Low‑Temperature CO₂‑Based PEALD of SiO₂ for Moisture‑Sensitive Applications

Abstract

We present a high‑performance, low‑temperature plasma‑enhanced atomic layer deposition (PEALD) process for silicon dioxide (SiO₂) that employs carbon dioxide (CO₂) as the oxidant and bis(tertiary‑butylamino)silane (BTBAS) as the silicon precursor. The process operates at 90 °C, producing SiO₂ films with a self‑limiting growth‑per‑cycle (GPC) of ~1.15 Å / cycle, a density of ~2.1 g / cm³, a refractive index of ~1.46 at 632 nm, and a low tensile residual stress of ~30 MPa. Chemical analysis reveals impurity levels of ~2.4 at % hydrogen, ~0.17 at % nitrogen, and carbon concentrations below the detection limit of time‑of‑flight elastic recoil detection analysis (TOF‑ERDA). These results establish CO₂ as a viable, non‑reactive oxidant for PEALD processes on moisture‑ and oxygen‑sensitive substrates.

Background

Silicon dioxide is indispensable across microelectronics, MEMS, photovoltaics, and optics. While conventional deposition methods such as thermal oxidation, PECVD, and PVD provide acceptable film quality, atomic layer deposition uniquely offers sub‑nanometer thickness control, exceptional uniformity, and conformality—key advantages for advanced device architectures.

Traditional SiO₂ ALD recipes rely on chlorosilanes or aminosilanes paired with oxidants like H₂O, H₂O₂, or O₃. These processes typically require temperatures above 150 °C, limiting their applicability to thermally sensitive materials. PEALD techniques lower the substrate temperature to below 100 °C, yet the widely used H₂O and O₂ oxidants can still degrade moisture‑ or oxygen‑sensitive substrates. CO₂, being chemically inert at low temperatures, offers a safer alternative by minimizing unintended oxidation. Previous work demonstrated CO₂‑based PEALD SiO₂ at 250–400 °C, but a low‑temperature implementation had not been achieved.

Here we report a CO₂‑based PEALD process that delivers high‑quality SiO₂ at 90 °C. We systematically investigate the influence of precursor pulse/purge times and plasma power on film growth, and we present comprehensive characterizations of the films’ structural, compositional, optical, and mechanical properties.

Methods

Film Preparation

SiO₂ layers were deposited on Si(100) and c‑Al₂O₃ substrates in a Beneq TFS 200 reactor equipped with a remote, capacitively coupled 13.56 MHz rf plasma source. CO₂ (99.5 %, Air Products) was introduced as the oxidant at a flow of 75 sccm mixed with N₂ (200 sccm) for the plasma region, while N₂ (99.999 %, AGA) at 600 sccm served as carrier and purge gas. BTBAS (97 %, Strem Chemicals) was pulsed at 21 °C and delivered with an N₂ booster during the precursor pulse. Process parameters are summarized in Table 1; the reactor pressure remained ~1 hPa throughout. A saturated GPC of ~1.15 Å / cycle was consistently obtained at 90 °C.

Film Characterization

Film thickness was measured by a SENTECH SE400adv ellipsometer using a 632.8 nm HeNe laser at 70°. GPC was calculated from thickness divided by the number of ALD cycles, with the deviation reflecting thickness non‑uniformity. Chemical composition was probed by glow‑discharge optical emission spectroscopy (GDOES), time‑of‑flight elastic recoil detection analysis (TOF‑ERDA), and attenuated total reflectance FTIR (ATR‑FTIR). GDOES employed a 4‑mm anode and 35 W pulsed rf power. TOF‑ERDA used 40 MeV Br⁺ ions at 40° detection angle; depth resolution was considered in the analysis. ATR‑FTIR spectra were recorded on a Nicolet 380 with a diamond crystal, using 2 cm⁻¹ resolution over 800–4000 cm⁻¹. X‑ray reflectivity (XRR) was performed on a Philips X’Pert Pro with Cu‑Kα₁ radiation; densities were extracted via in‑house fitting software. Optical transmittance of 150‑nm SiO₂ on sapphire was measured with a PerkinElmer Lambda 900, and refractive indices were derived by Cauchy fitting. Residual stress was determined by wafer curvature (TOHO FLX‑2320‑S) before and after growth, analyzed with Stoney’s equation; 50‑nm films were used for this measurement.

Results and Discussion

Film Growth

The dependence of GPC on BTBAS pulse and purge times was first examined at a fixed CO₂ plasma power of 180 W, 3 s exposure, and 2 s purge. Figure 1a shows that a 0.1 s BTBAS pulse suffices to reach saturation (≈1.15 Å / cycle). Short purge times (≤3 s) maintain this saturation, confirming efficient removal of residual precursor species. Figure 1b demonstrates that varying the BTBAS purge time from 0.5 to 3 s does not affect GPC, although uniformity improves with longer purge.

GPC of PEALD SiO₂ films as a function of BTBAS pulse (a) and purge (b) times. The applied plasma power was 180 W.

Next, we investigated the influence of CO₂ plasma exposure and purge times while keeping BTBAS pulse/purge at 0.3 s/3 s. Figure 2a illustrates that a 3 s exposure at 180 or 300 W yields the saturated GPC, whereas 1 s exposure is insufficient due to inadequate O‑radical generation. At 50 W, GPC increases up to 6 s exposure before plateauing. Figure 2b confirms that CO₂ purge times between 0.5 and 3 s have negligible impact on GPC, underscoring efficient precursor removal by the reactor’s cross‑flow design.

GPC of PEALD SiO₂ films as a function of CO₂ plasma exposure (a) at 50, 180, and 300 W, and purge (b) at 180 W.

Using the optimized parameters (BTBAS 0.3 s/3 s, CO₂ 3 s exposure/2 s purge, 180 W), the deposition rate reaches 50 nm / h; a 100 nm / h rate is attainable with shorter precursor and purge times.

Film Properties

Density measurements via XRR (Fig. 3) show a modest increase from 1.9 to 2.1 g / cm³ as CO₂ exposure extends from 1 to 6 s, consistent with film densification at higher plasma powers. The measured density (2.11 g / cm³) aligns with values reported for low‑temperature O₂‑based PEALD SiO₂ but remains below the 2.3 g / cm³ reported at 400 °C.

Density of SiO₂ films grown with plasma exposure times of 1, 3, and 6 s.

GDOES provided qualitative compositional insights: Si, O, H, N, and C were detected, with H content stable across exposure times and N decreasing at 6 s exposure, suggesting effective impurity removal during densification. C signals were uniform, likely due to surface contamination rather than bulk incorporation.

Qualitative chemical composition of SiO₂ films grown with plasma exposure times of 1, 3, and 6 s measured by GDOES.

TOF‑ERDA depth profiling of a 50‑nm film (Fig. 5a) confirms low bulk impurity levels: ~2.4 at % H, ~0.17 at % N, and C below detection limits. The Si/O ratio (~0.48) indicates a slightly oxygen‑rich film, consistent with residual –OH groups expected from the combustion‑like oxidation mechanism.

a TOF‑ERDA depth profile and b ATR‑FTIR transmission spectrum of the SiO₂ film (50‑nm target thickness).

ATR‑FTIR (Fig. 5b) reveals broad O‑H stretching (3200–3800 cm⁻¹) and a shoulder at ~900 cm⁻¹, confirming –OH presence. Si‑O‑Si stretches appear near 1108 and 1226 cm⁻¹, slightly higher than literature values, possibly due to residual stress effects. Minor CH vibrations (970, 1301, 1450 cm⁻¹) arise from surface contamination.

Optical measurements (Fig. 6) show a transmittance spectrum with a refractive index of 1.456 at 632 nm and negligible extinction coefficient, reflecting the low carbon content and high film quality.

a Transmittance spectrum and Cauchy fitting; b Refractive index dispersion of a 150‑nm SiO₂ film on sapphire.

Residual stress (Fig. 7) is low and tensile (~30 MPa) at 90 °C, contrasting with the 150 MPa compressive stress observed at 400 °C. This near‑zero stress likely originates from intrinsic film stress rather than thermal mismatch, aligning with trends reported by Putkonen and Shestaeva.

Residual stress of SiO₂ films versus growth temperature, comparing our 90 °C data (50‑nm film) with literature values.

Conclusions

The CO₂‑based PEALD process described here delivers high‑quality, low‑impurity SiO₂ films at 90 °C, suitable for moisture‑ and oxygen‑sensitive substrates. With a saturated GPC of ~1.15 Å / cycle, a density of ~2.1 g / cm³, and tensile residual stress around 30 MPa, the films meet industry standards. Importantly, the short 4‑s ALD cycle time supports high‑throughput manufacturing, making this approach attractive for next‑generation electronic and photonic devices.

Abbreviations

ALD

Atomic layer deposition

ATR‑FTIR

Attenuated total reflectance Fourier transform infrared spectroscopy

BTBAS

Bis(tertiary‑butylamino)silane

GDOES

Glow‑discharge optical emission spectroscopy

GPC

Growth‑per‑cycle

PEALD

Plasma‑enhanced atomic layer deposition

PECVD

Plasma‑enhanced chemical vapor deposition

PVD

Physical vapor deposition

rf

Radio frequency

TOF‑ERDA

Time‑of‑flight elastic recoil detection analysis

XRR

X‑ray reflectivity