- Nano Express
- Open Access
Investigation on Surface Polarization of Al2O3-capped GaN/AlGaN/GaN Heterostructure by Angle-Resolved X-ray Photoelectron Spectroscopy
Nanoscale Research Lettersvolume 12, Article number: 499 (2017)
The surface polarization of Ga-face gallium nitride (GaN) (2 nm)/AlGaN (22 nm)/GaN channel (150 nm)/buffer/Si with Al2O3 capping layer is investigated by angle-resolved X-ray photoelectron spectroscopy (ARXPS). It is found that the energy band varies from upward bending to downward bending in the interface region, which is believed to be corresponding to the polarization variation. An interfacial layer is formed between top GaN and Al2O3 due to the occurrence of Ga–N bond break and Ga–O bond forming during Al2O3 deposition via the atomic layer deposition (ALD). This interfacial layer is believed to eliminate the GaN polarization, thus reducing the polarization-induced negative charges. Furthermore, this interfacial layer plays a key role for the introduction of the positive charges which lead the energy band downward. Finally, a N2 annealing at 400 °C is observed to enhance the interfacial layer growth thus increasing the density of positive charges.
Gallium nitride (GaN) is considered as one of the most attractive semiconductor materials in many fields ranging from LED industries to power electronic industries [1, 2]. The popularity is due to a number of advantages with respect to silicon: high breakdown electrical field, high electron mobility, and excellent thermal stability [3, 4]. GaN high-electron-mobility transistor (HEMT) is widely studied for high-power and high-frequency application [1, 5, 6]. In the HEMT, if the Schottky gate is adopted, this gate interface brings in large interface states which exacerbate the large leakage current and low breakdown field . An insulator induced as the surface passivation layer and gate dielectric could help mitigate the above issues [8,9,10].
Al2O3 is preferred for such an insulator application due to its large band gap, high dielectric constant, and more negative Gibbs free energy comparing to Ga2O3, so it is believed that Al2O3 could passivate the surface states and improve the electric breakdown field . However, an interfacial layer is formed inevitably at the GaN/Al2O3 interface after the deposition of the Al2O3 [11, 12]. This interfacial layer is believed to be correlated with reliability of the threshold voltage and property of two-dimensional electron gas (2DEG) and plays a key role to control the band bending [2, 13,14,15,16].
Although the interfacial layer has been studied by several research groups, the role played by the interfacial layer has not been exploited in depth [12, 17]. Therefore, in this paper, we use the angle-resolved X-ray photoelectron spectroscopy (ARXPS) to detect the gradual changes of the band bending and obtain the atomic structure of the interfacial layer . Different thicknesses of Al2O3 are deposited on GaN samples by atomic layer deposition (ALD). ALD takes advantage of a low-temperature layer-by-layer deposition technology, which limits the thermal reaction between GaN and Al2O3. Thus, ALD is favored by the industries to deposit high-k dielectrics due to its high conformability and uniformity, accuracy on thickness control, high film quality, and low defect density . This enables a smooth and low-defect Al2O3/GaN interface. After Al2O3 deposition, a 400 °C post-deposition annealing (PDA) sample is also prepared to intensify the interfacial layer reaction, enhancing the interfacial layer formation. Based on the ARXPS results, it is found that the band is bending upward initially from the GaN substrate to the near interface due to the polarization-induced negative charges. However, as an increase in detection angle θ, the band bends downward gradually because of the formation of positive charges [5, 11,12,13].
The Ga-face GaN/AlGaN/GaN-on-Si(111) wafer was purchased from a commercial company (Enkris.com). The epitaxial wurtzite structure comprises a 2-nm GaN layer on top of a 22-nm AlGaN layer, and the two epitaxial layers are grown on a 150-nm i-GaN layer. A buffer layer serves as a transition layer connecting the GaN epilayer and the Si substrate. Three samples, S1, S2, and S3, are prepared. All samples were first decreased for 5 min in acetone, followed by immersion in isopropyl alcohol and a rinse in flowing deionized (DI) water. The native oxide was then etched away by dipping it into a dilute HCl solution (HCl:H2O = 1:10) for ~ 1 min, followed by a rinse in DI water. The Al2O3 are deposited by ALD on top of GaN, with trimethyl aluminum (TMA) and H2O as the metal precursor and oxidant, respectively. The Al2O3 thickness is 1 nm for sample S1 and 3 nm for samples S2 and S3. The thickness is measured by the ellipsometer. Moreover, S3 was subjected to PDA in N2 at 400 °C for 5 min.
ARXPS measurements were carried out in a Thermo Fisher Scientific Theta Probe system equipped with a monochromatic, microfocused Al Kα (1486.6 eV) X-ray source and a hemispherical electron energy analyzer. The binding energy (BE) calibration was performed using pure Ni, Au, Ag, and Cu standard samples by setting the Ni Fermi edge, Au 4f 7/2, Ag 3d 5/2, and Cu 2p 3/2 peaks at 0.00 ± 0.02, 83.98 ± 0.02, 368.26 ± 0.02, and 932.67 ± 0.02 eV, respectively. The FWHM of a given component spectrum was allowed to vary within a narrow range only (± 0.1 eV). The lowest number of component spectra was used to obtain acceptably low residual values . The XPS spectra were recorded at different detection angles (θ), ranging from 27.5° to 72.5° with respect to the sample normal in parallel without tilting the sample. To remove for possible positive charging-induced BE shift, the XPS spectra obtained were referenced to the C 1s peak at 285.0 eV. Quantitative analysis, including element/bond ratio determination, was achieved using relative sensitivity factors and algorithms embedded in the Avantage software .
Results and discussion
The Ga 3d core-level spectra for S1–S3 at different photoelectron detection angles are depicted in Fig. 1a–c, respectively. For S1, each Ga 3d spectrum can be fitted with two peaks, corresponding to the Ga–N and Ga–O bonds. The Ga–O bonds are due to the oxide formation as a result of the oxygenant exposure in the ALD, and the oxygenant penetrated into the initial thin Al2O3 layer . For S2 and S3, three peaks can be identified, noted as Ga–N, Ga–O, and O 2s, respectively. The O 2s peak is attributed to the Ga–O and Al–O bonds, and it becomes obvious when the detection angle is becoming larger. As this article is not focused on this peak, it will not be discussed further. Figure 2 presents the BE of Ga–N peaks as a function of θ for S1–S3. A 0.2-eV decrease is obtained from θ = 27.5° to 72.5° for S1. It suggests an upward band bending, which is consistent with the publications [3, 11]. For S2, the BE presents a decrease of 0.1 eV, indicating a mild upward band bending near the interface comparing to S1 or a flat energy band without band bending in consideration of experimental error. However, for S3, there is a 0.2-eV increase in the BE, which is in contrast to samples S1 and S2, advising a downward band bending. Figure 3 records the Al 2p spectra for all the samples, and there is no change in the BEs. Moreover, the peak is noted as the Al–O bond, and consequently, the AlGaN layer has negligible influence on the Ga–N BE variation. Table 1 summarizes the BEs of Ga 3d and Al 2p at different detection angles for all the samples, with the error of ± 0.1 eV.
The ratios of the Ga–O to Ga–N peak for all the samples are given in Table 2. The ratio is around 0.2 for samples S1 and S2, which is consistent with previous results . However, after the PDA treatment, the ratio increases to ~ 0.3 and denotes an increase of the GaO x interfacial layer. Moreover, the Ga/N ratio is also given in Table 2. The ratio is calculated by comparing the integrated intensities of the Ga 3d and N 1s peaks with atomic sensitivity factors . For samples S1 and S2, the ratio around 1.7 suggests a Ga-rich interfacial layer. However, after the N2 annealing, the ratio decreases to ~ 1.0. Furthermore, the sampling depth is also given in Table 2 for each angle. Because of the exponential attenuation of photoelectrons, 63 and 95% of detected electrons originate from within a distance of 1λ (i.e., λ represents electron’s inelastic mean free path (IMFP)) and 3λ, respectively, of the surface. Therefore, the XPS sampling depth is defined as 3λ nanometers underneath the sample surface. In our case, Al2O3 is the capping layer and the λ of Ga 3d photoelectrons in Al2O3 is estimated as ~ 3.4 nm. For a rough estimation, the sampling depth at different angles is given as 3λcosθ. However, the actual Ga–N BE sampling depth should consider the thickness of Al2O3, so the sampling depth is estimated as 3λcosθ minus the capping Al2O3 thickness. Because the GaO x layer is on top of GaN heterostructure, the signal of this layer is included for every detection angle. However, with the increase of the detection angle, the signal intensity of the Ga–N bond is decreased, resulting in the increase of the Ga–O/Ga–N ratio. Comparing S2 to S3, the increase of the Ga–O/Ga–N ratio and the decrease of the Ga/N ratio suggest the Ga-rich layer has been oxidized to form GaO x .
To illustrate the experiment data presented here, a model is schematically proposed in Fig. 4. The Fermi energy level of the GaN/AlGaN/GaN substrate is calibrated as 0 eV during XPS measurement . The conduction band maximum (CBM), valence band maximum (VBM), and core level are given. The BE is the energy difference between the core level and Fermi level. In the ALD deposition, O from the oxygenant could replace N in the Ga–N bond to oxidize the GaN and the surrounding N atoms could form N2 molecules , which leads to the formation of the Ga-rich layer and the gallium oxide (GaO x ) interfacial layer [11, 18]. This is supported by the Ga/N ratio which is larger than 1 in Table 2. This ratio indicates the change of GaN stoichiometry, and the intrinsic spontaneous polarization effect of GaN should disappear [21,22,23]. As a result, the Ga-rich layer, as the GaN-to-GaO x transition layer, eliminates the polarization-induced negative charges and results in a flat conduction band , as shown in Fig. 4.
Furthermore, in the annealing process, the Ga-rich surface is oxidized to form a thicker GaO x layer. Because the oxidation is a kinetically limited reaction and is restricted to about two surface monolayers, the bulk will not be strongly perturbed . Therefore, the Ga–N bond signal is mainly from the unoxidized underlying bulk, resulting in the Ga/N stoichiometric ratio of 1 for S3. The GaO x layer has been reported to bring in positive charges which may be interfacial fixed charges with energy states between the conduction band minima of the native oxide and GaN, which would bend the band downward [4, 11, 13, 14]. Therefore, the conduction band of the Ga-rich layer starts to decrease in the region near the GaO x layer. A thicker GaO x is expected to bring in larger density of positive charges. With respect to the constant BE of Ga–O and Al–O in S3, it indicates that the positive charge should locate at the interface of the Ga-rich layer/GaO x layer. The positive charges and polarization-induced negative charges build an internal electric filed which modified the band bending from upward band bending to downward band bending, shown in Fig. 4. Because of the downward band bending, the BE increases with the increase of detection angle.
The GaO x interfacial layer brings in positive charges which increase interface barrier height ɸb. ɸb is defined as the energy difference between the Fermi level and the conduction band minimum at the surface or interface . As a result, after the A2O3 deposition, the mobility of the 2DEG is increased and the electron density of 2DEG is decreased [16, 25, 26].
With the increase of the Al2O3 thickness, the XPS signal reflects more on the interface region between the capped Al2O3 and GaN/AlGaN/GaN, which is validated by the XPS sampling depth shown in Table 2. This explains that only part of the band bending profile could be detected for S2 . As a result, the BE variation is 0.1 eV, smaller than 0.2 eV of S1. For S3, with a thicker interfacial layer, the density of positive charges is increased resulting in a downward band bending.
In summary, the interface polarization of Al2O3-capped GaN/AlGaN/GaN is investigated by the ARXPS. The intrinsic polarization of GaN is eliminated because of a Ga-rich layer and a GaO x layer formation. Moreover, the Ga–O bonds from the GaO x layer bring in positive interfacial fixed charges. Due to this polarization variation, the band varies from the upward bending to the downward bending in the interface region.
Chien CY, Wu WH, You YH, Lin JH, Lee CY, Hsu WC, Kuan CH, Lin RM (2017) Breaking through the multi-mesa-channel width limited of normally off GaN HEMTs through modulation of the via-hole-length. Nanoscale Res Lett 12:1
Ťapajna M, Válik L, Gucmann F, Gregušová D, Fröhlich K, Haščík Š, Dobročka E, Tóth L, Pécz B, Kuzmík J (2017) Low-temperature atomic layer deposition-grown Al2O3 gate dielectric for GaN/AlGaN/GaN MOS HEMTs: impact of deposition conditions on interface state density. J Vac Sci Technol B 35:1
Duan TL, Pan JS, Ang DS (2013) Interfacial chemistry and valence band offset between GaN and Al2O3 studied by X-ray photoelectron spectroscopy. Appl Phys Lett 102:20
Shih HY, Chu FC, Das A, Lee CY, Chen MJ, Lin RM (2016) Atomic layer deposition of gallium oxide films as gate dielectrics in AlGaN/GaN metal–oxide–semiconductor high-electron-mobility transistors. Nanoscale Res Lett 11:1
Joglekar S, Azize M, Jones EJ, Piedra D, Gradečak S, Palacios T (2016) Impact of Al2O3 passivation on AlGaN/GaN nanoribbon high-electron-mobility transistors. IEEE Trans Electron Devices 63:1
Wang CH, Ho SY, Huang JJ (2016) Suppression of current collapse in enhancement-mode AlGaN/GaN high electron mobility transistors. IEEE Electron Device Lett 37:1
Oyama S, Hashizume T, Hasegawa H (2002) Mechanism of current leakage through metal/n-GaN interfaces. Appl Surf Sci 190:1–4
Hao Y, Yang L, Ma XH, Ma JG, Cao MY, Pan CY, Wang C, Zhang JC (2011) High-performance microwave gate-recessed AlGaN/AlN/GaN MOS-HEMT with 73% power-added efficiency. IEEE Electron Device Lett 32:5
Maeda N, Hiroki M, Watanabe N, Oda Y, Yokoyama H, Yagi T, Makimoto T, Enoki T, Kobayashi T (2007) Systematic study of insulator deposition effect (Si3N4, SiO2, AlN, and Al2O3) on electrical properties in AlGaN/GaN heterostructures. Jpn J Appl Phys Part 1 46:2
Liu C, Chor EF, Tan LS (2006) Investigations of HfO2/AlGaN/GaN metal-oxide-semiconductor high electron mobility transistors. Appl Phys Lett 88:17
Duan TL, Pan JS, Ang DS (2016) Investigation of surface band bending of Ga-face GaN by angle-resolved X-ray photoelectron spectroscopy. ECS J Solid State Sci Technol. 5:9
Coan MR, Woo JH, Johnson D, Gatabi IR, Harris HR (2012) Band offset measurements of the GaN/dielectric interfaces. J Appl Phys 112:2
Ganguly S, Verma J, Li GW, Zimmermann T, Xing HL, Jena D (2011) Presence and origin of interface charges at atomic-layer deposited Al2O3/III-nitride heterojunctions. Appl Phys Lett 99:19
Esposto M, Krishnamoorthy S, Nath DN, Bajaj S, Hung T-H, Rajan S (2011) Electrical properties of atomic layer deposited aluminum oxide on gallium nitride. Appl Phys Lett 99:13
Gregušová D, Jurkovič M, Haščík Š, Blaho M, Seifertová A, Fedor J, Ťapajna M, Fröhlich K, Vogrinčič P, Liday J, Derluyn J, Germain M, Kuzmik J (2014) Adjustment of threshold voltage in AlN/AlGaN/GaN high-electron mobility transistors by plasma oxidation and Al2O3 atomic layer deposition overgrowth. Appl Phys Lett 104:1
Bakeroot B, You S, Wu TL, Hu J, Hove MV, Jaeger BD, Geens K, Stoffels S, Decoutere S (2014) On the origin of the two-dimensional electron gas at AlGaN/GaN heterojunctions and its influence on recessed-gate metal-insulator-semiconductor high electron mobility transistors. J Appl Phys 116:13
Cook TE, Fulton CC, Mecouch WJ, Davis RF, Lucovsky G, Nemanich RJ (2003) Band offset measurements of the GaN (0001)/HfO2 interface. J Appl Phys 94:11
Craft HS, Rice AL, Collazo R, Sitar Z, Maria JP (2011) Spectroscopic measurements of the surface stoichiometry of chemical vapor deposited GaN. Appl Phys Lett 98:8
Duan TL, Yu HY, Wu L, Wang ZR, Foo YL, Pan JS (2011) Investigation of HfO2 high-k dielectrics electronic structure on SiO2/Si substrate by x-ray photoelectron spectroscopy. Appl Phys Lett 99:1
Sun Q, Selloni A, Myers TH, Doolittle WA (2006) Oxygen adsorption and incorporation at irradiated GaN (0001) and GaN (0001-) surfaces: first-principles density-functional calculations. Phys Rev B 74:19
Li SB, Ware M, Wu J, Minor P, Wang ZM, Wu ZM, Jiang YD, Salamo GJ (2012) Polarization induced pn-junction without dopant in graded AlGaN coherently strained on GaN. Appl Phys Lett 101:12
Li SB, Zhang T, Wu J, Yang YJ, Wang ZM, Wu ZM, Chen Z, Jiang YD (2013) Polarization induced hole doping in graded AlxGa1−xN (x = 0.7 ∼ 1) layer grown by molecular beam epitaxy. Appl Phys Lett 102:6
Li SB, Ware ME, Wu J, Kunets VP, Hawkridge M, Minor P, Wang ZM, Wu ZM, Jiang YD, Salamo GJ (2012) Polarization doping: reservoir effects of the substrate in AlGaN graded layers. J Appl Phys 112:5
Watkins NJ, Wicks GW, Gao Y (1999) Oxidation study of GaN using x-ray photoemission spectroscopy. Appl Phys Lett 75:17
Duan TL, Pan JS, Ang DS (2015) Effect of post-deposition annealing on the interface electronic structures of Al2O3-capped GaN and GaN/AlGaN/GaN heterostructure. ECS J Solid State Sci Technol 4:9
Ambacher O, Smart J, Shealy JR, Weimann NG, Chu K, Murphy M, Schaff WJ, Eastman LF, Dimitrov R, Wittmer L, Stutzmann M, Rieger W, Hilsenbeck J (1999) Two-dimensional electron gases induced by spontaneous and piezoelectric polarization charges in N- and Ga-face AlGaN/GaN heterostructures. J Appl Phys 85:6
Wang XL, Xiang JJ, Wang WW, Zhang J, Han K, Yang H, Ma XL, Zhao C, Chen DP, Ye TC (2013) A possible origin of core-level shift in SiO2/Si stacks. Appl Phys Lett 102:4
This study was funded by the Key Laboratories of third-generation semiconductor devices in Shenzhen, Shenzhen Science and Technology Innovation Committee, China (Grant No. ZDSYS20140509142721434), and research on low-cost GaN power devices fabrication and the system integration, Shenzhen Science and Technology Innovation Committee, China (Grant No. JCYJ20160226192639004).
Ethics Approval and Consent to Participate
Consent for Publication
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.