1 Introduction

SiO2/Si material structure has dominated the semiconductor industries for complementary metal oxide (CMOS) process for decades. The reason being SiO2 has excellent insulating properties and low interface trap level density around the order of approximately 1010 cm−2 eV−1 with Si surface [1,2,3,4]. To improve device performance, the geometry of device has been shrank continuously for past few decades. Besides dimensions shrinkage, alternative material structures were also studied intensively [5, 6]. The paragraph below will give an in-depth explanation.

Looking at current Si substrate, to improve device performance, besides shrinking the channel length, another option is changing the material which has higher mobility than silicon. Comparing both essential devices which comprise p-MOSFET and n-MOSFET, the mobility of majority carrier of p-MOSFET is lower than its majority carrier of n-MOSFET. Hence, it is more essential to improve from p-MOSFET aspect [7]. Ge was selected in this study because it has the highest hole mobility in all of the known semiconductor materials [8]. The hole mobility of Ge is 1900 cm2 eV−1 s−1, more than 4 times compared with Si [9].

As newer-generation chip dimensions are shrinking, gate oxide thickness is reducing. However, SiO2 encounters physical limitation with tunneling effect when thickness is down to 1.2 nm [10]. Therefore, alternative high κ materials were widely study as SiO2 replacement, such as HfO2, ZrO2 [11], Sm2O3 [12], Nd2O3 [13], TiO2 [14], Ta2O5 [15], Al2O3 [16], Y2O3 [17], and other. In this study, ZrO2 have been selected for its high κ value, large-energy bandgap, and its stability in the form of cubic or tetragonal polymorphs [18]. Another benefit is the interfacial layer (IL) free between ZrO2 gate dielectric and Ge wafer during thermal treatment [9] which will prevent electrical breakdown prematurely [19]. The phenomenon of IL free may due to the Ge diffusion into ZrO2 dielectric [9]. The details of the behavior still remains an open question. A possible mechanism of Ge atomic diffusion and its rearrangement in ZrO2 has been proposed and explicated in this paper.

Compared with the counterpart Si, Ge is considerably unstable at high-temperature process. As studied previously [20], it is observed that Zr/Ge was unstable at oxidation temperature above 600 °C at O2 ambient. On the other hand, NO and N2O are common gases used for oxidation and nitridation process in semiconductor industries [21]. N2O is the preferred gas because it is less toxic as compared with NO [22]. Another reason for NO was not selected in this study because of its activation energies are lower than O2. This study will focus on changing to N2O ambient conditions where activation energies are higher than O2 [23] and study the stability under different temperature.

2 Experimental procedures

To prepare impurity-free samples, germanium wafer [n-type, (100)-oriented, 0.01–0.05 Ω cm, 1 cm × 1 cm] were dipped cleaned by hydrofluoric acid (HF) solution (1:50 HF:H2O) and DI water. By using TF 450 physical vapor deposition (PVD) radio frequency (RF) sputtering system, metallic zirconium (Zr) film was sputtered from a Zr target (Kurt J. Lesker, USA, 99.999% purity) at room temperature. Pressure, RF power, and argon gas flow rate were set at 3 × 10−5 mbar, 170 W, and 25 cm3 min−1, respectively. Experimental oxidation/nitridation temperatures were set at 300, 400, 500, 600, 700, and 800 °C using Carbolite CTF tube furnace. Samples of sputtered Zr on Ge wafer were placed into the tube furnace with N2O flow rate of 150 mL min−1 at each experimental temperature for 15 min. The oxidized/nitrided samples were cooled down in the tube furnace before removal.

The properties of the samples were characterized by various measurement methods which are as follows: film crystallinity was characterized using a PANalytical Empyrean X-ray diffractometer (XRD) system in the 2\(\theta\) scan range from 10° to 80°. Copper radiation (Cu Kα) with wavelength (λ) of 0.15406 nm was used as X-ray source. The chemical compositions of the films were analyzed by X-ray photoelectron spectrometer (XPS) on AXIS ULTRA DLD (Al-Ka radiation). The X-ray tube was operated at 80 W. The analyzed area had a diameter of 300 µm. During XPS measurements, the pressure in UHV chamber was about 3 × 10− 9 Torr. Survey scan was measured at 160 eV (energy resolution 1 eV per step) pass energy, this was for elemental analysis. Narrow scan (with high resolution) was measured at 20 eV (energy resolution 0.1 eV per step) pass energy, and it was for chemical states analysis. All data were recorded at a 45° take-off angle. To study geometry structure, film samples were sent for deposition of platinum (Pt) on the surface for protection from ion bombardment damage caused by the focused ion beam during lamella preparation. Followed by cross-sectioning using TECNAI G2 F20 high-resolution transmission electron microscope (HRTEM), the interplanar spacing (d) of the polycrystalline structure was measured from HRTEM images using ImageJ software. Olympus BX61 optical microscope were used to examine samples surface and the picture capture by Color view (soft imaging system).

To study the electrical properties of the films, MOS capacitor test structures with area of 300 µm × 300 µm were fabricated. Aluminum (Al 99.9995% purity, 100 nm; Kurt J. Lesker, USA) was sputtered through a mask using a TF 450 PVD RF sputtering system. The back side of the Ge was also sputtered with 100-nm-thick Al film to form an Ohmic back contact. BPW-800 8″ probe station with a Keithley 4200 semiconductor characterization system (SCS) were used to measure current–voltage (I–V).

3 Results and discussion

3.1 J–E and J–V characteristics

The J–E and J–V characteristics were used to study the effects of oxidation/nitridation temperature on the electrical properties of the samples. Figure 1 shows the J–E and J–V characteristics of samples oxidized/nitridation at various temperatures. For samples of 300, 400, 500, and 600 °C, the thickness of the dielectric will be discussed in Sect. 3.6, the J–E curve using left axis and bottom axis at which the arrows pointing indicates. In the results, all the samples shows typical J–E curve. The electrical breakdown fields were 6.0, 6.9, 5.7, and 5.7 MV cm−1, respectively. Electrical characterization revealed that 400 °C samples yielded the highest electrical breakdown fields. For samples of 700 and 800 °C, due to lack of values of dielectric thickness, the J–V curve were shown in Fig. 1. The J–V curve using right axis and top axis at which the arrows pointing indicates. The 700 °C samples shows an almost short circuit in the J–V curve, and the 800 °C samples shows an almost open circuit in J–V curve measurement results. For further study, a physical analysis will be performed in next section.

Fig. 1
figure 1

J–E and J–V characteristics of samples oxidized/nitridation at various temperatures

Full size image

3.2 Optical microscope analysis

Optical microscope was used to examine samples surface condition at various temperatures as shown in Fig. 2. Figure 2a shows the 300–600 °C samples, the surfaces were found to be smooth and free from defects. Figure 2b shows the 700 °C samples with scattered blisters. Figure 2c shows the 800 °C samples. For indication of surface roughness, another lower magnification for sample 800 °C is shown in Fig. 2d.

Fig. 2
figure 2

Optical microscope of samples surface. a Samples oxidized/nitridation at 300–600 °C, b samples oxidized/nitridation at 700 °C, c samples oxidized/nitridation at 800 °C and d samples oxidized/nitridation at 800 °C with lower magnification

Full size image

Comparing 600 °C samples between N2O ambient and O2 ambient, the N2O ambient samples were smooth and defect-free. This might be because the activation energies of N2O is higher than O2. For the 700 °C samples, the surface was similar for both N2O and O2 ambient. This could be results of similar reaction where N2O or O2 atoms diffused through the ZrO2 layer, forming GeO2 under N2O or O2 ambient. For 800 °C, Rough surface with uneven thick films on the sample were observed. There have been lack of literatures on the liquid state of Ge3N4. It has been reported that the Ge3N4 will decompose around 700–900 °C [24, 25]. In this study, based on the on the J–V result and the OM analysis, it is assumed that some form of liquidation occurred during oxidized/nitridation at 800 °C and trapping GeO2 in a form of solid powder on sample surface. These mixtures of liquid and powder further solidified on sample surface after cooling to room temperature. It was not observed at O2 ambient. Hence, Ge3N4 was the possible material based on this assumption. Thus, explaining the difference between J–V curve results where short circuits were observed for samples under O2 ambient, and open circuits were observed for samples under N2O ambient, all under 800 °C.

3.3 Raman spectroscopy analysis

Figure 3 shows the Raman analysis of oxidized/nitridation samples at various temperatures. The background noise for 700 and 800 °C samples were observed to be higher than the other samples during measurement. As a result, an extra Y-axis was added in Fig. 3 right hand sites. The peak that belonged to Ge was detected at 300 cm−1 for all samples. The peak that belonged to monoclinic ZrO2 was detected at 181 and 346 cm−1 [26,27,28]. The peaks that belonged to tetragonal ZrO2 were detected at 404 and 460 cm−1 [29]. The peak that belonged to Orthorhombic ZrO2 was detected at 415 cm−1 [30]. The peaks that belonged to cubic ZrO2 were detected at 454 cm−1 [29]. The peak that belonged to GeO2 were detected at 167 cm−1 for 700 and 800 °C samples. The numbers of GeO2 peak increased as the oxidation/nitridation temperature increased. Extra GeO2 peaks found in 800 °C samples were 263, 444, and 515 cm−1 [31, 32]. Ge3N4 peaks were detected at 202 and 455 cm−1 for 800 °C samples [33, 34]. These results were in conformity with the J–V characteristics where GeO2 was detected for the samples oxidized/nitridation above 700 °C and Ge3N4 was detected in 800 °C samples.

Fig. 3
figure 3

Raman spectra of samples oxidized/nitridation at various temperatures

Full size image

3.4 XRD analysis

Figure 4 shows XRD patterns of samples oxidized/nitridation at various temperatures, from 49° to 56°. Several structures of ZrO2 were detected. The orthorhombic ZrO2 was found at around 51.21°, 52.71°, and 52.87°, corresponding to (151), (020), and (142) planes, respectively. These peaks are matched by ICSD with reference code 98-017-3960 and 98-005-6696. The tetragonal ZrO2 was found at around 51.91° and 52.04° corresponding to (112) and (020) planes, respectively. These peaks are matched by ICSD with reference code 98-009-2092. The cubic ZrO2 was found at around 51.32° corresponding to (022) planes. This peak is matched by ICSD with reference code 98-018-5123. The monoclinic ZrO2 was found at around 51.03° corresponding to \({\text{(12}}\overline {\text{2}} {\text{)}}\) planes. This peak is matched by ICSD with reference code 98-065-9226. On the other hand, the hexagonal O0.27Zr was found at around 51.41° corresponding to (026) planes. This peak is matched by ICSD with reference code 98-004-2986. The tetragonal GeO8Zr3 was found at around 51.61° corresponding to (220) planes. This peak is matched by ICSD with reference code 98-002-9263. The tetragonal Ge4Zr5 was found at around 51.78°, corresponding to (117) planes. This peak is matched by ICSD with reference code 98-063-8154.

Fig. 4
figure 4

XRD patterns of samples oxidized/nitridation at various temperatures form 49° to 56°

Full size image

Figure 5 shows XRD patterns of samples oxidized/nitridation at various temperatures, from 10° to 80°. The peaks of hexagonal GeO2, h-GeO2 that were found in the 700 and 800 °C samples were around 20.55°, 25.94°, 35.99°, 38.02°, 39.47°, 41.80°, 44.90°, 48.27°, 48.66°, 53.35°, 56.31°, 58.84°, 61.65°, 66.04°, 67.03°, 73.81°, 76.32°, 77.27°, and 78.50° corresponding to (010), (011), (110), (102), (111), (020), (021), (003), (112), (202), (210), (211), (113), (122), (031), (302), (220), (213), and (221) planes, respectively. These peaks are matched by ICSD with reference code 98-001-6577. The peaks of orthorhombic Ge3N4, o-Ge3N4 that were found in the 700 and 800 °C samples were around 20.30°, 23.84°, 25.34°, 35.27°, 41.27°, 41.39°, 58.79°, 65.54°, 65.99°, 66.93°, 70.70°, and 73.89° corresponding to (101), (011), (111), (311), (202), (112), (430), (701), (313), (123), (413), and (721) planes, respectively. These peaks are matched by ICSD with reference code 98-015-6339. The GeO2 and Ge3N4 were observed for sample 700 and 800 °C. It could be contributed to the diffusion of N2O through ZrO2 layer, which reacted with Ge and formed crystallization of GeO2 and Ge3N4. These results were in conformity with the J-V characteristics, Raman spectroscopy analysis and OM analysis.

Fig. 5
figure 5

XRD patterns of oxidized/nitridation at various temperatures form 10° to 80°

Full size image

Williamson–Hall (W–H) plot analysis is used for evaluating the crystallite size (D) and microstrain (ɛ) [35,36,37]. The W–H as Eq. (1):

$$({\beta _S}+{\beta _D})\cos \theta =\frac{{K\lambda }}{D}+4\varepsilon ~{\text{sin}}\theta$$
(1)

where λ is the wavelength of the radiation; K is a constant, which is equal to 0.9; \(\theta\) is the peak position; βD is the peak width at half-maximum intensity; and βS is the peak width at half-maximum intensity. Based on Eq. (1), one plots a graph of (βD + βS) cos\({\text{~}}\theta\) versus 4 sin \(\theta\). Microstrain is the gradient, whereas \(K\lambda\)/D is the intercept of the graph, from which D can be calculated as Eq. (2):

$$D=\frac{{K\lambda }}{{y~intercept}}$$
(2)

Crystallite sizes of ZrO2, O0.27Zr, Ge4Zr5, and GeO8Zr3 were not calculated by using W–H plot because the number of peaks was insufficient for study purposes. Figure 6 shows the W–H plot of h-GeO2 and O-Ge3N4 for 700 °C and 800 °C samples. According to Fig. 6, the points were well distributed in the four groups.

Fig. 6
figure 6

W–H plot of GeO2 and Ge3N4 for samples 700 and 800 °C

Full size image

The crystallite size of h-GeO2 (type 1) of 700 and 800 °C samples were 26.15 and 26.21 nm, respectively. The microstrain of h-GeO2 (type 1) of the 700 and 800 °C samples were − 0.0036 and − 0.0034, respectively. The crystallite size of h-GeO2 (type 2) of the 700 and 800 °C samples were 19.30 and 20.02 nm, respectively. The microstrain of h-GeO2 (type 2) of the 700 and 800 °C samples were − 0.0089 and − 0.0076, respectively. These results were similar with the O2 ambient.

The crystallite size of O-Ge3N4 for sample of 700 °C was not calculated by using W–H plot because the number of peaks was insufficient for study purposes. The crystallite size of O-Ge3N4 (type 1) of the 800 °C samples were 17.55 nm. The microstrain of O-Ge3N4 (type 1) of the 800 °C samples were − 0.0048. The crystallite size of O-Ge3N4 (type 2) of the 800 °C samples were 13.29 nm. The microstrain of O-Ge3N4 (type 2) of the 800 °C samples was − 0.012.

Two type of GeO2 and Ge3N4 crystallite size and microstrain were found indicates that two generating sources for GeO2 and Ge3N4. This inference will be verified and discussed in a later section.

3.5 Thermal oxidation/nitridation mechanism

Figure 7a shows the schematic diagram of sputtered Zr film on Ge wafer before thermal oxidation. For the 300, 400, 500, and 600 °C samples, the N2O directly reacted with the Zr film which was deposit by sputtering according to Eqs. (3) and (4) [38] and formed the ZrO2 dielectric as shown in Fig. 7b.

Fig. 7
figure 7

Schematic diagram of thermal oxidation/nitridation mechanism. a Sputtering Zr on Ge wafer, b Zr oxidation to be ZrO2 (for sample 300, 400, 500 and 600 °C), and c N2O pass ZrO2, GeO2 and Ge3N4 generated (for sample 700 and 800 °C)

Full size image
$${\text{Z}}{{\text{r}}^+}+{{\text{N}}_2}{\text{O}} \to {\text{Zr}}{{\text{O}}^+}+{{\text{N}}_2}$$
(3)
$${\text{Zr}}{{\text{O}}^+}+{{\text{N}}_2}{\text{O}} \to {\text{ZrO}}_{2}^{+}+{{\text{N}}_2}$$
(4)

For the 700 and 800 °C samples, the samples thermally oxidized/nitridation according to the 3-step reactions, as shown in Fig. 7c. In high-temperature ambient, the N2O received enough energy to diffuse through ZrO2 dielectric layer and reacted with the Ge wafer. GeO2 film formed on the surface of Ge wafer was assumed as equation Eqs. (5) and (6). After the GeO2 formed, GeO2 will react with Ge wafer, and GeO gas was generated according to Eq. (7) [9, 39]. The GeO gas diffused outward and reacted with N2O in atmosphere. After that, another GeO2 film was formed on the ZrO2 assumed to be according to Eq. (8).

$${\text{G}}{{\text{e}}^+}+{{\text{N}}_2}{\text{O}} \to {\text{Ge}}{{\text{O}}^+}+{{\text{N}}_2}$$
(5)
$${\text{G}}{{\text{e}}^+}+{{\text{N}}_2}{\text{O}} \to {\text{GeO}}_{2}^{+}+{{\text{N}}_2}$$
(6)
$$3{\text{Ge}}{{\text{O}}_2}+3{\text{Ge}} \to 6{\text{GeO}}\left( {\text{g}} \right)$$
(7)
$${\text{GeO}}\left( {\text{g}} \right)+{{\text{N}}_2}{\text{O}} \to {\text{Ge}}{{\text{O}}_2}+{{\text{N}}_2}$$
(8)

Because the Ge3N4 was found at Raman and XRD measurements, the reactive equation were assumes to be Eqs. (9) and (10).

$$4{\text{Ge}}+2{{\text{N}}_2}{\text{O}} \to {\text{Ge}}{{\text{O}}_2}+{\text{G}}{{\text{e}}_3}{{\text{N}}_4}$$
(9)
$$~8{\text{GeO}}+~2{{\text{N}}_2}{\text{O}} \to ~5{\text{Ge}}{{\text{O}}_2}+~{\text{G}}{{\text{e}}_3}{{\text{N}}_4}$$
(10)

From the analysis of the two layers of GeO2 and the two layers of Ge3N4, which are similar with the O2 ambient, the absolute value of microstrain for the top layer will be less than the bottom layer. Therefore, h-GeO2 (type 1) and O-Ge3N4 (type 1) at Fig. 6 were assumed to be the top layers, and the h-GeO2 (type 2) and O-Ge3N4 (type 2) as the bottom layers.

3.6 HRTEM analysis

HRTEM analysis was performed to study the cross-section of the samples. Figure 8 shows the cross-section of the HRTEM images of the 400 and 500 °C samples. The thicknesses of ZrO2 dielectric for the 400 and 500 °C samples were 5.6 and 6.8 nm, respectively. Assuming the thicknesses are in linear relationship with oxidation temperatures [40], by using 400 and 500 °C thicknesses, a regression line was shown at Fig. 9. Therefore, the thicknesses of the oxidized Zr thin films were approximately ranging from 4.4 to 8 nm from the 300 to 600 °C samples. The thicknesses for 700 and 800 °C will not be analyzed here because the film conditions were not appropriate for analysis. The ZrO2 dielectric layer exhibits mainly an amorphous state with partial crystallized structure, and no IL was found between Ge wafer and ZrO2 dielectric layer. The phenomenon of IL free will be discussed at Sect. 3.8.

Fig. 8
figure 8

Cross-sectional HRTEM images of samples oxidized/nitridation at temperatures 400 and 500 °C

Full size image
Fig. 9
figure 9

Thickness of ZrO2 dielectric of samples oxidized/nitridation at various temperatures

Full size image

3.7 XPS analysis

XPS analysis was performed for the evaluations of the structural and chemical properties of the ZrO2 dielectric. Core-level spectra of Zr 3d, Zr 3p, Zr 3s, Ge 3d, Ge 3p, Ge 3s, C 1s, N 1s, and O 1s were detected in the XPS wide scans. Narrow scans on Zr 3d, Ge 3p, and O 1s were also performed as a function of etching time for different oxidation/nitridation temperatures (300–600 °C) as shown in Figs. 10, 11, 12 and 13. Each spectrum was calibrated in respect to the carbon 1s (284.8 eV) state [41].

Fig. 10
figure 10

Evolution of Zr 3d core level XPS spectra as a function of etching time for different oxidation/nitridation temperatures

Full size image
Fig. 11
figure 11

Evolution of Ge 3p core level XPS spectra as a function of etching time for different oxidation/nitridation temperatures

Full size image
Fig. 12
figure 12

Evolution of O 1s core level XPS spectra as a function of etching time for different oxidation/nitridation temperatures

Full size image
Fig. 13
figure 13

Evolution of N 1s core level XPS spectra as a function of etching time for different oxidation/nitridation temperatures

Full size image

Figure 10 shows the narrow scan of Zr 3d for samples oxidized/nitridation at various temperatures from 300 to 600 °C. The positions of the peaks that belonged to Zr 3d5/2 were around 182.37–183.25 eV. The positions of peaks that belonged to Zr 3d3/2 were around 184.68–185.55 eV [42,43,44,45]. The peak positions of Zr 3d at the ZrO2 surface were obviously lower than the inside for all samples. The detail mechanism will be discussed later. The peaks that belonged to Ge 3 s were observed for all samples. The peaks’ height increased as etch time increases. As observed, after being etched around 840 s, the peaks that belonged to Zr 3d disappeared and left only Ge 3s peaks.

Figure 11 shows the narrow scan of Ge 3p. The peaks that belonged to GeOX of Ge 3p1/2 and Ge 3p3/2 were around 128.81–127.29 eV and 124.64–123.08 eV, respectively [46, 47], before etch. After being etched for 420–480 s, the peaks of Ge 3p1/2 and Ge 3p3/2 which belong to Ge were around 126 and 122 eV, respectively [44, 47]. After being etched for 420–480 s, a weak peak found at 118–119.7 eV, indicating that Ge-N bonds exist for all samples.

Figure 12 shows the narrow scan of O 1s. The peaks that belonged to GeO2 and ZrO2 were found before etch for all samples at around 531.88–532.29 eV [48] and 530.67–531.12 eV [44, 49], respectively. After a certain level of etching, the peak of GeO2 disappeared and left only the peak that belongs to ZrO2. The details of the mechanism will be discussed in Sect. 3.8.

Figure 13 shows the narrow scan of N 1s, no peak was found at the surface for all samples. After etch, a weak peak was found at around 404 eV which was assumed to be Ge–N. The same peaks were also detected at a deeper level of Ge wafer.

Figure 14 shows Zr 3d3/2 narrow scan peak position of samples oxidized/nitridation at various temperatures at various etching times. The fully oxidized line was plotted at 185.3 eV [45, 50]. The surface of ZrO2 dielectric was under oxidation for all samples, it may be that Ge atom accumulation at the surface and combined with oxygen atom of existing ZrO2 and induced the ZrO2 under oxidation. For the 300 °C sample, mainly ZrO2 under oxidation and the oxidation level increase when the temperature increases except the for 500 °C samples. This phenomenon was similar with the O2 ambient. For the 600 °C sample, the result shows mainly ZrO2 overoxidation. The sample of 400 °C was the sample nearest to the full oxidation line, thus having the highest electrical breakdown fields compared with the other samples. For 300 and 500 °C samples, the ZrO2 overoxidization is near Ge interface region. It could be that the Ge atoms diffused up during the cooling process, leaving excess O atoms causing ZrO2 overoxidation. The details of the mechanism will be discussed at Sect. 3.8.

Fig. 14
figure 14

XPS Zr 3d3/2 narrow scan peak position of samples oxidized/nitridation at various temperatures and various etching time

Full size image

Figure 15 shows O 1s narrow scan of samples oxidized/nitridation at 400 °C. The back color line shows the original measurement line including noise. The peak position of GeO2 was located at around 532.06 eV which was labeled as a green color line. The peak position of ZrO2 was located at around 531.05 eV which was labeled as a red color line. The blue color shows the peak merge with ZrO2 and GeO2. Hence, when GeO2 increase, the peak will shift to the left and shift to right when the ZrO2 increases. Figure 16a was drawn based on this concept used to monitor the Ge atom arrangement in the ZrO2. The GeO2 and ZrO2 binding energy were assumed to be 532.2 eV [51, 52] and 530.8 eV [53], respectively. As shown in Fig. 16a, the sample of 600 °C has the highest Ge atom in the ZrO2 and sequentially, were 500, 400 and 300 °C. Obviously, the number of Ge atom diffused to ZrO2 increase as the oxidation temperature increases. For the 500 and 600 °C samples, the concentration of Ge atoms increases at the interface of ZrO2 and Ge wafer. Figure 16b shows the 400 °C sample cross-sectional HRTEM images, which correspond with Fig. 16a. A possible of mechanism will be discussed at the next section.

Fig. 15
figure 15

XPS O 1s narrow scan of 400 °C sample

Full size image
Fig. 16
figure 16

a XPS O 1s narrow scan peak position of samples oxidized/nitridation at various temperatures and various etching time and b 400 °C sample cross-sectional HRTEM images

Full size image

3.8 Ge diffused and rearrange mechanism

When Ge atoms diffuse to ZrO2 region, it should be travelling in random thermal motion [54]. However, the XPS result has shown that the Ge atoms were accumulating at both sides of ZrO2, indicating that another factor does exist to influence the way the Ge atom behaves. The factor could be the thermal gradient during cooling appear to make Ge rearrange. As ZrO2 has low thermal conductivity of less than 4 W m−1 K−1 above 500 °C [55], high thermal gradient formed during the cooling process. Figure 17a shows the schematic diagram during thermal process, the Ge atoms diffuse to ZrO2 film with Ge concentration being highest near Ge surface and lowest Ge concentration at ZrO2 surface [56]. Figure 17b shows the schematic diagram of Ge distribution during cooling process. The Ge atom at the center region was at random thermal motion due to higher temperature at the center of ZrO2. The atoms diffused near the edge of the ZrO2 which stayed stagnant due to the high thermal gradient. Hence, the Ge atoms accumulate at both sides of ZrO2 eventually. As explained above, the thermal motion of Ge atoms during cooling will result in an IL-free phenomenon.

Fig. 17
figure 17

Schematic diagram of Ge diffuse a during thermal process and b during cooling process

Full size image

4 Conclusions

The activation energies of N2O are higher than O2, thus explaining the stable film performance at N2O ambient than O2 ambient at temperature above 600 °C. The ZrO2 dielectric film was unstable for temperatures above 700 °C on Ge wafer in N2O ambient. Ge diffusion was observed during oxidation process. A possible mechanism of Ge atomic diffusion and its rearrangement in ZrO2 has been proposed and explicated. The sample of 400 °C was closest to the full oxidation, and electrical characterization has shown that the 400 °C sample has the highest electrical breakdown fields.