3.1.1.

CuPc/ITO interface

In Fig. 1, the UPS spectra of CuPc films on ITO are presented as a function of the thickness of the CuPc films. All the spectra have been normalized to the same height for visual clarity. In Fig. 1(a) and 1(b), the cut-off and the HOMO regions are presented, respectively. The cut-off binding energy (BE) of the ITO substrate was measured to be 16.57 eV, which corresponds to a surface work function of 4.65 eV, obtained by subtracting the cut-off BE from the photon energy of the excitation source (21.22 eV). With increasing thickness of the CuPc film, we observed cut-off shift towards the lower WF (or higher cut-off BE) values. The shift saturated around 32 Å CuPc thickness. From Fig. 1(b), the HOMO onset for 8 Å CuPc film was measured at 0.31 eV. With further increase in the CuPc thickness, the occupied level peaks remained more or less unchanged on the HOMO peaks, as illustrated by the dashed line in the figure.

Fig. 1

The UPS spectra of CuPc on ITO as a function of CuPc thickness. The cut-off and the HOMO regions are presented in Fig. 1(a), and 1(b), respectively.

3.1.2.

$\mathrm{CuPc}/\mathrm{Mo}{\mathrm{O}}_x$ interface

In Fig. 2, the UPS data for the cut-off and the HOMO regions are plotted in Fig. 2(a) and 2(b), respectively. The ITO WF was measured to be 4.59 eV and, with the deposition of a 100 Å $\mathrm{Mo}{\mathrm{O}}_x$ film, the WF was measured to be 6.82 eV. With the deposition of CuPc on $\mathrm{Mo}{\mathrm{O}}_x/\mathrm{ITO}$, the WF first decreased rapidly, then the shift in WF became more gradual. In Fig. 2(b), the valence band peak of $\mathrm{Mo}{\mathrm{O}}_x$ was observed at $\sim 3.8\mathrm{eV}$. The $\mathrm{Mo}{\mathrm{O}}_x$ peak shows a gradual shift towards the higher BE with increasing thickness of CuPc. This shift was measured to be $\sim 0.2\mathrm{eV}$ up to 8 Å of CuPc deposition. With further CuPc deposition, all the occupied level peaks were observed to be continuously shifting towards the higher BE. The HOMO onset BE for 8 Å, 64 Å, and 200 Å thick CuPc films were measured to be 0.21, 0.36, and 0.53 eV, respectively.

Fig. 2

The UPS spectra of CuPc on $\mathrm{Mo}{\mathrm{O}}_x/\mathrm{ITO}$ as a function of CuPc thickness. The cut-off and the HOMO regions are presented in Fig. 1(a) and 1(b), respectively.

3.1.3.

Energy level alignment

The energy level alignment diagrams for the CuPc/ITO and the $\mathrm{CuPc}/\mathrm{Mo}{\mathrm{O}}_x$ interfaces are presented in Fig. 3(a) and 3(b), respectively. The band gap of 3.2 eV for $\mathrm{Mo}{\mathrm{O}}_x$ and 1.7 eV for CuPc hasbeen reported in our earlier works.^26–30 In Fig. 3(a), the ITO WF was measured to be 4.65 eV. At the CuPc/ITO interface, we observed an interface dipole of 0.33 eV, due to transfer of holes from ITO to the CuPc side. After the formation of the interface dipole, a near flat band situation was observed. The initial hole injection barrier was measured to be 0.31 eV, with the ionization potential (IP) of 4.65 eV. In Fig. 3(b), the ITO WF was measured to be 4.59 eV, which became 6.82 eV with the deposition of 100 Å $\mathrm{Mo}{\mathrm{O}}_x$. With the deposition of CuPc on $\mathrm{Mo}{\mathrm{O}}_x/\mathrm{ITO}$, the energy levels of $\mathrm{Mo}{\mathrm{O}}_x$ shift towards the higher BE and a large dipole moment at the interface is observed, due to the rapid transfer of holes from $\mathrm{Mo}{\mathrm{O}}_x$ side to the organic side. The conservative estimate of the shift is 0.2 eV in the $\mathrm{Mo}{\mathrm{O}}_x$ energy levels towards the higher BE. The estimate is made from the shift of $\mathrm{Mo}{\mathrm{O}}_x$ valence band peak towards the higher BE with the deposition of CuPc, as depicted in Fig. 2(b). The large interface dipole of 2.97 eV is observed at the interface. The initial hole injection barrier is observed to be 0.21 eV. With further depsition of CuPc, the energy levels gradually relax back to their normal values, thus, creating a band bending like region in the organic side. At the thickness of 200 Å CuPc, the HOMO onset value was measured to be 0.53 eV.

Fig. 3

The energy level alignment diagrams for, (a) CuPc on ITO, and (b) CuPc on $\mathrm{Mo}{\mathrm{O}}_x/\mathrm{ITO}$ interfaces.

Kim et al. have reported 30 percent improvement in the fill factor (FF) and 37 percent improvement in the overall power conversion efficiency, with a 100 Å $\mathrm{Mo}{\mathrm{O}}_x$ inter-layer between ITO anode and a CuPc layer.²⁴ We assign the improvement to the lower hole injection barrier at the $\mathrm{CuPc}/\mathrm{Mo}{\mathrm{O}}_x$ interface and the presence of a drift electric field. There are two possible improvement mechanisms in the presence of the drift electric field. The first is that the drift field is a steering field which is helpful for easier extraction of holes at the anode.²⁴ The second is that the field facilitates the absorption of photons incident on the material through transparent anode side, which is supported by the reported improved absorption with $\mathrm{Mo}{\mathrm{O}}_x$ inter-layer.³¹

3.2.1.

Evolution of gap states with oxygen exposure

In our earlier report, we have demonstrated the effect of oxygen exposure on valence band region and WF of $\mathrm{Mo}{\mathrm{O}}_x$ thin films.²⁵ High WF ($\sim 6.8\mathrm{eV}$), of thermally evaporated $\mathrm{Mo}{\mathrm{O}}_x$ films, decreases sharply with oxygen exposure before reaching a saturation ($>5.8\mathrm{eV}$). In Fig. 4, the evolution of gap states in the valence band region is presented with increasing oxygen exposure on a 50 Å $\mathrm{Mo}{\mathrm{O}}_x$ film. The spectra of gap states are obtained by subtracting valence region spectrum of evaporated $\mathrm{Mo}{\mathrm{O}}_x$ film from the background of the valence band region of individual exposed spectrum. Before background subtraction, the dominant oxygen $2p$ peak intensities were normalized and peak position of the evaporated spectrum was shifted to match the peak positions of each of the exposed spectrum. In the figure, we observe the growth of a gap state at $<1.0\mathrm{eV}$ which is marked as ‘A’. The gap state is dominant throughout the measurements. Initially, the intensity of the gap state ‘A’ increases rapidly with the exposures, then it saturates around ${10}^{10}$ Langmuir (L) exposure. One Langmuir is equal to the ${10}^{-6}$ Torr-sec exposure. In other words, an exposure (in L) can be calculated by multiplying the pressure (in torr) and the duration of the exposure (in seconds) with ${10}^6$. Beyond ${10}^{10}$ L exposures, the intensity starts to decrease. A low intensity gap state, around 2.3 eV, is also observed at an early stage of exposure. The intensity of gap state ‘B’ is observed to be maximum at ${10}^6$ L exposure, and thereafter it starts to diminish. The third gap state, marked as ‘C’, develops at a mature exposure stage around $\sim 2.8\mathrm{eV}$. At $5\times {10}^{14}$ L exposure, the intensity of the gap state ‘C’ is comparable to the gap state ‘A’. In the next section, we will discuss the effect of air exposure on $\mathrm{Mo}{\mathrm{O}}_x$ film.

Fig. 4

The evolution of gap states with oxygen exposure.

3.2.2.

Evolution of gap states with air exposure

We have reported the effect of air exposures of $\mathrm{Mo}{\mathrm{O}}_x$ film on the valence region and core levels in our earlier reports.^25,32,33 We have demonstrated that the effect of air exposure on WF is substantial, on the valence band region it is significant, and on the core levels it is almost negligible. The WF decreases sharply with the air exposure before reaching a saturation ($>5.4\mathrm{eV}$). In Fig. 5, the growth of gap states in the valence band region is presented with increasing air exposure on a 50 Å $\mathrm{Mo}{\mathrm{O}}_x$ film. The presented spectra are obtained by subtracting valence region spectrum of thermally evaporated $\mathrm{Mo}{\mathrm{O}}_x$ film from a background, as described in Sec. 3.2.1. We observed the growth of a gap state around 1.0 eV, marked as ‘D’ in the figure, which is similar to the gap state ‘A’ in Sec. 3.2.1. Initially, the intensity of the state ‘D’ increases sharply with air exposures, then it saturates around ${10}^8$ L exposure. The gap state ‘D’ is dominant until ${10}^{12}$ L exposure. Another gap state ‘E’, at $\sim 2.3\mathrm{eV}$, is also observed which is similar to the state ‘B’ of the previous section. Initially, the intensity of the state ‘E’ is low. The gap state, marked as ‘F’, develops at a mature exposure stage around $\sim 2.9\mathrm{eV}$. At the final air exposure, the intensity of the ‘F’ state is dominant in the spectrum. In the next section, we will compare the evolutions of gap states with the exposures.

Fig. 5

The evolution of gap states with air exposure.

3.2.3.

Comparison of growths of gap states

In Fig. 6, relative intensities of the gap states are presented as a function of the exposures. The intensities of the gap states with the oxygen exposures are presented in Fig. 6(a), and with the air exposures are presented in Fig. 6(b). Comparing the evolution of states ‘A’ and ‘D’, both the gap states can be assigned to the effect of oxygen on the $\mathrm{Mo}{\mathrm{O}}_x$ films. While comparing the evolution of states ‘C’ and ‘F’, the origin of these states can be attributed to the effect of moisture on the $\mathrm{Mo}{\mathrm{O}}_x$ films. Earlier, we have reported that the WF reduction in exposed $\mathrm{Mo}{\mathrm{O}}_x$ films is a two stage process. The first stage of reduction is caused by adsorption of oxygen on the $\mathrm{Mo}{\mathrm{O}}_x$ films.²⁵ This stage reduces the WF from 6.8 to $\sim 5.8\mathrm{eV}$. The second stage of WF drop is caused by the adsorption of moisture on the $\mathrm{Mo}{\mathrm{O}}_x$ films. It is interesting to note, that a gap state similar to the ‘A’ and the ‘D’ states, also around 1 eV, has been reported earlier by Nakayama et al. on the initial deposition of $\mathrm{Mo}{\mathrm{O}}_x$ on poly(dioctylfluorine-alt-benzothiadiazole) (F8BT).²⁸ Therefore, it can be concluded that the first stage WF drop, caused by oxygen exposure, must be related to the charge transfer between adsorbed oxygen and the surface of $\mathrm{Mo}{\mathrm{O}}_x$ film.

Fig. 6

The comparison of gap states with air and oxygen exposures.

3.2.4.

Correlation of gap states with device performance and work function

In one of our earlier reports, we have demonstrated a substantial degradation in the device performance with 15 min ($\sim 7\times {10}^{11}\mathrm{L}$) of air exposure of the $\mathrm{Mo}{\mathrm{O}}_x$ film.²⁵ The exposures over ${10}^{11}$ L also mark the appearance of the gap state ‘F’ in Fig. 6. Since the growth of the gap state ‘F’ is caused by moisture, as discussed earlier, we attribute the major degradation in the device to the gap state ‘F’. In another report, a similar degradation was observed in the OPV performance by 30 min ($\sim 1.4\times {10}^{12}L$) of air exposure.³³ It should be noted, that both Nakayama et al.²⁸ and Maria et al have attributed the improvement of device performance to the presence of the gap state similar to the D state around 1 eV.²⁹ In our understanding, the WF reduction, from high 6.8 eV to moderate ($\sim 5.8\mathrm{eV}$), is coupled with the growth of gap state ‘D’ (or ‘A’ for ${\mathrm{O}}_2$ exposure) and may not be significantly deleterious to devices since the reduced WF is still higher than the ionization potential of most of the organic semiconductors. The appearance of the gap state ‘F’, which further reduces the WF from $\sim 5.8$ to 5.4 eV, is probably the most harmful to devices. Meyer et. al., have also reported no significant effect on the device performance with an air exposure of 3 min (${10}^{11}$ L), probably just before the appearance of the gap state F.³⁴ However, at present, an exact correlation of gap states is not well established and further investigations are required for this to be accomplished.

3.3.1.

Thermally evaporated $\mathrm{Mo}{\mathrm{O}}_x$ film

In Fig. 7, the XPS data of Mo $3{\mathrm{d}}_{5/2}$ and O 1s core levels, for 1, 2, 5, 10, 30, and 50 nm thicknesses, are plotted. All the spectra have been normalized to the same height for visual clarity. In Fig. 7(a), the 3d core level of molybdenum consists of two spin orbit interaction split peak, $3{\mathrm{d}}_{5/2}$ and $3{\mathrm{d}}_{5/2}$ with the peak separation of 3.15 eV and the intensity ratio of $3:2$. The $3{\mathrm{d}}_{5/2}$ peak, for 1 nm thickness, was observed at the binding energy (BE) of 231.97 eV. With the increasing thickness of $\mathrm{Mo}{\mathrm{O}}_x$ film, the 3d peaks kept gradually shifting towards the lower BE before reaching a saturation at around 30 nm. The overall peak shift, measured in the range of investigation, was 0.23 eV. The full width at half the maximum height (FWHM) of 3d peaks were measured to be $\sim 1.35\mathrm{eV}$. In Fig. 7(b), the oxygen 1s core levels, with a dominant peak at the BE of 529.81 eV and a very low intensity peak around $\sim 530.90\mathrm{eV}$, are observed for 1 nm thickness. With increasing thickness of the $\mathrm{Mo}{\mathrm{O}}_x$ film, the dominant oxygen peak kept gradually shifting towards the lower BE with the saturation setting in around 30 nm. The overall peak shift measured from 1 to 50 nm $\mathrm{Mo}{\mathrm{O}}_x$ film thickness was 0.26 eV. The FWHM for the dominant peak was $\sim 1.4\mathrm{eV}$ up to 5 nm, which increased to $\sim 1.55\mathrm{eV}$ for thicker films. A shift of $\sim 0.2\mathrm{eV}$ in the $\mathrm{Mo}{\mathrm{O}}_x$ work function (WF) and the valence band peak towards the lower BE has been reported with increasing film thickness from 2 to 30 nm on indium tin oxide (ITO) substrate (19). The shift of core levels, observed in the present case, are consistent with the WF shift which further suggests that the shift extend to deep core levels rather than just shallow lying valence band and vacuum level. We also observed Au 4f peaks, originating from substrate, for thin $\mathrm{Mo}{\mathrm{O}}_x$ depositions up to 5 nmwhich will be discussed later in this work.

Fig. 7

XPS spectra of (a) Mo 3d, and (b) O 1s core levels as a function of increasing $\mathrm{Mo}{\mathrm{O}}_x$ film thicknesses.

3.3.2.

Annealing of $\mathrm{Mo}{\mathrm{O}}_x$ films

In Fig. 8, the XPS data of Mo 3d and O 1s core levels are plotted for 1, 2, 5, 10, 30, 50 nm thicknesses, after annealing in air at 300  °C for 20 h. All the spectra have been normalized to the same height for visual clarity. In Fig. 8(a), the $3{\mathrm{d}}_{5/2}$ peak for 1 nm thickness was observed at the BE of 231.13 eV. With increasing thickness of $\mathrm{Mo}{\mathrm{O}}_x$ film, a complex peak shift pattern was observed. We also observed diffusion of gold from substrate towards the top surface through $\mathrm{Mo}{\mathrm{O}}_x$ film due to annealing for long time at high temperature. The complex peak shift pattern is originated due to diffusion of gold which will be discussed in the next section. The FWHM of 3d peaks was measured to be $\sim 1.35\mathrm{eV}$. In Fig. 8(b), the oxygen 1s core level, with a dominant peak at the BE of 530.08 eV and a low intensity peak around $\sim 531.44\mathrm{eV}$, was observed for 1 nm thickness. The complex peak shift, after annealing, was also present in O 1s core levels with increasing thickness of the $\mathrm{Mo}{\mathrm{O}}_x$ film.

Fig. 8

XPS spectra of (a) Mo 3d, and (b) O 1s core levels after the annealing, as a function of increasing $\mathrm{Mo}{\mathrm{O}}_x$ film thicknesses.

3.3.3.

Comparison of oxygen to molybdenum ratio

XPS spectra of Au 4f peaks for as deposited and after air annealing are presented in Fig. 9 in order to understand the changes brought in by the annealing. In Fig. 9(a), the 4f core levels of gold consist of two spin orbit interaction split peaks $4{\mathrm{f}}_{7/2}$ and $4{\mathrm{f}}_{5/2}$ with the peak separation of 4.3 eV and the intensity ratio of $4:3$. The intensity of 4f peaks kept continuously decreasing and, at 5 nm thickness of $\mathrm{Mo}{\mathrm{O}}_x$ layer, the intensity of this peak is insignificant. After annealing, the intensity of Au 4f peaks became more than 10 times higher for 1 nm $\mathrm{Mo}{\mathrm{O}}_x$ thickness, which strongly indicates the diffusion of Au from substrate towards the top surface. With increasing $\mathrm{Mo}{\mathrm{O}}_x$ thickness, the intensity of Au 4f peaks kept continuously decreasing. The intensity of Au 4f peaks at 10 nm $\mathrm{Mo}{\mathrm{O}}_x$ thickness, after annealing, is about four times higher that of 1 nm $\mathrm{Mo}{\mathrm{O}}_x$ for as deposited film. At the 30 nm film thickness, the intensity of Au 4f peaks are low, indicating that the gold diffusion does not significantly affect the surface for such thicknesses. As mentioned earlier in the annealing section, the complex peak shift in the Mo 3d and the O 1s peaks observed after the annealing is most likely originated due to the diffusion of gold.

Fig. 9

XPS spectra of Au 4f for (a) a deposit, and (b) after annealing. After annealing intensity of Au 4f peaks increased, indicating diffusion of gold from substrate towards the top surface.

In Fig. 10, the intensity ratio of oxygen to molybdenum are plotted for 1, 2, 5, 10, 30, and 50 nm thicknesses of $\mathrm{Mo}{\mathrm{O}}_x$ films before and after the annealing. As obvious from figure, we measured the oxygen to molybdenum ratio about 2.45, with small variation, for deposited film. After annealing, the ratio was initially higher than the stoichiometric value of 3.0 which quickly drops and saturates around $\sim 2.35$. Higher value of oxygen to molybdenum ratio, up to 10 nm, is due to higher oxygen content at these thicknesses. As discussed in the previous figure, the annealing of $\mathrm{Mo}{\mathrm{O}}_x$ thin film with gold substrate causes diffusion of gold atoms towards the surface and substantially enhances the contribution of gold in the surface chemical composition. The contribution becomes insignificant for the 30 nm thick film. Chemisorption of oxygen molecules on gold at high temperature has been reported earlier.³⁹ Therefore, the enhancement of oxygen intensity for the air annealed $\mathrm{Mo}{\mathrm{O}}_x$ film at the early stage is due to the chemisorption of oxygen on the diffused gold atoms. Since we do not have significant diffusion of gold on surface (or sub surface) for thicknesses $\ge 30\mathrm{nm}$, the contribution from the chemisorbed oxygen on gold is insignificant. The measured oxygen to molybdenum ratio, in the saturated thickness region, is slightly less ($\sim 2.35$) for the annealed $\mathrm{Mo}{\mathrm{O}}_x$ film than the as deposit film ($\sim 2.45$). Thus, the current investigation establishes that the air annealing does not significantly suppress oxygen vacancies present in the thermally evaporated $\mathrm{Mo}{\mathrm{O}}_x$ films.

Fig. 10

Oxygen to molybdenum ratio for as deposited and air annealed film with increasing thickness of $\mathrm{Mo}{\mathrm{O}}_x$ film. High ratio after the annealing for thin $\mathrm{Mo}{\mathrm{O}}_x$ film is due to the absorption of oxygen and moisture on diffused gold atoms.