International Journal of Energy Research

Volume 2025, Issue 1 5517224

Research Article

Open Access

Synergistic Effects of Nitrogen–Oxygen–Nitrogen, Forming Gas–Oxygen–Forming Gas, and Argon–Oxygen–Argon Annealing Ambient on the Structural and Electrical Characteristics of Thulium Oxide Passivation Layers on Silicon Substrate

Junchen Deng

orcid.org/0000-0002-0361-2147

Institute of Nano Optoelectronics Research and Technology (INOR) , Universiti Sains Malaysia , 11800 , Penang , Malaysia , usm.my

Bailie School of Petroleum Engineering , Lanzhou City University , Lanzhou , 730070 , China , lzcu.edu.cn

Search for more papers by this author

Hock Jin Quah,

Corresponding Author

Hock Jin Quah

orcid.org/0000-0002-7338-9489

Institute of Nano Optoelectronics Research and Technology (INOR) , Universiti Sains Malaysia , 11800 , Penang , Malaysia , usm.my

Search for more papers by this author

Junchen Deng,

Junchen Deng

orcid.org/0000-0002-0361-2147

Institute of Nano Optoelectronics Research and Technology (INOR) , Universiti Sains Malaysia , 11800 , Penang , Malaysia , usm.my

Bailie School of Petroleum Engineering , Lanzhou City University , Lanzhou , 730070 , China , lzcu.edu.cn

Search for more papers by this author

Hock Jin Quah,

Corresponding Author

Hock Jin Quah

orcid.org/0000-0002-7338-9489

Institute of Nano Optoelectronics Research and Technology (INOR) , Universiti Sains Malaysia , 11800 , Penang , Malaysia , usm.my

Search for more papers by this author

First published: 02 February 2025

https://doi.org/10.1155/er/5517224

Citations: 1

Academic Editor: Sejoon Lee

Share a link

Email
Wechat
Bluesky

Abstract

A comprehensive probe was conducted to compare the impact of postdeposition annealing at 700°C in different ambient of nitrogen–oxygen–nitrogen (NON), forming gas–oxygen–forming gas (FGOFG), and argon–oxygen–argon (ArOAr) on the passivating characteristics of thulium oxide (Tm₂O₃) on the silicon (Si) substrate. The nitrogen ions have been incorporated in Tm₂O₃ passivation layers after annealing in NON and FGOFG ambient, of which NON ambient has impeded the growth of silicon oxide (SiO₂) interfacial layer (3.258 nm). Although a thicker SiO₂ interfacial layer (4.026 nm) was formed after annealing in FGOFG ambient, the attainment of the highest k value (16.8) indicated that the existence of hydrogen ions has assisted the improvement in the overall k value of Tm₂O₃ passivation layers. Furthermore, the capacitance–voltage characteristic revealed that the FGOFG ambient was effective in reducing the effective oxide charge (1.32 × 10¹² cm⁻²), while NON ambient was effective in passivating the slow trap density (STD) (3.20 × 10¹¹ cm⁻²). The Terman, Hill–Coleman, and high–low frequency methods have demonstrated the acquisition of the best interface quality during annealing in FGOFG ambient. As a result, the FGOFG annealing process has led to the maximum breakdown electric field of 4.03 MV/cm and the minimum leakage current density for the Tm₂O₃ passivation layer.

1. Introduction

With the birth and development of Industry 4.0, almost all aspects of human life and production will be linked together through a range of electronic devices, wherein the silicon (Si)-based metal–oxide–semiconductor (MOS) devices have played an important role in fulfilling the requirement of high-performance and advanced technology [1]. For many decades, Si has stood out from the crowd to be the most widely used semiconductor material due to its low cost and ability of attaining excellent passivating layer through the thermally grown native oxide of Si (silicon oxide [SiO₂]) that would ensure the Si-based MOS devices to operate reliably [2]. In order to further enhanced the performance and energy efficiency of Si-based MOS devices, the dimension of Si-based MOS devices has been shrunk throughout the years wherein further reduction in the thickness of SiO₂ to the range of 1 nm would result in the intolerable high gate leakage current governed by direct tunneling mechanism [3]. Thus, substituting SiO₂ with alternative materials possessing higher dielectric constant (k) could provide similar capacitance with a greater physical thickness that would circumvent the detrimental effect of direct tunneling mechanism, which would further enhanced the performance as well as energy efficiency of Si-based MOS devices [1]. To date, numerous research efforts have been carried out with regard to the employment of various alternative high-k materials, such as HfO₂ [4], Al₂O₃ [5], ZrO₂ [6], Ga₂O₃ [7], Y₂O₃ [8], CeO₂ [9], Yb₂O₃ [10], La₂O₃ [11], Ta₂O₅ [12], and thulium oxide (Tm₂O₃) [13, 14] which could potentially addressed aforementioned detrimental effects. Out of all these high-k materials, rare earth Tm₂O₃ has emerged as a promising option for application in Si-based MOS devices as a passivation layer by virtue of intriguing features such as high-k value (9.4–18.0) [14–18], wide band gap (E_g) (5.2–6.8 eV) [16, 19–21], large conduction band offset with relative to Si (2.00–2.95 eV) [19, 22], and excellent thermal stability with Si [23].

Previous attempts have disclosed that the as-deposited Tm₂O₃ passivation layers on Si substrate were comprised of a high density of intrinsic defects regarding to the Si dangling bonds and oxygen vacancies formed at the interface [24–26]. Similarly to other high-k materials, Tm₂O₃ has a high coordination number making relaxation as well as reorientation of Tm₂O₃ bonding to be very challenging unless postdeposition annealing process was employed [3, 27]. Consequently, the postdeposition annealing was performed to decrease the number of defect states in the Tm₂O₃ passivation layer, giving rise to the improved threshold voltage, carrier mobility, and reliability of the Si-based MOS devices [3, 15, 17, 24, 25, 28]. Different forms of ambient have been used for annealing after deposition process, including oxygen [16, 24, 25, 29, 30], nitrogen [17, 28, 30], argon [31], ozone [30], and forming gas (FG) (a mixture of hydrogen and nitrogen gases) [15, 16, 30] to enhance the passivating characteristics of the Tm₂O₃ passivation layer, which include the leakage current density, interface quality, and k value. Of this investigated postdeposition annealing ambient, it was found that using an oxygen ambient during annealing at 800°C has helped to low the interface trap density of the as-deposited Tm₂O₃ passivation layer on Si from 8.1 × 10¹² to 1.6 × 10¹² cm⁻²eV⁻¹ [24]. The beneficial impact of the oxygen ambient on interface quality was further confirmed for the Tm₂O₃ passivation layer on a germanium (Ge) substrate, wherein the oxygen annealing has resulted in a lower interface trap density of 3.7 × 10¹¹ cm⁻²eV⁻¹ than nitrogen annealing of 8.9 × 10¹¹ cm⁻²eV⁻¹ at similar annealing temperature of 500°C [30]. Additionally, the compositional homogeneity of the as-deposited Tm₂O₃ passivation layer was restored when it was subjected to postdeposition annealing at temperatures ranging from 500 to 800°C in an oxygen ambient. The smaller flat-band voltage shift (∆V_FB) observed in the annealed samples indicated a lower defect density compared to the as-deposited sample [24]. Nevertheless, one of the hindrance of using oxygen ambient is the acquisition of a higher capacitance equivalent thickness when compared with nitrogen ambient, suggesting the formation of thicker interfacial layer [30]. Further investigation based on X-ray photoelectron spectroscopy core level spectra of Ge 3 p has revealed the increment in ratio between the peak areas associated with GeO_x and Ge, which has reinforced the increased formation of the GeO_x interfacial layer that likely contributes to the reduction in the overall k value of the Tm₂O₃ passivation layer [30].

Additionally, the employment of nitrogen or FG (97% nitrogen and 3% hydrogen) during annealing process for Tm₂O₃ passivation layers has contributed to the acquisition of higher k values of 16 [17] and 18 [15], respectively, when compared with oxygen annealed Tm₂O₃ passivation layer (k = 14.0–15.0) [16]. Therefore, the annealing process in nitrogen- and/or hydrogen-containing ambient has facilitated in the suppression of oxygen diffusion to the Si/Tm₂O₃ interface, thereby forming a thinner interfacial layer. Similar results have been observed for the HfO₂ passivation layer [32, 33]. Besides, argon gas was also known as an oxygen-deficient ambient that was capable of providing suppressive effect toward the growth of interfacial layer during annealing process [34]. While the growth of interfacial layer was retarded during annealing in an oxygen-deficient ambient, the oxygen-deficient annealing, which includes the nitrogen and FG annealing, may result in a degradation of the interface quality of Tm₂O₃ passivation layer, whereby a lower interface trap density was attained by Tm₂O₃ passivation layer annealed in oxygen ambient [30]. This indicated that oxygen ambient was a prerequisite to enhance the interface quality, fix defects associated to oxygen, and restore the compositional homogeneity of Tm₂O₃ passivation layers [30, 31]. In order to take advantage on both oxygen and oxygen-deficient ambient, the postdeposition annealing was conducted in a combination of nitrogen–oxygen–nitrogen (NON) ambient. Specifically, nitrogen gas was introduced during the heating and cooling phases, while oxygen gas was introduced during the dwelling phase [35, 36]. The beneficial effect of using NON ambient was revealed, whereby the Tm₂O₃ passivation layer annealed in NON ambient at 700°C has obtained a higher k value of 16.0 [35] when compared with oxygen ambient (k = 14.0–15.0) [16]. Therefore, this research work intends to innovate by performing a systematic investigation on the effects of postdeposition annealing at 700°C in NON [35], forming gas–oxygen–forming gas (FGOFG), and argon–oxygen–argon (ArOAr) ambient on the structural, morphological, optical, and MOS characteristics of the Tm₂O₃ passivation layers.

2. Experimental Procedure

The 4-inch phosphorus doped (~10¹⁵ cm⁻³) n-type Si (100) wafer (resistivity of 1–10 Ω·cm) was cut into smaller dimension of 1 cm × 1 cm. These samples were cleaned in accordance with the procedures of the Radio Corporation of America (RCA). Subsequently, the native oxide of SiO₂ was removed by dipping these cleaned samples in a diluted hydrofluoric solution. These samples were immediately loaded into a radio frequency (RF) magnetron sputtering machine (HVV AUTO A500) for the deposition of Tm₂O₃ passivation layers on Si substrate. The Tm₂O₃ sputtering target employed in this study was of 99.99% purity. Initially, the plasma cleaning was performed in argon (Ar; 99.9999%) ambient with flow rate of 15 sccm for 10 min and a working pressure of 7.6 × 10⁻³ mbar using RF power of 100 W. Then, the shutter of the RF magnetron sputtering machine was opened to allow deposition of Tm₂O₃ passivation layer to be carried out for 40 min. After the deposition process, the postdeposition annealing was conducted in a horizontal tube furnace at 700°C and atmospheric pressure, with different ambient (NON, FGOFG, and ArOAr). The samples were ramped up from 25 to 700°C at a heating rate of 10° C/min under the flow of N₂/FG/Ar gas at 100 sccm. Once the postdeposition annealing temperature has been raised to 700°C, a dwelling time of 30 min was employed with the flow of O₂ gas at 100 sccm. Subsequently, the horizontal tube furnace was naturally cooled down from 700 to 25°C under the flow of N₂/FG/Ar gas at 100 sccm. In this work, FG with a mixture of 10% of hydrogen and 90% of nitrogen was being used. In order to fabricate into a MOS test structure, aluminum (Al) electrodes with diameter of 0.2 mm were deposited on Tm₂O₃ passivation layers via a shadow mask using a thermal evaporator machine (AUTO 306), while a blanket of Al was evaporated on the backside of Si substrate with a contact resistivity of 0.12 ± 0.02 Ω·cm².

Grazing incidence X-ray diffraction (GIXRD) with a step size of 0.05° and a step time of 1.5 s were used to characterize the crystalline phases and orientations of the investigated Tm₂O₃ passivation layers. The thickness of the Tm₂O₃ passivation layer and the SiO₂ interfacial layer was determined by performing X-ray reflectivity (XRR) on the investigated passivation layers with a step size of 0.01° and a step time of 1.5 s. Both GIXRD and XRR measurements were carried out using Bruker D8 Discover with a Cu K_α radiation (λ = 1.5406 Å) that was operated using operating voltage and current of 40 kV and 40 mA, respectively. Surface morphologies, cross-section images, and atomic percentage of elements of the investigated Tm₂O₃ passivation layers were obtained using field-emission scanning electron microscopy (FESEM) (FEI Nova NanoSEM 450) that was equipped with energy-dispersive X-ray (EDX) (JSM-6460 LV) spectroscopy. The three-dimensional root mean square (RMS) roughness and topographies of the examined Tm₂O₃ passivation layers were investigated using atomic force microscopy (AFM) (Dimension Edge, BRUKER). The UV-Vis spectrophotometer (Cary 5000) was employed to measure the diffused reflectance spectra of the Tm₂O₃ passivation layers. Keithley 4200-SCS parameter analyzer was employed to perform current–voltage (I–V) measurements as well as capacitance–voltage (C–V) and conductance–voltage (G–V) measurements in a frequency ranging from 1 MHz to 10 kHz.

3. Results and Discussion

The crystal structure and characteristics of the Tm₂O₃ passivation layers annealed in NON [35], FGOFG, and ArOAr ambient at 700°C were determined by analyzing the typical GIXRD spectra shown in Figure 1. Polycrystalline phase was observed in all of the investigated Tm₂O₃ passivation layers. The Tm₂O₃ phase was oriented in the (222), (400), (440), and (622) planes, which were assigned to the International Center for Diffraction Data (ICDD) file number 00-041-1090 for cubic Tm₂O₃ phase. Reference to this standard Tm₂O₃ sample, it was revealed that NON annealing resulted in the least variance in the shifting of detected Tm₂O₃ peaks to wider diffraction angle. This may imply that nitrogen ions have been incorporated into Tm₂O₃ lattice and attached to the oxygen vacancies, while the adjacent oxygen ions were released in order to achieve charge compensation. A greater shifting of Tm₂O₃ diffraction peak was observed for the sample annealed in FGOFG ambient. The observation of larger variation in the shifting of detected GIXRD peaks during annealing in FGOFG ambient could be linked to the existence of both hydrogen and nitrogen gases, wherein the smaller-radius hydrogen ions (0.038 nm) incorporated into the Tm₂O₃ lattice would contribute to a contraction in the lattice parameter a of the investigated passivation layer [37]. Among these investigated samples, the largest shifting in the detected Tm₂O₃ GIXRD peaks was observed for the sample annealed in in ArOAr ambient, implying that this sample consisted of a significant number of oxygen vacancies (Table 1).

Details are in the caption following the image — **Figure 1**
Open in figure viewer PowerPoint

GIXRD patterns of Tm₂O₃ passivation layers experienced NON [35], FGOFG, and ArOAr annealing at 700°C. ArOAr, argon–oxygen–argon; FGOFG, forming gas–oxygen–forming gas; GIXRD, grazing incidence X-ray diffraction; NON, nitrogen–oxygen–nitrogen; Tm₂O₃, thulium oxide.

Table 1. GIXRD diffraction angle of Tm₂O₃ passivation layers experienced NON [35], FGOFG, and ArOAr annealing at 700°C.

Ambient	Orientation
Ambient	(222)	(400)	(440)	(622)
NON	29.598	34.230	49.257	58.497
FGOFG	29.605	34.242	49.270	58.518
ArOAr	29.653	34.273	49.299	58.524
ICDD File No. 00-041-1090	29.475	34.156	49.099	58.312

Abbreviations: ArOAr, argon–oxygen–argon; FGOFG, forming gas–oxygen–forming gas; GIXRD, grazing incidence X-ray diffraction; ICDD, International Center for Diffraction Data; NON, nitrogen–oxygen–nitrogen; Tm₂O₃, thulium oxide.

The lattice parameter a of the investigated Tm₂O₃ passivation layers was determined according to the fitted results of the obtained GIXRD spectra, as indicated by the following equations [38]:

(1)

(2)

where λ, n, θ, d_hkl, and hkl denote X-ray radiation wavelength, diffraction order, diffraction angle, interplanar spacing, and Miller’s index, respectively. All of the investigated samples have attained a smaller a (Figure 2) compared to the standard Tm₂O₃ sample (1.0487 nm), which has supported the fact that the employment of these annealing ambient (NON, FGOFG, and ArOAr) have triggered a lattice contraction. The lattice parameter was expected to be potentially influenced by the generation of oxygen vacancies and the incorporation of the foreign ions in the Tm₂O₃ lattice. It was anticipated that the lattice will expand and contract in response to the incorporation of the larger radius nitrogen ions (0.146 nm) and smaller radius nitrogen ions (0.038 nm), respectively [37]. Among the samples analyzed, it has been demonstrated that the Tm₂O₃ passivation layer, which was annealed in an ArOAr environment, possesses the smallest a (Figure 2). Due to the annealing ambient lack hydrogen and nitrogen ions, the largest lattice contraction observed in this sample was attributed to a significant oxygen vacancies formed after annealing process. Nonetheless, this does not match with the EDX result, which indicates that this sample acquired the highest at% of oxygen (Table 2). One of the possible reasons contributing to this discrepancy was the generation of SiO₂ interfacial layer stimulated during annealing in ArOAr ambient, wherein the absence of nitrogen and FG during the heating and cooling stage would cause more oxygen ions to diffuse from the Tm₂O₃ passivation layer to the Tm₂O₃/Si interface, forming the SiO₂ interfacial layer. Hence, it was postulated that attainment of the highest at% of oxygen for the sample annealed in ArOAr ambient was due to the oxygen ions being accumulated at Si/Tm₂O₃ interface as well as the oxygen ions from the SiO₂ interfacial layer.

Table 2. Atomic percentage (at%) of oxygen, nitrogen, and thulium in Tm₂O₃ passivation layers experienced NON [35], FGOFG, and ArOAr annealing at 700°C.

Ambient	at%
Ambient	Tm	O	N
NON	40.06	54.36	5.59
FGOFG	38.95	56.13	4.92
ArOAr	41.57	58.43	—

Abbreviations: ArOAr, argon–oxygen–argon; FGOFG, forming gas–oxygen–forming gas; NON, nitrogen–oxygen–nitrogen; Tm₂O₃, thulium oxide.

When FGOFG ambient was employed during annealing, the EDX results of this Tm₂O₃ passivation layer demonstrated a lower at% of oxygen (56.13 at%) which implied that the existence of hydrogen and nitrogen gases could have possibly impeded the reaction of the oxygen and Si to form the SiO₂ interfacial layer. It was hypothesized that, during the annealing process, the hydrogen and nitrogen from the FG would diffuse toward the Tm₂O₃/Si interface and form a barrier layer that would restrict the diffusion of oxygen ions towards the Tm₂O₃/Si interface. Besides, these hydrogen and nitrogen ions could also attach to the oxygen vacancies in the Tm₂O₃ passivation layer, wherein the smaller radius hydrogen ions should diffuse faster than nitrogen ions to attach to the oxygen vacancies. This would contribute to a smaller lattice parameter a for the Tm₂O₃ passivation layer annealed in FGOFG ambient than in NON ambient (Figure 2). The probability of nitrogen ions attaching to the oxygen vacancies and diffusing to the interface to form a barrier layer should be reduced, as a portion of the oxygen vacancies in the Tm₂O₃ passivation layer were attached to the hydrogen ions during annealing in FGOFG ambient. Consequently, a lower nitrogen at% of 4.92% was observed for this sample. Hence, FGOFG annealing was predicted to produce a thicker SiO₂ interfacial layer than NON annealing. Of these investigated samples, the largest lattice parameter a (1.0447 nm) was attained by the sample annealed in NON ambient. Although it was anticipated that the incorporation of nitrogen ions would result in a greater number of oxygen vacancies due to the release of nearby oxygen ions to achieve charge neutrality, the acquisition of the largest lattice parameter a for the sample annealed in NON ambient has suggested that nitrogen ions with larger ionic radius of 0.146 nm than oxygen ions (0.138 nm) have partially occupied the oxygen vacancies in the Tm₂O₃ passivation layer resulting in the expansion of lattice parameter a. Furthermore, the acquisition of the highest at% of nitrogen as well as the lowest at% of oxygen suggested that more nitrogen ions have been accumulated at the Tm₂O₃/Si interface and formed a barrier layer with a stronger impeding effect on the formation of SiO₂ interfacial layer.

Two layers model on Si substrate was used in XRR fitting for determining the thickness of the Tm₂O₃ passivation layer and SiO₂ interfacial layer by employing Bruker DIFFRAC Leptos (7.10.0.12) software (Figure 3). The thickness of the SiO₂ interfacial layer formed after annealing process in different ambient as shown in Figure 4 has indicated that the annealing ambient comprising of nitrogen and/or hydrogen gases were more efficient in preventing the oxygen ions from interacting with the Si surface. Therefore, the sample annealed in ArOAr ambient attained the thickest SiO₂ interfacial layer (10.893 nm), which agreed with previous postulation whereby the observation of the greatest oxygen at% (58.43 at%) for this passivation layer might be because of the generation of the thickest SiO₂ interfacial layer. Besides, it was also disclosed that NON annealing was more effective in inhibiting the reaction between oxygen ions and Si (3.258 nm) than that of FGOFG annealing (4.026 nm). The reason for this observation might be linked to the larger ionic radius of nitrogen ions accumulating and forming the barrier layer at the Tm₂O₃/Si interface during NON annealing whereby more energy was needed for the oxygen ions to diffuse through this barrier layer and react with Si surface when compared with FGOFG annealing. When FGOFG ambient was employed, both hydrogen and nitrogen ions might be capable to diffuse to the interface. The smaller ionic radius of the hydrogen ion would enhance the susceptibility of oxygen ions to diffuse through the hydrogen ions and react with substrate. Another justification for the increased thickness of SiO₂ after FGOFG annealing was due to the hydrogen ions having a greater tendency to create weaker bonding with Si dangling links when compared to the nitrogen ions. This has caused the oxygen ions to easily break the bonding between hydrogen ions and Si dangling bonds, resulting in the thickening of the SiO₂ interfacial layer [39, 40]. Hence, the reaction between oxygen and Si surface was weaker for the samples annealed in NON and FGOFG ambient than that in ArOAr ambient as the oxygen vacancies sites available for the oxygen ions to hop to the interface was being reduced during annealing in NON and FGOFG ambient. In addition, the SiO₂ interfacial layer of all annealed Tm₂O₃ passivation layers have increased from 0.217 nm for the as-deposited sample [35] to the range of 3.258–10.893 nm.

Besides, the effect of different annealing ambient on the thickness was evaluated by determining the thickness from XRR measurements. It was discovered that all of the annealed samples were undergoing densification process and demonstrated a thinner thickness ranging from 45.621 to 51.156 nm when compared with the as-deposited sample (61.687 nm) [35]. Among these investigated passivation layers, it was determined that NON ambient was less effective in densifying the Tm₂O₃ passivation layer, which might be associated with the occupancy of oxygen vacancies by nitrogen ions. With the nitrogen ions incorporated into Tm₂O₃ lattice, the adjacent oxygen ions in the Tm₂O₃ lattice were being released to achieve charge neutrality, whereby the densification of Tm₂O₃ passivation layer was restricted. Moreover, it was mentioned earlier that the larger radius nitrogen ions attached to the oxygen vacancies would hinder the oxygen diffusion. Thus, the densification of Tm₂O₃ passivation layer during annealing in NON ambient was hindered by the release of the adjacent oxygen and the incorporation of nitrogen. Despite the introduction of both hydrogen and nitrogen ions has taken place during annealing in an FGOFG ambient, the formation of the thinnest passivation layer has suggested that hydrogen ions with smaller ionic radius are more likely to bond with oxygen vacancies. This bonding increased the likelihood of oxygen ions diffusing through and filling these vacancies within the Tm₂O₃ lattice. Furthermore, the simultaneous occupancy of oxygen vacancies by both hydrogen and nitrogen ions during annealing in an FGOFG ambient did not cause the release of neighboring oxygen ions from the Tm₂O₃ lattice for charge neutrality. This is because the excess negative charges introduced by nitrogen ions are balanced by the positive charges generated when hydrogen ions were attached to the oxygen vacancies.

During postdeposition annealing in oxygen-deficient ambient, such as nitrogen, FG, and argon, oxygen will diffuse outward from the Tm₂O₃ lattice due to the difference in concentration gradient between the Tm₂O₃ layer and the annealing ambient. The outward diffusion of oxygen would result in the formation of doubly positively charged oxygen vacancies in the lattice since the two electrons carried by oxygen anion (O²⁻) that should occupy the lattice were missing. The two electrons were transferred to the neighboring Tm³⁺ cations, thus yielding a reduction of the oxidation state from Tm³⁺ to Tm²⁺. The processes were illustrated by the following half equations:

(3)

(4)

The N₂ molecules from the nitrogen gas ambient would be adsorbed on the oxygen vacancies sites in the Tm₂O₃ lattice and dissociated into nitrogen anions (N³⁻). The N³⁻ anions might substitute into the oxygen vacancies in the lattice since both N³⁻ (143 pm) and O²⁻ (140 pm) possessed similar ionic radius. Nonetheless, the substitution of N³⁻ in one oxygen vacancy would trigger charge imbalance in the lattice, according to Equation (5) and thus inducing the formation of more oxygen vacancies in neighboring anion sites for charge neutrality:

(5)

However, when the Tm₂O₃ passivation layer was annealed in FG ambient consisting of 10% hydrogen and 90% nitrogen, the N₂ and H₂ molecules from the FG ambient would be adsorbed on the oxygen vacancies sites in the Tm₂O₃ lattice and dissociated into N³⁻ and H⁺ ions. The relatively smaller radius of hydrogen ions was reactive, and the incorporation of the H⁺ ions with the

lattice would form an electroneutral product, based upon the suggestion equation below:

(6)

Therefore, the densification of Tm₂O₃ passivation layer during annealing in FGOFG ambient was less affected by the oxygen vacancies. Since the formation of SiO₂ interfacial layer was more excessive in ArOAr ambient, the formation of more oxygen vacancies in the Tm₂O₃ passivation layer has reduced the densification process of Tm₂O₃ passivation layer, wherein a slightly thicker Tm₂O₃ passivation layer was acquired when compared with the sample annealed in FGOFG ambient. Overall, it was determined that the densification of the Tm₂O₃ passivation layer was hindered during annealing in NON because nitrogen ions attached to the oxygen vacancies have caused the formation of more oxygen vacancies. The cross-sectional images of the samples annealed in NON, FGOFG, and ArOAr ambient were shown in Figure 5, where in 10 measurements were taken using ImageJ software to determine total oxide thickness. Furthermore, the total thickness derived from XRR measurements and cross-sectional FESEM characterization were found to be comparable and exhibit a similar trend.

The effects of the NON, FGOFG, and ArOAr annealing on the microstrain (ε) and crystallize size (D) of the studied Tm₂O₃ passivation layers were also determined using the Williamson–Hall (W-H) approach expressed by the following equation [41]:

(7)

where the β_i, θ_i, and λ denote integral breadth, diffraction angle, and X-ray wavelength. For the extracted ε and D of the investigated samples as shown in Figure 6, it was noticed that the NON annealing produced the smallest D for the Tm₂O₃ passivation layer due to both the restriction effect of the incorporated nitrogen ions and the induced extra oxygen vacancies. Although the FGOFG annealing has introduced both hydrogen and nitrogen ions into Tm₂O₃ lattice, the attainment of the largest D indicated that the largest D has been attained by this passivation layer as the release of oxygen ions was not encountered because these oxygen vacancies were occupied by both hydrogen and nitrogen ions, whereby the oxygen ions were more likely to replace the hydrogen ions during dwelling stage. An analysis comparing the Tm₂O₃ passivation layers annealed in FGOFG and ArOAr ambient has revealed that the Tm₂O₃ passivation layer treated with ArOAr annealing process has demonstrated a greater quantity of oxygen vacancies. The increase of oxygen vacancies has triggered oxygen ions to diffuse to the Tm₂O₃/Si interface, facilitating the reaction with Si to form SiO₂ and a reduction in D value. The Tm₂O₃ passivation layer was induced with compressive ε when annealed in the NON ambient, while tensile ε was induced when annealed in the ArOAr and FGOFG ambients, leading to the acquisition of positive and negative ε values (Figure 6), respectively. The acquisition of compressive ε for the sample annealed in NON ambient can be ascribed to the introduction of nitrogen ions and the release of adjacent oxygen ions. Furthermore, the higher tensile ε was observed for the sample subjected to FGOFG annealing because of the simultaneous introduction of both hydrogen and nitrogen ions into the Tm₂O₃ lattice in which the adjacent oxygen ions abstained from being released have triggered the generation of additional tensile ε compared to the sample annealed in ArOAr ambient. This indicates that the concentration of oxygen vacancies has governed the acquisition of a lower tensile ε in the sample annealed in ArOAr ambient.

The evaluation on the growth orientation was conducted by extracting the coefficient of texture (T_hkl), in accordance with the following equation [42]:

(8)

where n is the number of Tm₂O₃ reflection peaks and the I_m(hkl) and I_o(hkl) are the relative reflective intensities from (hkl) plane of the annealed and standard Tm₂O₃, respectively. It could be conveniently perceived from the calculated T_hkl values (Figure 7) that the samples annealed in NON and FGOFG ambient demonstrated a T₂₂₂ and T₄₄₀ valuer greater than 1, suggesting that these samples were grown preferentially in (222) and (440) planes. Nonetheless, only T₂₂₂ exceeding 1 was observed after ArOAr annealing, implying that Tm₂O₃ phase with (440) orientation was no longer predominating.

Furthermore, AFM was employed to evaluate the surface topographies of the Tm₂O₃ passivation layers during annealing in the NON, FGOFG, and ArOAr ambient, with a scanning area of 10 μm × 10 μm. The FGOFG annealing process, as illustrated in Figure 8, led to the uniform growth of protrusions with a comparable height across the surface, as well as the lowest RMS roughness (0.997 nm). The consistent densification throughout the Tm₂O₃ passivation layer during annealing in FGOFG ambient due to lesser release of the adjacent oxygen ions could explain this observation. By comparing with ArOAr annealing, the NON annealing would cause a weak densification on Tm₂O₃ passivation layer, while a protrusion with comparable height (Figure 8a) was formed due to the accumulation of the nitrogen ions at the Tm₂O₃/Si interface slowing down the reaction between oxygen ions and Si surface. Hence, the NON annealing resulted in a lower RMS roughness for the sample annealed in ArOAr ambient. Based on the fact of forming an excessively thick SiO₂ interfacial layer after annealing in ArOAr ambient, uneven protrusions (Figure 8c) formed on this Tm₂O₃ passivation layer due to oxygen deficiency resulted in the highest RMS roughness (Figure 8d). Furthermore, surface morphologies of the Tm₂O₃ passivation layer annealing in NON, FGOFG, and ArOAr ambient were evaluated using FESEM, where no cracks and voids were formed on the smooth surface for all samples (Figure 9).

The Kubelka–Munk function was used to extract the direct (E^D) and indirect (E^ID) bandgap of the investigated Tm₂O₃ passivation layers from the diffused reflectance spectra obtained using a UV-Vis spectrophotometer. This function can be represented by the following equation [43]:

(9)

where R denotes diffuse reflectance and F (R) is the function of R used to extract the band gap. The curves of [F (R) × hv]² and [F (R) × hv]^1/2 with respect to hv was plotted (not shown) to determine direct E_g (E^D) and indirect E_g (E^ID), respectively. Based on the literature, it was reported that the E^D of bulk Tm₂O₃ was 5.400 eV [44], while the E^D of Tm₂O₃ film deposited on Si substrate was 5.760 eV [20]. The E^D and E^ID of the investigated Tm₂O₃ passivation layers were shown in Figure 10. It was revealed in this work that the E^D of Tm₂O₃ passivation layer annealed in NON (5.088 eV) and FGOFG (5.236 eV) ambient was narrower than the reported E^D for bulk Tm₂O₃ (5.400 eV) [44] and Tm₂O₃ film (5.760 eV) [20], which could be associated to the incorporation of nitrogen ions into the Tm₂O₃ lattice when FG and N₂ were employed. Nonetheless, the Tm₂O₃ passivation layer annealed in ArOAr has attained a larger E^D of 5.434 eV, which was comparable with the E^D for bulk Tm₂O₃ (5.400 eV) [44]. In addition, it was noticed that NON annealing has attained the smallest E^D because of the introduction of more nitrogen ions into the Tm₂O₃ lattice when compared with the sample annealed in FGOFG ambient. Therefore, the band gap of this sample was narrowed by a higher concentration of interband state caused by the nitrogen incorporation during NON annealing. Similar finding has been revealed by the nitrogen incorporated HfO₂ passivation layer, whereby narrowing of band gap has taken placed as a result of the introduction of N2p states that have triggered occurrence of conduction band narrowing as well as valance band broadening [45]. Nevertheless, it was perceived that a smaller E^ID of 4.119 eV was acquired by the Tm₂O₃ passivation layer annealed in FGOFG ambient than NON ambient (E^ID = 4.364 eV), which could be linked to fact that the hydrogen ions from the FGOFG annealing ambient were induced into Tm₂O₃ lattice. This could be an indication that these induced hydrogen ions would give rise to a higher concentration of shallower energy level when compared with the nitrogen ions that would narrow the E^ID of the FGOFG annealed Tm₂O₃ passivation layer [46]. Since no foreign ions were introduced into the Tm₂O₃ passivation layer annealed in ArOAr ambient, the largest E^D (5.434 eV) and E^ID (4.722 eV) have been obtained by this sample. Despite the samples annealed in FGOFG and NON ambient having narrower E^D and E^ID values than the sample annealed in ArOAr ambient, the samples annealed in these ambient have nevertheless achieved E^D and E^ID values that are large enough to offer a sufficient conduction and valence band offset with respect to the Si substrate.

The C–V measured for the investigated Tm₂O₃ passivation layers subjected to NON, FGOFG, and ArOAr annealing was presented in Figure 11. It was worth noting that self-written code was employed in accordance with the theory described in the [47] for the fitting of the measured C–V curves with respect to ideal C–V curve. Based on Figure 11, the effect of different annealing ambient on the dielectric constant (k) values was calculated using the following equation [48]:

(10)

where ε₀ is the permittivity of free space, C_ox is the accumulation capacitance that measured at accumulation level, A is area of metal contact, and t_ox denote total oxide thickness of Tm₂O₃ passivation layers that measured from the cross-section FESEM characterization (Figure 4). Of these investigated samples, the lowest k value of 15.2 (Figure 12) was obtained for the sample annealed in an ArOAr ambient. This result further endorses the notion that the thick SiO₂ interfacial layer (10.893 nm) generated in this sample was a result of the lack of hydrogen and nitrogen ions. Hence, the samples annealed in FGOFG and NON ambient obtained higher k values of 16.8 and 16.0, respectively, as the XRR results have indicated the formation of thinner SiO₂ interfacial layer ranging from 3.258 to 4.026 nm. Although a thicker SiO₂ interfacial layer (4.026 nm) was formed during annealing in FGOFG ambient, the attainment of a higher dielectric constant of 16.8 by this passivation layer when compared with that annealed in NON ambient (16.0) has denoted that the minute variation in the of SiO₂ interfacial layer’s thickness was considered to have no substantial impact on the k value of the investigated samples. Furthermore, the lower k value observed in the sample annealed in NON ambient, compared to FGOFG ambient, might be due to the higher concentration of nitrogen ions and oxygen vacancies present in the samples annealed in NON ambient. Previous studies have shown that the presence of nitrogen ions in high-k materials and the formation of oxygen vacancies in high-k passivation layers have triggered an increment in dielectric [49–51].

It was divulged that negative flatband voltage shift (ΔV_FB) was measured from Tm₂O₃ passivation layers annealed in NON, FGOFG, and ArOAr ambient and presented as effective oxide charge (Q_eff) determined by the following equation [52]:

(11)

where q denotes electronic charge while the ΔV_FB was calculated by comparing the shift in V_FB between the ideal and measured C–V curves [47]. According to the determined Q_eff shown in Figure 13, the sample annealed in NON ambient has attained the largest Q_eff value (4.89 × 10¹² cm⁻²) denoting the presence of the highest density of positive charged tarps in this sample. The obtained largest Q_eff for the sample annealed in NON ambient agreed with earlier explanation, whereby a higher concentration of positively charged oxygen vacancies was formed contributing to the attainment of smallest D for this passivation layer. However, the sample annealed in FGOFG ambient achieved the lowest Q_eff value of 1.32 × 10¹² cm⁻², indicating that the occupation of oxygen vacancies by hydrogen and nitrogen ions has impeded the released of adjacent oxygen ions. Furthermore, the likelihood of inward diffusing oxygen ions to disrupt the hydrogen ion interaction with oxygen vacancies was increased. Consequently, a greater number of oxygen ions might be diffused and occupied the oxygen vacancies in the sample that was annealed in the FGOFG ambient. A marginal increment in Q_eff value (1.34 × 10¹² cm⁻²) was gained by the sample annealed in ArOAr ambient when compared with FGOFG ambient suggesting that the absence of hydrogen and nitrogen ions has triggered a higher concentration of oxygen ions to diffuse to the interface contributing to the generation of thickest SiO₂ interfacial layer (10.893 nm). Hence, by comparing with FGOFG annealing, the ArOAr annealing induced more oxygen vacancies in the Tm₂O₃ passivation layer.

The slow trap density (STD) of the investigated samples was determined from the bidirectional C–V curves using the following equation [53]:

(12)

where ΔV denote the difference in flat-band voltage between forward and reverse bias. It can be conveniently observed from Figure 13 that the sample annealed in ArOAr ambient demonstrates the highest STD due to the presence of the abundant of oxygen vacancies and the absence of nitrogen and/or hydrogen ions in passivating the Si dangling bonds, wherein these positively charged oxygen vacancies as well as Si dangling bonds located near to the interface would be able to capture and release the injected electrons when forward and reverse biases were applied. It was noticed that the sample annealed in FGOFG ambient has resulted in a higher STD than that in NON ambient because of the attachment of hydrogen ions to the Si dangling bonds. When positive bias was applied to the sample annealed in FGOFG ambient, the injected electrons might have gained sufficient energy to break the bonding of hydrogen ions to the Si dangling bonds. Therefore, these Si dangling bonds were capturing the subsequent injected electrons when further increment in forward bias was applied. Then during the reverse bias, the electrons would de-trap from the Si dangling bonds and demonstrated a higher STD for Tm₂O₃ passivation layer annealed in FGOFG ambient than NON ambient. Since the employment of NON annealing has stimulated the accumulation of more nitrogen ions at the Tm₂O₃/Si interface, the injected electrons during forward bias would not be attracted to these negatively charged nitrogen ions, and thus, the probability of electrons to be trapped would be lower, thus resulting in the lowest STD value.

The interface quality with regard to interface trap density (D_it) of the Tm₂O₃ passivation layers annealed in NON, FGOFG, and ArOAr ambient was determined using the following equation [54]:

(13)

where Φ_s represents the responding Si surface potential for a given voltage while the ΔV_g represents voltage difference between the gate voltage acquired experimentally (V_g) and ideally from the ideal C–V curve (V_g(ideal)) [47]. As shown in Figure 14, the Tm₂O₃ passivation layer after NON annealing demonstrated the highest D_it as a result of the highest content of nitrogen ions on the Si/Tm₂O₃ interface. This observation was consistent with the degrading effect of nitrogen annealing on interface quality of Al₂O₃ passivation layer where the D_it was increased from ~1.5 × 10¹³ to ~3.0 × 10¹³ cm^–2 after annealing in nitrogen [55]. On the other hand, the nitrogen barrier layer might prevent oxygen ions from reacting with the Si surface, thereby contributing to a higher D_it for the samples annealed in NON and FGOFG ambient. Hence, it was observed that an improvement in D_it was achieved after ArOAr annealing, wherein the thickest SiO₂ interfacial layer would annihilate the interface traps and improve the interface quality. The sample experiencing FGOFG annealing demonstrated a lower D_it than NON ambient, which might be attributed to the accumulation of lesser nitrogen ions at the interface as well as the possibility of oxygen ions to break the attachment of hydrogen ions with Si dangling bonds. As a result, the reaction between these oxygen ions and Si would be stimulated to form interfacial layer and reduce the D_it. Although the sample annealed in FGOFG ambient has a higher D_it ranging from 1.4 to 7.0 × 10¹² eV^–1cm^–2 than the sample annealed in ArOAr ambient (0.2 to 5.5 × 10¹² eV^–1cm^–2), the range of D_it values obtained by the sample annealed in FGOFG ambient was comparable to the reported D_it values for HfO₂ (3.0 × 10¹² eV^–1cm^–2) [56], Y₂O₃ (ranging from 2.0 to 8.0 × 10¹² eV^–1cm^–2) [57], and ZrO₂ (ranging from 3.0 to 7.0 × 10¹² eV^–1cm^–2) [58] passivation layers deposited on Si substrate.

Besides, based on the C–V and G–V characteristics of the investigated samples that were measured at different frequencies (10–1000 kHz), the investigation on the interface quality was also conducted using Hill–Coleman and high–low frequency methods. To eliminate the effect of the series resistance in the measuring circuit, the obtained C–V and G–V curves were corrected using the method presented in [47, 59]. Based on the C_c–V and G_c–V results as shown in Figure 15, the D_it was calculated using the Hill–Coleman method [59]:

(14)

where G_c,max is the maximum conductance in G_c–V result, C_c,max denotes the corrected capacitance at the voltage corresponding to G_c,max, and C_ox represent the corrected capacitance of the corresponding passivation layers that calculated from G_m,acc and C_m,acc using the method presented in [47, 59].

Figure 16 depicts the calculated D_it (Hill–Coleman mothed) for the investigated samples. Overall, it was determined that both Hill–Coleman and Terman’s methods have revealed similar trend for the sample annealed in NON ambient, wherein this passivation layer has attained the highest D_it values. Nonetheless, a discrepancy between the trend obtained from Hill–Coleman and Terman’s methods for both Tm₂O₃ passivation layers annealed in ArOAr as well as FGOFG ambient has revealed that the acquisition of a greater D_it (Hill–Coleman method) value by the sample annealed in ArOAr ambient might be associated with the presence of the greatest STD for this passivation layer. The attainment of the lowest D_it (Hill–Coleman method) value for the Tm₂O₃ passivation layer experiencing FGOFG annealing has further suggested the hydrogen in FG was more effective in eliminating the interface related defects. It was also revealed in other research work, whereby the FG annealing for Al₂O₃ passivation layer has resulted in an interface state density of ~5 × 10¹² cm⁻² which is lower than nitrogen annealing (~3 × 10¹³ cm⁻²) [55].

In addition, based on the corrected C–V results, the D_it were also calculated using high–low frequency C–V method [60]:

(15)

where C_LF is the corrected capacitance calculated from the capacitance measured at 10 kHz and C_HF is corrected capacitance calculated from the capacitance measured at 1 MHz. Figure 17 displays the calculated D_it from the high–low frequency C–V method for the investigated passivation layers, wherein the E_c–E_t was derive from the surface potential [61] which was determined using Berglund’s integral as described in the references [62]. From Figure 17, it was deduced that Tm₂O₃ passivation layers annealed in FGOFG and ArOAr ambient have demonstrated a higher D_it near to the conduction band edge and decreased as the energy level moved away from the conduction band edge. This observation could be associated to the ability of oxygen ions to passivate the traps located further away from the conduction band edge during annealing in FGOFG and ArOAr ambient. Besides, the availability of hydrogen ions during annealing in FGOFG ambient has resulted in the acquisition of the lowest D_it at E_c–E_t level larger than 0.38 eV in which these hydrogen ions would be able to diffuse to a higher energy level to passivate these traps. Nonetheless, an increment in D_it was noticed when the E_c–E_t was enhanced beyond 0.22 eV, which could be linked to the formation of nitrogen barrier layer that have impeded the oxygen ions from repairing the interface related traps. It has been discovered that the sample annealed in NON ambient has exhibited the lowest D_it at E_c–E_t values below 0.22 eV. This indicates that oxygen ions have acquired sufficient energy to surpass the nitrogen barrier layer and rectify the interface-related traps within this specific range of E_c–E_t.

The leakage current density–electric field (J–E) characteristics of the investigated Tm₂O₃ passivation layers was presented in Figure 18. Among the investigated passivation layers, the sample which underwent FGOFG annealing showed the lowest J and the highest E_B (4.03 MV/cm). This can be attributed to its highest k value, the lowest Q_eff, the lowest D_it calculated using the Hill–Coleman method, and lowest D_it at E_c–E_t level greater than 0.38 eV calculated using the high–low frequency method. Although the sample experiencing NON annealing ambient has attained the lowest STD as well as the lowest D_it at E_c–E_t smaller than 0.22 eV (calculated using high-low frequency method), the attainment of a lower E_B (2.83 MV/cm) by this sample when compared with the sample annealed in FGOFG ambient has suggested that the influence of these two parameters toward the E_B could be less significant. It was postulated that for the acquisition of the highest Q_eff, the highest D_it calculated from both Terman and Hill–Coleman methods by the sample annealed in NON ambient has also contributed to the deterioration in J–E characteristic. Nonetheless, it was observed that the Tm₂O₃ passivation layer experiencing ArOAr annealing ambient has demonstrated a soft breakdown at 1.69 MV/cm, which might be a sign of the breakdown of the thicker SiO₂ interfacial layer (10.893 nm). The observation of a hard breakdown at 2.10 MV/cm by the sample annealed in ArOAr ambient has suggested that the formation of a higher concentration of oxygen vacancies has triggered degradation in the passivating properties of this Tm₂O₃ passivation layer.

4. Conclusion

The effects of NON, FGOFG, and ArOAr annealing ambient on the passivating characteristics of the Tm₂O₃ passivation layer have been comprehensively evaluated. It was concluded that annealing in NON ambient was the most effective ambient in impeding the formation of SiO₂ interfacial layer as the XRR measurement has revealed the formation of thinnest SiO₂ interfacial layer of 3.258 nm. In addition, the EDX measurement revealed that the highest at% of nitrogen (5.59%) was acquired during annealing in an ambient of NON. This finding further confirmed that a larger number of nitrogen ions have been accumulated at the interface between the Tm₂O₃ passivation layer and the Si substrate, effectively reducing the formation of the SiO₂ interfacial layer. However, the incorporated nitrogen ions would narrow the band gap by inducing more interband state when compare with FGOFG and ArOAr annealing, where the highest direct band gap (E^D = 5.434 eV) and indirect band gap (E^ID = 4.722 eV) were attained from the Tm₂O₃ passivation layer annealed in ArOAr ambient. Significantly, the FGOFG annealing would prompt the Tm₂O₃ passivation layer gain the highest k value (16.8) owing to its capacity to prevent the formation of both the SiO₂ interfacial layer and oxygen vacancies. With the both beneficial effects of hydrogen and nitrogen, the FGOFG annealing demonstrated the lowest concentration of effective oxide charge (Q_eff) of 1.32 × 10¹² cm⁻² and the lowest interface trap density (D_it), which have been proved by Terman, Hill–Coleman, and high–low frequency methods. Additionally, the excellent interface quality together with an acceptable thickness of SiO₂ interfacial layer contributed to a lower leakage current density and a higher breakdown electric felid (4.03 MV/cm) for the Tm₂O₃ passivation layer annealed in FGOFG ambient.

Conflicts of Interest

The authors declare no conflicts of interest.

Funding

The authors would like to acknowledge the financial support from the Ministry of Higher Education Malaysia for Fundamental Research Grant Scheme (FRGS) with Project Code FRGS/1/2023/STG05/USM/02/8.

Acknowledgments

The authors would like to acknowledge the financial support from the Ministry of Higher Education Malaysia for Fundamental Research Grant Scheme (FRGS) with Project Code FRGS/1/2023/STG05/USM/02/8.

Open Research

Data Availability Statement

The raw/processed data required to reproduce these findings cannot be shared at this time as the data also form part of an ongoing study.

References

1 Robertson J. and Wallace R. M., High-K Materials and Metal Gates for CMOS Applications, Materials Science and Engineering R: Reports. (2015) 88, 1–41.
10.1016/j.mser.2014.11.001
Web of Science® Google Scholar
2 Busch B. W., Pluchery O., and Chabal Y. J., et al.Materials Characterization of Alternative Gate Dielectrics, Mrs Bulletin. (2002) 27, 206–211.
10.1557/mrs2002.72
CAS Web of Science® Google Scholar
3 Robertson J., High Dielectric Constant Oxides, The European Physical Journal Applied Physics. (2006) 184, 177–184.
Google Scholar
4 Hoshino Y., Kido Y., Yamamoto K., Hayashi S., and Niwa M., Characterization and Control of the HfO₂/Si(001) Interfaces, Applied Physics Letters. (2002) 81, 2650–2652.
10.1063/1.1510941
CAS Web of Science® Google Scholar
5 Vitanov P., Harizanova A., Ivanova T., and Dimitrova T., Chemical Deposition of Al₂O₃ Thin Films on Si Substrates, Thin Solid Films. (2009) 517, 6327–6330.
10.1016/j.tsf.2009.02.085
CAS Web of Science® Google Scholar
6 Bera M. K., Chakraborty S., and Saha S., et al.High Frequency Characterization and Continuum Modeling of Ultrathin High-k (ZrO₂) Gate Dielectrics on Strained-Si, Thin Solid Films. (2006) 504, 183–187.
10.1016/j.tsf.2005.09.083
CAS Web of Science® Google Scholar
7 Passlack M., Hunt N. E. J., and Schubert E. F., et al.Dielectric Properties of Electron-Beam Deposited Ga₂O₃ Films, Applied Physics Letters. (1994) 64, 2715–2717.
10.1063/1.111452
CAS Web of Science® Google Scholar
8 Dimoulas A., Vellianitis G., Travlos A., Ioannou-Sougleridis V., and Nassiopoulou A. G., Structural and Electrical Quality of the High-k Dielectric Y₂O₃ on Si (001): Dependence on Growth Parameters, Journal of Applied Physics. (2002) 92, 426–431.
10.1063/1.1483379
CAS Web of Science® Google Scholar
9 Nishikawa Y., Fukushima N., Yasuda N., Nakayama K., and Ikegawa S., Electrical Properties of Single Crystalline CeO₂ High-k Gate Dielectrics Directly Grown on Si (111), Japanese Journal of Applied Physics. (2002) 41, 2480–2483.
10.1143/JJAP.41.2480
CAS Web of Science® Google Scholar
10 Kahraman A. and Yilmaz E., Proposal of Alternative Sensitive Region for MOS Based Radiation Sensors: Yb₂O₃, Journal of Vacuum Science & Technology A. (2017) 35, 061511.
10.1116/1.4993545
Google Scholar
11 Cheng J. B., Li A. D., and Shao Q. Y., et al.Growth and Characteristics of La₂O₃ Gate Dielectric Prepared by Low Pressure Metalorganic Chemical Vapor Deposition, Applied Surface Science. (2004) 233, 91–98.
10.1016/j.apsusc.2004.03.232
CAS Web of Science® Google Scholar
12 Atanassova E., Dimitrova T., and Koprinarova J., AES and XPS Study of Thin RF-Sputtered Ta₂O₅ Layers, Applied Surface Science. (1995) 84, 193–202.
10.1016/0169-4332(94)00538-9
CAS Web of Science® Google Scholar
13 Iwai H., Ohmi S., and Akama S., et al. Advanced Gate Dielectric Materials for Sub-100 Nm CMOS, International Electron Devices Meeting, 2002, 625–628.
10.1109/IEDM.2002.1175917
Google Scholar
14 Päiväsaari J., Putkonen M., and Niinistö L., A Comparative Study on Lanthanide Oxide Thin Films Grown by Atomic Layer Deposition, Thin Solid Films. (2005) 472, 275–281.
10.1016/j.tsf.2004.06.160
CAS Web of Science® Google Scholar
15 Kouda M., Kawanago T., and Ahmet P., et al.Interface and Electrical Properties of Tm₂O₃ Gate Dielectrics for Gate Oxide Scaling in MOS Devices, Journal of Vacuum Science & Technology B. (2011) 29, 062202.
10.1116/1.3660800
Google Scholar
16 Mitrovic I. Z., Althobaiti M., and Weerakkody A. D., et al.Interface Engineering of Ge Using Thulium Oxide: Band Line-up Study, Microelectronic Engineering. (2013) 109, 204–207.
10.1016/j.mee.2013.03.160
CAS Google Scholar
17 Litta E. D., Hellström P.-E., Henkel C., Valerio S., Hallén A., and Östling M., High-Deposition-Rate Atomic Layer Deposition of Thulium Oxide From TmCp₃ and H₂O, Journal of the Electrochemical Society. (2013) 160, D538–D542.
10.1149/2.056311jes
CAS Google Scholar
18 Shannon R. D., Dielectric Polarizabilities of Ions in Oxides and Fluorides, Journal of Applied Physics. (1993) 73, 348–366.
10.1063/1.353856
CAS PubMed Web of Science® Google Scholar
19 Wang J. J., Fang Z. B., Ji T., Ren W. Y., Zhu Y. Y., and He G., Band Offsets of Epitaxial Tm₂O₃ High-k Dielectric Films on Si Substrates by X-Ray Photoelectron Spectroscopy, Applied Surface Science. (2012) 258, 6107–6110.
10.1016/j.apsusc.2012.03.013
CAS Google Scholar
20 Wang J., Ji T., Zhu Y., Fang Z., and Ren W., Band Gap and Structure Characterization of Tm₂O₃ Films, Journal of Rare Earths. (2012) 30, 233–235.
10.1016/S1002-0721(12)60029-5
CAS Google Scholar
21 Mitrovic I. Z., Hall S., and Althobaiti M., et al.Atomic-Layer Deposited Thulium Oxide as a Passivation Layer on Germanium, Journal of Applied Physics. (2015) 117, 18–23.
10.1063/1.4922121
Google Scholar
22 Yao B., Zhu Y., and Fang Z., et al.Energy Band Diagram of Metal/Tm₂O₃/Si Stack Structure Acquired From the Study of Leakage Current Mechanisms, Thin Solid Films. (2016) 611, 52–55.
10.1016/j.tsf.2016.05.015
CAS Google Scholar
23 Zdanowicz T., Electrical Properties of Thulium Oxide Thin Films, Thin Solid Films. (1988) 164, 175–182.
10.1016/0040-6090(88)90130-7
Google Scholar
24 Pan T. M. and Yen L. C., Influence of Postdeposition Annealing on Structural Properties and Electrical Characteristics of Thin Tm₂O₃ and Tm₂Ti₂O₇ Dielectrics, Applied Surface Science. (2010) 256, 2786–2791.
10.1016/j.apsusc.2009.11.029
CAS Google Scholar
25 Pan T. M., Da Lee C., and Wu M. H., High-k Tm₂O₃ Sensing Membrane-Based Electrolyte-Insulator-Semiconductor for pH Detection, Journal of Physical Chemistry C. (2009) 113, 21937–21940.
10.1021/jp908129k
CAS Web of Science® Google Scholar
26 Guha S. and Narayanan V., Oxygen Vacancies in High Dielectric Constant Oxide-Semiconductor Films, Physical Review Letters. (2007) 98, no. 19, 8, https://doi.org/10.1103/PhysRevLett.98.196101, 2-s2.0-34547277537, 196101.
10.1103/PhysRevLett.98.196101
CAS PubMed Web of Science® Google Scholar
27 Shekkeer K. M. A., Cheong K. Y., and Quah H. J., Effects of Varying Annealing Ambient Towards Performance of Ternary Ga_xCe_yO_z Passivation Layers for Metal-Oxide-Semiconductor Capacitor, International Journal of Energy Research. (2024) 2024, 1–15.
10.1155/2024/2936714
Google Scholar
28 Litta E. D., Hellström P.-E., Henkel C., and Östling M., Thulium Silicate Interfacial Layer for Scalable High-k/Metal Gate Stacks, IEEE Trans. Electron Devices. (2013) 60, 3271–3276.
10.1109/TED.2013.2275744
CAS Google Scholar
29 Ji T., Cui J., and Fang Z. B., et al.Single Crystalline Tm₂O₃ Films Grown on Si (0 0 1) by Atomic Oxygen Assisted Molecular Beam Epitaxy, Journal of Crystal Growth. (2011) 321, 171–175.
10.1016/j.jcrysgro.2011.02.038
CAS Google Scholar
30 Žurauskaitė L., Jones L., Dhanak V. R., Mitrovic I. Z., Hellström P.-E., and Östling M., Investigation of Tm₂O₃ As a Gate Dielectric for Ge MOS Devices, ECS Transactions. (2018) 86, 67–73.
10.1149/08607.0067ecst
CAS Google Scholar
31 Meeuwissen A., Bosco G. B. F., and van der Kolk E., Optical and Structural Characterization of Tm₂O₃, TmN, and TmO_xN_y Thin Films Grown by Direct-Current Reactive Magnetron Sputtering, Thin Solid Films. (2021) 717, 138450.
10.1016/j.tsf.2020.138450
CAS Google Scholar
32 Wang H., Wang Y., and Zhang J., et al.Interface Control and Leakage Current Conduction Mechanism in HfO₂ Film Prepared by Pulsed Laser Deposition, Applied Physics Letters. (2008) 93, 19–22.
10.1063/1.3033526
Google Scholar
33 Wang Y., Wang H., Ye C., Zhang J., Wang H., and Jiang Y., Interfacial Reaction and Electrical Properties of HfO₂ Film Gate Dielectric Prepared by Pulsed Laser Deposition in Nitrogen: Role of Rapid Thermal Annealing and Gate Electrode, ACS Applied Materials & Interfaces. (2011) 3, no. 10, 3813–3818, https://doi.org/10.1021/am2008695, 2-s2.0-84866140260.
10.1021/am2008695
CAS PubMed Web of Science® Google Scholar
34 Chen G. L., Sivertsen J. M., and Judy J. H., The Effect of Oxygen Partial Pressure on the Properties of RF-Sputtered Al₂O₃ Thin Films, Materials Letters. (1984) 2, 196–201.
10.1016/0167-577X(84)90023-5
CAS Google Scholar
35 Deng J. and Quah H. J., Structural, Morphological, and Metal-Oxide-Semiconductor Characteristics of Thulium Oxide Passivation Layer Grown in Nitrogen-Oxygen-Nitrogen Ambient, Sustainable Materials Technology. 35, e00534.
Google Scholar
36 Deng J. and Quah H. J., Innovative Approach of Growing Thulium Oxide Passivation Layer in Nitrogen/Oxygen/Nitrogen Ambient, Materials Letters. (2022) 323, 132572.
10.1016/j.matlet.2022.132572
CAS Google Scholar
37 Shannon R. D., Revised Effective Ionic Radii and Systematic Studies of Interatomic Distances in Halides and Chalcogenides, Acta Crystallographica. (1976) 32, 751–767.
10.1107/S0567739476001551
CAS Web of Science® Google Scholar
38 Shekkeer K. M. A., Cheong K. Y., and Quah H. J., Effects of Post-Deposition Annealing of Cerium Oxide Passivation Layer in Nitrogen-Oxygen-Nitrogen Ambient, International Journal of Energy Research. (2022) 46, 14814–14826.
10.1002/er.8184
Google Scholar
39 Nachtigall P., Jordan K. D., and Janda K. C., Calculation of the Si-H Bond Energies for the Monohydride Phase of Si(100), Journal of Chemical Physics. (1991) 95, 8652–8654.
10.1063/1.461245
CAS Web of Science® Google Scholar
40 Chen L., Xu L., Li D., and Lin B., Mechanism of Selective Si₃N₄ Etching over SiO₂ in Hydrogen-Containing Fluorocarbon Plasma, Microelectronic Engineering. (2009) 86, 2354–2357.
10.1016/j.mee.2009.04.016
CAS Web of Science® Google Scholar
41 Zak A. K., Majid W. H. A., Abrishami M. E., and Yousefi R., X-Ray Analysis of ZnO Nanoparticles by Williamson-Hall and Size-Strain Plot Methods, Solid State Sciences. (2011) 13, 251–256.
10.1016/j.solidstatesciences.2010.11.024
CAS Google Scholar
42 Zabidi A. R. M. and Lim W. F., Formation of Cerium Oxide Film via Post-Sputter Oxidation of Cerium in Nitrogen/Oxygen/Nitrogen Ambient, Journal of Alloys and Compounds. (2021) 851, 156786.
10.1016/j.jallcom.2020.156786
Google Scholar
43 López R. and Gómez R., Band-Gap Energy Estimation From Diffuse Reflectance Measurements on Sol-Gel and Commercial TiO₂: A Comparative Study, Journal of Sol-Gel Science and Technology. (2012) 61, 1–7.
10.1007/s10971-011-2582-9
CAS Web of Science® Google Scholar
44 Prokofiev A. V., Shelykh A. I., and Melekh B. T., Periodicity in the Band Gap Variation of Ln₂X₃ (X = O, S, Se) in the Lanthanide Series, Journal of Alloys and Compounds. (1996) 242, 41–44.
10.1016/0925-8388(96)02293-1
CAS Web of Science® Google Scholar
45 He G., Fang Q., Li G. H., Zhang J. P., and Zhang L. D., Structural and Optical Properties of Nitrogen-Incorporated HfO₂ Gate Dielectrics Deposited by Reactive Sputtering, Applied Surface Science. (2007) 253, 8483–8488.
10.1016/j.apsusc.2007.04.026
CAS Web of Science® Google Scholar
46 Kaviani M., Afanas’Ev V. V., and Shluger A. L., Interactions of Hydrogen With Amorphous Hafnium Oxide, Physical Review B. (2017) 95, 1–7.
10.1103/PhysRevB.95.075117
Google Scholar
47 Nicollian E. H. and Brews J. R., MOS (Metal Oxide Semiconductor) Physics and Technology, 1982, John Wiley & Sons.
Google Scholar
48 Quah H. J. and Cheong K. Y., Retardation Mechanism of Ultrathin Al₂O₃ Interlayer on Y₂O₃ Passivated Gallium Nitride Surface, ACS Applied Materials & Interfaces. (2014) 6, no. 10, 7797–7805, https://doi.org/10.1021/am501075s, 2-s2.0-84901684277.
10.1021/am501075s
CAS PubMed Web of Science® Google Scholar
49 Huang F., Fan S., and Tian Y., et al.Influence of N-Doping on Dielectric Properties of Carbon-Coated Copper Nanocomposites in the Microwave and Terahertz Ranges, Journal of Materials. (2022) 8, 1131–1140.
Google Scholar
50 Liu Y., Ding M., and Su J., et al.Dielectric Properties of Nitrogen-Doped Polycrystalline Diamond Films in Ka Band, Diamond and Related Materials. (2017) 76, 68–73.
10.1016/j.diamond.2017.04.009
CAS Web of Science® Google Scholar
51 Arjmand M. and Sundararaj U., Effects of Nitrogen Doping on X-Band Dielectric Properties of Carbon Nanotube/Polymer Nanocomposites, ACS Applied Materials & Interfaces. (2015) 7, no. 32, 17844–17850, https://doi.org/10.1021/acsami.5b04211, 2-s2.0-84939824984.
10.1021/acsami.5b04211
CAS PubMed Web of Science® Google Scholar
52 Quah H. J., Hassan Z., and Lim W. F., Passivation of Silicon Substrate Using Two-Step Grown Ternary Aluminium Doped Zirconium Oxide, Applied Surface Science. (2019) 493, 411–422.
10.1016/j.apsusc.2019.07.023
CAS Web of Science® Google Scholar
53 Quah H. J., Hassan Z., and Lim W. F., A Two-Step Growth Route of Ternary Aluminium Doped Zirconium Oxide Film on Silicon, Journal of Alloys and Compounds. (2019) 777, 736–748.
10.1016/j.jallcom.2018.10.359
CAS Web of Science® Google Scholar
54 Terman L. M., An Investigation of Surface States at a Silicon/Silicon Oxide Interface Employing Metal-Oxide-Silicon Diodes, Solid-State Electronics. (1962) 5, 285–299.
10.1016/0038-1101(62)90111-9
CAS Web of Science® Google Scholar
55 Long R. D., Jackson C. M., and Yang J., et al.Interface Trap Evaluation of Pd/Al₂O₃/GaN Metal Oxide Semiconductor Capacitors and the Influence of Near-Interface Hydrogen, Applied Physics Letters. (2013) 103, 201607.
10.1063/1.4827102
CAS Web of Science® Google Scholar
56 Fedorenko Y. G., Truong L., Afanas’ev V. V., and Stesmans A., Energy Distribution of the (100) Si/HfO₂ Interface States, Applied Physics Letters. (2004) 84, 4771–4773.
10.1063/1.1758302
CAS Web of Science® Google Scholar
57 Evangelou E. K., Wiemer C., Fanciulli M., Sethu M., and Cranton W., Electrical and Structural Characteristics of Yttrium Oxide Films Deposited by Rf-Magnetron Sputtering on n-Si, Journal of Applied Physics. (2003) 94, 318–325.
10.1063/1.1580644
CAS Web of Science® Google Scholar
58 Chim W. K., Ng T. H., and Koh B. H., et al.Interfacial and Bulk Properties of Zirconium Dioxide as a Gate Dielectric in Metal-Insulator-Semiconductor Structures and Current Transport Mechanisms, Journal of Applied Physics. (2003) 93, 4788–4793.
10.1063/1.1561995
CAS Web of Science® Google Scholar
59 Hill W. A. and Coleman C. C., A Single-Frequency Approximation for Interface-State Density Determination, Solid-State Electronics. (1980) 23, 987–993.
10.1016/0038-1101(80)90064-7
CAS Web of Science® Google Scholar
60 Castagné R. and Vapaille A., Description of the SiO₂-Si Interface Properties by Means of Very Low Frequency MOS Capacitance Measurements, Surface Science. (1971) 28, 157–193.
10.1016/0039-6028(71)90092-6
CAS Web of Science® Google Scholar
61 Ouisse T., Bécourt N., Jaussaud C., and Templier F., Low-Frequency, High-Temperature Conductance and Capacitance Measurements on Metal-Oxide-Silicon Carbide Capacitors, Journal of Applied Physics. (1994) 75, 604–607.
10.1063/1.355846
CAS Web of Science® Google Scholar
62 Berglund C. N., Surface States at Steam-Grown Silicon-Silicon Dioxide Interfaces, IEEE Transactions on Electron Devices. (1966) ED-13, no. 10, 701–705, https://doi.org/10.1109/T-ED.1966.15827, 2-s2.0-84896741951.
10.1109/T-ED.1966.15827
Web of Science® Google Scholar

Citing Literature

All articles

Synergistic Effects of Nitrogen–Oxygen–Nitrogen, Forming Gas–Oxygen–Forming Gas, and Argon–Oxygen–Argon Annealing Ambient on the Structural and Electrical Characteristics of Thulium Oxide Passivation Layers on Silicon Substrate

Abstract

1. Introduction

2. Experimental Procedure

3. Results and Discussion

4. Conclusion

Conflicts of Interest

Funding

Acknowledgments

Open Research

Data Availability Statement

References

Citing Literature

Figures

References

Information

About Wiley Online Library

Help & Support

Opportunities

Connect with Wiley

Synergistic Effects of Nitrogen–Oxygen–Nitrogen, Forming Gas–Oxygen–Forming Gas, and Argon–Oxygen–Argon Annealing Ambient on the Structural and Electrical Characteristics of Thulium Oxide Passivation Layers on Silicon Substrate

Abstract

1. Introduction

2. Experimental Procedure

3. Results and Discussion

4. Conclusion

Conflicts of Interest

Funding

Acknowledgments

Open Research

Data Availability Statement

References

Citing Literature

Figures

References

Related

Information