Band Alignment of AlN/InGaZnO Heterojunction for Thin-Film Transistor Application

Abstract

Uncrystallized indium-gallium-zinc-oxide (InGaZnO) thin-film transistors (TFTs) combined with an aluminum nitride (AlN) dielectric have been used to promote performance and steadiness. However, the high deposition temperature of AlN films limits their application in InGaZnO flexible TFTs. In this work, AlN layers were deposited via low-temperature plasma-enhanced atomic layer deposition (PEALD), and InGaZnO films were fabricated via high-power impulse magnetron sputtering (HIPIMS). The band alignment of the AlN/InGaZnO heterojunction was studied using the X-ray photoemission spectrum and ultraviolet visible transmittance spectrum. It was found that the AlN/InGaZnO system exhibited a staggered band alignment with a valence band offset ΔE_v of −1.25 ± 0.05 eV and a conduction band offset ΔE_c of 4.01 ± 0.05 eV. The results imply that PEALD AlN could be more useful for surface passivation than a gate dielectric to promote InGaZnO device reliability under atmospheric exposure.

1. Introduction

Indium-gallium-zinc-oxide (InGaZnO, or IGZO) has been deemed as one of the promising channel candidates for flexible thin-film transistors (TFTs) due to its high mobility, excellent optical transparency, and the capacity to grow on flexible substrates at low temperatures with decent quality [1,2,3]. Meanwhile, IGZO FETs have also made impressive progress in emerging technologies, such as flat planer displays [2], flexible circuits [4,5], and electronic paper [6]. In spite of the great number of studies on InGaZnO TFTs, their low operation voltage and long-term reliability are issues that still need to be explored [7,8,9,10]. These issues have been attributed to the charge trapping at IGZO/dielectric interface or the conductivity modifications of the IGZO channel due to the exposure to the hydrogen or water in the atmosphere [7,8,9,10]. The high-k gate dielectric was one of the effective solutions for reducing the threshold voltage and subthreshold swing, which promote capacitance coupling at the IGZO/dielectric interface [3,8]. However, the small bandgap (E_g) of the high-k dielectric results in an insufficient band offset (ΔE) at the InGaZnO/dielectric interface, which is unable to prevent the injection of electrons and holes [10,11,12]. There are many reported studies on the band alignments between InGaZnO and insulators, such as SiO₂ [13], Al₂O₃ [14], HfO₂ [15], HfSiO [11], HfTiO [16], HfLaO [17], ZrSiO_x [18], and Sc₂O₃ [8]. Even if a specific dielectric is unsuitable for the gate of IGZO TFTs, it may play role in surface passivation, which is vital for IGZO TFTs due to their reliability being hindered by exposure to the hydrogen or water in the atmosphere [10,19].

To the dielectrics mentioned above, aluminum nitride (AlN) is a viable alternative solution due to its high dielectric constant (~9.5 [20]) and E_g (~6.2 eV) [21], superior thermal conductivity of ~17 W·m⁻¹K⁻¹ [22], low thermal expansion coefficient (~5.3 × 10⁻⁶ K⁻¹ [23]), and suitable thermal stability [24]. InGaZnO TFTs using AlN have been reported and achieved acceptable device performance with a reduced self-heating effect [25,26]. However, to the best of our knowledge, no study has been conducted on the band offsets in the AlN/InGaZnO heterojunction. In this work, by using X-ray photoelectron spectroscopy, the band offsets in the InGaZnO/AlN heterojunction were analyzed, and InGaZnO and AlN film were fabricated by high-power impulse magnetron sputtering (HIPIMS) and plasma-enhanced atomic layer deposition (PEALD), respectively.

2. Materials and Methods

2.1. Fabrication of Samples

The InGaZnO films were prepared with a physical vapor deposition system (Ljuhv SP122I, Jhubei city, Taiwan, China). The system at our institute has been used to deposit TCO films such as IGZO and ITO in previous studies [7,27,28,29]. In this work, InGaZnO was deposited by HIPIMS on c-plane sapphire wafers (α-Al₂O₃ (0001)) and quartz substrates based in a ceramic target of InGaZnO (99.99% purity, In₂O₃:Ga₂O₃:ZnO = 1:1:1). The radio frequency (RF) power and operating pressure were, respectively, 50 W and 40 mTorr, in a pure Ar ambient environment. Further deposition details of InGaZnO can be found in a previous work [7]. Prior to deposition, the wafers were divided into 10 mm × 10 mm pieces, which were ultrasonically washed with normal organic cleaners (acetone, isopropanol, and ultrapure water for 10 min each). HIPIMS-InGaZnO films exhibited an amorphous state according to X-ray diffraction [7]. For uncrystallized InGaZnO films, high-temperature deposition or annealing are improper because InGaZnO is widely used for low-temperature applications with wafers like plastic and tape [7].

In this work, a Beneq TFS-200 reactor and the PEALD technique were utilized to grow AlN at 185 °C. Given the requirements for a flexible thin-film transistor as well as the AlN film quality, a low deposition temperature was used in this work, which has been reported in the literature [30]. The precursors were trimethylaluminum (TMA) and ammonia (NH₃), using Al and N sources, respectively. Further AlN deposition details can be found in the literature [30]. In this work, three samples (1#–3#) were used for XPS experiments: (1) 100 nm-thick IGZO on sapphire; (2) 3 nm-thick AlN grown on IGZO; (3) 40 nm-thick AlN grown on IGZO.

2.2. Characterizations

A mature method [31] was used to calculate the valence band offset (ΔE_c) and conduction band offset (ΔE_v) based in X-ray photoelectron spectroscopy (XPS) [3,8,17,18,21,31]. A Thermo ESCALAB 250 X-ray photoelectron spectrometer with a monochromatic Al -Kα X-ray source (energy 1486.6 eV) was used for ex situ XPS measurements. The X-ray source possessed a power of 300W, and the detection region was set was a spot with a radius of 650 μm and a take-off angle of 90°. XPS survey scans with 1.0 eV/step were used to detect the chemical state of samples 1–3#. High resolution scans with 0.05 eV/step and pass energy of 30 eV were utilized to evaluate the binding energy of specific elements. In addition, the valence band scans shared the same settings as the high-resolution scans. It should be noted that the whole XPS spectra were corrected using the C 1s peak (using ~284.8 eV), which stems from surface carbon contamination to compensate for the charging effect [3]. A 30 s Ar ion sputtering was used for sample 3#, which was followed immediately by XPS measurements to eliminate the potential influence coming from the atmosphere, water vapor, etc. The surface micromorphologies of the fabricated films were detected with a Bruker Dimension Icon atomic force microscope (AFM). The E_g of AlN was evaluated using the sample fabricated on quartz using the UV–visible transmittance spectrum (JASCO, wavelength range of 180–600 nm).

3. Results and Discussion

3.1. Chemical Bonding State Analysis

XPS survey scans with an 89.45 eV pass energy were obtained to analyze the different chemical composition states in the InGaZnO, 3 nm AlN on InGaZnO, and 40 nm AlN films. The O 1s, Ga 2p, C 1s, Zn 2p, N 1s, Al 2p, In 3p, In 3d, and In 4d peaks were seen in the survey spectra of samples 1#–3# (Figure 1), which originated from In, Zn, Ga, Al, C, and O.

Table 1. Relative atomic ratios of the specific elements in IGZO targets, samples 1# and 3#.

Sample	In (%)	Zn (%)	Ga (%)	O (%)	Al (%)	N (%)
IGZO Target	16.67	8.33	16.67	58.33	NR	NR
1#	14.34	12.44	17.11	48.67	7.44	NR
3#	NR	NR	NR	1.63	51.15	47.22

provides the atomic ratios of the different elements in the targets and samples 1# and 3#. Specifically, for the HIPIMS InGaZnO films, the Ga ratio and Zn ratio were higher than those in the target materials, and the In ratio and O ratio were lower than those in the target materials, which is similar with reported HIPIMS InGaZnO films [6].

Firstly, the high Zn ratio could be attributed to the much larger sputtering yield (Zn: 3.68 [6]), which accelerated the Zn sputtering off from the target to the substrate surface and increased the Zn percentage in the InGaZnO films. Secondly, compared with Ga, In experienced more collision scattering during sputtering due to its longer mean free path [6]. In addition, the actual Al: N ratio of sample 3# (~1.08:1) was different from the ideal ratio (1:1). It may have resulted from the high hygroscopicity of Al₂O₃ (Al-O) and sample 3# being exposed to atmospheric water vapor or hydrogen before the XPS measurements. In addition, the surface morphology of InGaZnO and AlN is shown in Figure 2. Both the InGaZnO and AlN films had a flat surface with a relatively low root mean square (RMS) roughness of 1.02 and 0.32 nm, respectively. A flat surface is beneficial for suppressing surface recombination and leakage current, which thus increase the performance of IGZO TFT devices.

3.2. Energy Gap of AlN, IGZO

The E_g of the InGaZnO and AlN was, respectively, evaluated using the XPS O 1s CL spectrum and ultraviolet-to-visible transmittance spectrum. According to the O 1s spectra in Figure 3, the E_g of InGaZnO and AlN was 3.42 and 6.18 ± 0.05 eV, respectively. As shown in Figure 3a, the energy loss structure on the high- energy side was used to evaluate the E_g of the InGaZnO films [32,33]. The measured value for IGZO is in agreement with the reported values (~3.4–3.6 eV) [3,34,35] but is higher than in some reports (~3.2 eV) [8,18]. In addition, the E_g of the direct bandgap semiconductor AlN was extracted from the Tauc plot ((αhν)² vs. hν) based on the UV–visible transmittance spectrum, which is close to the reported values (~6.2 eV [20,21,36,37,38]).

3.3. Band Alignment Analysis

Figure 4 shows the valence band spectra for the thick InGaZnO and AlN films, and the valence band maximum (VBM) of InGaZnO and AlN is obtained (2.32 and 1.73 ± 0.05 eV, respectively). The VBM values were extracted by linear extrapolation [8,16,17,18] based on Figure 4, and the values are similar to those reported for sputtered InGaZnO [8,16,17,18] and ALD AlN films [21,30].

Using the evaluation method proposed by Kraut et al. [31], the valence band offset (ΔE_v) of the AlN/InGaZnO heterojunction can be calculated as

$$\begin{array}{c}\Delta {\mathrm{E}}_{\mathrm{v}}={({\mathrm{E}}_{\mathrm{C}\mathrm{o}\mathrm{r}\mathrm{e}}-{\mathrm{E}}_{\mathrm{V}\mathrm{B}\mathrm{M}})}_{\mathrm{I}\mathrm{n}\mathrm{G}\mathrm{a}\mathrm{Z}\mathrm{n}\mathrm{O}}-{({\mathrm{E}}_{\mathrm{C}\mathrm{o}\mathrm{r}\mathrm{e}}-{\mathrm{E}}_{\mathrm{V}\mathrm{B}\mathrm{M}})}_{\mathrm{A}\mathrm{l}\mathrm{N}}\hfill \\ -{({\mathrm{E}}_{\mathrm{C}\mathrm{o}\mathrm{r}\mathrm{e}}^{\mathrm{I}\mathrm{n}\mathrm{G}\mathrm{a}\mathrm{Z}\mathrm{n}\mathrm{O}}-{\mathrm{E}}_{\mathrm{C}\mathrm{o}\mathrm{r}\mathrm{e}}^{\mathrm{A}\mathrm{l}\mathrm{N}})}_{\mathrm{A}\mathrm{l}\mathrm{N}/\mathrm{I}\mathrm{n}\mathrm{G}\mathrm{a}\mathrm{Z}\mathrm{n}\mathrm{O}}\hfill \end{array}$$

(1)

where E_core and E_VBM are, respectively, the core level (CL) positions and the VBM of these bulk materials, combined with the CL difference of the thin AlN/InGaZnO film.

According to Kraut’s method, we further measured the core-level spectra of samples 1#–3# to evaluate the actual band alignment and the corresponding band offsets (ΔE_v and ΔE_c). The high-resolution core-level spectra are provided in Figure 5 for (a) InGaZnO VBM with CLs and (b) AlN VBM with CLs, as well as in Figure 6 for the InGaZnO-AlN CLs. Figure 5 and Figure 6 were utilized to evaluate the specific CL peak positions.

Table 2. Summary of XPS data on IGZO, AlN, and AlN/IGZO samples. Peak position values and VBM values are ± 0.05 eV.

1# Thick IGZO			3# Thick AlN		2# Thin AlN on IGZO		Valence Band Offset	Average ΔEv	Conduction Band Offset
IGZO Metal Core	Metal Core Level	Metal Core-IGZO VBM	Al 2p Core Level	Al2p-AlN VBM		ΔCL IGZO-AlN
Zn2p3	1021.74	1019.42	74.08	72.35		948.35	−1.28	−1.25	4.04
Ga2p3	1117.68	1115.36				1044.30	−1.19		3.95
In3d5	444.49	442.17				371.10	−1.28		4.04

exhibits the extracted values; thus, the valence band offset ΔE_v with different CLs (Zn, Ga, O) were calculated as −1.28, −1.19, and −1.28 ± 0.05 eV, respectively. The average ΔE_v of AlN/InGaZnO was calculated as nearly −1.25 eV.

Subsequently, the conduction band offset ∆E_c of AlN/IGZO was calculated to be 3.86, 3.87, and 3.92 ± 0.05 eV with different CLs (Zn, Ga, O) using the following equation [3,9,10,11,12,13,14]:

$$\Delta {\mathrm{E}}_{\mathrm{c}}={\mathrm{E}}_{\mathrm{g}}(\mathrm{A}\mathrm{l}\mathrm{N})-{\mathrm{E}}_{\mathrm{g}}(\mathrm{I}\mathrm{G}\mathrm{Z}\mathrm{O})-\Delta {\mathrm{E}}_{\mathrm{v}}$$

(2)

Figure 7 exhibits the abbreviated band diagram and the complete band diagram of the AlN/InGaZnO heterojunction. These results demonstrate that a staggered alignment (or type-II) existed at the AlN/InGaZnO heterojunction with an average ∆E_v of −1.25 ± 0.05 eV and an average ∆E_c of 4.01 ± 0.05 eV. The PEALD AlN film had a large E_g and ∆E_c but a negative ∆E_v, which could provide sufficient electron confinement combined with undesirable hole confinement. These results imply that the PEALD AlN is not the perfect candidate as a gate dielectric in InGaZnO TFTs because of the inability to confine holes may result in serious device instability issues owing to the various hole defects or acceptor traps. Additionally, AlN still plays a vital role in the surface passivation of InGaZnO devices, which promotes device stability during atmospheric exposure [11,12].

4. Conclusions

The AlN/InGaZnO heterojunction was found to have a staggered alignment. AlN and InGaZnO were, respectively, produced via PEALD and HIPIMS. The corresponding ∆E_c and ∆E_v were evaluated to be 4.01 ± 0.05 eV and −1.25 ± 0.05 eV, respectively. These results suggest that AlN could provide a sufficient barrier for electrons but cannot hinder holes on InGaZnO, which form the threshold voltage and cause the long-term instability of InGaZnO TFTs. In addition, they indicate that AlN is more suitable for surface passivation to prevent InGaZnO surfaces from exposure to atmosphere hydrogen and oxygen. This accurate determination provides useful information for the further development of transparent TFTs.