3.1.

Effect of MoO_x Insertion Layer between Indium Tin Oxice and AlPc-Cl

It is well known that the control of the electronic structure and the complex interactions at the electrode/organic interface can significantly affect the device efficiency and lifetime.¹⁴⁻¹⁸ For example, a great deal of work has been performed to modify a conducting ITO anode^{26, 27} surface, which includes oxygen plasma, thermal annealing, and various kinds of insertion layers as well. A description of an organic semiconductor/electrode interface and energy level alignment can be found in recent review articles.^{28, 29} In a recent work from one of us, the effect of molybdenum oxide (MoO_x) insertion layer on both small molecule organic photovoltaic cells as well as polymer photovoltaic cells was studied. The results suggest that with the presence of MoO_x insertion layer, the power conversion efficiencies were enhanced by a maximum of 38% due to a significant enhancement in the fill factor.¹⁹ In the early studies of OLED devices, the improvement from MoO_x insertion layer is usually attributed largely to the reduction of the hole injection barrier at the organic hole transport layer/ITO anode interface. However, this may not explain the role of MoO_x insertion layer in OPV, because one is not injecting any charges in this case, rather extracting holes from the organic hole transport layer to the ITO anode. Therefore, further investigations are necessary to understand the origin of such improvement. In this section, we will discuss the electronic energy-level evolution at the chloro-aluminum pthalocyanine (AlPc-Cl)/ITO interface with and without a 10-nm MoO_x insertion layer and based on energy-level alignment, we will explain mechanism encouraging OPV device efficiency enhancement. We found that deposition of MoO_x abruptly increases the surface WF due to which all the energy levels of subsequently deposited AlPc-Cl shift toward the lower binding energy (BE) at the interface, and as a result, there is a strong hole accumulation at the AlPc-Cl/MoO_x interface. With increasing AlPc-Cl thickness, the energy levels shift back gradually toward the lower BE. This relaxation creates a band-bending-like region in the hole transport layer (AlPc-Cl) and introduces the hole drift component to the already existing hole diffusion current.

The electronic energy-level evolution of the cutoff and the HOMO region for AlPc-Cl deposited on ITO are shown in Figs. 1 and 1, respectively. The ITO substrate used in this experiment are ex situ OP treated, and WF values may be different from the in situ treated ones, which is consistent with earlier reports.²⁶ The ITO WF was measured to be 4.56 eV. In Fig. 1, we can see fast saturation of WF as overlaying AlPc-Cl thickness increases. We also note the overall WF shift is ∼0.2 eV and at the 16-Å AlPc-Cl deposition, the WF shift nearly saturated. At the 64 Å of AlPc-Cl deposition, the work function was measured to be 4.33 eV. From the data presented in Fig. 1, it is clear that the frontier-occupied levels were fully developed after 32 Å of AlPc-Cl deposition. Four peaks belonging to the AlPc-Cl are clearly visible in the scanned BE range. Binding energies of these peak are measured to be 1.45, 3.63, 5.23, and 6.72 eV at 64 Å of AlPc-Cl deposition.

Fig. 1

UPS data for 64, 32, 16, 8, 4, and 2 Å AlPc-Cl and ITO for (a) the cutoff region and (b) the HOMO region.

The electronic energy-level evolution shown in Fig. 2 addresses the question of efficiency enhancement reported with an MoO_x interlayer.¹⁹ All the spectra have been normalized to the same height for visual clarity. In Fig. 2, we observe an abrupt WF shift on the deposition of 100 Å of MoO_x. The ex situ OP treated ITO WF was measured to be 4.58 eV, and the deposition of MoO_x moved the surface WF value to 6.55 eV, which is ∼2 eV higher than the ITO surface. With the deposition of AlPc-Cl on MoO_x/ITO, the WF at first decreased rapidly until 16 Å and then the decrease became slower and more gradual. At 228 Å AlPc-Cl deposition, the surface WF was measured to be 4.92 eV. The HOMO features of AlPc-Cl were observed to develop between 8 and 16 Å. There was a shift ∼0.2 eV toward the higher BE in the MoO_x characteristic peak as indicated in Fig. 2, before AlPc-Cl feature were developed. Four strong peaks belonging to the AlPc-Cl occupied levels were observed in the experimental scan range of BE from –1 to 7 eV with respect to the Fermi level of the system. The BE of the occupied level peaks of AlPc-Cl were observed to be about 0.67, 2.81, 4.31, and 5.87 eV for 32 Å of AlPc-Cl. The other interesting effect due to the presence of MoO_x interlayer was that all the peaks in the HOMO region of AlPc-Cl kept gradually shifting toward the higher BE. The BE for four occupied level peaks were measured to be 1.11, 32.3, 4.85, and 6.31 eV for 228 Å AlPc-Cl thickness. The shift in the HOMO peak of AlPc-Cl in the investigated thickness range was ∼0.4 eV, as presented in Fig. 2.

Fig. 2

Electronic energy-level evolution (UPS data) for 228, 128, 64, 32, 16, 8, 4, and 2 Å AlPc-Cl, and ITO with 100 Å MoO_x interlayer for (a) the cutoff region and (b) the HOMO region.

The energy-level alignment diagrams for AlPc-Cl/ITO interface without and with a 100 Å MoO_x insertion layer are depicted in Figs. 3 and 3, respectively. A bandgap of 1.3 and 3.2 eV is determined for AlPc-Cl and MoO_x, respectively, from our UPS and IEPS data (not shown).²¹⁻²³ At the AlPc-Cl/ITO interface, an abrupt vacuum level shift of 0.2 eV is observed, which can be attributed as an interface dipole. Increasing thickness of AlPc-Cl results in a flat-band situation, in which the HOMO lies 1.0 eV below the Fermi level of the system. On the other hand, the insertion of the MoO_x layer leads to a large increase in the surface WF of ∼2 eV as shown in Fig. 3. Here, the valence-band maximum of the MoO_x layer is measured to be ∼2.53 eV below the Fermi level. The large WF of the metal oxide therefore causes an upward shift of all the energy levels of AlPcCl, which results in a strong hole accumulation at the AlPcCl/MoO_x interface. This hole accumulation is also reflected in the shift of 0.2 eV in the MoO_x peak toward higher binding energy, as a result of the electron transfer from AlPc-Cl to MoO_x. The net interface dipole is calculated to be 0.5 eV at the AlPc-Cl/MoO_x interface after the MoO_x shift is subtracted from the WF change at the interface. At the final thickness, we probed (∼200 Å), the onset of HOMO level of AlPcCl reached only ∼0.5 eV below the Fermi level, which is in sharp contrast to the sample without the MoO_x insertion layer. In the OPV cells with the MoO_x interlayer, the built-in field due to band-bending-like region enhances hole extraction from the AlPc-Cl layer to the anode, resulting in a net reduction of the cell series resistance and, thus, improving the device efficiency of OPV cells.^{19, 21}

Fig. 3

The energy level alignment at AlPc-Cl/ITO interface (a) without and (b) with 100 Å MoO_x insertion layer. Strong band bending is observed in the presence of the MoO_x interlayer between ITO and AlPc-Cl.

3.2.

Thickness Dependence of MoO_x Insertion Layer

After establishing the advantageous nature of the transition metal oxide insertion layer between conducting ITO and organic semiconductor, it is crucial to further conduct a detailed investigation of the anode/organic interface with different thicknesses of the insertion layer and find out the optimum thickness. It is important to point out at this point that, with the transition metal oxide insertion layer, there are two competing mechanism of opposite nature. The first, which is beneficial in enhancing the device performance, is high WF of the insertion layer that causes the energy levels of the organic material to shift toward the lower binding energy and thus reducing hole-injection barrier. The other is its own resistance, which will increase with the thickness of the insertion layer. In this section, we will discuss this issue and explore the optimum thickness of the insertion layer on the basis of energy-level alignment.

In order to avoid variations originating from different sample preparation conditions, we designed a substrate with a gradually changing thickness of MoO_x. The ITO substrate used in this experiment was OP treated in situ. The surface WF uniformity of OP-treated ITO was measured with UPS, and the uniform region with WF of 5.22 ± 0.1 eV was selected for further sample preparation and measurements. We deposited 1.25 × 6 mm strips of MoO_x with thicknesses ranging from 0 to 300 Å in seven steps onto the OP-treated ITO substrate. A thin metal shutter parallel to the substrate surface with a sharp edge was attached to a micrometer, about 1 mm below the substrate to expose the specific area of substrate for MoO_x deposition. The shutter was adjusted to fabricate MoO_x films of stepped thicknesses. Subsequently, AlPc-Cl was deposited layer by layer (other modes of growth are not completely ruled out) on the stepped MoO_x substrate.

In Fig. 4, the UPS spectra of the MoO_x/ITO interface are presented as a function of thickness of the MoO_x layer. The cutoff and HOMO regions of the UPS data are plotted in Figs. 4 and 4, respectively. The schematics of MoO_x with stepped thicknesses 0, 10, 20, 50, 100, 200, and 300 Å are shown in the inset of Fig. 4. The WF increase for the 10 and 20 Å MoO_x on ITO was measured to be 0.66 and 1.42 eV, respectively, and it remained almost the same for further MoO_x deposition, indicating saturation at ∼20 Å. From Fig. 4, a characteristic peak of MoO_x can be observed to develop and saturate at 4.0 eV. It is quite evident that in the HOMO region, the MoO_x features became prominent after 20 Å, and after 50 Å, all spectra remained almost the same. It further supports the notion of the electronic energy-level saturation between 20 and 50 Å. It is not surprising that the WF saturation occurs sooner than the HOMO because the former is related only to the surface properties of the material.

Fig. 4

UPS data for 300, 200, 100, 50, 20, and 10 Å MoO_x and in situ OP-treated ITO for (a) the cutoff region and (b) the HOMO region. Inset: Schematics of gradual thickness of MoO_x.

In Fig. 5, the WF versus MoO_x interlayer thickness (█) is plotted for 4 Å (•), 8 Å (▴), 32 Å (▾), and 128 Å (⧫) AlPc-Cl coverages. It is simple to deduce the saturation of the WF of the MoO_x interlayer at the thickness (█ data points) of 20 Å. At 128-Å AlPc-Cl coverages (⧫), the WF of AlPc-Cl saturates around 4.9 eV, irrespective of MoO_x interlayer thickness, representing the saturation of band-bending-like situation in AlPc-Cl. The 32-Å AlPc-Cl coverages have a similar trend, with WF values of <0.2 eV higher than that of the 128-Å coverage. The 4- and 8-Å AlPc-Cl data show clear MoO_x interlayer dependence. In the absence of the MoO_x interlayer, the variation in WF values is <0.2 eV for all four AlPc-Cl coverages. With increasing interlayer thickness, the WF saturation requires more AlPc-Cl to be deposited. We observed that the BE drop across MoO_x interlayer is ∼0.2 eV by AlPc-Cl, regardless of the thickness of the MoO_x. The data in Fig. 5 indicate that, for the final 128 Å thickness of AlPc-Cl, there is about the same WF difference from MoO_x for interlayer ≥20 Å, but it clearly shows slower relaxation in WF for thicker MoO_x interlayer. Because the drift field is the gradient of built-in potential, swifter relaxation results in larger drift field at the interface. This in turn would result in higher hole current. Because of the trade-off between the drift-induced enhancement and increased resistivity of MoO_x, 20 Å is most likely the optimum choice of insertion-layer thickness. The prediction of optimum thickness solely on a device-performance basis was recently reported by Cattin et al.,³⁰ which is quite consistent with our UPS results. In Fig. 5, AFM data are presented for 228-Å-thick AlPc-Cl film. The scan area shown is 2 × 2 μm. The root-mean-square (rms) roughness of the film was measured to be 27 Å.

Fig. 5

(a) Work function versus MoO_x interlayer thickness for 4 Å (•), 8 Å (▴), 32 Å (▾) and 128 Å (⧫) AlPc-Cl coverages. (b) AFM data for 2 × 2μm area of 228-Å-thick AlPc-Cl film.

The establishment of the fact that the high WF of the insertion layer is crucial in the uplift of all the frontier energy levels of the organic hole transport layer toward the lower BE at the organic/insertion layer interface raises a question. If the insertion layer surface WF is smaller, then what impact this can have on the device performance? This question needs to be addressed because very recent reports^{24, 25, 31} have established that the WF of vacuum-grown MoO_x film reduces to much lower values after the vacuum is break, which is usually the case with device fabrication. Mayer et al. have reported³¹ hole-injection properties remain unaffected, while both our OPV²⁴ and OLED²⁵ results—investigated at two different collaborating laboratories—indicate that a vacuum break before the deposition of the organic hole transport layer adversely affects the device performance. In Secs. 3.2 and 3.3, we will discuss the issue of WF reduction due to controlled air and oxygen exposures.

3.3.

Effect of Air Exposure to MoO_x Film

In this section, we will address another crucial issue that has been a cause of concern and much debate among experts. The beneficial nature of the MoO_x insertion layer is well established, but there were two completely different explanations for this enhancement. The pivotal point in resolving the discrepancy is the WF of MoO_x insertion layer. The reported values of WF are in the range of 5.3–6.9 eV.^{17, 19−23} It is imperative to resolve the discrepancy in order to understand how to control the device fabrication and to obtain consistent device performance. In this section, we report the effect of air exposure to MoO_x films. We found that the vacuum-deposited MoO_x films started with high initial WF values, but with air exposure, the surface WF gradually reduced to substantially lower values.

In Fig. 6, the UPS spectra of the MoO_x on ITO are presented as a function of the air exposure. Figures 6 and 6 show the cutoff and valence-band regions of the UPS data, respectively. All the spectra have been normalized to the same height for visual clarity. Short bars are placed at the centroid of MoO_x HOMO peaks to track the shift in the peak BE values. The in situ OP-treated ITO WF was measured to be 5.50 eV, which increased to 6.75 eV upon the deposition of 50 Å MoO_x. From Fig. 6, a characteristic peak of MoO_x can be observed at ∼3.91 eV and HOMO energy-level onset was measured to be 2.69 eV. With air exposure, the surface WF kept continuously decreasing and, at the final step of 2 × 10¹⁴ L exposure, the surface WF was measured to be 5.32 eV, which is 1.44 eV lower than the initial surface WF. The HOMO peak has smaller but continuous shift toward the higher BE, and at the 2 × 10¹⁴ L exposure, the HOMO peak was measured to be 4.33 eV, ∼0.42 eV higher than the initial unexposed value. In Fig. 5, AFM data for 50 Å MoO_x film is presented. The scan area shown is 5 × 5 μm. The rms roughness of the film measured was ∼46 Å.

Fig. 6

UPS data for ITO; 50 Å MoO_x; and 10², 10⁴, 10⁶, 10⁸, 10¹⁰, 10¹¹, 10¹², 5 × 10¹², 10¹³, 5 × 10¹³, 10¹⁴, and 2 × 10¹⁴ L air exposure in (a) the cutoff region and (b) the HOMO region. Substantial to low WF reduction is observed, depending on the amount of air exposure. (c) AFM image of 50-Å MoO_x film is presented.

In Fig. 7, the binding-energy evolution of the core levels of Mo 3d_5/2 and 3d_3/2 [Fig. 7] and O1s [Fig. 7] are presented as a growing amount of air exposure. It is straightforward to conclude that no significant core level peak shift has occurred due to the air exposure. The full width at half the maximum (FWHM) peak height of O 1s peak was 1.8 eV. Not only the peak BE, but also the FWHM remained the same for O 1s core level. The FWHM for Mo 3d 5/2 peak in the beginning was found to be 1.6 eV. With air exposure, however, FWHM for the Mo 3d 5/2 peak kept increasing. At the final 2 × 10¹⁴ L air exposure, it was calculated to be 1.9 eV. The separation between spin-orbit splitting levels Mo 3d_5/2 and 3d_3/2 is 3.10 eV, which is consistent with reported values in the literature.³³ It is interesting to note that the core level positions remain unchanged during the exposure, in contrast to the valence peak shift of 0.42 eV toward higher binding energy. This indicates that the shift is confined to the valence states due to the chemisorption of the gas molecules.

Fig. 7

XPS data for, (a) the Mo 3d and (b) the O 1s core level for, ITO, 50 Å MoO_x, and 10², 10⁴, 10⁶, 10⁸, 10¹⁰, 10¹¹, 10¹², 5 × 10¹², 10¹³, 5 × 10¹³, 10¹⁴, and 2 × 10¹⁴ L air exposure.

In Fig. 8, the growth of a gap state near the Fermi energy level is shown as a function of air exposure. The broad peak at ∼2.0 eV is a satellite of the HOMO from the He β line at 23.09 eV from our unfiltered He I (21.22 eV) ultraviolet photon source.³³ With the air exposure, the gap state at ∼1.1 eV grows. The gap-state peak is clearly visible from 10⁴ L onward, which becomes more prominent with further air exposure. It can be understood that the electron transfer from the adsorbates to MoO_x dopes the latter at the surface, and as a result, the HOMO of MoO₃ shifts downward by 0.43 eV, as shown in Fig. 6. From Fig. 8, it is clear that the gap state is not intrinsic to the MoO_x but originates from electron transfer from adsorbates sticking to the MoO_x film at the surface. The effect of this gap state on device performance has not been determined.

Fig. 8

Near HOMO region UPS data for 50 Å MoO_x and 10², 10⁴, 10⁶, 10⁸, 10¹⁰, 10¹¹, 10¹², 5 × 10¹², 10¹³, 5 × 10¹³, 10¹⁴, and 2 × 10¹⁴ L air exposure.

3.4.

Effect of Oxygen Exposure to MoO_x Film

After establishing the WF reduction of MoO_x film on exposure to air and resolving the discrepancies in the reported MoO_x WF values in the literature, in this section we will discuss the effect of oxygen exposure on MoO_x films. An oxygen exposure experiment along with the previous air exposure experiment would allow us to decouple the effect of oxygen molecules and humidity on the surface of MoO_x film. We found that the vacuum-deposited MoO_x film started with high initial WF values, but with oxygen exposure, the surface WF gradually reduced to substantially lower values, but the near saturation of the WF was observed higher than the air-exposed MoO_x films.

In Fig. 9, the energy-level evolution of 50-Å MoO_x film on ITO are presented as an increasing amount of oxygen exposure. The cutoff and valence-band regions of the UPS data are presented in Figs.9 and 9, respectively. All the spectra have been normalized to the same height for visual clarity. Short bars are placed at the centroid of MoO_x HOMO peaks to track the shift in the peak BE values. The in situ OP-treated ITO WF was found to be 5.58 eV. The deposition of 50 Å MoO_x lead to a high increment of 1.16 eV in the surface WF. From Fig. 9, a characteristic peak of MoO_x can be observed at ∼3.9 eV and HOMO energy-level onset was measured to be 2.66 eV. With oxygen exposure, a continuous surface WF reduction was observed, which saturated around ≥10¹³ L oxygen exposure. At the final step of 5 × 10¹⁴ L exposure, the surface WF was measured to be 5.75 eV, which is 1 eV lower than the initial MoO_x surface WF. The HOMO peak as well has smaller but continuous shift towards the higher BE, and at the 5 × 10¹⁴ L oxygen exposure, the HOMO peak was measured to be 4.04 eV. With the oxygen exposure, the surface WF kept continuously decreasing in a similar fashion as in the air exposure experiment except for exposures of >10¹⁴ L.

Fig. 9

Electronic energy-level evolution of oxygen-exposed MoO_x film. The UPS data presented is for ITO, 50 Å MoOx, and 1, 10², 10⁴, 10⁶, 10⁸, 10¹⁰, 10¹¹, 10¹², 5 × 10¹², 5 × 10¹³, and 5 × 10¹⁴ L oxygen exposure (a) the cutoff region and (b) the HOMO region. Continuous reduction in surface WF is demonstrated.

In Fig. 10, the core energy-level evolution of (a) Mo 3d_5/2 and 3d_3/2, and (b) O1s are exhibited as a growing amount of oxygen exposure. Once again, there was no appreciable shift in the peak binding position of both Mo 3d and O 1s core levels. No significant peak broadening was observed in O 1s core levels from the initial thermally evaporated value of 1.78 eV. However, some peak broadening was again observed in Mo 3d_5/2 peak. The initial FWHM for Mo 3d_5/2 peak was 1.63 eV, which became 2.00 eV for 5 × 10¹² L oxygen exposure. The separation between spin-orbit splitting levels Mo 3d_5/2 and 3d_3/2 were 3.10 eV, which is consistent with reported values in the literature.³²

Fig. 10

Core energy-level evolution of MoO_x film. XPS data for, (a) the Mo 3d and (b) the O 1s core level for, ITO, 50 Å MoO_x, and 1, 10², 10⁴, 10⁶, 10⁸, 10¹⁰, 10¹¹, 10¹², 5 × 10¹², 5 × 10¹³, and 5 × 10¹⁴ L oxygen exposure.

In Fig. 11, the growth of a gap state close to the Fermi energy level is shown as a function of oxygen exposure. The growth of a gap state at ∼1.1 eV is observed with the oxygen exposure. The gap-state peak is clearly visible from 10⁴ L onward, which becomes more prominent with further air exposure. Nakayama et al. reported a similar interface electronic state at ∼1 eV on initial deposition of MoO_x on poly(dioctylfluorine-alt-benzothiadiazole) (F8BT), which attenuated after 23 Å of MoO_x.³⁴ Our data in Fig. 3 suggest that the gap state observed by Nakayama et al.³⁴ is from electron transfer from F8BT to MoO_x. At present, it is not clear what effects this gap state has on the device performance. It is also interesting to note that Kroger et al. meticulously ruled out the existence of a gap state in MoO_x.^{20, 23} Our results indicate that the gap state indeed is not intrinsic to MoO_x. Instead, it is by charge transfer to MoO_x at the interface from interaction with other materials.

Fig. 11

Near HOMO region UPS data in situ OP-treated ITO, 50 Å MoOx, and 1, 10², 10⁴, 10⁶, 10⁸, 10¹⁰, 5 × 10¹², and 5 × 10¹⁴ L oxygen exposure. The growth of gap state at ∼1.1 is clearly visible.

The reduction in the WF and growth of the gap state are consistent to a physical model of electron transfer from chemisorbed oxygen to MoO_x, forming an interface dipole layer pointing toward the MoO_x surface. Because the oxygen ionization potential is high (12.063 eV),³⁵ the dielectric screening by the MoO_x has to be included to facilitate such a charge transfer.³⁶ Below a ≤10¹³-L exposure charge transfer from an oxygen molecule seems to be the dominant mechanism. Beyond 10¹³ L of air exposure, the reduction of the work function is most likely from the polarization of the adsorbed water molecule because the gap-state intensity no longer increases at this stage, as shown in Fig. 11.

The UPS and XPS studies illustrated the evolution of the electronic structure of the MoO_x during the interface formation with organic materials and with exposure to oxygen or air. It can be concluded that the high WF of MoO_x is the dominant factor that brings the HOMO level of the organic toward the Fermi level of the anode and creates a band-bending-like region that encourages holes to drift toward the anode. The reduction of WF by gas exposure can be clearly correlated to the deteriorating device performance.²¹ The low LUMO energy of MoO_x should also be an important factor contributing to the carrier transport.²⁰ However, its significance is secondary in comparison to the WF as the gas exposure brings the frontier orbitals even lower by ∼0.4 eV while the device performance is deteriorated.²¹ It can also be concluded that the gap state is associated with the electron doping of MoO_x by organic materials or other adsorbates. It grows and saturates at high gas exposures, opposite to the trend of the device performance. It therefore remains questionable how much the gap state contributes to improvement of the device performance.