Band Offset Measurements in Atomic-Layer-Deposited Al2O3/Zn0.8Al0.2O Heterojunction Studied by X-ray Photoelectron Spectroscopy

Pure aluminum oxide (Al2O3) and zinc aluminum oxide (ZnxAl1-xO) thin films were deposited by atomic layer deposition (ALD). The microstructure and optical band gaps (E g) of the ZnxAl1-xO (0.2 ≤ x ≤ 1) films were studied by X-ray diffractometer and Tauc method. The band offsets and alignment of atomic-layer-deposited Al2O3/Zn0.8Al0.2O heterojunction were investigated in detail using charge-corrected X-ray photoelectron spectroscopy. In this work, different methodologies were adopted to recover the actual position of the core levels in insulator materials which were easily affected by differential charging phenomena. Valence band offset (ΔE V) and conduction band offset (ΔE C) for the interface of the Al2O3/Zn0.8Al0.2O heterojunction have been constructed. An accurate value of ΔE V = 0.82 ± 0.12 eV was obtained from various combinations of core levels of heterojunction with varied Al2O3 thickness. Given the experimental E g of 6.8 eV for Al2O3 and 5.29 eV for Zn0.8Al0.2O, a type-I heterojunction with a ΔE C of 0.69 ± 0.12 eV was found. The precise determination of the band alignment of Al2O3/Zn0.8Al0.2O heterojunction is of particular importance for gaining insight to the design of various electronic devices based on such heterointerface.


Background
Nano-thick oxide films with high resistance have attracted much attention as the most promising conductive layer for the applications of microchannel plate (MCP) as electron multiplier [1,2], resistive memories [3], and electronoptical micro-electro mechanical systems (MEMS) [4]. A large research effort has been devoted to the novel idea of adjusting resistivity of such thin films due to the abovementioned large potential applications in a special environment. MCP is a thin glass plate with thickness of about 500 μm consisting of several millions pores of a cylinder geometry with a 4-25-μm diameter and with a bias angle usually 5°-13°to the normal of the plate surface, and the high aspect ratio in each pore is about 20:1-100:1 [5,6]. For recent MCP fabrication, two kinds of nano-thick layers are deposited on the MCP pore surfaces to conduct an electron multiplication function [1,2]. The first layer is a conductive layer for supplying electrons, and the second layer is a secondary electron emission (SEE) layer for generating electrons. The three-dimensional surfaces and high aspect ratio of MCP should be firstly taken into consideration for depositing uniform thickness and composition of thin films. So far, the only effective approach growing high-quality thin films is the atomic layer deposition (ALD) technique based on sequential self-terminating gas-solid reactions [7].
ZnO is an n-type semiconductor with a direct bandgap of around 3.37 eV and a large exciton binding energy of 60 meV at room temperature [8,9]. A lot of elements such as Mg [10,11], Cd [12], Ga [13], W [14], and Mo [15] were used to doping in ZnO in order to tune its optical and electrical properties for special applications. In electron multiplier application, such as MCP, zinc aluminum oxide (Zn x Al 1-x O) films have been investigated because of their thermal stability in a special application environment and low cost of industrialization. The properties of Zn x Al 1-x O films can be controlled by changing the Al content, paving a way to design optoelectronic and photonic devices based on this material. Usually, high-resistivity Zn x Al 1-x O thin films as a conductive layer with x at the range of 0.7-0.85 have been applied in the field of electron multiplier [16]. For SEE layers, borondoped diamond with hydrogen-terminated material has higher SEE coefficient than that of other traditional insulators. This provides a strong impetus for the development of electron multipliers. However, in the presence of degradation due to electron beam-induced contamination, these must be seriously regarded as preliminary [17]. From a practical point of view, two kinds of traditional insulators used as SEE layers in MCP are magnesium oxide (MgO) and Al 2 O 3 thin films [18]. Although pure MgO has higher SEE coefficient than that of Al 2 O 3 , it is limited in the application on MCP because it is highly deliquescent and its surface is rather reactive with atmospheric moisture and carbon dioxide as demonstrated by our previous work [19], which probably results in degraded SEE performance. However, the physical and chemical properties of Al 2 O 3 are very stable even after long-term exposure to the atmosphere. Therefore, Al 2 O 3 is one of the most commonly used SEE materials in MCP application.
According to the structure of MCP, the Al 2 O 3 and Zn x Al 1-x O thin films have different band gaps (E g ) resulting in band offsets in the heterointerface. Therefore, the determination of the band offsets at Al 2 O 3 /Zn x Al 1-x O interface is of importance because valence band offset (ΔE V ) and conduction band offset (ΔE C ) can deteriorate or promote SEE performance and also have a great influence on the performance of electron multiplier.
Generally, Kraut's method is widely used to calculate the valence band maximum (VBM) and the conduction band minimum (CBM) of semiconductor/semiconductor heterojunctions [20]. However, in the case of insulator/ semiconductor or, in more serious cases, insulator/insulator heterojunctions, the positive charges generated during X-ray bombardment accumulate in the insulators and induce a strong modification of the kinetic energy of the emitted photoelectrons which is the so-called differential charging effect [21]. Although it is probably dealt with using a neutralizing electron gun [22], the use of C 1s peak recalibration [23], and zero charging method [24][25][26], a careful evaluation of the experimental result is necessary due to the differential charging effect during X-ray irradiation [19].
In this work, we will study the structure and optical E g of Zn x Al 1-x O (0.2 ≤ x ≤ 1) thin films firstly, and then, we especially determine the ΔE V and ΔE C of the Al 2 O 3 / Zn 0.8 Al 0.2 O heterojunction by using high-resolution Xray photoelectron spectroscopy (XPS).

Characterization
Optical transmittance spectra in a wavelength range from 185 to 700 nm were carried out by using a doublebeam UV-Vis-IR spectrophotometer (Agilent Cary 5000) at room temperature in air. The crystal structure of the films were characterized by X-ray diffraction (XRD, Bruker D8) using Cu K α radiation (40 kV, 40 mA, λ = 1.54056 Å). The film thickness was measured by Spectroscopic Ellipsometry (Sopra GES5E) where the incident angle was fixed at 75°, and the wavelength region from 230 to 900 nm was scanned with 5-nm steps. And the ellipsometric thicknesses of samples ALD03, ALD04, ALD05, and ALD06 were 3.01, 4.02, 5.01, and 8.01 nm, respectively. The XPS (PHI Quantera SXM) is used to analyze both the core levels (CLs) and valence band spectra of the samples. Charge neutralization was performed with an electron flood gun, and all XPS spectra were calibrated by the C 1s peak at 284.6 eV. In order to avoid differential charging effect, during the measurement, the spectra were taken after a few minutes of X-ray irradiation. All the samples are measured under the same conditions in order to acquire reliable data.

Calculations
The ΔE V of the Al 2 O 3 /Zn 0.8 Al 0.2 O heterojunction can be calculated from Kraut's formula where ΔE CL ¼ E Þ were the Al 2 O 3 and Zn 0.8 Al 0.2 O bulk constants, which were obtained on the respective thick films. The VBM values were determined by linear extrapolation of the leading edge to the baseline of the valence band spectra. A root sum square relationship is used to combine the uncertainties in the different binding energies to determine the uncertainty of calculated results [26].

Structure and Band Gaps of Zn x Al 1-x O Samples
The XRD patterns of the as-deposited 80-nm-thick Zn x Al 1-x O (x = 0.2, 0.6, 0.8, 0.9, 1) thin films grown on quartz and Si substrates are shown in Fig. 1a, b, respectively. For the pure ZnO grown on quartz substrates in Fig. 1a, the strong peaks at 32.4°and 34.8°and the relatively weak peaks at 36.5°and 57.2°come from hexagonal ZnO phase, indicating the polycrystalline nature of the ZnO layer. And strong (0 0 2) peak shows the preferential orientation growth of ALD ZnO. However, the above characteristic peaks become weak for Zn 0.9 Al 0.1 O sample and disappear for Zn x Al 1-x O (x ≤ 0.8) samples, which suggests that ZnO crystallization is suppressed with Al concentration increase. Besides, the broad peak ranging from 20°to 30°is the typical pattern of the quartz substrate. For Si substrate, the strong peaks around 28.4°and 58.9°are easily detected (data not shown). These peaks are corresponding to the diffractions originated from Si (1 1 1) and Si (2 2 2) crystal planes. In addition, the relatively weak peaks in Fig. 1b at 2θ = 32.6°, 33.2°, 35.4°, 35.9°, 38.8°, 39.2°, and 42.8°in the diffractograms that arise from the Si substrate itself are also observed. These unknown peaks may be related to the process conditions for producing crystalline silicon and are observed in previous work [27,28]. Except for diffraction peaks from the Si substrate, no other diffraction peaks from the Zn x Al 1-x O (x ≤ 0.9) samples are detected. Only (0 0 2) and weak (1 1 0) peaks appear in the pure ZnO sample. From the above results, the crystal quality of the Zn x Al 1-x O film is a serious decline with the increasing concentration of Al content. It is well known that the particle size of Al ions is less than that of Zn ions. Zn is easily substituted by Al when doping concentration of Al increases. This results in weakened ZnO crystallinity, so the structure of Zn x Al 1-x O  [29]. Taken into consideration, Zn x Al 1-x O layer growth appears to be substrate sensitive and Al doping concentration has an influence on the crystallization of the films. Figure 2a shows transmission spectra of the Zn x Al 1-x O samples deposited on quartz substrate. The average transmittance is above 80% in the visible wavelength for all samples. It is found that ZnO film exhibits abrupt absorption edge which appears at~390 nm corresponding to the fundamental E g of ZnO. A blue shift of the absorption edge is apparently observed when Al concentration increases. The E g of Zn x Al 1-x O thin films can be obtained by fitting the sharp absorption edges. The relationship between absorption coefficient (α) and E g of direct band gap semiconductor is given by Tauc equation [30], (αhv) 2 = B(hv−E g ), where hν is the photon energy and B is a constant. The dependence of (αhν) 2 on photon energy is shown in Fig. 2b. The E g is obtained by the extrapolations of the liner regions of the optical absorption edges. The E g of pure ZnO thin film deposited by ALD is 3.26 eV, which is consistent with the previous reports [31,32]. With the Zn concentration x decreases from 0.9 to 0.2, the E g of Zn x Al 1-x O thin films increases from 4.11 to 6.51 eV. It is directly demonstrated that the E g of Zn x Al 1-x O thin films can be adjusted in a large range by controlling the Al doping concentration, which makes it a suitable candidate for application in many scientific research fields [33,34]. For the new type of MCP, the properties of Zn 0.8 Al 0.2 O thin film are suitable for conductive layer proved by previous study [2]. Therefore, the E g of atomic-layer-deposited Zn 0.8 Al 0.2 O thin film is 5.29 eV, which is sufficient to make a band gap discontinuity in Al 2 O 3 /Zn 0.8 Al 0.2 O heterojunction and is used for calculating the ΔE C value later.

Valence and Conduction Band Offset Measurements of Al 2 O 3 /Zn 0.8 Al 0.2 O Heterojunction
The XPS spectra of survey scan, CLs, and VBM region for Zn 0.8 Al 0.2 O and Al 2 O 3 samples are shown in Fig. 3.
In this study, we find that the CLs positions of the Zn 0.8 Al 0.2 O and Al 2 O 3 thin films do not change as a function of X-ray irradiation time for 15 min (data not shown), because of operating a low energy electron flood gun. Figure 3a, e shows the whole scanning spectrum of the thick Zn 0.8 Al 0.2 O and Al 2 O 3 thin films, respectively. The C 1s peak at 284.6 eV appeared due to some surface contamination, and the Ar 2p peak at 242.1 eV appeared because of residual inert gas composition in the ultrahigh vacuum chamber. The peaks in Fig. 3a located 660 Table 2 Fig. 4. The Al 2p, Zn 2p 1/2 , and Zn 2p 3/2 XPS spectra in Fig. 4(a, e, i), (b, f, j), and (c, g, k), respectively, are fitted by a single contribution, attributed to the bonding configurations Al-O and Zn-O. For the O 1s XPS spectrum in Fig. 4d, h, l, an additional low-intensity higher binding energy component is observed. The extra component is attributed to metal (Al, Zn)-O bonding at the interface and/or inelastic losses to free carries in the Al 2 O 3 layer, similar results obtained by previous study [19]. With the increase of the Al 2 O 3 thickness, the intensity of Zn 2p 1/2 peak is weakened while the energy resolution is deteriorated shown in Fig. 4f. It is difficult to observe and fit for Al 2 O 3 thickness of 5 nm as shown in Fig. 4j. So, the peak position of Zn 2p 1/2 in 5-nm Al 2 O 3 sample listed by a bold number in Table 2 Table 3 lists the ΔE V values for all Al 2 O 3 samples with thickness of 3-5 nm, and the error in each value is ± 0.07 eV. Therefore, the averaged ΔE V value is 0.87 ± 0.22 eV. It is important to note that the calculation does not include the italic numbers in Table 3 because of the big error fitting of CLs of Zn 2p 1/2 in the 5-nm Al 2 O 3 /Zn 0.8 Al 0.2 O sample.
However, there are obvious considerable CL shifts up to 0.6 eV sensitive to the thicknesses of the Al 2 O 3 and Zn 0.8 Al 0.2 O layers from the given experimental data in Table 2, and different ΔE V values are obtained in the   Table 3. It is directly proved that the charging phenomenon generated by the X-ray irradiation results in adverse effects on the ΔE V determination when taking XPS measurement on insulator/semiconductor heterojunction in spite of operating neutralizing electron gun. As has been widely reported, the influences of differential charging on the band offsets determination cannot be neglected even in very thin oxides. Therefore, zero charging method is adopted in this study in order to eliminate charginginduced errors and recover the accurate ΔE V value. The error in the ΔE V measurement is resulting from the differential charging effect that prevents the correct determination of the energy differences, such as between  the Al 2p and Zn 2p 3/2 signals even in very thin Al 2 O 3 films in heterojunction. In Fig. 5, the binding energies of the Al 2p, Zn 2p 3/2 , and Zn 2p 1/2 CLs for the 3, 4, 5, and 8 nm Al 2 O 3 films are plotted as a function of X-ray irradiation time. The binding energies of Al 2p, Zn 2p 3/2 , and Zn 2p 1/2 CLs of the 3-nm Al 2 O 3 sample in Fig. 5a increase slowly until they stabilize on a steady state value of 74.63 ± 0.01, 1021.77 ± 0.01, and 1044.83 ± 0.02 eV, respectively. Similar time dependencies are observed in the 4-, 5-, and 8-nm Al 2 O 3 films, as shown in Fig. 5b-d. The results show that CL steady state spectra are obtained after stabilization of the signals in the heterojunctionconsidered charge saturated when X-ray irradiation time is more than 25 min. Therefore, X-ray irradiation time is one of the most important parameters to determine the insulator/semiconductor band offsets. Layer thickness dependence in peak positions is mainly resulting from the differential charging effects. True peak positions can be acquired by extrapolating the measured binding energies to zero oxide thickness and ideally to zero charge, similar results are reported for SiO 2 /Si [25], HfO 2 /Si [26], and MgO/Zn 0.8 Al 0.2 O [19] systems. The CLs positions of the Al 2p, Zn 2p 1/2 , and Zn 2p 3/2 are plotted as a function of the Al 2 O 3 film thickness, as shown in Fig. 6. By linear fitting of the experimental data, the CLs positions of the Al 2p, Zn 2p 1/2 , and Zn 2p 3/2 peaks are determined to be 74.51 ± 0.03, 1044.77 ± 0.06, and 1021.7 ± 0.04 eV, respectively. In order to correct the ΔE V of the Al 2 O 3 /Zn 0.8 Al 0.2 O heterojunction, we calculate the energy differences between the extrapolated (Al 2p, Zn 2p 1/2 ) and (Al 2p, Zn 2p 3/2 ) at zero thickness. The values are 970.26 ± 0.07 and 947.19 ± 0.05 eV, respectively. Inserting these values in Eq. (1), the ΔE V are calculated to be 0.83 ± 0.09 and 0.8 ± 0.08 eV, which is in good agreement using the two combinations of CLs of the Al 2 O 3 /Zn 0.8 Al 0.2 O heterojunction. Therefore, the averaged ΔE V value is 0.82 ± 0.12 eV.
There are three possible reasons that affect the ΔE V values in addition to the XPS method itself. Firstly, the oxide stoichiometry of the Al 2 O 3 thin films measured by XPS is almost the same in the different Al 2 O 3 samples with thickness of 3-8 nm. Therefore, the composition of the Al 2 O 3 film is independent of thickness and the binding energy shifts in Fig. 5 is related to the differential charging effect occurring in the Al 2 O 3 /Zn 0.8 Al 0.2 O heterojunction during X-ray irradiation. Secondly, band bending at the heterointerface could induce a systematic error in determination of ΔE V , and we check that this error is much smaller than the average standard deviation of ± 0.03 eV given above. Finally, the strain existing in the Al 2 O 3 overlayer of the heterojunction will induce  [40]. The heterojunction underlayer Zn 0.8 Al 0.2 O is thick enough, and the structure of both materials is amorphous. Therefore, the strain-induced piezoelectric field is not taken into consideration in this study.
To infer the ΔE C based on the value of ΔE V , we need to know the E g of the ultrathin Al 2 O 3 layer, which can be estimated from O 1s core-level binding energy spectrum of atomic-layer-deposited Al 2 O 3 thin films with energy loss structure. The binding energy is calculated from the difference in the total photoelectron energy minus the kinetic energy due to the loss in photoelectron energy by inelastic collision processes within the sample. The minimum inelastic loss is equal to the band gap energy, and the most cited value of E g is 6.8 eV [41][42][43]. Together with the Zn 0.8 Al 0.2 O E g of 5.29 eV at room temperature, the ΔE C can be simply derived by the equation, ΔE C = E g (Al 2 O 3 ) − ΔE V − E g (Zn 0.8 Al 0.2 O), where E g (Al 2 O 3 ) and E g (Zn 0.8 Al 0.2 O) are the band gaps of the Al 2 O 3 and Zn 0.8 Al 0.2 O thin films, respectively. The ΔE C is calculated to be 0.69 ± 0.12 eV, which means that the barrier height for transport of electrons is smaller than that of holes. The band alignment of the Al 2 O 3 /Zn 0.8 Al 0.2 O heterojunction obtained from XPS measurements is shown in Fig. 7. The CBM of Al 2 O 3 is higher than that of Zn 0.8 Al 0.2 O; however, the VBM of Al 2 O 3 is lower than that of Zn 0.8 Al 0.2 O. Therefore, a nested type-I band alignment with a ratio ΔE C /ΔE V of about 1:1.2 is obtained.
Usually, the MCP gain under direct current (DC) mode is limited by the space charge density without consideration of ion feedback, and the recharge time constant or dead time (τ) is several milliseconds [44]. When operating a MCP as a DC current amplifier, the gain is constant until the output current (I oc ) exceeds about 10% of the strip current through the plate. However, the MCP works in a highly saturated state under a photon-counting mode, and the electron avalanche multiplication is done within several nanoseconds that is a million times faster than τ [44][45][46]. The peak output current in pulsed operation exceeding the I oc by several orders of magnitude is observed. Therefore, anode signal charges probably come from the tunneling electrons in the Al 2 O 3 /Zn x Al 1-x O heterojunction of the inner wall of the MCP. For photon-counting mode, both ΔE V and ΔE C should be sufficiently large, which can prevent the thermal excitation of electrons generated from the SEE layer into the electron multiplier system that probably produces high electronics dark noise and result in a reduced signal to noise ratio. The present result has no effects on the MCP operating under DC mode which is determined by space charge saturation, but has negative effects on the photon-counting mode which needs a type-II heterojunction to improve tunneling probability for excellent performance. The relationship between the Al 2 O 3 /Zn x Al 1-x O heterojunction and charge replenishment  mechanism under photon-counting mode needs further study. Therefore, the band alignment of the Al 2 O 3 / Zn x Al 1-x O heterojunction should be constructed and adjusted by appropriately changing the ratio of Al and Zn elements under the premise of meeting the requirements of the electron multiplier.

Conclusions
The structure and optical band gaps of Zn x Al 1-x O (0.2 ≤ x ≤ 1) films deposited by atomic layer deposition are investigated. And the band offset measurements of the Al 2 O 3 /Zn 0.8 Al 0.2 O heterojunction have been determined by XPS with zero charging method. The results show that X-ray irradiation time is one of the most important parameters to determine the band offsets. The layer thickness dependence in peak positions is mainly derived from the differential charging effects, and true peak positions are obtained by extrapolating the measured binding energies to zero oxide thickness and ideally to zero charge. The ΔE V value is obtained to be 0.82 ± 0.12 eV, and the corresponding ΔE C is calculated to be 0.69 ± 0.12 eV. Therefore, a nested type-I band alignment is obtained. Understanding of the band alignment parameters of the Al 2 O 3 /Zn 0.8 Al 0.2 O interface will facilitate the knowledge of their carrier transport mechanism and design of corresponding hybrid devices, especially in the research process of electron multipliers.