped SnO2 single crystals at elevated temperatures under different oxygen partial pressures. Almost no change of conductivity with xO2 is observed for SnO2 :Sb. This is in strong contrast to the behavior observed for ZnO:Al [22,25] and ITO [23–25]. The pronounced reduction of conductivity with increasing oxygen pressure can be explained by the formation of compensating defects, which can be identified as zinc vacancies and oxygen interstitials for ZnO:Al and ITO, respectively [68]. Apparently, such compensating defects remain energetically unfavorable in doped SnO2 , even at high oxygen pressures. This is in agreement with available defect calculations for SnO2 [69]. 5.3. Surface vs. bulk properties As discussed in the previous section, the electrical conductivity and optical bandgap of SnO2 :Sb are literally not affected by variations of the oxygen partial pressure in the sputter atmosphere. This can be attributed to the high formation energies of compensating acceptor defects. It is thus evident that the bulk Fermi level position of SnO2 :Sb changes only slightly with xO2 . However, photoelectron spectroscopy on SnO2 :Sb surfaces reveal a clear decrease in surface Fermi level position with increasing oxygen content in the sputter gas. Moreover, the line shapes of core level emissions suggest high carrier concentrations for reduced and low for oxidized SnO2 :Sb specimens [38,39]. In addition, UP spectra, show a distinct Fermi edge for reduced SnO2 :Sb specimens, which is a clear indication of a high carrier concentration at the surface. With increasing oxygen content in the sputter gas, the intensity of those emissions are depressed and vanish eventually for highly oxidized films, which exhibit a surface Fermi level below the conduction band minimum. A possible explanation for the apparent discrepancy between bulk and surface Fermi level position for SnO2 :Sb films, deposited with higher xO2 , could be the formation of a depletion layer (band bending) at the oxidized surface of the SnO2 :Sb films. This would result in a low Fermi level position at the surface, as determined from surface sensitive PES measurements. The bulk of the material, however, still remains highly doped, i.e., shows a high Fermi level position as observed in electrical conductivity and optical transmission measurements. A model for surface band bending on SnO2 (1 1 0) depending on the oxidation state of the surface was proposed by Cavicchi et al. [43]. In that model, flat band conditions are only observed

C. Körber et al. / Sensors and Actuators B 139 (2009) 665–672

671

pressure in the sputter atmosphere strongly affect the conductivity of undoped SnO2 , literally no effect was observed for SnO2 :Sb. This was attributed to different defect mechanisms governing the defect chemistry of undoped and Sb-doped SnO2 , respectively. Thus SnO2 :Sb lacks compensating acceptor defects even under the most oxidizing conditions, used for sputter deposition of the presented films. A combination of surface sensitive photoelectron spectroscopy and bulk sensitive electrical four-point conductivity and optical transmission measurements suggest a depletion layer at the surface of oxidized SnO2 :Sb. Acknowledgements

Fig. 8. Schematic energy diagrams for SnO2 and SnO2 :Sb films, deposited under oxidizing and reducing conditions, respectively. Bulk and surface Fermi level positions are indicated for each situation.

The support of the German Science Foundation under Grant No. KL1225/4 and in the context of the Sonderforschungsbereich 595 (Electrical Fatigue of Functional Materials) is acknowledged. The authors also want to thank Natalie Gorczak and André Wachau for the theoretical calculation of carrier concentrations. References

for a perfectly oxidized stoichiometric surface. Removal of oxygen at the surface leads to an accumulation layer, i.e., downward band bending. The formation of a depletion layer, as also observed in our investigations for oxidized films, is attributed to additional adsorbed oxygen. The quantitative comparison of bulk and surface Fermi level of the SnO2 :Sb films indicates flat band conditions for deposition with low xO2 and the presence of a depletion layer with a bend bending of ∼0.4 eV for deposition with high xO2 . A comparison of surface and bulk Fermi level positions for the undoped SnO2 films is not directly possible as the optical absorption edge can not be lower than the fundamental band gap. The overall agreement between calculated conductivity and surface Fermi level positions for the undoped films (Fig. 6) suggests, however, flat band conditions in all cases. The presence/absence of a depletion layer for the doped/undoped films can be explained if the energy levels of the acceptor states causing the depletion, e.g., adsorbed O2 − or O2 2 −, are close to the conduction band as a depletion layer is only possible for a higher bulk Fermi level position. Under oxidizing conditions this is the case for SnO2 :Sb but not for SnO2 . The anticipated situation is sketched in Fig. 8. However, an alternative explanation for the different behavior of undoped and Sb-doped SnO2 would be the implantation of high energetic oxygen ions in interstitial positions at the surface, that may occur during sputter deposition [70–72]. With increasing oxygen content in the sputter gas, more oxygen ions are accelerated towards the film surface. Yet, as discussed above, the formation of compensating oxygen interstitials appears to be energetically unfavored in SnO2 . 6. Summary and conclusions Sputter deposited undoped and Sb-doped SnO2 thin films have been investigated, using in situ photoemission spectroscopy, electrical conductivity, and optical transmission measurements. A variation of oxygen partial pressure during sputter deposition strongly affects the work function and ionization potential of intrinsic and Sb-doped SnO2 . That effect was attributed to an oxidation/reduction of the SnO2 surface. The observed doping efficiency of Sb in SnO2 is rather low, i.e., the actual electron concentration is approx. one order of magnitude smaller than the nominal dopant concentration. This is tentatively attributed to the occurrence of Sb3+ , incorporated into SnO2 instead of Sb5+ . Yet, surface enrichment of Sb was not observed. Whereas changes of oxygen partial

[1] D.S. Ginley, C. Bright, Transparent conducting oxides, MRS Bulletin 25 (2000) 15–18. [2] H.L. Hartnagel, A.L. Dawar, A.K. Jain, C. Jagadish, Semiconducting Transparent Thin Films, Institute of Physics Publishing, Bristol, 1995. [3] T. Minami, Transparent conducting oxide semiconductors for transparent electrodes, Semiconductor Science and Technology 20 (2005) 35–44. [4] B. Stjerna, E. Olsson, C.G. Granqvist, Optical and electrical properties of radio frequency sputtered tin oxide films doped with oxygen vacancies, F, Sb, or Mo, Journal of Applied Physics 76 (1994) 3797–3817. [5] X.T. Hao, L.W. Tan, K.S. Ong, F.R. Zhu, High performance low temperature transparent conducting aluminum-doped ZnO thin films and applications, Journal of Crystal Growth 287 (2006) 44–47. [6] L. Kerkache, A. Layadi, E. Dogheche, D. Remiens, Physical properties of rf sputtered ITO thin films and annealing effect, Journal of Physics D—Applied Physics 39 (2006) 184–189. [7] K. Maknys, A.G. Ulyashin, H. Stiebig, A.Y. Kuznetsov, B.G. Svensson, Analysis of ITO thin layers and interfaces in heterojunction solar cells structures by AFM, SCM and SSRM methods, Thin Solid Films 511 (2006) 98–102. [8] K. Suzuki, M. Mizuhashi, Structural, electrical and optical properties of rfmagnetron-sputtered SnO2 –Sb film, Thin Solid Films 97 (1982) 119–127. [9] P.A. Cox, R.G. Egdell, C. Harding, W.R. Patterson, P.J. Taverner, Surface properties of antimony doped tin(IV) oxide: a study by electron spectroscopy, Surface Science 123 (1982) 179–203. [10] V. Dusastre, D.E. Williams, Sb(III) as a surface site for water adsorption on Sn(Sb)O2 , and its effect on catalytic activity and sensor behavior, Journal of Physical Chemistry B 102 (1998) 6732–6737. [11] R.G. Egdell, W.R. Flavell, P. Tavener, Antimony-doped tin(IV) oxide—surface composition and electronic structure, Journal of Solid State Chemistry 51 (1984) 345–354. [12] C. McGinley, H. Borchert, M. Pflughoefft, S. Al Moussalami, A.R.B. de Castro, M. Haase, H. Weller, T. Moller, Dopant atom distribution and spatial confinement of conduction electrons in Sb-doped SnO2 nanoparticles, Physical Review B 64 (2001) 2453121–2453129. [13] C.S. Rastomjee, R.G. Egdell, M.J. Lee, T.J. Tate, Observation of conduction electrons in Sb-implanted SnO2 by ultraviolet photoemission spectroscopy, Surface Science 259 (1991) 769–773. [14] B. Slater, C.R.A. Catlow, D.H. Gay, D.E. Williams, V. Dusastre, Study of surface segregation of antimony on SnO2 surfaces by computer simulation techniques, Journal of Physical Chemistry B 103 (1999) 10644–10650. [15] D. Szczuko, J. Werner, S. Oswald, G. Behr, K. Wetzig, XPS investigations of surface segregation of doping elements in SnO2 , Applied Surface Science 179 (2001) 301–306. [16] S.P. Harvey, T.O. Mason, C. Körber, Y. Gassenbauer, A. Klein, Evidence for surface dipole modifications in In2 O3 -based transparent conductors, Applied Physics Letters 92 (2008) 2521061–2521063. [17] I.G. Hill, D. Milliron, J. Schwartz, A. Kahn, Organic semiconductor interfaces: electronic structure and transport properties, Applied Surface Science 166 (2000) 354–362. [18] Th. Kugler, W.R. Salaneck, H. Rost, A.B. Holmes, Polymer band alignment at the interface with indium tin oxide: consequences for light emitting diodes, Chemical Physics Letters 310 (1999) 391–396. [19] B. Lagel, M.M. Beerbom, B.V. Doran, M. Lagel, A. Cascio, R. Schlaf, Investigation of the poly[2-methoxy-5-(2 -ethyl-hexyloxy)-1,4-phenylene vinylene]/indium tin oxide interface using photoemission spectroscopy, Journal of Applied Physics 98 (2005) 023512. [20] M.G. Mason, L.S. Hung, C.W. Tang, S.T. Lee, K.W. Wong, M. Wang, Characterization of treated indium–tin-oxide surfaces used in electroluminescent devices, Journal of Applied Physics 86 (1999) 1688–1692.

672

C. Körber et al. / Sensors and Actuators B 139 (2009) 665–672

[21] K. Sugiyama, H. Ishii, Y. Ouchi, K. Seki, Dependence of indium–tin-oxide work function on surface cleaning method as studied by ultraviolet and X-ray photoemission spectroscopies, Journal of Applied Physics 87 (2000) 295–298. [22] A. Klein, F. Säuberlich, Surfaces and interfaces of sputter-deposited ZnO films, in: K. Ellmer, A. Klein, B. Rech (Eds.), Transparent Conductive Zinc Oxide: Basics and Applications in Thin Film Solar Cells, Springer–Verlag, Berlin, 2008, pp. 125–185. [23] Y. Gassenbauer, A. Klein, Electronic surface properties of rf-magnetron sputtered In2 O3 :Sn, Solid State Ionics 173 (2004) 141–145. [24] Y. Gassenbauer, A. Klein, Electronic and chemical properties of ITO surfaces and ITO/ZnPc interfaces studied in situ by photoelectron spectroscopy, Journal of Physical Chemistry B 110 (2006) 4793–4801. [25] A. Klein, C. Körber, A. Wachau, F. Säuberlich, Y. Gassenbauer, R. Schafranek, Surface potentials of magnetron sputtered transparent conducting oxides, Thin Solid Films, submitted for publication. [26] M. Batzill, Surface science studies of gas sensing materials: SnO2 , Sensors 6 (2006) 1345–1366. [27] W. Göpel, K.D. Schierbaum, SnO2 sensors: current status and future prospects, Sensors and Actuators B, Chemical 26–27 (1995) 1–12. [28] K. Ihokura, J. Watson, The Stannic Oxide Gas Sensor—Principles and Applications, CRC Press, Boca Raton, FL, 1994. [29] J.F. Mcaleer, P.T. Moseley, J.O.W. Norris, D.E. Williams, P. Taylor, B.C. Tofield, Tin oxide based gas sensors, Materials Chemistry and Physics 17 (1987) 577–583. [30] K.D. Schierbaum, H.D. Wiemhöfer, W. Göpel, Defect structure and sensing mechanism of SnO2 gas sensors: comparative electrical and spectroscopic studies, Solid State Ionics 28–30 (1988) 1631–1636. [31] U. Weimar, W. Göpel, AC measurements on tin oxide sensors to improve selectivities and sensitivities, Sensors and Actuators B, Chemical 26 (1995) 13–18. [32] N. Yamazoe, J. Fuchigami, M. Kishikawa, T. Seiyama, Interactions of tin oxide surface with O2 ,H2 O and H2 , Surface Science 86 (1979) 335–344. [33] N. Yamazoe, G. Sakai, K. Shimanoe, Oxide semiconductor gas sensors, Catalysis Surveys from Asia 7 (2003) 63–75. [34] W. Göpel, K.D. Schierbaum, H.D. Wiemhöfer, J. Maier, Defect chemistry of tin(IV)-oxide in bulk and boundary layers, Solid State Ionics 32/33 (1989) 440–443. [35] N. Yamazoe, New approaches for improving semiconductor gas sensors, Sensors and Actuators B, Chemical 5 (1991) 7–19. [36] A. Klein, T. Mayer, A. Thissen, W. Jaegermann, Photoelectron spectroscopy in materials science and physical chemistry, Bunsen-Magazin 10 (2008) 124–139. [37] D. Ensling, A. Thissen, Y. Gassenbauer, A. Klein, W. Jaegermann, In-situ preparation and analysis of functional oxides, Advanced Engineering Materials 7 (2005) 945–949. [38] R.G. Egdell, J. Rebane, T.J. Walker, D.S.L. Law, Competition between initial- and final-state effects in valence- and core-level X-ray photoemission of Sb-doped SnO2 , Physical Review B 59 (1999) 1792–1799. [39] R.G. Egdell, T.J. Walker, G. Beamson, The screening response of a dilute electron gas in core level photoemission from Sb-doped SnO2 , Journal of Electron Spectroscopy and Related Phenomena 128 (2003) 59–66. [40] Y. Gassenbauer, R. Schafranek, A. Klein, S. Zafeiratos, M. Hävecker, A. KnopGericke, R. Schlögl, Surface potential changes of semiconducting oxides monitored by high-pressure photoelectron spectroscopy: importance of electron concentration at the surface, Solid State Ionics 177 (2006) 3123–3127. [41] M. Batzill, U. Diebold, Surface studies of gas sensing metal oxides, Physical Chemistry Chemical Physics 9 (2007) 2307–2318. [42] M. Batzill, K. Katsiev, J.M. Burst, U. Diebold, A.M. Chaka, B. Delley, Gas-phasedependent properties of SnO2 (1 1 0), (1 0 0), and (1 0 1) single-crystal surfaces: structure, composition, and electronic properties, Physical Review B 72 (2005) 165414. [43] R. Cavicchi, M. Tarlov, S. Semancik, Preparation of well ordered, oxygen-rich SnO2 2(1 1 0) surfaces via oxygen plasma treatment, Journal of Vacuum Science and Technology A 8 (1990) 2347. [44] D.F. Cox, T.B. Fryberger, J.W. Erickson, S. Semancik, Summary abstract: surface properties of clean and gas-dosed SnO2 (1 1 0), Journal of Vacuum Science and Technology A 5 (1987) 1170. [45] S. Semancik, T.B. Fryberger, Model studies of SnO2 -based gas sensors—vacancy defects and pd additive effects, Sensors and Actuators B, Chemical 1 (1990) 97–102. [46] M. Batzill, U. Diebold, The surface and materials science of tin oxide, Progress in Surface Science 79 (2005) 47–154. [47] D.F. Cox, T.B. Fryberger, S. Semancik, Surface reconstructions of oxygen deficient SnO2 (1 1 0), Surface Science 224 (1989) 121–142. [48] D.F. Cox, T.B. Fryberger, S. Semancik, Oxygen vacancies and defect electronic states on the SnO2 (1 1 0)- 1 × 1 surface, Physical Review B 38 (1988) 2072. [49] J. Robertson, Electronic structure of SnO2 , GeO2 , PbO2 , TeO2 and MgF2 , Journal of Physics C 12 (1979) 4767. [50] O. Madelung, Semiconductors—Basic Data, second ed., Springer, Berlin, Heidelberg, New York, 1996. [51] E. Burstein, Anomalous optical absorption limit in InSb, Physical Review 93 (1954) 632–633.

[52] T.S. Moss, The interpretation of the properties of indium antimonide, Proceedings of the Physical Society of London Section B 67 (1954) 775–782. [53] C.L. Lau, G.K. Wertheim, Oxidation of tin—esca study, Journal of Vacuum Science & Technology 15 (1978) 622–624. [54] C. Körber, S.P. Harvey, T.O. Mason, A. Klein, Barrier heights at the SnO2 /Pt interface: in situ photoemission and electrical properties, Surface Science 602 (2008) 3246–3252. [55] J.W. Erickson, T.B. Fryberger, S. Semancik, Metal–semiconductor interfaces on SnO2 (1 1 0), Journal of Vacuum Science & Technology A—Vacuum Surfaces and Films 6 (1988) 1593–1598. [56] E. de Fresart, J. Darville, J.M. Gilles, Influence of the surface reconstruction on the work function and surface conductance of (1 1 0)SnO2 , Applied Surface Science 11–12 (1982) 637–651. [57] K. Jacobi, G. Zwicker, A. Gutmann, Work function, electron affinity and band bending of zinc oxide surfaces, Surface Science 141 (1984) 109–125. [58] H. Moormann, D. Kohl, G. Heiland, Work function and band bending on clean cleaved zinc oxide surfaces, Surface Science 80 (1979) 261–264. [59] W. Ranke, Ultraviolet photoelectron spectroscopy investigation of electron affinity and polarity on a cylindrical gaas single crystal, Physical Review B 27 (1983) 7807–7810. [60] J. Fritsche, D. Kraft, A. Thissen, T. Mayer, A. Klein, W. Jaegermann, Interface engineering of chalcogenide semiconductors in thin film solar cells: CdTe as an example, Materials Research Society Symposium Proceedings 668 (2001) H6.6.1–H6612.6612.6612. [61] A.V. Chadwick, EXAFS studies of dopant sites in metal-oxides, Solid State Ionics 63–65 (1993) 721–727. [62] J. Rockenberger, U. zum Felde, M. Tischer, L. Troger, M. Haase, H. Weller, Near edge X-ray absorption fine structure measurements (XANES) and extended Xray absorption fine structure measurements (EXAFS) of the valence state and coordination of antimony in doped nanocrystalline SnO2 , Journal of Chemical Physics 112 (2000) 4296–4304. [63] F.J. Berry, B.J. Laundy, Sb-121 mossbauer study of the effects of calcination on the structure of tin–antimony oxides, Journal of the Chemical Society-Dalton Transactions 6 (1981) 1442–1444. [64] J.A. Marley, R.C. Dockerty, Electrical properties of stannic oxide single crystals, Physical Review 140 (1965) A304. [65] J. Robertson, Defect levels of SnO2 , Physical Review B 30 (1984) 3520. [66] C.G. Fonstad, R.H. Redike, Electrical properties of high-quality stannic oxide crystals, Journal of Applied Physics 42 (1971) 2911–2918. [67] S. Samson, C.G. Fonstad, Defect structure and electronic donor levels in stannic oxide crystals, Journal of Applied Physics 44 (1973) 4618–4621. [68] S. Lany, A. Zunger, Dopability, intrinsic conductivity, and nonstoichiometry of transparent conducting oxides, Physical Review Letters 98 (2007) 045501. [69] C. Kilic, A. Zunger, Origins of coexistence of conductivity and transparency in SnO2 , Physical Review Letters 88 (2002) 095501. [70] Y. Gassenbauer, A. Wachau, A. Klein, Chemical and Electronic Properties of the ITO/Al2 O3 Interface, Physical Chemistry Chemical Physics 11 (2009) 3049. [71] D.M. Mattox, Particle bombardment effects on thin-film deposition: a review, Journal of Vacuum Science and Technology A 7 (1989) 1105–1114. [72] L.W. Rieth, P.H. Holloway, Influence of negative ion resputtering on ZnO:Al thin films, Journal of Vacuum Science and Technology A 22 (2004) 20–29.

Biographies Christoph Körber studied materials science and obtained his diploma in 2006 from the Department of Materials and Geo Science at Technische Universität, Germany. Since 2006 he is a PhD student in the Surface Science Division of the Department of Materials and Geo Science at Technische Universität Darmstadt. His current research interests are electronic and chemical surface and interface properties, as well as the defect structure, of transparent conducting oxides. Péter Ágoston studied materials science and obtained his diploma in 2007 from the Department of Materials and Geo Science at Technische Universität Darmstadt, Germany. Since 2007 he is a PhD student with the Materials Modeling Division of the Department of Materials and Geo Science at Technische Universität Darmstadt. His current research interests are first-principles electronic structure calculations of point defect and surface properties in transparent conducting oxides. Andreas Klein studied physics at the Universities of Tübingen and Konstanz and obtained his diploma in 1989. He finished his PhD in 1994 at University of Konstanz. From 1994 to 1997 he was working as research assistant at the Hahn-MeitnerInstitut, Berlin and moved to Darmstadt in 1998. He completed his Habilitation in Materials Science in 2003 and became außerplanmäßiger Professor at Technische Universität Darmstadt in 2008. His research interests are electronic properties of surfaces and interfaces with a focus on thin film solar cells, transparent conducting oxides and ferroelectric oxides.