2.1. Fundamental band gaps
A composite of all XPS/UPS-derived data is shown in
Figure 2a–d as plots of work function (
)
vs. Fermi level (
) for each oxide system. For SnO
(
Figure 2a) and ZnO (
Figure 2b), the data for undoped and doped oxides (ATO, AZO, respectively) have been superimposed. In the case of In
O
, for the sake of clarity, the data for undoped In
O
(
Figure 2c) and doped ITO (
Figure 2d) have been plotted separately. In the case of SnO
and ZnO, closed triangles represent
in situ sputter-deposited
undoped specimens, with relative size indicative of the substrate temperature,
i.e., ranging from low (25
C) to high (500
C). (The data for doped films and bulk specimens are represented by closed diamonds and squares, respectively, to be discussed further below.) In the case of In
O
, closed
vs. open triangles refer to reactively-evaporated thin films and magnetron-sputtered thin films, and in the case of ITO to magnetron sputtered films with 10 and 2 wt% SnO
doping, respectively, again with relative size indicative of the substrate temperature,
i.e., ranging from low (25
C) to high (400
C). (The data for bulk specimens are represented by closed squares and the crossed symbols will be discussed further below.)
Figure 2.
Work function versus Fermi level position plots. Solid lines follow constant ionization potentials, dotted vertical lines mark the intrinsic bandgap, and symbol size corresponds to substrate temperature during deposition (smallest = room temperature, largest = C (C for (b)). (a) upper left graph RF magnetron sputtered SnO (triangles) and SnO:Sb (diamonds) grown at C, and sintered ceramic SnO:Sb (squares); (b) lower left graph DC magnetron sputtered ZnO (triangles) and ZnO:Al (diamonds) thin films, and sintered ceramic ZnO:Al (bowtie); (c) upper right graph InO reactively evaporated (open triangles), RF magnetron sputtered (closed triangles), sintered (filled squares), and air-annealed (crossed squares); (d) lower right ITO sintered ceramic samples (filled squares), air-annealed thin films (crossed squares), ozone-trated thin films (crossed circle), and RF magnetron sputtered 2% doped ITO (open triangles) and 10% doped ITO (filled triangles).
Figure 2.
Work function versus Fermi level position plots. Solid lines follow constant ionization potentials, dotted vertical lines mark the intrinsic bandgap, and symbol size corresponds to substrate temperature during deposition (smallest = room temperature, largest = C (C for (b)). (a) upper left graph RF magnetron sputtered SnO (triangles) and SnO:Sb (diamonds) grown at C, and sintered ceramic SnO:Sb (squares); (b) lower left graph DC magnetron sputtered ZnO (triangles) and ZnO:Al (diamonds) thin films, and sintered ceramic ZnO:Al (bowtie); (c) upper right graph InO reactively evaporated (open triangles), RF magnetron sputtered (closed triangles), sintered (filled squares), and air-annealed (crossed squares); (d) lower right ITO sintered ceramic samples (filled squares), air-annealed thin films (crossed squares), ozone-trated thin films (crossed circle), and RF magnetron sputtered 2% doped ITO (open triangles) and 10% doped ITO (filled triangles).
Focusing on the
undoped results for the moment, we have superimposed published values of fundamental band gap of ∼3.6 eV for SnO
[
21] and ∼3.3 eV for ZnO [
22] on
Figure 2a and b, respectively. In the case of In
O
, a band gap of ∼2.8 ± 0.2 eV (justified below) has been similarly superimposed on
Figure 2c and d. It can be seen that, with the exception of a handful of In
O
thin films, the undoped specimens (ZnO, SnO
, In
O
) tend to have Fermi levels less than the fundamental band gap. In other words, intrinsic defect formation (e.g., by oxygen vacancies) is insufficient to impart detectable degeneracy to the basis oxides; aliovalent donor-doping is required to shift the Fermi level significantly above the CBM (AZO, ATO, ITO). Furthermore, whereas weak metal-like Fermi edge emissions were consistently observed for
doped films whose Fermi levels were significantly in excess of their fundamental band gaps (e.g., see
Figure 4 in [
17]), such emissions were only observed for In
O
specimens with Fermi levels above the fundamental gap (∼2.9 eV), and never for undoped ZnO or SnO
specimens. Therefore, the data in
Figure 2 are consistent with the band gaps of 3.6 eV, 3.3 eV, and 2.8 eV for SnO
, ZnO, and In
O
, as shown.
The fundamental band gap of In
O
/ITO requires additional explanation and justification. It is widely known that highly degenerate thin films of ITO can exhibit band edge absorption in the range of 3.5–4.0 eV, which is much larger than the 2.8 eV fundamental band gap shown in
Figure 2c and d, even accounting for Burstein-Moss shift. One possible explanation is the presence of an indirect gap that is significantly smaller than the direct band gap. However, recent density functional theory calculations indicated that the energy difference between the overall VBM and the highest occupied level at the point is less than 50 meV [
23]. More recent resonant X-ray emission spectroscopy results were found to be consistent with a direct band gap for In
O
[
24]. The discrepancy has been resolved by other theoretical work [
25], which showed that the separation between weak and strong optical onsets (of ∼0.85–0.9 eV) arises from the fact that transitions from the highest valence bands into the conduction band are either symmetry-forbidden or have very low dipole intensity. Therefore, the onset of strong optical absorption in degenerate ITO films occurs above 3.65–3.70 eV (the fundamental gap of
eV plus the 0.85–0.90 eV separation described above). That same work set an upper limit of
eV for the fundamental band gap of In
O
[
25].
In
Figure 3 are plotted the binding energy of the In-
core level
vs. the Fermi level (
) for all the In
O
and ITO specimens in the present work. The open squares represent undoped In
O
(both films and sintered ceramics), the closed circles represent sintered ITO ceramics, the closed triangles 10% doped ITO films, the open triangles represent 2% doped ITO films, the cross-hatched squares represent ozone treated ITO films, and the cross-hatched circles represent ITO films annealed at 400
C. It can be seen that at low Fermi levels, there is a parallel change in the In-
core level and in the Fermi level; this is indicated in
Figure 3 by the solid line with slope 1. Deviation from linearity occurs in the vicinity of
eV. As mentioned above, this deviation from linearity is accompanied by the appearance of metal-like emissions at the Fermi edge in UPS spectra (in specimens with Fermi levels of ∼2.9 eV or greater). Therefore, the deviation from linearity can be explained by screening of the photoemission core hole by free electrons in the conduction band [
15,
26,
27,
28]. The dashed line, representing the behavior of the degenerate specimens, roughly intersects the solid line at a Fermi level of ∼2.8 eV, which is the value we have used for the fundamental band gap of In
O
/ITO in
Figure 2. Given the scatter of the data in
Figure 3, we cannot at present be more precise regarding this value. Nevertheless, the data in
Figure 3 provide support for the assignment of the fundamental band gap of In
O
/ITO at approximately this value.
Figure 3.
In 3d5/2 core level binding energy vs. Fermi level position for InO and ITO. The solid line marks a slope of one. Squares represent InO films and sintered ceramic samples, solid circles represent sintered ceramic ITO, solid triangles represent 10% doped ITO films, open triangles represent 2% doped ITO films, cross circles represent air-annealed ITO films (at C), and crossed squares represent ozone treated ITO films. Symbol size reflects the substrate temperature during deposition, which ranges from room temperature (small) to C (large).
Figure 3.
In 3d5/2 core level binding energy vs. Fermi level position for InO and ITO. The solid line marks a slope of one. Squares represent InO films and sintered ceramic samples, solid circles represent sintered ceramic ITO, solid triangles represent 10% doped ITO films, open triangles represent 2% doped ITO films, cross circles represent air-annealed ITO films (at C), and crossed squares represent ozone treated ITO films. Symbol size reflects the substrate temperature during deposition, which ranges from room temperature (small) to C (large).
2.2. Fermi level manipulation
If we consider the range of Fermi levels for donor-doped SnO
, ZnO, and In
O
in
Figure 2, it is clear that the behavior in ATO is markedly different from that of AZO and ITO. Whereas the Fermi levels of AZO films range from ∼2.4 eV to as high as ∼3.8 eV (
Figure 2b), and the Fermi levels of ITO films range from ∼2.0 eV to ∼3.5 eV (
Figure 2d), the Fermi levels of ATO films are constrained to a relatively narrow range of ∼3.3–3.75 eV (
Figure 2a). Slightly larger values of Fermi level of ∼4.0 eV are obtained for fluorine doped SnO
, which is consistent with the higher conductivity of this material [
1]. The variation of Fermi level can be understood by considering the prevailing defect models for the three materials.
In
Figure 4a–c are plotted the schematic Brouwer diagrams for ATO, AZO, and ITO, respectively. Brouwer diagrams are log-log plots of defect concentrations
vs. a control variable, in this case the oxygen partial pressure. The Brouwer diagram for AZO (
Figure 4b) is adapted from [
29,
30]. It reflects the fact that the formation of "killer" defects, in this case zinc vacancies, results in ionic compensation of the donor species (
) rather than the desired carrier generation [
31,
32,
33,
34]; the electroneutrality condition in the high-pO
regime is
. As pO
is reduced, however, the electron population steadily rises until it becomes the prevalent species compensating the donors; the electroneutrality condition in the low-pO
regime becomes
, as shown. What this means is that the electron population in AZO can be modified by orders of magnitude through control of the pO
during thin film deposition, with high oxygen contents yielding films with very small carrier contents (and Fermi levels in the band gap) and low oxygen contents yielding degenerate films (and Fermi levels above the CBM).
The Brouwer diagram for ITO in
Figure 4c has been published previously [
17] and agrees well with DFT calculations [
29,
35]. It is well known that the prevailing defect type in Sn-doped In
O
is the so-called Frank-Köstlin (F-K) associate [
36,
37]. This neutral associate consists of two Sn-donors and an oxygen interstitial acceptor, or
in Kröger-Vink notation. At high oxygen pressures, most of the Sn-donors are tied up in these neutral clusters, thereby reducing the overall electron population. Under reducing conditions, however, the interstitial oxygen can be removed from these associates, which activates the Sn-donors and increases the electron content (the middle regime in
Figure 4c). Under the most reducing conditions (the leftmost regime in
Figure 4c), oxygen vacancies can also be formed and further increase the electron population. What this means is that the electron population in ITO can be modified by orders of magnitude through control of the pO
during thin film deposition [
13,
15,
16,
17], with high oxygen contents yielding films with very small carrier contents (and Fermi levels in the band gap) and low oxygen contents yielding degenerate films (and Fermi levels above the CBM).
In contrast, the Brouwer diagram for ATO in
Figure 4a reflects the absence of either a global “killer" defect (like zinc vacancies in AZO) or a local compensating defect (like oxygen interstitials in the F-K clusters of ITO). The origin of this difference is related to the higher formation enthalpies of the possible “killer" defects, Sn-vacancies and O-interstitials, which are related to the high charge state of Sn and the rutile crystal structure of SnO
[
38,
39]. Instead, the high-pO
electroneutrality regime represents straightforward donor-doping,
i.e.,
, and the low-pO
regime represents the possibility of additional electrons owing to intrinsic defects (e.g., oxygen vacancies). What this means is that the electron population should be high under all pO
conditions, which sets ATO apart from the other two systems. This explains why there is very little change of carrier concentration with pO
during deposition of the ATO films [
40].
It should be noted that the observed segregation of dopant species to the surfaces of ITO (Sn) in films grown under reducing conditions [
15] may also contribute to increased carrier doping of their surfaces. In other words, the low pO
values during deposition yields additional carriers at the surface in two ways: (1) increased donor concentrations (owing to the segregation effect) and (2) activation of those donors (as per the Brouwer diagrams, traversing from right-to-left in
Figure 4b and c).
Figure 4.
Schematic Brouwer diagrams (log concentration versus log pO) for (a) SnO:Sb (ATO), (b) ZnO:Al (AZO), and (c) InO:Sn (ITO). Vertical dotted lines separate different electroneutrality regimes, as denoted at the top of each column. Fractions indicate line slopes. Concentrations of electrons (n), doubly charged O vacancies (), doubly charged Zn vacancies (), Sb donors (), Al donors (), Sn donors (), and Frank-Köstlin defects () are shown where relevant.
Figure 4.
Schematic Brouwer diagrams (log concentration versus log pO) for (a) SnO:Sb (ATO), (b) ZnO:Al (AZO), and (c) InO:Sn (ITO). Vertical dotted lines separate different electroneutrality regimes, as denoted at the top of each column. Fractions indicate line slopes. Concentrations of electrons (n), doubly charged O vacancies (), doubly charged Zn vacancies (), Sb donors (), Al donors (), Sn donors (), and Frank-Köstlin defects () are shown where relevant.
2.3. Work functions
The lines superimposed on
Figure 2a–d represent constant ionization potential, which is the sum of the Fermi level (
) and the work function (
). If the ionization potential of a given oxide maintains a fixed value (
), all the work function
vs. Fermi level data should fall on a single line, whose slope is
and whose y-intercept is the ionization potential (
):
In terms of the schematic diagram in
Figure 1b, with a fixed ionization potential any shift in Fermi level will result in an equal and opposite change of work function; we are simply moving the Fermi level up or down between the VBM and the vacuum level (
).
From the data plotted in
Figure 2, however, it is apparent that the ionization potentials of the oxides studied are not constants, but rather can be modified during deposition and/or by post-deposition treatments (e.g., ITO, see below). In our previous work we discussed at length the potential origins of surface dipole modifications for the three oxide systems under consideration [
12]. What follows is a brief discussion of these factors, system-by-system, to provide a framework for our discussion of how the work functions of these important TCOs can be manipulated to advantage.
In the case of ATO (Sb-doped SnO
), the data in
Figure 2a do not follow the trend of Equation
1,
i.e., they fall between lines rather than on a single line of constant
. As mentioned in the previous section, for ATO there is little or no variation of Fermi level with oxygen content employed during sputter deposition [
40]. This is explained by the defect model in
Figure 4a; the absence of "killer" defects in SnO
accounts for the small overall changes in doping level with oxygen content during deposition. On the other hand, there are large (∼1.0 eV) changes in ionization potential (and work function) with oxygen content during deposition. In
Figure 4a, the data on the
eV line were achieved for deposition under oxidizing conditions, whereas the data on the
eV line were achieved for deposition under reducing conditions, with a gradual transition for intermediate pO
values. In our prior work, this was attributed to changes in surface orientation (texture) and/or surface termination [
12]. For example, prior work on SnO
surfaces yielded work functions of ∼5.7 eV after oxygen annealing
vs. ∼4.7 eV after reduction annealing [
41]. The variation of work function (∼1 eV) was attributed to different terminations of the
surface. Whereas oxygen annealing produces a stoichiometric surface, reduction annealing yields Sn in the Sn
oxidation state, owing to the removal of bridging and in-plane oxygens [
41,
42,
43]. Comparable swings of work function (oxidized
vs. reduced) have been reported for the SnO
surface, again attributed to changes in surface terminations [
44,
45,
46]. On the other hand, an increase in ionization potential has also been linked with a change of crystallographic orientation from
to
[
47]. Therefore, an influence of surface orientation on the work function cannot be ruled out. Regardless of mechanism (change of surface orientation and/or surface termination), the ionization potential (and work function) of SnO
-based TCOs can be significantly modified by as much as
eV by control of the oxygen content during sputter deposition. It should be noted that the data for sintered ceramic specimens tend toward the higher ionization potential (
eV) line. This is to be expected, since the specimens were sintered in air (
i.e., oxidizing conditions).
In contrast with ATO, the data for Al-doped ZnO (AZO) in
Figure 2b more closely approximate a line (or band) of constant ionization potential,
i.e., they roughly behave according to Equation
1. This means that the changes in work function essentially reflect the corresponding shift of the Fermi level with doping. As discussed in the previous section, the electron population can be altered over orders of magnitude by controlling the oxygen content during sputter deposition. In
Figure 2b, films grown under oxidizing conditions exhibit small Fermi levels (high work functions) and those grown under reducing conditions exhibit the opposite (high Fermi levels, low work functions). This behavior is consistent with the Brouwer diagram of
Figure 4b, with redox conditions during deposition controlling the overall carrier content.
In addition, however, there are subtle changes in ionization potential for AZO films processed under oxidizing
vs. reducing conditions. For example, it can be observed in
Figure 2b that the ZnO/AZO data are bracketed by two lines of constant ionization potential. Although the effect is not as pronounced as with the SnO
/ATO films, AZO films deposited under oxidizing conditions tend to approach the
eV line (at small Fermi levels), whereas those grown under reducing conditions tend to approach the
eV line (at large Fermi levels). The significance of these two lines is that the lower line (
eV) has been established for the polar, zinc-terminated
surface of ZnO, whereas the upper line (
eV) has been established for the polar oxygen-terminated
surface and the nonpolar
surface of ZnO [
48,
49,
50]. These results suggest that processing conditions not only alter the carrier content (Fermi level), but also the surface dipole (and therefore the ionization potential). It should be noted that the datum for a sintered ceramic specimen falls near the higher ionization potential (
eV) line. Since large ionization potentials are seen for all orientations except
[
51], we conclude that the ceramic specimens are dominated by orientations other than
.
There is one significant difference between the AZO behavior and that of the other two systems. Whereas the surface dipole of either ATO or ITO can be modified by post-deposition treatments (i.e., by ozone treatments and/or oxidation at intermediate temperatures), the ionization potential of AZO films is fixed during deposition and appears to be unchangeable. This is to be expected, since a change of surface dipole requires a change of grain orientation, which is virtually impossible to achieve by post-deposition treatments.
The work function
vs. Fermi level data for thin film ITO (
Figure 2d) closely follows Equation
1,
i.e., at roughly constant ionization potential (
eV). This is consistent with prior studies [
14,
15,
17,
52]. The movement right or left along this line indicates a shift of the Fermi level between more or less constant values of
and the vacuum level (see
Figure 1b). The large range of achievable work functions (∼4.2 to ∼5.6 eV) derives from the correspondingly large range of achievable Fermi level positions, as discussed in terms of the Brouwer diagram (
Figure 4c) in the prior section. It should be stressed that the larger values of work function along this line (to the left of the fundamental band gap) are not desirable, in that films with Fermi levels much below the fundamental band gap are not degenerately doped,
i.e., they have low conductivities and therefore are not suitable as TCOs.
We do not comment further about the behavior of undoped In
O
films (
Figure 2c), given that these films are insufficiently doped/conductive for TCO applications, other than to note that the data fall on a line of lesser ionization potential (
eV). This may be attributable to the absence of Sn, which when present seems to be associated with higher ionization potentials,
i.e.,
(In
O
)
(ITO)
(SnO
/ATO).
What is noteworthy about the ITO films is the ability to modify their ionization potentials/work functions through post-deposition treatments. For example, the crossed squares in
Figure 2d represent thin film specimens, whose original data lie on the
eV line, which have subsequently been air-annealed at
C. There are insignificant changes in Fermi level, but as can be seen, their work functions (and ionization potentials) increase significantly. Similarly, the crossed circle represents an ITO film, originally lying on the
eV line, which was subsequently subjected to ozone treatment at room temperature. This datum falls on the upper line of
eV, which is also where the data for sintered ceramic specimens lie.
The explanations for the post-deposition modifications of ionization potential in ITO films are not as compelling as for the SnO
-based and ZnO-based films, owing to a lack of experimental and theoretical studies of specific crystallographic orientations/terminations. Plausible explanations for the origin of the observed surface dipole changes in In
O
-based materials include modification of the surface dipole through change of surface terminations (as in the case of SnO
) and/or possibly through the incorporation/implantation of surface species during oxidation or ozone treatments. The bixbyite structure of In
O
is unique in its abundance of "structural oxygen vacancies" (
i.e., oxygen interstitial positions) [
37], which can accommodate additional species, whether in the bulk (see
Figure 4c) or in the surface. Regardless of explanation, it is well established that post-deposition oxidation treatments can be employed to increase the work function of ITO-based TCOs [
53,
54,
55,
56,
57].
2.4. Energy band alignment
The energy band alignment is another important factor in determining electrical properties of heterojunction devices [
58]. It is here necessary to distinguish between TCO contacts to organic and to inorganic semiconductors. The energy band alignment with organic materials is largely governed by alignment of the vacuum level, so the work function of the TCO determines the energy barriers at the contacts [
6,
7]. In that case, modification of band alignment can be obtained following the general dependencies outlined in the previous section. In contrast, the presence of intrinsic and extrinsic interface states at interfaces between inorganic materials lead to interface dipoles, which can result in large deviations from the vacuum level alignment [
59,
60]. Furthermore, TCO interfaces in thin film solar cells are polycrystalline and the TCO and its contact partner have different crystal structures and/or different lattice constants. To date, no general understanding of such interfaces exists.
The almost exclusively used contact partner for the TCO in thin film solar cells with CdTe and Cu(In,Ga)(S,Se)
absorber materials is CdS [
61]. The main role of this interface is the collection of electrons, which is achieved by bringing the Fermi level close to the conduction band. This is, however, not sufficient for high efficiency. Neither in CdTe nor in Cu(In,Ga)(S,Se)
thin film solar cells are high efficiencies obtained by direct contact of the absorber (CdTe or Cu(In,Ga)(S,Se)
) with the TCO, although the condition of a Fermi level close to the absorber conduction band can be fulfilled. The insertion of the CdS “buffer" layer most likely results in a strongly reduced minority carrier recombination at the absorber/buffer interface, compared to the absorber/TCO interface.
Several studies related to TCO/CdS interfaces employing stepwise evaporation of CdS onto TCO and stepwise sputter deposition of TCOs onto CdS, as well as sputter depth profiles, have been performed [
62,
63,
64,
65,
66,
67,
68]. An extensive investigation of the CdS/ZnO interface is described in [
18]. It turns out that the band alignment can depend significantly on processing, which is mainly related to two effects: (1) due to the dissimilar structure of the TCO and CdS, amorphous phases may occur at the interface; (2) the Fermi level in CdS seems to be restricted to a range 1.8–2.2 eV above the valence band maximum. This Fermi level pinning is particularly important as the Fermi level in TCOs can vary by as much as 1 eV (see
Section 2.2), which can consequently lead to an apparent dependence of band alignment on doping. The effect is most pronounced at interfaces between ZnO and In
S
, an alternative buffer layer material for Cu(In,Ga)(S,Se)
thin film solar cells [
18,
60].
A summary of experimentally determined energy band alignments at TCO/CdS interfaces is given in
Figure 5. For the CdS/ZnO interface we have chosen a valence band offset of
eV. This value is consistently derived from CdS deposition onto both undoped and Al-doped ZnO, indicating that Fermi level pinning is not affecting the alignment [
18]. Since the valence bands of the three basis TCOs are derived mainly from O
orbitals, it is expected that the valence bands are at comparable energy and that the valence band offsets between the TCOs and CdS are comparable. This expectation is fulfilled for ZnO and SnO
, but the valence band offsets between CdS and ITO and In
O
are considerably smaller than those between CdS and ZnO and SnO
. The deviation can again be attributed to the Fermi level pinning in CdS and to the smaller energy gap of In
O
. Assuming a valence band energy for In
O
similar to those of SnO
and ZnO as done in the final graph of
Figure 5, it is not possible to find a Fermi level position consistent with the allowed range in In
O
(see above) and the 1.8–2.2 eV range found for CdS. In consequence, a local dipole occurs at the interface, shifting the energy bands of In
O
upwards. An interface experiment between SnO
and In
O
has independently verified their similar valence band energies and this interpretation of measurements. One can thus conclude that valence band offsets between the three basis TCOs and any inorganic semiconductor are generically of similar magnitude, if they comply with the possible range of Fermi levels.
Figure 5.
Energy band alignment at TCO/CdS interfaces as determined from photoelectron spectroscopy using stepwise deposition experiments and sputter depth profiles. All values are given in electronvolts. The first four plots show the interpretation of experimental data, whereas the last depicts why it is not possible to find a Fermi level position consistent with the allowed Fermi level range in InO and the eV range found for CdS. In consequence, a local dipole occurs at the interface, shifting the energy bands of InO upwards as shown in the experimental InO/CdS band diagram.
Figure 5.
Energy band alignment at TCO/CdS interfaces as determined from photoelectron spectroscopy using stepwise deposition experiments and sputter depth profiles. All values are given in electronvolts. The first four plots show the interpretation of experimental data, whereas the last depicts why it is not possible to find a Fermi level position consistent with the allowed Fermi level range in InO and the eV range found for CdS. In consequence, a local dipole occurs at the interface, shifting the energy bands of InO upwards as shown in the experimental InO/CdS band diagram.
2.5. Practical ramifications
Key parameters of interest to photovoltaic applications of TCOs include high electronic conductivity, good visible transparency (in thin film form), work function and energy band alignment. These properties can be judged on the basis of the modified work function
vs. Fermi level diagram shown in
Figure 6. In this diagram, TCO-appropriate properties are represented as shaded parallelogram boxes for each oxide. The left (low Fermi level) side of each box corresponds to the fundamental band gap of the corresponding oxide. A fundamental band gap of
eV or greater is a requisite for full optical transparency. (As discussed previously, In
O
and ITO retain full optical transparency in spite of having a smaller fundamental band gap (∼2.8 eV), owing to weak (or symmetry-forbidden) transitions from the top of the valence band.) The left side of each parallelogram is important for another reason, since it represents the onset of significant electronic conductivity (through degenerate doping). It further represents an estimate for the conduction band alignment at interfaces between TCOs and inorganic semiconductors, as it can be assumed that the valence band maxima of the TCOs are at comparable energy. As SnO
exhibits the highest conduction band energy, it may in principle provide the best electrical contact to n-type conductors. It is recalled however, that this is only true if strong interfacial dipoles and/or Fermi level pinning are not present [
18,
60]. The right side of each box represents a generous Burstein-Moss shift of Fermi level by doping to ∼0.5 eV above the fundamental band gap [
19,
20]. (The right boundary is
eV above the fundamental gap in the case of ITO, to account for the larger Fermi levels observed in thin films.) The top and bottom of each box represent the potential modifications in surface dipole found in the present work by control of the oxygen content during film deposition. In the case of SnO
and ZnO, however, the upper and lower limits are consistent with work function variations in the literature for specific orientations and/or surface terminations. In addition, the upper limits in each instance are consistent with the data for sintered ceramic specimens, which can be expected to represent relaxed, fully-oxidized surfaces.
Figure 6.
Work function versus Fermi level position for ATO, AZO, and ITO. The left side of each parallelogram corresponds to the bandgap, the right side corresponds to the maximum Burstein-Moss shift, and the top and bottom lines correspond to known ranges in ionization potential.
Figure 6.
Work function versus Fermi level position for ATO, AZO, and ITO. The left side of each parallelogram corresponds to the bandgap, the right side corresponds to the maximum Burstein-Moss shift, and the top and bottom lines correspond to known ranges in ionization potential.
The boxes in
Figure 6 are therefore reasonable representations of the Fermi levels and work functions available for each type of TCO. For ATO, the range of work functions is from
eV to
eV. For AZO, the range of work functions is significantly lower, from
eV to
eV. And for ITO, the range of work functions is from
eV to
eV. As mentioned previously, the ionization potentials (work functions) of both ATO and ITO films can be significantly increased by post-deposition treatments (ozone, oxidation at intermediate temperatures), but never above the values indicated by the boxes in
Figure 6. It should be stressed that changes in “effective" work function, over and above the “intrinsic" (oxygen-related) modifications in the present work, can be achieved by surface chemical treatments as pointed out by Armstrong
et al. [
69].