H₂ Plasma and PMA Effects on PEALD-Al₂O₃ Films with Different O₂ Plasma Exposure Times for CIS Passivation Layers

Abstract

In this study, the electrical properties of Al₂O₃ film were analyzed and optimized to improve the properties of the passivation layer of CMOS image sensors (CISs). During Al₂O₃ deposition processing, the O₂ plasma exposure time was adjusted, and H₂ plasma treatment as well as post-metallization annealing (PMA) were performed as posttreatments. The flat-band voltage (V_fb) was significantly shifted (ΔV_fb = 2.54 V) in the case of the Al₂O₃ film with a shorter O₂ plasma exposure time; however, with a longer O₂ plasma exposure time, V_fb was slightly shifted (ΔV_fb = 0.61 V) owing to the reduction in the carbon impurity content. Additionally, the as-deposited Al₂O₃ sample with a shorter O₂ plasma exposure time had a larger number of interface traps (interface trap density, D_it = 8.98 × 10¹³ eV⁻¹·cm⁻²). However, D_it was reduced to 1.12 × 10¹² eV⁻¹·cm⁻² by increasing the O₂ plasma exposure time and further reduced after PMA. Consequently, we fabricated an Al₂O₃ film suitable for application as a CIS passivation layer with a reduced number of interface traps. However, the Al₂O₃ film with increased O₂ plasma exposure time deteriorated owing to plasma damage after H₂ plasma treatment, which is a method of reducing carbon impurity content. This deterioration was validated using the C–V hump and breakdown characteristics.

1. Introduction

Recently, the importance of CMOS image sensor (CIS) technology has rapidly increased owing to its relevance in mobile products and autonomous driving. As electronic products become ever-smaller in size, smaller CIS devices are also required. Therefore, CIS devices must be scaled, similar to other semiconductor devices. The pixel size of the CIS image sensor has been rapidly scaled, limiting the number of photons entering the pixel. In addition, as a result of scaling, light reflection occurred, causing light loss and cross-talk issues [1]. Therefore, a backside illumination-type CIS device that illuminates the rear side of the device was developed [2]. However, the backside illumination structure is adversely affected by dark currents and noise. Hence, in order to decrease dark current and increase quantum efficiency, research on the development of high-k materials for application as a CIS passivation dielectric layer is necessary.

Al₂O₃, which is a high-k dielectric material, has a wide energy bandgap and high thermal stability; therefore, it is suitable for application as a passivation dielectric film for CIS [3,4]. In addition, unlike other dielectric films, Al₂O₃ has negative fixed charges and shows excellent passivation characteristics [5]. In most semiconductor devices such as complementary metal oxide semiconductors (CMOSs), fixed charges act as defects [6]. Thus, many studies have been conducted to control these negative fixed charges [7]. However, in the CIS device, a passivation dielectric layer is required to contain high fixed charges for field effect passivation. Therefore, Al₂O₃ is a suitable dielectric material as a passivation layer of CIS. However, a dielectric film with fewer impurities is required for fabricating a more precise CIS device, and defects in Al₂O₃ must be further cured. In particular, for application as a dielectric film, the interface trap density (D_it) should be reduced to increase the amount of light absorbed. There are several causes of trap generation in the interface area between Al₂O₃ dielectric and substrate. If the Al₂O₃ dielectric is deposited on the silicon substrate, the hydroxyl group (-OH) and Si are bonded, which may act as an interface trap [8]. In another case, carbon impurities generated during the Al₂O₃ deposition process act as interface traps.

Carbon impurities were generated after the Al₂O₃ film was deposited via plasma-enhanced atomic layer deposition (PEALD) using trimethylaluminum as a precursor [9]. These impurities act as traps inside the Al₂O₃ and in the interface region. Previously, residual carbon was removed using the H₂ plasma treatment of an Al₂O₃ film [10]. The quality of the dielectric and interface areas increased with a decrease in carbon impurity contents. In addition, posttreatments provided sufficient fixed charges for the Al₂O₃ dielectric to be used as the passivation layer of the CIS [11,12]. However, a low D_it is required for next-generation CIS devices. Although well-known defects, such as oxygen vacancies, have been investigated [13], limited studies have been conducted to reduce residual carbon contents, except by changing the precursor [14].

In this study, the oxygen plasma exposure time was adjusted during Al₂O₃ deposition to reduce the residual carbon content. The increased O₂ plasma exposure time sufficiently decreased the D_it of the Al₂O₃ gate stack. Consequently, it showed a considerably lower D_it compared with that of the sample processed via rapid thermal annealing and H₂ plasma treatment on Al₂O₃, which exhibited the lowest D_it in a previous study [10]. In particular, D_it was the lowest after post-metallization annealing (PMA) to Al₂O₃ samples with increased O₂ plasma exposure time. In addition, a positive shift in flat-band voltage (ΔV_fb) was prevented by reducing carbon generation. However, D_it increases when H₂ plasma treatment is performed after Al₂O₃ deposition. Plasma damage and residual hydrogen impurities were caused by excessive H₂ plasma treatment on Al₂O₃ dielectric and were validated using the C–V hump occurring in the capacitance vs. voltage (C–V_G) measurement curve.

2. Experimental Materials and Methods

As shown in Figure 1, an Al₂O₃ film was deposited on a Si substrate at 275 °C using PEALD. Substrate included moderately doped p-type Si (1–30 Ω·cm, (100)) with a doping concentration of ~1.3 × 10¹⁶ cm⁻³. Prior to deposition of the Al₂O₃ layer, Si substrates were cleaned by dipping in a NH₄OH:H₂O₂:H₂O mixture (1:1:5 by volume), known as Standard Clean 1 (SC1), for 10 min at 70 °C, followed by dipping in dilute HF (100:1) for 1 min to remove native oxides. For deposition of Al₂O₃ dielectric, a commercial 200 mm wafer plasma-enhanced vapor deposition (PECVD; Quros Plus 200) was used. As a precursor, Trimethylaluminum (TMA, Al(CH₃)₃) (Up chemical co. Ltd., Pyeongtaek, Gyeonggi-do, Republic of Korea; 99.9999%) was supplied. For sequential surface reactions, O₂ plasma was supplied with TMA. The O₂ plasma exposure times were 3 and 7 s. During the deposition, an Al(CH₃)₃ container temperature of 25 °C, an Ar purge flow rate of 500 sccm, an O₂ flow rate of 100 sccm and a chamber pressure of 0.4 mTorr were used. Al electrode with a diameter of 300 µm and an area of 7.06 × 10⁴ µm² was deposited on the Al₂O₃ dielectric using an e-beam evaporator. The thickness of the Al₂O₃ film was measured using transmission electron microscopy (TEM; JEM-2100F; JEOL KOREA LTD., Seoul, Republic of Korea) and ellipsometry (M-2000; J. A. Woollam Co., Anyang, Gyeonggi-do, Republic of Korea). After Al₂O₃ deposition, H₂ plasma treatment and PMA were performed separately depending on the sample (

Table 1. Al₂O₃ samples under deposition conditions (O₂ plasma exposure time: 3 and 7 s) and posttreatment conditions (H₂ plasma treatment and post-metallization annealing (PMA)).

Samples	O2 Plasma Exposure Time (s)	H2 Plasma Treatment	PMA
S1_as_dep	3	X	X
S1_H2_plamsa	3	O	X
S1_PMA	3	X	400 °C; 30 min
S1_H2 plasma + PMA	3	O	400 °C; 30 min
S2_as_dep	7	X	X
S2_H2_plamsa	7	O	X
S2_PMA	7	X	400 °C; 30 min
S2_H2 plasma + PMA	7	O	400 °C; 30 min

). H₂ plasma treatment was performed with a H₂ gas flow rate ratio {[H₂] = ([H₂] + [Ar])} of 0.89 in a PECVD chamber for 15 min. PMA was performed at 400 °C under a N₂ gas flow in a furnace for 30 min. The N₂ gas flow rate {[N₂] = ([N₂] + [H₂])} was 0.95 (gas pump: 100 sccm; pressure: 0.7 atm). Under the N₂ gas flow, the temperature increased from 25 °C to 400 °C in 1 h and then decreased from 400 °C to 25 °C in 2 h. Secondary ion mass spectrometry (SIMS) measurements were conducted on a circular area with a diameter of 33 µm using the Cs+ software. Selective area diffraction pattern (SADP) analysis was carried out to determine crystallinity of the Al₂O₃ film. The capacitance and conductance were measured using a B1520A multifrequency capacitance measurement unit at various frequencies (1 kHz–1 MHz). The leakage current and breakdown field were measured using a Keithley 4200-SCS instrument (Tektronix KOREA, Seoul, Republic of Korea). D_it (≈2.5(qA)⁻¹($G_p/\omega$)_max) was calculated following the well-known conductance method [15]:

$$G_p/\omega =C_{OX}^2{G}_M{\omega }^{-1}/\left\{{(G_M/\omega )}^2+{\left(C_{OX}-C_M\right)}^2\right\}$$

(1)

where q = 1.6 × 10¹⁹ C; A is the area of the electrode; ($G_p/\omega$)_max is the normalized parallel conductance peak; $C_{OX}$ is the capacitance in strong accumulation; $C_M$ is the measured capacitance; and $G_M$ is the measured conductance.

3. Results and Discussion

3.1. Post-Metallization Annealing

Al₂O₃ was deposited via PEALD using trimethylaluminum as the precursor and O₂ plasma. A flux of O* radicals reacts with methyl groups and is effused in the CO_X (x = 1–2) state [16]. However, residual carbon is generated when a sufficient reaction is not performed and acts as a defect in the inner and interfacial regions of Al₂O₃. Accordingly, the O₂ plasma exposure time was increased to 7 s to ensure a sufficient response.

The TEM image of the as-deposited Al₂O₃ film is shown in Figure 2a. An Al₂O₃ film with a thickness of 30 nm was deposited on the Si substrate, and Al electrode was deposited on the Al₂O₃ dielectric. No interfacial layer (IL) was formed at the interface between Si and Al₂O₃. Additionally, based on the SADP in Figure 2a, the as-deposited Al₂O₃ is in an amorphous state.

PMA was performed at 400 °C for 30 min after Al₂O₃ film deposition. After PMA on the Al₂O₃ film, oxygen in the dielectric film diffused toward the Si substrate. Accordingly, Si and oxygen form a bond in the SiO_X (x = 1–2) state, thereby forming an IL with a thickness of 2.5 nm [8,17] (Figure 2b). As IL was formed between Al₂O₃ and Si, the thickness of Al₂O₃ decreased from 28.7 to 26.9 nm after PMA. Furthermore, as shown in SADP, amorphous Al₂O₃ is converted to polycrystalline Al₂O₃ via PMA [18].

The normalized capacitance vs. voltage curves before and after PMA of S1 and S2 are shown in Figure 3. The graphical ((C_OX/C_MOS)² − 1)(V_G) method [19] was applied to the normalized capacitance vs. voltage curve to extract V_fb. The V_fb of as-deposited S1 was 1.65 V, showing a considerable flat-band voltage shift (ΔV_fb ≈ 2.54 V) compared with the theoretical value of Al₂O₃ dielectric (V_fb ≈ −0.89 V). This V_fb shift resulted from defects, such as carbon impurities that occur during Al₂O₃ deposition via PEALD. However, in the case of S2 samples with an increased O₂ plasma exposure time, the V_fb of S2_as_dep is 0.61 V, exhibiting a smaller ΔV_fb compared with that of S1. This is because the amount of negatively charged defects inside S2 is smaller than that of S1.

V_fb increased by 0.54 V after PMA in the case of Al₂O₃ samples with short O₂ plasma exposure times. Internal defects that form bonds with carbon impurities have a negative charge and diffuse toward Si [10,14]. However, in the case of S2 samples with long O₂ plasma exposure times, the change in V_fb was as small as 0.2 V owing to a decrease in the defects that can be diffused.

The permittivity of Al₂O₃ samples before and after PMA is shown in Figure 4. The permittivity is 9.5 in the case of the as-deposited S1 sample, which is similar to the generally known permittivity value of amorphous Al₂O₃ (6–9) [18,20]. However, an IL of SiO_X (x = 1–2) is formed between Al₂O₃ and Si after PMA, slightly decreasing the permittivity. The permittivity of the as-deposited S2 sample is 12.5, which is considerably higher than that of the S1 sample. This is because of the decrease in the content of various defects and the increase in the internal carbon concentration owing to the longer O₂ plasma exposure time. After PMA on the as-deposited S2 sample, the permittivity decreases to 10.5 because an IL of SiOx (x = 1–2) is formed between Al₂O₃ and Si like the S1 sample. However, the S2_PMA sample still showed a higher permittivity than the S1 samples with shorter O₂ plasma exposure time.

The decrease in the carbon impurity content with increasing O₂ plasma exposure time was validated using SIMS depth profiling. As shown in Figure 5, the amount of carbon impurities in the Al₂O₃ film deposited with an O₂ plasma exposure time of 7 s is considerably less than that of the Al₂O₃ sample deposited with a shorter O₂ plasma exposure time. As the O₂ plasma exposure time increased, more carbon was effused into the CO_X (x = 1–2) gas state through numerous reactions between the oxygen plasma and carbon [16]. If the O₂ plasma exposure time is more than 7 s, there is a possibility of improvement as much as carbon is reduced. However, there is a limit to effuse through the reaction with carbon, and the improvement effect is expected to be saturated as carbon is reduced.

To apply Al₂O₃ as a passivation dielectric film, the quality of the interface region between Si and Al₂O₃ is crucial. Carbon in Al₂O₃ acts as an interface trap in the interface region between the Al₂O₃ dielectric and Si substrate [21]. The parallel conductance versus frequency plots of the Al₂O₃ films with various D_it values are shown in Figure 6. D_it was measured using the conductance method [13]. The D_it of the S1_as_dep sample was 8.98 × 10¹³ eV⁻¹·cm⁻², whereas that of the S2_as_dep sample was 1.12 × 10¹² eV⁻¹·cm⁻². The interface traps of the S2 sample decreased with a decrease in the carbon impurity content in the interface area with increasing O₂ plasma exposure time. After PMA, the interface region between the Al₂O₃ dielectric and Si was improved due to various reasons. First, an IL was formed after the application of PMA to the Al₂O₃ gate stack. Therefore, the number of hydroxyl groups is reduced, thereby decreasing the number of interface traps [22]. For other reason, as crystallization of Al₂O₃ occurred due to PMA, defects and dangling bonds acting as traps in the interface region were removed. In addition, crystallization of the Al₂O₃ dielectric stabilized the bond between the Al₂O₃ and Si substrate [10].

In summary, the number of interface traps of the S2_PMA sample, in which the concentrations of both carbon impurities and hydroxyl groups were reduced, were the lowest in this study (D_it = 1.35 × 10¹¹ eV⁻¹·cm⁻²).

The interface improvement owing to the increase in the O₂ plasma exposure time was also validated using the breakdown characteristics. The gate leakage current with an increase in the electrical field of the S1 and S2 Al₂O₃ samples is shown in Figure 7a. In the case of S1_as_dep, the breakdown occurred at 9.73 MV/cm. The breakdown characteristics improved after PMA was performed owing to the formation of an IL, which occurred at 11.47 MV/cm. However, breakdown did not occur until the application of the maximum electric field (14 MV/cm) of the 4200-SCS equipment in the case of the S2 sample. Furthermore, breakdown did not occur in the case of the S2_as_dep sample without the IL. This was because of the reduction in the impurity content in the interface area with an increase in the O₂ plasma exposure time.

In addition, the FN plots to validate the improvement in the interface quality are shown in Figure 7b. The FN plot is analyzed using the leakage current density caused by FN tunneling, J_FN, and can be described as follows:

$$J_{FN}=A{E}^2\exp \left(-B/E\right),$$

(2)

where

$$A=q^3{m}_0/\left(8\pi h{m}^{*}{\mathsf{\Phi }}_B\right),$$

and

$$B=4{(2m^{*})}^{\frac{1}{2}}{(q{\mathsf{\Phi }}_B)}^{\frac{3}{2}}/\left(3qh/2\pi \right),$$

where $A$ is the Richardson’s constant; $q$ is the electronic charge; $h$ is Planck’s constant; $m_0$ is the free electron mass; $m^{*}$ is the effective electron mass in the oxide; and ${\mathsf{\Phi }}_B$ is the barrier height [23]. The steeper the slope in the FN plot, the larger the FN barrier height ${\mathsf{\Phi }}_B$ [4]. Since the absolute value of the slope of the S2_as_dep sample (slope = −182.06) is larger than that of the S1_as_dep sample (slope = −103.28), it means that the barrier height is higher in S2_as_dep. Therefore, the FN plot shows that the interface region of Al₂O₃/Si was improved in the S2 sample with increased O₂ plasma exposure time.

In summary, the increase in the O₂ plasma exposure time decreases the carbon content in Al₂O₃, which reduces D_it, improves the breakdown field, and prevents the V_fb shift. However, the H₂ plasma treatment decreased the quality of the oxide and interface owing to the increase in the O₂ plasma exposure time, which is discussed later.

3.2. H₂ Plasma Treatment

H₂ plasma treatment significantly decreased the carbon impurity content in Al₂O₃ in previous studies [10], thereby preventing the V_fb shift and improving the breakdown characteristics. However, further improvements in the interface quality is required for next-generation CIS devices. Therefore, we analyzed the effects of the H₂ plasma treatment on Al₂O₃ films with increasing O₂ plasma exposure time.

D_it values depending on various treatments on the Al₂O₃ samples are shown in Figure 8. The average D_it of the sample with the H₂ plasma treatment was 4.45 × 10¹² eV⁻¹·cm⁻², which was significantly smaller than that of as-deposited S1. However, the average D_it of the sample with the H₂ plasma treatment was 1.13 × 10¹² eV⁻¹·cm⁻² in the case of S2 samples with an increased O₂ plasma exposure time, which increased compared with the average D_it of as-deposited S2 (D_{it,as_dep S2} = 5.79 × 10¹¹ eV⁻¹·cm⁻²). A similar trend was observed after PMA. D_it was higher in the S2 sample with the H₂ plasma treatment and PMA than that of the S2 sample treated with only PMA.

A large amount of carbon impurities was removed owing to the increased O₂ plasma exposure time in the S2 sample. Therefore, there are not enough carbon impurities for the reaction with the H₂ plasma. As a result, owing to the excessive postprocessing H₂ plasma treatment on the S2 sample, H impurities remained inside the Al₂O₃ film [24]. In addition, additional H₂ plasma treatment for carbon impurities, whose content was reduced owing to an increase in the O₂ plasma exposure time, had a more significant effect on the formation of defects owing to damage due to the plasma treatment compared with the effects of curing defects owing to carbon content reduction [25]. In conclusion, in the case of the S2 sample with increased O₂ plasma exposure time, excessive postprocessing H₂ plasma treatment caused residual H impurities and plasma damage, which contributed to increase D_it by forming dangling bonds in interface region.

Using the capacitance vs. voltage curve, the plasma damage to the gate stack was confirmed. The normalized capacitance before and after the H₂ plasma treatment in the S2 sample with an increased O₂ plasma exposure time is shown in Figure 9. In contrast to the S2_as_dep sample, the C–V hump occurs near V_fb in the S2_H₂ plasma sample. Therefore, the hydrogen plasma, which should be effused via the reaction with carbon, damaged the Al₂O₃ dielectric.

The formation of defects in the oxide and interface regions of the S2 sample owing to the H₂ plasma treatment resulted in more leakage flow in the gate stack. In contrast to the S2_as_dep sample, where breakdown does not occur even under the electric field limit of the 4200-SCS equipment (E_field = 14 MV/cm), the breakdown occurs at 11.2 MV/cm in the S2_H₂ plasma sample (Figure 10). As a result, in the case of the Al₂O₃ film with increased O₂ plasma exposure time, H₂ plasma treatment rather deteriorates the interface quality between Al₂O₃ dielectric and Si.

In summary, H₂ plasma treatment has different effects depending on the O₂ plasma exposure time during deposition of the Al₂O₃ dielectric. H₂ plasma treatment was effective for S1 samples with a large amount of carbon impurities because of the short O₂ plasma exposure time. Due to the reduction of carbon impurities, the D_it of the S1 sample was greatly reduced after H₂ plasma treatment. However, the treatment effects on S2 samples was rather poor, resulting in reduced carbon content owing to the long O₂ plasma exposure time. H₂ plasma treatment produced residual H impurities in the S2 samples and also caused plasma damage. Therefore, H₂ plasma treatment rather increased D_it in the Al₂O₃ with increased O₂ plasma exposure time.

4. Conclusions

The criterion of fixed charges in the Al₂O₃ film for application as a CIS passivation layer was satisfied in a previous study; however, the issue of interface traps remained unresolved. Further improvement in the interface area is required for Al₂O₃ to be used as a passivation dielectric layer. Therefore, this study investigated the conditions to reduce defect contents and the D_it of the Al₂O₃ film. The carbon content inside the Al₂O₃ was significantly decreased by adjusting the O₂ plasma exposure time to induce more reactions during dielectric deposition. D_it was significantly decreased owing to the reduction in the amount of carbon impurities, and the improvement in the interface region was validated using the breakdown characteristics. Moreover, H₂ plasma treatment effectively reduced D_it in Al₂O₃ films with a short O₂ plasma exposure time during deposition. However, H₂ plasma treatment of the Al₂O₃ film deposited with a long O₂ plasma exposure time rather increased D_it due to plasma damage. PMA slightly decreased the permittivity after Al₂O₃ deposition; however, D_it significantly decreased. In particular, in the case of Al₂O₃ samples with increased O₂ plasma exposure time, after PMA, it had the lowest D_it, which is suitable for use as a passivation layer for CIS.