Electrical Properties and Interfacial Issues of HfO₂/Ge MIS Capacitors Characterized by the Thickness of La₂O₃ Interlayer

Abstract

Effects of the La₂O₃ passivation layer thickness on the interfacial properties of high-k/Ge interface are investigated systematically. In a very thin range (0~15 cycles), the increase of La₂O₃ passivation layer deposition cycles improves the surface smoothness of HfO₂/Ge structures. The capacitance-voltage (C-V) characteristics show that the thickness of La₂O₃ passivation layer can affect the shift of flat band voltage (V_FB), hysteretic behaviors, and the shapes of the dual-swept C-V curves. Moreover, significant improvements in the gate leakage current and breakdown characteristics are also achieved with the increase of La₂O₃ interlayer thickness.

1. Introduction

Along with the continuing scaling of complementary metal oxide semiconductor (CMOS) technology according to Moore’s Law, insulation oxides with higher permittivity have been introduced to replace SiO₂ for acceptable gate leakage current density and low power consumption [1]. However, other than reducing feature size, technological progress to gain high-speed operation is also needed with the development of integrated circuit (IC) technology. Considering this, high mobility semiconductors have been considered as alternative channel materials for obtaining high drive current. Among the high mobility semiconductors, germanium (Ge) is compatible with standard Si CMOS process, and both the electron and hole bulk mobility are higher than those of Si substrate [2,3]. Therefore, Ge has been regarded as one of the most promising alternative channel materials. Unfortunately, different from the perfect interfacial properties of SiO₂/Si interface, the poor thermal stability of GeO₂/Ge interface results in the deterioration of interfacial properties to a great extent [4]. When most gate dielectric materials deposited on the unpassivated Ge substrate, the generation of unstable Ge oxides is unavoidable during the high temperature post-deposition annealing (PDA) process, and the decomposition or desorption from GeO₂ to volatile GeO would bring in a great deal of defects. Consequently, to get acceptable electrical properties of high-k/Ge structures, a suitable surface passivation treatment is a necessary technical issue prior to the deposition of high-k gate dielectrics on Ge substrates [5].

Owing to the much better thermodynamic stability than that of GeO₂/Ge, fewer interface traps exist at the interface between La-based oxides and Ge substrates [6]. Consequently, La-based oxides have been considered as one kind of the alternative gate dielectric materials deposited on Ge to realize good electrical performance of Ge-based metal-insulator-semiconductor (MIS) devices [7,8]. However, the hygroscopicity of La₂O₃ layer limits its application as gate insulators [9]. In a previous study, we have proved that a ~2 nm La₂O₃ interlayer could effectively improve the electrical performance of Ge MIS devices, resulting in more than one order of magnitude decrease in the gate leakage current density [10]. In this work, a further investigation was performed on the impacts of La₂O₃ passivation layer thickness on the interfacial properties of HfO₂ insulators and Ge substrates.

2. Materials and Methods

La₂O₃ and HfO₂ were deposited on Sb-doped n-type Ge (100) wafers with a doping concentration of 1 × 10¹⁶ cm⁻³ by atomic layer deposition (ALD) method (R-150, Picosun, Espoo, Finland). Prior to the deposition, all the wafers were treated using acetone and alcohol, followed by dipping into HF (2%) and deionized water by cyclic cleaning for 5 times to remove the defective native Ge oxide. Tris(isopropyl-cyclopentadienyl) lanthanum (La(^i-PrCp)₃) and tetrakis (ethylmethylamino) hafnium (TEMAH) were used as La and Hf precursor, while H₂O was used as oxidant with reference to the reported literatures [11,12]. Under the chamber temperature of 300 °C, La₂O₃ layers were deposited by alternately introducing La(^i-PrCp)₃ and H₂O precursors to the reactor chamber using high purity N₂ (>99.999%) as the carrier gas with a typical ALD growth cycle of 0.3 s La(^i-PrCp)₃ pulse/4 s N₂ purge/0.3 s H₂O pulse/9 s N₂ purge. Using these process parameters, the steady-state growth rate of La₂O₃ was approximately 0.85 Å/cycle. For an ALD growth cycle of HfO₂, TEMAH was pulsed into the chamber by carrier gas for 0.3 s with a 8 s N₂ purge, and then H₂O was pulsed for 0.1 s followed by a 8 s N₂ purge. As a result, a stable growth rate of 0.75 Å/cycle was obtained for HfO₂. The thickness of the La₂O₃ passivation layer was tuned by varying the number of ALD cycles, and the cycle numbers of 5, 10, and 15 cycles were chosen. Then ~6 nm HfO₂ was deposited as the gate insulator materials after the deposition of La₂O₃ passivation layer. For comparison, the sample with only ~6 nm HfO₂ as the gate oxide was also fabricated as a control sample. After the deposition of gate insulators, rapid thermal annealing (RTA) was carried out at 400 °C for 90 s in N₂ ambient for all the wafers. For simplicity, the control sample is assigned as S1, and samples with 5, 10, and 15 ALD cycles of La₂O₃ passivation layers are assigned as S2, S3, and S4, respectively.

The physical thickness of the deposited films was optically measured using the Woollam M2000U spectroscopic ellipsometry (SE, Woollam Co. Inc., Lincoln, NE, USA), and the ellipsometry data were fitted using a Gen-Osc mode consisting of Gaussian and Tauc-Lorentz oscillators. The chemical bonding states of the interfaces between the deposited films and Ge substrates were examined by X-ray photoelectron spectroscopy (XPS, Axis Ultra DLD, Kratos Analytical, UK) measurements. Cross-sectional high-resolution transmission electron microscopy (HRTEM, TECNAI F20 system, ThermoFisher Scientific, Waltham, MA, USA) and energy dispersive X-ray spectroscopy (EDX) line scan measurements were performed to observe the microstructures and atomic compositions of the deposited films. The surface morphology of the films was monitored using atomic force microscopy (AFM, Dimension 3100, Veeco Digital Instruments by Bruker, Billerica, MA, USA) in tapping mode. The electrical properties of the HfO₂/La₂O₃ gate stacks were evaluated using MIS capacitor structures. MIS capacitors were fabricated by ion-beam etching (IBE, IBE-A-150, Beijing Chuangshi Weina Technology Co., Ltd., Beijing, China) of e-beam evaporated 150 nm Al through a shadow mask with a diameter of 300 µm as the top gate electrode and 100 nm Al as the back ohmic contact. Then, the electrical properties of the fabricated MIS capacitors were measured using an Agilent B1500A parameter analyzer (Santa Clara, CA, USA).

3. Results and Discussion

3.1. Chemical Bonding States of the Interfaces between the Deposited Films and Ge Substrates

The chemical bonding states near the interface of HfO₂ and HfO₂/La₂O₃ gate stacks deposited on Ge substrate were investigated by XPS measurements. Prior to the measurements, the wafers S1~S4 were etched by Ar⁺ ion beam bombardment for 25 s, 30 s, 30 s, and 35 s, respectively, under the etch rate of ~0.20 nm/s to remove the influence of the surface impurities and to analyze the chemical bonding states in the interfacial region. The XPS data were calibrated by setting the C 1s peak originated from the carbon impurities in the spectra at 284.6 eV for all the wafers. Figure 1 shows the variations of the O 1s XPS spectra for the HfO₂ film and HfO₂/La₂O₃ gate stacks deposited on Ge substrates. For the control sample (S1), the O 1s spectra consist of three Gaussian–Lorentzian line shape peaks. These peaks are located at 530.2, 531.0, and 531.9 eV, corresponding to the chemical bonding states of Hf–O–Hf, Hf–O–Ge, and Ge–O–Ge, respectively [13,14]. The existence of Hf–O–Ge and Ge–O–Ge peaks indicate that HfGeO_x and GeO_x were generated at the HfO₂/Ge interface during the deposition of HfO₂ and post-deposition annealing (PDA) process. As for the HfO₂/La₂O₃/Ge structures, the O 1s core level spectra are deconvoluted into four peaks, three of which are at 529.0 (I), 530.2 (II), and 531.9 (IV) eV. These peaks correspond to the chemical bonds of La–O–La, Hf–O–Hf, and Ge–O–Ge, respectively [15]. The other peak (III) at around 530.9 eV originates from the Hf–O–Ge and/or La–O–Ge bonding states, indicating the generation of germanate (LaGeO_x and HfGeO_x) compound at the interface of HfO₂/La₂O₃ gate stacks and Ge substrates. It is worth noting that the position of peak (III) shifts to lower binding energies from sample S1 to S4. This phenomenon is attributed to the transformation from Hf-rich to La-rich interfacial layer (IL) with the increase of La₂O₃ passivation layer thickness, owing to the stronger reaction between La₂O₃ and outdiffused Ge atoms than that of HfO₂ with outdiffused Ge atoms [16,17].

To further discuss the variation tendency of Ge oxides at high k dielectrics/Ge interfaces with the increase of La₂O₃ passivation layer thickness, investigations on the Ge 3d XPS spectra are shown in Figure 2. A substrate peak (Ge 3d⁰) located at ~28.6 eV with additional peaks at higher binding energies in relation to the IL (consisting of Ge oxides and germanate) could be observed in all spectra. The Ge 3d⁰ substrate peak is fitted with a doublet of Ge 3d_5/2 and Ge 3d_3/2 with spin-orbit splitting of 0.6 eV and intensity ratio of 3:2, respectively [18]. The Ge oxides (GeO_x) consist of four peaks (Ge¹⁺, Ge²⁺, Ge³⁺, Ge⁴⁺), which are at higher binding energy respect to the Ge 3d⁰ with energy shifts of 0.8, 1.8, 2.6, and 3.4 eV, respectively [19]. Compared with the control sample shown in Figure 2a, the intensity of the Ge²⁺(GeO) peak decreases with the increase of La₂O₃ passivation layer thickness, while the intensity of the LaGeO_x peak shows an increasing trend. Consequently, with the increase of La₂O₃ passivation layer thickness, the variation tendency of the interfacial components extracted from the Ge 3d XPS spectra is consistent with the results extracted from the O 1s XPS spectra.

3.2. Microstructures of the Deposited Films on Ge Substrates

Additional microstructure information of the deposited films is provided by cross-sectional HRTEM analysis as shown in Figure 3. For the control sample without La₂O₃ passivation layer, the HfO₂ layer exhibits an amorphous structure as no nanometer-sized crystal or long-range ordered crystal region is observed. However, in sample S4 with 15 ALD cycles of La₂O₃ passivation layer (as shown in Figure 3b), nanometer-sized crystals could be observed in the HfO₂ layer, indicating the incomplete crystallization state of the HfO₂ film. It is noteworthy that upon the same RTA condition at 400 °C for 90 s in N₂ ambient, the HfO₂ films in sample S1 and S4 show different crystalline characteristics. This difference could be attributed to the RTA-induced Ge diffusion from the substrate into the HfO₂ layer as shown in the EDX profiles near the interfaces between the deposited films and Ge substrates (Figure 4). The EDX data were dealt with normalization method after eliminating the influence of impurity elements, and the depth values were calibrated by HRTEM results. During the RTA process, in sample S1 as shown in Figure 4a, the substrate Ge atoms would diffuse easily to the upper HfO₂ layer and react with the oxygen nearby the HfO₂/Ge interface, which would prevent the formation of crystalline HfO₂ precipitates [20] and cause the formation of IL (~0.8 nm) consisting of HfGeO_x and GeO_x, as analyzed in the above XPS results. As for the HfO₂/La₂O₃/Ge structure in sample S4 (Figure 4b), due to the high affinity of La₂O₃ for Ge atoms [21], the out-diffused Ge atoms would react with La₂O₃ passivation layer, leading to the spontaneous formation of stable LaGeO_x IL. As a result, few Ge atoms would diffuse into the HfO₂ layer, and the crystallization phenomenon occurs after the RTA treatment.

3.3. Surface Morphology of HfO₂ and HfO₂/La₂O₃ Gate Stacks on Ge Substrates

The AFM images of HfO₂/La₂O₃ gate stacks on Ge substrates with 0, 5, 10, and 15 ALD cycles of La₂O₃ interfacial passivation layer are shown in Figure 5. The scan size of each AFM image was set as 10 × 10 µm². The AFM images were captured from the AFM data by the software named Gwyddion. The root mean square (RMS) roughness of samples S1~S4 is extracted to be 0.63, 0.51, 0.28, and 0.25 nm, respectively. The small RMS roughness values of all the samples illustrate the smooth and crack-free surfaces of the deposited HfO₂/La₂O₃ gate stacks, indicating that ALD is a good deposition method for high-k dielectric films and stacks. On the basis of the smooth surfaces, a decrease in the RMS roughness values is observed after inserting a La₂O₃ interfacial passivation layer, and the RMS roughness values decrease with the increase of La₂O₃ layer thickness. It has been reported that the outdiffused Ge atoms from substrates and the desorption of volatile species GeO during the high temperature annealing process could bring in large roughness [22]. After inserting La₂O₃ interfacial passivation layer, the stable LaGeO_x layer could suppress the formation of unstable GeO_x layer, which contributes to the decrease of the RMS roughness value. With the increase of La₂O₃ thickness, more La₂O₃ reacts with the outdiffused Ge atoms and unstable Ge oxides, which effectively suppresses desorption from GeO₂ to volatile GeO, resulting in a smoother surface. It is found that when the thickness of La₂O₃ layer increase above 15 cycles, the variation in the RMS roughness values becomes negligible. The variation tendency of RMS roughness illustrates that 15 ALD cycles of La₂O₃ layer are thick enough to improve the surface smoothness for HfO₂/Ge structures.

3.4. Electrical Performance of Al/HfO₂/Ge and Al/HfO₂/La₂O₃/Ge MIS Capacitors

The capacitance-voltage (C-V) and conductance-voltage (G-V) characteristics of the fabricated Al/HfO₂/n-Ge and Al/HfO₂/La₂O₃/n-Ge MIS capacitors are shown in Figure 6. Dual-swept C-V curves were obtained by sweeping from inversion to accumulation (forward) and sweeping back (backward) at the frequency of 100 kHz. G-V curves were obtained simultaneously with the C-V curves measured forward. To quantitatively characterize the effects of La₂O₃ passivation layer thickness on electrical performance of HfO₂/Ge MOS structures, some parameters are extracted from the dual-swept C-V curves. The accumulation capacitance (C_ac) values for gate stacks S1~S4 are obtained to be 1.35, 1.48, 1.69, and 1.87 μF/cm², respectively. Then, the oxide capacitance (C_ox) and dielectric constant values of gate stacks could be calculated following the equations [23,24]:

$$C_{\mathrm{ox}}=C_{\mathrm{ac}}[1+{(\frac{G_{\mathrm{ac}}}{\omega C_{\mathrm{ac}}})}^2]$$

(1)

$$CET=\frac{{\epsilon }_0{\epsilon }_{{\mathrm{SiO}}_2}A}{C_{\mathrm{ox}}}$$

(2)

$$k=\frac{{\epsilon }_{{\mathrm{SiO}}_2}{t}_{\mathrm{ox}}}{CET}$$

(3)

where C_ac is the capacitance value at accumulation region, G_ac is the conductance corresponding to the accumulation region of the C-V curves, ω is the angular frequency, C_ox is the oxide capacitance of gate stacks, A is the area of the top electrode, CET is the capacitance equivalent thickness of deposited gate stacks (including IL), t_ox is the measured thickness of deposited gate stacks (including IL), ε₀ and ε_SiO2 are the permittivity values of vacuum and SiO₂, respectively. The physical thickness of gate stacks (including IL) in S1, S2, S3, and S4 is optically measured to be 6.71, 7.89, 8.26, and 8.95 nm, respectively. The corresponding 95% confidence interval for the average thickness of the deposited films was shown in Table S1 in the Supplementary Materials to evaluate the discrete degree of thickness testing data. Therefore, according to Equations (1)–(3), the effective k values of S1~S4 are achieved to be 10.69, 13.43, 15.97, and 18.46, respectively. It is found that the dielectric constant value of HfO₂/La₂O₃ gate stacks increases with the increase of La₂O₃ layer thickness. The improvement on the value of dielectric constant for the gate stacks is attributed to the effective suppression of Ge atoms outdiffusion benefiting from the generation of stable LaGeO_x, since the k value of La₂O₃ is much higher than that of GeO_x [25]. In addition, as analyzed in the EDX profiles, the HfO₂ quality is improved by La₂O₃ passivation, which should be the other cause that increases the dielectric constant of the deposited films.

Moreover, as shown in Figure 6, varying degrees of anomalous humps in the weak inversion region of C-V curves could be observed. Compared with the control sample (Figure 6a), the anomalous hump phenomenon turns more serious after inserting 5 ALD cycles of La₂O₃ passivation layer. While with the further increase of La₂O₃ passivation layer thickness, the humps of C-V curves are obviously reduced. Besides, as the thickness of La₂O₃ passivation layer increases, different degrees of frequency dispersion phenomenon at the weak inversion regions could be observed from the multi-frequency C-V curves in Figure 7, and the degrees of frequency dispersion phenomenon show a similar variation tendency with the anomalous humps in the C-V curves measured at 100 kHz. It has been reported that the existence of slow interface traps (Q_it) should be responsible for the anomalous phenomenon at the weak inversion regions of C-V curves [26,27]. Considering this, the interface state density (D_it) values of S1~S4 are discussed using the following relation of single-frequency approximation method through the forward swept C-V and G-V curves [28]:

$$D_{\mathrm{it}}=\frac{2}{qA}\frac{\frac{G_{\max }}{\omega }}{\left[{\left(\frac{G_{\max }}{\omega C_{\mathrm{ox}}}\right)}^2+{\left(1-\frac{C_{\max }}{C_{\mathrm{ox}}}\right)}^2\right]}$$

(4)

where q is the elementary charge (1.602 × 10⁻¹⁹ C), A is the area of the top electrode, C_ox is the gate oxide capacitance as defined in Equation (1), G_max is the peak value of G-V curve, and C_max is the capacitance corresponding to G_max at the same gate applied voltage. Following Equation (4), the D_it values for the fabricated MIS capacitors using S1~S4 as insulators are calculated to be 9.18 × 10¹², 9.72 × 10¹², 3.95 × 10¹², and 2.71 × 10¹² eV⁻¹cm⁻², respectively. The D_it value increases slightly after inserting 5 ALD cycles of La₂O₃ passivation layer, which is in consistent with the variation tendency of the humps in C-V curves. The increment of D_it is reported to be caused by an incomplete reaction of the precursor molecules during the first few ALD cycles [29], which brings in extra interface states at the interface. As the ALD cycles of La₂O₃ passivation layer increase, the improvement of the La₂O₃ passivation layer on the interfacial properties of gate insulators/Ge interfaces plays a more and more important role, contributing to the decrease of interface traps for the samples S3 and S4. For accuracy, the D_it values were also extracted using conductance method [30] measured from 1 kHz to 1 MHz, as shown in Figure 8. It could be observed that for each energy state, the D_it value extracted by conductance method increases slightly after inserting 5 ALD cycles of La₂O₃ passivation layer, while with the further increase of La₂O₃ passivation layer thickness, the D_it value shows decrease trend. This result is in good consistency with the single-frequency approximation method result. The contribution of the La₂O₃ interfacial passivation layer to the amount of interface traps at the La₂O₃/Ge interface could be explained as follows: during the ALD process of La(^i-PrCp)₃ and H₂O precursors and the high temperature PDA process, La₂O₃ is more likely to react with outdiffused Ge atoms and Ge oxides nearby the interface to form stable LaGeO_x compound. The stable LaGeO_x compound could effectively suppress Ge outdiffusion, and further inhibit the decomposition or desorption from GeO₂ to volatile GeO (following the reaction equation of GeO₂ + Ge → 2GeO), contributing to the reduction of structural defects and dangling bonds [31,32]. The D_it value (2.71 × 10¹² eV⁻¹cm⁻²) of sample S4 achieved in our work is comparable with the results obtained in the reported literatures at La₂O₃/Ge interfaces of 3 × 10¹² [8] and 3.5 × 10¹² eV⁻¹cm⁻² [25], indicating that a La₂O₃ passivation layer with 15 ALD cycles (~1.3 nm) can effectively suppress the generation of interface states at high k dielectrics/Ge interfaces.

It is also observed from the C-V curves that the flat band voltages (V_FB) translate toward positive voltages with the increase of La₂O₃ passivation layer deposition cycles. To further investigate this phenomenon in detail, the V_FB values of the capacitors were extracted using the Hauser NCSU CVC simulation software taking into account of quantum-mechanical effects [33]. Considering the work function difference between the top electrode aluminum and the n-type Ge substrate with a doping concentration of 1 × 10¹⁶ cm⁻³, the ideal V_FB should be 0.03 V. However, the actual V_FB swept forward for the control sample is −0.206 V. The negative V_FB shift indicates the existence of effective positive oxide charges in the HfO₂ film. The effective positive oxide charges may be attributed to the existence of positive fixed oxide charges and oxide trapped charges mainly consisting of oxygen vacancies and structural defects in the gate insulators and/or nearby the insulators/Ge interface. Compared with the control sample, the forward swept V_FB values of MIS capacitors S2~S4 were extracted to be −0.083, −0.071, and −0.048 V, separately. The positive shifts of V_FB with the increase of La₂O₃ passivation layer thickness reveal the reduction of oxide charges after the Ge surface passivation treatment using La₂O₃ as the passivation layer. Considering this, the trapped charge density (N_ot) is estimated using the midgap charge separation method through the C-V hysteresis characteristics following the Equation (5) [34]:

$$N_{\mathrm{ot}}=\frac{\Delta V_{\mathrm{FB}}{C}_{\mathrm{ox}}}{qA}$$

(5)

where ΔV_FB is the hysteresis width of V_FB, C_ox is the oxide capacitance, q is the elementary charge (1.602 × 10⁻¹⁹ C), and A is the area of the electrode. For the control sample, using the exacted ΔV_FB value of 685 mV, a trapped oxide charge density of 6.03 × 10¹² cm⁻² is obtained according to Equation (5). After the insertion of La₂O₃ passivation layer, the trapped oxide charge densities of samples S2, S3 and S4 are calculated to be 4.74 × 10¹², 2.87 × 10¹², and 1.79 × 10¹² cm⁻², respectively. These parameter results concerning the electrical properties are summarized in

Table 1. The electrical parameters extracted from the fabricated MIS capacitors without and with La₂O₃ interfacial passivation layers.

Sample	Cox (µF/cm2)	k	△VFB (mV)	Dit (eV−1cm−2)	Not (cm−2)
S1	1.411	10.69	685	9.18 × 1012	6.03 × 1012
S2	1.510	13.43	504	9.72 × 1012	4.74 × 1012
S3	1.712	15.97	269	3.95 × 1012	2.87 × 1012
S4	1.883	18.46	152	2.71 × 1012	1.79 × 1012

. As suspected, the trapped oxide charge density reduces with the increase of La₂O₃ thickness. We ascribe the decrease of trapped charges to the reduction of oxygen vacancies and structural defects in the gate stack region nearby the interface. At the HfO₂/Ge interface, the outdiffused Ge atoms will react with GeO₂ to form unstable GeO, which could bring in a large number of oxygen vacancies and structural defects. The LaGeO_x compound formed by La₂O₃ and Ge could effectively suppress the outdiffusion of Ge atoms and the desorption of volatile GeO, contributing to the reduction of N_ot values.

Figure 9 shows the gate leakage current density as a function of the applied effective field for samples S1~S4. It is observed that the gate leakage current density of gate insulator/Ge structures obviously decreases with the increase of La₂O₃ layer thickness. At the applied effective field of 3 MV/cm, compared with the control sample, in the sample S4 the gate leakage current density decreases of almost an order of magnitude. The obvious decrease in the gate leakage current density could be ascribed to the reduction in the structural defects in the HfO₂/La₂O₃ gate stack benefiting from the generation of stable LaGeO_x component. Moreover, the gate leakage current density-voltage (J-V) characteristics show different breakdown behaviors, which suggests that the incorporation of La₂O₃ interfacial passivation layer improves the breakdown field strength of HfO₂/Ge MIS capacitors. The improvement of the breakdown field strength is reported to be associated with a reduction in defects in gate dielectrics [35]. As mentioned above, the existence of La₂O₃ interfacial passivation layer could effectively suppress the generation of structural defects, oxygen vacancies and dangling bonds. The reduction of structural defects, oxygen vacancies and dangling bonds decreases the possibility to create a conduction path by forming a continuous chain connecting the gate to the semiconductor, resulting in higher breakdown field.

4. Conclusions

In summary, we investigate the effect of La₂O₃ passivation layer thickness on the interfacial and electrical properties of HfO₂/Ge MIS structures. It is found that the gate leakage current density of the MIS capacitor after inserting 15 ALD cycles (~1.3 nm) of La₂O₃ passivation layer gains almost one order of magnitude decrease than that of the Al/HfO₂/Ge MIS sample. Besides, a relatively low D_it value of 2.71 × 10¹² eV⁻¹cm⁻² is achieved. The thickness of the La₂O₃ passivation layer also affects the dielectric constants, V_FB shifts, hysteresis behaviors and the shapes of dual-swept C-V curves. The improvement on the interfacial and electrical properties with the increase of La₂O₃ passivation layer deposition cycles is related to the formation of a stable LaGeO_x layer on the Ge surface, which restrains the decomposition or desorption from GeO₂ to volatile GeO, contributing to the decrease of structural defects. These results provide us with quite an effective method for realizing high-quality dielectric/Ge interfaces for Ge-based MIS devices.