X-ray Diffraction

A x-ray diffraction process is shown in Fig.3-8, where the incident X-ray in a crystal results in the reflective x-ray that following the bragg law:

n= 2dsin 

where n is the order of diffraction corresponding to n =1, 2, 3,… ..,is wavelength of x-ray,

d is distance between crystal planes,  is the angle between incident x-ray and crystal plane.

Figure 3.8 Schematic diagram shows the XRD process

XRD generally used to identify the crystalline materials or to determine the crystal phase of materials. In this study, we used XRD to provide the lattice constant of crystalline core-materials. By comparing the experimental XRD data and the standard data from International Center Diffraction Data (ICDD) or formerly known as Joint Committee on Powder Diffraction Standards (JCPDS), we can evaluate the lattice strain for the electrode.

The standard database is available from the diffraction equipment manufacturers or from ICDD.

Chapter 4

Interfacial Energy Levels of Al

O

Films on TiO

Electrodes of Dye-Sensitized Solar Cells

To achieve a comprehensive study of the physical, chemical, and electrical properties of the Al2O3 overlayers, we studied the ALD Al2O3/TiO2 electrodes on the interfacial energy level. We examine low-temperature ALD process for forming the Al2O3 barriers to achieve compatibility with low-temperature DSSCs fabrication processes. The interfacial energy level related to power conversion efficiency (PCE) were investigated by using TEM, ultraviolet photoelectron spectroscopy (UPS), x-ray photoelectron spectroscopy (XPS), and reflective electron energy loss spectroscopy (REELS). The nano-structure and the influence of ALD Al2O3layers on nanoporous TiO2 electrodes in DSSCs were also studied using these characterization techniques.

4.1 Microstructure and Chemical Analysis

The TEM images in Fig. 4.1(a) show the nanocrystalline TiO2 without the ALD Al2O3

overlayer; Fig. 4.1(b) shows the nanocrystalline TiO2 with the 10-cycles ALD Al2O3

overlayers. Both of the samples presented spherical-like, uniform, interconnected TiO2

nanoparticles that were about 20 nm in size. The uniform, over-coated Al2O3 layers can be seen in part of the TEM image have been identified as amorphous structure by XRD. The thicknesses of 1-cycle, 2-cycle, 5-cycle, 10-cycle, and 20-cycle ALD Al2O3 overlayers determined by TEM, which are about 0.2 nm, 0.5 nm, 1.0 nm, 1.9 nm and 2.8nm, respectively, indicated a significant discrepancy of deposition rate (nm/cycle) for different cycles. The

discrepancy may be ascribed to the growth-per-cycle (CM) of the ALD Al2O3 layers, as shown in the equation (4-1) [81]:

CM = h x (NA / M) (4-1) where the CM defined as the number of atoms M (here is Al) adsorbed per unit surface area per cycle, h is the thickness increment per cycle, is the density of layer and NA is the number of atoms per mole (6.02214 x 10²³ mol^-1) , M is the molar mass of layer. By substituting the M=50.98 g/mol and = 2.9 x 10^-21 g/nm³ [68] into the equation, Fig. 4.1(c) shows the growth-per-cycle of Al2O3 layers via the Al2O3 thickness increment per cycle. The growth-per-cycle increased to a maximum and then decreased to the steady value, referred to as substrate-inhibited growth of the ALD Al2O3 layers. This result is consistent to Puurunen’s study suggested island growth as the origin of this type of substrate-inhibited growth in ALD Al2O3 layers [82].

Figure 4.1 (a) TEM image of the nanocrystalline TiO2 without ALD Al2O3 overlayer. (b) TEM image of the nanocrystalline TiO2 with 10-cycles ALD Al2O3 over-layer. (c) Result of the growth-per-cycle of Al2O3 layers via the Al2O3 thickness increment per cycle.

After the peak fitting was performed by using the Shirley background subtraction and Gaussian/Lorentzian functions, the normalized XPS spectrum of the as-received TiO2

electrode shown in Fig. 4.2(a) indicates that the Ti 2p3/2 core level of the nanoporous TiO2

electrode located at 458.9 eV lies within the range of 458.8-459.4 eV reported in the literature for TiO2[83, 84]. Therefore, charging effects and subsequent shifts in the photoelectron peak positions were considered to be negligible. Following the ALD process, first, the binding energy of the Ti 2p3/2 core level was shifted by –0.4 eV to 458.4 eV and then shifted by +0.3 eV to 458.7 eV after the deposition of 10-cycle Al2O3 over-layers. These shifts of the Ti 2p3/2

core levels partially indicated an air-surface reaction existing within the 0.2-nm thickness of the Al2O3 over-layer. The interfacial reaction can be seen more clearly in the Al 2p spectra shown in Fig. 4.2(b). There are two groups of the Al 2p peaks, named S1 and S2, in the Figure. The S1 peaks located at 74.1 eV were identified as Al2O3 according to previous literature [84], and they shifted by 0.5 eV to 74.6 eV after 10-cycle deposition. Another peak, which was labeled S2 and shifted from 75.8 eV to 76.1 eV, was identified as Ti-O-Al(OH)2, according to our previous model [19]. In our previous model, the thickness of a 1-cycle Al2O3

overlayer equals that of the Ti-O-Al(OH)2 monolayer, or 0.3 nm, which is close to our measured values. It should be noted that the Ti-O-Al(OH)2 and Al2O3 were formed when the sample’s surface reacting to the atmosphere, there is another one like “ Ti-O-Al(OH)-O-Ti”, with only one hydroxyl on Al atom to maintain a constant growth rate each cycle, should exist in ALD vacuum chamber. The fact that the S2/S1 peak area ratios listed in table 4-1 were inversely proportional to the number of cycles indicates that the Ti-O-Al(OH)2 monolayer, which is located close to the Al2O3/TiO2 interface, could be a possible cause of the negative binding energy shifts of the Ti 2p3/2 core levels.

Figure 4.2 XPS results showing (a) Ti2p3/2 and (b) Al2p core level photoelectron peaks from ALD Al2O3coated TiO2electrodes.

Table 4-1 XPS results of the ALD Al2O3 overlayers on TiO2electrodes.

Deposition

2-cycles 458.3/ -0.2 74.3/ 0.2 76.1/ 0.2 0.08

5-cycles 458.7/ -0.1 74.5/ 0.4 76.0/ 0.1 0.08

10-cycle 458.7/ -0.1 74.6/ 0.5 76.0/ 0.1 0.03

4.2 Interfacial Energy Levels Analysis

UPS is a powerful technique for measuring the valence band maximum (VBM) and work function (WF) of ultra-thin film; in addition, it gives the most directive figure about the energy level alignment at the interface between two layers [76]. Figures 4.3(a) and 4.3(b) show the UPS spectra obtained from the as-received and ALD-deposited TiO electrodes. The

VBM relative to the Fermi-level (EF) was determined by the values of cutting points in the VB region of UPS assuming that EF is 0 eV, and the resulting values are listed in table 4-2.

Following the ALD process, the VBM was shifted by 0.9 eV from 3.7 eV to 4.9 eV, which is larger than the shifts in core levels, and it is believed that the VBM shifted toward the reported value of 5.6 eV for pure Al2O3 thin film [85]. On the other hand, the work function (WF) was measured from the values of cutting points in the secondary electron (SE) region of the UPS, as can be seen in Fig. 4.3(b). It should be noted that the WF of ultra-thin Al2O3

overlayers increased from 4.7 eV to 5.1 eV at the surface in the first ALD cycle process, which indicates that the band is bending downward across the surface to the interface, and the difference of WF, which induced a 0.4 eV built-in potential (Vbi), accelerated the electron transport from Al2O3 over-layers to the TiO2 electrode. The WF of ultra-thin Al2O3 overlayers decreased with subsequent 2-10 cycle ALD deposition and shifted to 4.4 eV after the 10-cycle ALD deposition process. A similar trend was found in the reference [86], where the interfacial dipole layers were regarded as the main factor that determined the work function of ultra-thin Al2O3 films on a Cu-9%Al surface. The thickness dependence of the WF for the Al2O3

overlayers revealed the interfacial charge stability of the dipole layers, and this is shown in Fig. 4.4.

Table 4-2 Thickness, energy levels of the ALD Al2O3 overlayers of the DSSCs.

Deposition

The REELS spectrum of as-received TiO2 in Fig. 4.3(c) shows that the onset of electron

signals located at 3.9 eV lies within the band gap value reported in the literature for nano-crystalline TiO2, which indicated the blue-shift or quantum confinement compared to 3.2 eV for a large grain size of TiO2 [87, 88]. By considering the depth sensitivity in the REELS spectra, we calculated the depth of electron penetration (X) from the following equation [89]:

X (m) =0.1E^1.5/d (4-2)

where E is the e-beam energy (KeV), and d is the sample density (g/cm³) in the REELS spectra. Substituting the experimental e-beam energy value (17 eV), ALD Al2O3 density value (2.9 g/cm³) [68], and TiO2 density value (3.893 g/cm³, JCPDS-ICDD No:31-1272) into the equation (4-2), the resulting depths of electron penetration in our REELS analysis were determined to be about 0.07 nm for Al2O3 and about 0.05 nm for TiO2 materials. This result may guarantee that our REELS spectra are depth-sensitive enough to the Al2O3 over-layers for measuring their band gap. After the deposition of the Al2O3 overlayers, the onset of electron signals shifted from 4.0 eV to 5.9 eV as the thicknesses of the Al2O3 over-layers increased as shown in Fig. 4.3(c). The trend are similar to the earlier reports that used REELS to measure the band gap of Al2O3ultra-thin films from 2.5 eV for the surface to 8.7 eV for the bulk state, and the lower band gap value is mainly due to structural surface states, such as defects, Al spz dangling bond [90], and the amorphous structure [91]. The present results are consistent with earlier findings in REELS studies that suggested that the band gap of the Al2O3 overlayer is positively correlated with its thickness at the sub-nanometer scale due to the fact that the ratio of reflective electrons from structural surface-states is a major portion of the REELS spectra in ultra-thin Al2O3 films.

Figure 4.3 (a) UPS valence band region and (b) UPS secondary electron region of the as-received and ALD Al2O3 deposited TiO2 electrode. (c) REELS spectra of the as-received and ALD deposited TiO2 electrode.

The work function (WF) changes as a function of the thickness of the Al2O3 over-layers (Fig. 4.4), indicate that the magnitude and the direction of the core level shifts are in excellent agreement with the decrease in WF. The measured WF of the pure, nanoporous TiO2

electrode was 4.7 eV, which is close to the WF of 4.8 eV of the pure TiO2(110) surface [92].

Following the deposition of the Al2O3 over-layers, the shifts of Ti 2p3/2 and Al 2p XPS peaks showed a similar trend to that of the WF. Significant changes of WF and Ti 2p3/2 core level by 0.4 eV at about a 0.2-nm thickness of Al2O3 deposition indicate the occurrence of an interfacial reaction between the Al2O3 over-layer and the TiO2 electrode. This could be due in part to the formation of Ti-O-Al(OH)2, as previously mentioned when we discussed Fig. 4.2.

The reduction of WF, when the thickness of the Al2O3 over-layers exceed 0.2 nm, was associated with the increase of binding energy of Ti 2p3/2 and Al 2p core levels. These results

were consistent with previous studies [77, 86], and they could be attributed to the interfacial Ti-O-Al(OH)2 and the formation of dipole layers, which stabilized after the formation of two Al2O3layers.

Figure 4.4 Changes of the work function, Ti2p3/2 XPS peak and Al2p XPS peak with various thicknesses of Al2O3layers.

We have repeated the UPS/XPS and REELS measurements several times, indicating the data’s error less than 0.1 eV. The accuracy or resolution of our data was calibrated before every spectral experiment by following analysis: The “resolution” of our electron spectroscopy determined as the energy between 80% to 20% intensity of pure Ag Fermi-edge was ~98 meV. But the real ability to resolve signals was determined as <20 meV using the same spectral condition in this study and tested a pure Ag surface from sample bias in the range of -5.00 eV to -5.04 eV by increments of 0.02 eV. The UPS deviations were smaller than 0.005 eV, and it could be assured that the resolving power was < 0.02 eV. Therefore, the

larger data’s error was mainly resulted from the deviations of samples preparation.

4.3 Energy Level Alignment at the Interfaces of Al

O

/ TiO

Electrodes

The measured power conversion efficiency (PCE) of the DSSCs in Table 4-3 shows that the optimal ALD Al2O3 layer thickness is one monolayer, or 0.2 nm, which is considerably thinner than the reported 0.9-2.5 nm for sol-gel Al2O3 films. As discussed in our previous work, we ascribe this discrepancy to the limited film-thickness resolution of the sol-gel method, in which the minimum achievable Al2O3 thickness was ~1 nm, and the poorer infiltrating ability of the liquid precursors of the sol-gel process, which tends to cause overestimation of the Al2O3 thickness (i.e., the sol-gel films may accidentally be thinner on the electrode surfaces that are more difficult for the precursors to reach, where the thickness may fall into the desirable Al2O3layer thickness range.)

Table 4-3 Cell performance of the DSSCs containing the ALD Al2O3 overlayers with different thickness.

To discuss the effects of Al2O3 overlayers on the power conversion efficiency (PCE) of DSSCs, a schematic band diagram is presented in Fig. 4.5 with three interfacial energy parameters i.e., the difference of WF between Al2O3 and TiO2 overlayer (), charge

recombination barrier height (RB), and the interfacial energy barrier height between N719 dye and Al2O3 (*IB). The conduction band minimum (CBM) relative to the Fermi-level was calculated from the aforementioned VBM and band gap data as: CBM = Band gap - VBM. It is reasonable to correlate these three parameters with PCE based on the concept of electron transportation, which suggests positive values of , high RBvalues, and low *IBvalues, resulting in higher PCE for DSSCs. By assuming that the Lowest Unoccupied Molecular Orbital (LUMO) of N719 in the DSSCs was 0.4 eV higher than the conduction band minimum (CBM) of TiO2 [93], the interfacial energy barriers between N719 dye and the Al2O3 layer (*IB) can be calculated quantitatively to evaluate its effects on PCE.

Figure 4.5 Schematic band diagram with three interfacial energy parameters: the difference of WF between Al2O3and TiO2 overlayer (), the charge recombination barrier height (RB) and the interfacial energy barrier height between N719 dye and Al2O3 (IB).

As a result, Fig. 4.6 shows the relationship between the energy levels’ differences (,

RB, *IB) and PCE with various thicknesses of Al2O3 layers. The efficiency was increased initially by 13% at ~0.2-nm thickness of Al2O3 layer, and then it dropped dramatically to almost zero percent after the thickness of the Al2O3layers increased. The tunneling effect [94]

which will exponentially decrease the current from dye to TiO2electrode with the thickness of Al2O3overlayers increased was found from the open-circuit voltage (Voc), short-circuit current (Jsc), and fill factor (FF) listed in table 4-3. Accordingly, the drop of the PCE maybe ascribed to tunneling effect when the thickness of Al2O3overlayers increased and the initial increase of PCE and Jsc at ~0.2-nm thickness of Al2O3 layer maybe attributed in part to the reduction of recombination. On the other hand, the increase of Voc might exist few possible reasons: One possible reason is that the Al2O3overlayer provides a negative conduction band by the lower electron affinity [55]; however, the electron affinity of Al2O3 overlayer at ~0.2-nm thickness is higher than TiO2 electrode in the present study (see table 4-2); therefore, this reason can be excluded. Another ideal reason is more likely that the positive work function difference (_) provided a built-in potential (eVbi) at the semiconductors’ heterojunction [95] to increase the Voc of the DSSC. Therefore, the positive _ (~ 0.4 eV) value of Al2O3

overlayer may be regarded as a important factor to increase the PCE of the DSSCs.

It can be seen that the _ of 0.4 eV and RB of 0.1 eV, which were associated with 0 eV *IB, should be the key factors for the PCE increase after the first monolayer or ~0.2 nm of ALD Al2O3 layer deposition. The significant reduction of PCE was accompanied by decreasing values of  and increasing values of *IB at the following deposition of 2-cycle ALD Al2O3layers. This is consistent with the concept of semiconductor physics, that there will be no difficulty for electron transfer to the Al2O3layers, especially if *IB remains at 0 eV. The evidence suggests that the change in efficiency of the DSSCs that resulted from the

various thicknesses of the Al2O3 overlayers can be correlated with the difference of work function (_), recombination barrier height (RB), and interfacial energy barrier height (*IB).

Figure 4.6 Interfacial energy levels difference (, RB, *IB) and PCE via the thickness of Al2O3layers.

It should be noted that the linear and parallel increases of RB and *IB with the layer thickness is not related to a similar increase in . This can be attributed to the changes of the band gap by ultra-thin Al2O3 structure and the changes of the work function by the dipole layers and Ti-O-Al(OH)2, as the aforementioned description of Fig 4.3, 4.4 and 4.5. The RB

(~0.1 eV) of the thinnest Al2O3 layer may increase the electron concentration in the TiO2

electrode by reducing the electron transported across the Al2O3layer. The surface passivation

effect could be induced by the increase of the electron concentration in the TiO2 electrode, thus enhancing both the photovoltage and photocurrent of the DSSC [96].

The results suggest two mechanisms correlated with the paths of electron transfer which is illustrated in Fig. 4.7. The first mechanism revealed that electron injects from dye into TiO2

without any energy barrier and therefore improve the cells PCE due to the proper ultra-thin Al2O3 energy levels in Fig. 4.7(a). The second paths of electron transfer suggested that electron injects from dye into TiO2by tunneling effect in Fig. 4.7(b) when the Al2O3 structure is thicker and therefore decreases the cells PCE.

Figure 4.7 Schematic band diagrams illustrate that the paths of electron transfer from dye into TiO2(a) without any energy barrier and therefore improve the cells PCE due to the ultra-thin Al2O3energy levels, and (b) by tunneling effect when the Al2O3layer is thicker.

The dependence of the interfacial energy levels on the thickness of the Al2O3overlayers may explain the fluctuation of the PCE in the dye-sensitized TiO2 solar cells. It also has implications for the design and optimization of the DSSCs, especially those based on nanoporous TiO2 electrodes. Regarding thickness control, it is necessary to understand the

mechanism of growth of the ALD Al2O3 overlayers on the TiO2 electrodes. The mechanism may be related to the island growth modes resulting in a low coverage of ALD Al2O3

overlayer on nanoporous TiO2 electrode [82]; therefore, in this study, the ALD may not have so much advantage than sol-gel techniques due in part to the low coverage of ALD Al2O3

overlayers on nanoporous TiO2 electrodes was not optimized at present stage. However, since the coverage of ALD Al2O3 overlayers on nanoporous TiO2 electrodes is not known, further experiments are in progress to clarify this issue.

4.4 Summary

A low-temperature (150 °C) ALD process was used to obtain ultra-thin Al2O3

over-layers on a TiO2 electrode in dye-sensitized solar cells. The optimal Al2O3thickness was determined to be about 0.2 nm, and this thickness resulted in an enhancement of the power conversion efficiency (PCE) by ~13%, which is a significant increase over the DSSCs without Al2O3 overlayer. Thicker Al2O3 overlayers moved the valence band maximum and band gap toward those of pure Al2O3. The core level XPS peaks showed the formation of Ti-O-Al(OH)2

and dipole layers at the Al2O3/TiO2 interface and exhibited a strong influence on the work function of the Al2O3 overlayers. A work function difference (_) of 0.4 eV and a recombination barrier height (RB) of 0.1 eV were associated with the highest PCE after the deposition of the first ALD Al2O3 monolayer. A significant reduction of PCE occurred as

decreased and *IBincreased.

The results show that the paths of electron transfer from dye into TiO2 without any energy barrier and therefore improve the cells PCE due to the proper ultra-thin Al2O3energy

The results show that the paths of electron transfer from dye into TiO2 without any energy barrier and therefore improve the cells PCE due to the proper ultra-thin Al2O3energy