PECVD Reactor

1. Theory

• Definition: A PECVD reactor uses an RF plasma to break up precursor gases so that films deposit at 200 to 400 ${}^{\circ }$C instead of the 650 to 800 ${}^{\circ }$C of LPCVD nitride (KTH; LPCVD Furnace page).
• Hydrogen and stoichiometry: PECVD nitride contains much hydrogen and is usually non-stoichiometric; hydrogen content strongly affects stress, index and etch rate (KTH; Rasmussen et al.).
• Frequency and stress: 13.56 MHz films are tensile, 120 to 150 MPa, while 380 kHz films are compressive, 600 to 700 MPa, so mixing the two gives zero stress (Iliescu et al.).
• Composition and index: Index rises with the SiH${}_4$/NH${}_3$ ratio as films become silicon-rich (Rasmussen et al.).

2. Working Principle

• Parallel-plate reactor: HF power at 13.56 MHz on the top electrode, LF at 380 kHz, heated platen up to 300 ${}^{\circ }$C (CNS).
• Gases: SiH${}_4$, NH${}_3$, N${}_2$O, N${}_2$ and dopant gases for SiO${}_2$, Si${}_3$N${}_4$, low-stress nitride and a-Si (CNS).
• Process window: SiN${}_x$ at 600 W and about 800 to 900 mTorr (Iliescu et al.).
• Step coverage: Remote-plasma systems near 0.1 Torr deposit directionally with poorer step coverage than older systems above 1 Torr (Georgia Tech).

3. Applications

• Low-temperature dielectrics: SiO${}_2$ and nitride over metal, where LPCVD temperatures are too high (KTH; CNS).
• MEMS membranes and masks: Stress-tuned nitride and KOH-resistant masks (Iliescu et al.; Rasmussen et al.).
• Film monitoring: Index and stress checked on the Spectroscopic Ellipsometer and Wafer Curvature Stress Gauge pages (my reading).

Plasma, Frequency and Rate

Name	Formula	Related Formulae	Variables/Constants
Plasma-assisted temperatures	$T_{SiN,PECVD}=200\text{to}400{}^{\circ }C$	$T_{Si{O}_2}:Si{H}_4+2N_{2}O(200\text{to}350{}^{\circ }C),\text{TEOS}(300\text{to}360{}^{\circ }C)$	$\begin{array}{r}KTH:PECVDnitridehydrogen-richand\\ non-stoichiometric,against650to800{}^{\circ }C\\ forLPCVDnitride;CNS:platenupto300\\ {}^{\circ }C;Rasmussenetal.:300{}^{\circ }C;LPCVD\\ Furnacepage\end{array}$ $\begin{array}{r}(myreading:theplasmasuppliesthe\\ energythatheatsuppliesinLPCVD)\end{array}$
Two frequencies	$f_{HF}=13.56\mathrm{MHz},f_{LF}=380\mathrm{kHz}$	$\text{mixed: HF/LF power-on ratio}$	$\begin{array}{r}CNSandRasmussenetal.:bothsupplies;\\ Iliescuetal.:380\mathrm{kHz}inthetext,400\\ \mathrm{kHz}intheabstract\end{array}$ $\begin{array}{r}(myreading:ionscanfollowthelow\\ frequencyandbombardthefilm,which\\ compactsit)\end{array}$
Deposition rate	$R_{SiN}=10(HF),45(LF)nm/min$	$R_{SiN}=100\to 250\text{to}320nm/min(100\to 600\mathrm{W})$	$\begin{array}{r}CNS:Si{O}_{2}at56(HF)and79(LF)nm/min;\\ Iliescuetal.:HFpowerseries;Rasmussen\\ etal.:powerthemainfactor\end{array}$ $\begin{array}{r}(computed:a2.5to3.2timesrisefora6\\ timesriseinpower)\end{array}$

Stress Control

Name	Formula	Related Formulae	Variables/Constants
Frequency and stress	${\sigma }_{HF}=+120\text{to}+150\mathrm{MPa}$	${\sigma }_{LF}=-600\text{to}-700\mathrm{MPa}$	$\begin{array}{r}+:tensile,-:compressive;Iliescuet\\ al.:Si{N}_{x}at20to80\mathrm{W},zerostressby\\ alternatingorsimultaneousHFandLF;\\ CNS:mixed-frequencytailoring\end{array}$ $\begin{array}{r}(computed,assuminglinearmixing:zero\\ stressatanLFfractionof135/785=\\ 0.17)\end{array}$
Power and pressure	$\sigma (600\mathrm{W})\approx 4\mathrm{MPa}$	$p_{opt}\approx 900\mathrm{mTorr}(700\text{to}1100\mathrm{mTorr})$	$\begin{array}{r}Iliescuetal.:tensilestressfallswith\\ HFpower,higherat1100\mathrm{mTorr},recipe600\\ \mathrm{W},800\mathrm{mTorr},N_2/N{H}_3/Si{H}_{4}1200/30/60\mathrm{sccm};\\ Rasmussenetal.:stresslesscompressive\\ withpowerandpressure\end{array}$ $\begin{array}{r}(myreading:thetwostudiesagreein\\ direction,towardtensile,withpower)\end{array}$
Stress measurement	$\sigma \text{from wafer bow}$	$\text{limits thickness before bowing or cracking}$	$\begin{array}{r}GeorgiaTech:filmstresslimits\\ thickness;WaferCurvatureStressGauge\\ page\end{array}$ $\begin{array}{r}(myreading:stresstuningisthemain\\ reasonfordual-frequencyPECVDinMEMS)\end{array}$

Composition and Etch Resistance

Name	Formula	Related Formulae	Variables/Constants
Refractive index and composition	$n_{SiN}=1.86\text{to}1.93(LF),1.93\text{to}2.25(mixed)$	$n\uparrow \text{with}Si{H}_4/N{H}_3$	$\begin{array}{r}Rasmussenetal.:higherindexfromhigher\\ Sicontent;CNS:S{i}_3{N}_{4}1.977(LF),2.01\\ (HF),Si{O}_{2}1.48and1.485;Spectroscopic\\ Ellipsometerpage\end{array}$ $\begin{array}{r}(myreading:indexisthequickestin-line\\ checkoffilmstoichiometry)\end{array}$
Hydrogen bonding	${\nu }_{N-H}=3350,1150\mathrm{c}{\mathrm{m}}^{-1}$	${\nu }_{Si-H}\approx 2200\mathrm{c}{\mathrm{m}}^{-1}\text{(mostly undetected)}$	$\begin{array}{r}Rasmussenetal.:nitrogen-richfilms,\\ hydrogencontentstronglyaffecting\\ stress,indexandetchrate;KTH:PECVD\\ nitridecontainsmuchhydrogen\end{array}$ $\begin{array}{r}(myreading:FTIRpeakareastrackbonded\\ hydrogen)\end{array}$
Wet etch resistance	$R_{BOE}\downarrow \text{as}n\uparrow$	$R_{KOH}\uparrow \text{with power and pressure}$	$\begin{array}{r}Rasmussenetal.:LFandmixed-frequency\\ filmsmoreKOHresistantthanHFfilms,a\\ trade-offagainststress\end{array}$ $\begin{array}{r}(myreading:denser,Si-richerfilms\\ resistetchingbetter)\end{array}$

Important Questions

Ques: Why does PECVD deposit at lower temperature than LPCVD?

Because the plasma supplies the reaction energy. KTH gives PECVD nitride at 200 to 400 ${}^{\circ }\mathrm{C}$, PECVD oxide from SiH${}_4$ and N${}_2$O at 200 to 350 ${}^{\circ }\mathrm{C}$, and plasma TEOS at 300 to 360 ${}^{\circ }\mathrm{C}$, against 650 to 800 ${}^{\circ }\mathrm{C}$ for LPCVD dichlorosilane nitride (LPCVD Furnace page). Harvard CNS runs its platen up to 300 ${}^{\circ }\mathrm{C}$, and Rasmussen et al. deposited all their films at 300 ${}^{\circ }\mathrm{C}$. The cost is film quality: KTH states that PECVD nitride contains much hydrogen and is usually non-stoichiometric. Rasmussen et al. find that hydrogen content strongly affects stress, index and etch rate. In my reading, PECVD is chosen whenever metal or other temperature-limited layers are already on the wafer. Reference: KTH Royal Institute of Technology, Lecture 8 ‘CVD and ALD’: PECVD nitride at 200 to 400 C with much hydrogen and usually non-stoichiometric, PECVD oxide SiH4 + 2N2O at 200 to 350 C, plasma TEOS at 300 to 360 C, LPCVD nitride at 650 to 800 C (fetch-tool summary); Harvard University Center for Nanoscale Systems, ‘STS PECVD’ tool page (NNCI): HF 13.56 MHz top electrode and LF 380 kHz supply, substrate from room temperature to 300 C (recipes at 300 C platen, 250 C lid), gases SiH4, NH3, N2O, N2, O2, Ar, PH3/Ar and B2H6/H2, films SiO2, Si3N4, low-stress Si3N4, a-Si and doped versions, rates SiO2 79 (LF) and 56 (HF) nm/min, Si3N4 45 (LF) and 10 (HF) nm/min, indices SiO2 1.48 (LF) and 1.485 (HF), Si3N4 1.977 (LF) and 2.01 (HF), mixed-frequency stress tailoring by the HF/LF power-on ratio (fetch-tool summary); F. E. Rasmussen, B. Geilman, M. Heschel, O. Hansen and A. M. Jorgensen (Mikroelektronik Centret, DTU, and Hymite), ‘Development and characterisation of KOH resistant PECVD silicon nitride for microsystems applications’, ECS meeting abstract: all films at 300 C, 380 kHz, 13.56 MHz and mixed excitation, hydrogen content strongly affecting stress, index and etch rate, nitrogen-rich films with N-H peaks at 3350 and 1150 cm-1 and Si-H near 2200 cm-1 mostly undetectable, index rising with SiH4/NH3 ratio, 1.86 to 1.93 for LF and 1.93 to 2.25 for mixed frequency, BOE rate falling as index rises, stress less compressive with higher power and pressure, KOH etch rate rising with power and pressure, LF and mixed films more KOH resistant, power the main factor in rate (fetch-tool summary); LPCVD Furnace page of this vault.

Ques: How does RF frequency control PECVD nitride stress?

High frequency gives tensile films and low frequency compressive ones. Iliescu et al. report SiN${}_x$ at 20 to 80 $\mathrm{W}$ with 120 to 150 $\mathrm{MPa}$ tensile stress at 13.56 $\mathrm{MHz}$ and 600 to 700 $\mathrm{MPa}$ compressive stress at 380 $\mathrm{kHz}$. Alternating the two, or running both together, reaches zero residual stress. Harvard CNS tailors stress through the HF/LF power-on ratio. If stress mixed linearly, zero would fall at an LF fraction of about 0.17 (computed, assuming linear mixing). Iliescu et al. also find that HF stress falls from 100 to 150 $\mathrm{MPa}$ to about 4 $\mathrm{MPa}$ as power rises from 100 to 600 $\mathrm{W}$. Rasmussen et al. likewise find stress less compressive at higher power and pressure. In my reading, ions that follow the low frequency compact the film; the result is checked from wafer bow (Wafer Curvature Stress Gauge page). Reference: C. Iliescu, M. Avram, B. Chen, A. Popescu, V. Dumitrescu, D. P. Poenar, A. Sterian, D. Vrtacnik, S. Amon and P. Sterian, ‘Residual stress in thin films PECVD depositions: a review’, Journal of Optoelectronics and Advanced Materials 13(4), 387-394 (2011): SiNx at 20 to 80 W with HF 13.56 MHz giving 120 to 150 MPa tensile and LF 380 kHz (400 kHz in the abstract) giving 600 to 700 MPa compressive, alternating or simultaneous HF and LF reaching zero stress, HF power 100 to 600 W lowering tensile stress from 100 to 150 MPa to 4 to 60 MPa and raising rate from about 100 to 250 to 320 nm/min, about 4 MPa at 600 W, 700 to 1100 mTorr studied with about 900 mTorr optimal and higher stress at 1100 mTorr, higher SiH4 fraction raising rate and lowering stress, a-Si:H LF compressive 310 to 450 MPa, a-SiC refractive index 2.5 to 2.6, recommended SiNx 600 W alternating HF/LF at 800 mTorr with N2/NH3/SiH4 1200/30/60 sccm (fetch-tool summary); Harvard University Center for Nanoscale Systems, ‘STS PECVD’ tool page (NNCI): HF 13.56 MHz top electrode and LF 380 kHz supply, substrate from room temperature to 300 C (recipes at 300 C platen, 250 C lid), gases SiH4, NH3, N2O, N2, O2, Ar, PH3/Ar and B2H6/H2, films SiO2, Si3N4, low-stress Si3N4, a-Si and doped versions, rates SiO2 79 (LF) and 56 (HF) nm/min, Si3N4 45 (LF) and 10 (HF) nm/min, indices SiO2 1.48 (LF) and 1.485 (HF), Si3N4 1.977 (LF) and 2.01 (HF), mixed-frequency stress tailoring by the HF/LF power-on ratio (fetch-tool summary); F. E. Rasmussen, B. Geilman, M. Heschel, O. Hansen and A. M. Jorgensen (Mikroelektronik Centret, DTU, and Hymite), ‘Development and characterisation of KOH resistant PECVD silicon nitride for microsystems applications’, ECS meeting abstract: all films at 300 C, 380 kHz, 13.56 MHz and mixed excitation, hydrogen content strongly affecting stress, index and etch rate, nitrogen-rich films with N-H peaks at 3350 and 1150 cm-1 and Si-H near 2200 cm-1 mostly undetectable, index rising with SiH4/NH3 ratio, 1.86 to 1.93 for LF and 1.93 to 2.25 for mixed frequency, BOE rate falling as index rises, stress less compressive with higher power and pressure, KOH etch rate rising with power and pressure, LF and mixed films more KOH resistant, power the main factor in rate (fetch-tool summary); Wafer Curvature Stress Gauge page of this vault.

Ques: What does the refractive index tell you about PECVD nitride?

Its silicon content. Rasmussen et al. found the index rising with the SiH${}_4$/NH${}_3$ flow ratio: 1.86 to 1.93 for low-frequency films and 1.93 to 2.25 for mixed-frequency films, which they link to higher silicon content. Their films were nitrogen-rich, with FTIR dominated by N-H peaks at 3350 and 1150 $\mathrm{c}{\mathrm{m}}^{-1}$ and the Si-H peak near 2200 $\mathrm{c}{\mathrm{m}}^{-1}$ mostly undetectable. The buffered-oxide etch rate fell as the index rose. Harvard CNS lists its nitride at 1.977 (LF) and 2.01 (HF) and its oxide at 1.48 and 1.485. Because index tracks composition and etch rate, it is the quickest in-line check of a PECVD film, measured as on the Spectroscopic Ellipsometer page (my reading). Reference: F. E. Rasmussen, B. Geilman, M. Heschel, O. Hansen and A. M. Jorgensen (Mikroelektronik Centret, DTU, and Hymite), ‘Development and characterisation of KOH resistant PECVD silicon nitride for microsystems applications’, ECS meeting abstract: all films at 300 C, 380 kHz, 13.56 MHz and mixed excitation, hydrogen content strongly affecting stress, index and etch rate, nitrogen-rich films with N-H peaks at 3350 and 1150 cm-1 and Si-H near 2200 cm-1 mostly undetectable, index rising with SiH4/NH3 ratio, 1.86 to 1.93 for LF and 1.93 to 2.25 for mixed frequency, BOE rate falling as index rises, stress less compressive with higher power and pressure, KOH etch rate rising with power and pressure, LF and mixed films more KOH resistant, power the main factor in rate (fetch-tool summary); Harvard University Center for Nanoscale Systems, ‘STS PECVD’ tool page (NNCI): HF 13.56 MHz top electrode and LF 380 kHz supply, substrate from room temperature to 300 C (recipes at 300 C platen, 250 C lid), gases SiH4, NH3, N2O, N2, O2, Ar, PH3/Ar and B2H6/H2, films SiO2, Si3N4, low-stress Si3N4, a-Si and doped versions, rates SiO2 79 (LF) and 56 (HF) nm/min, Si3N4 45 (LF) and 10 (HF) nm/min, indices SiO2 1.48 (LF) and 1.485 (HF), Si3N4 1.977 (LF) and 2.01 (HF), mixed-frequency stress tailoring by the HF/LF power-on ratio (fetch-tool summary); Spectroscopic Ellipsometer page of this vault.

Ques: What deposition rate does PECVD nitride reach, and which figure is more likely?

Both figures are real; the difference is power. Harvard CNS lists Si${}_3$N${}_4$ at 10 $nm/min$ in HF mode and 45 $nm/min$ in LF mode, and SiO${}_2$ at 56 and 79 $nm/min$. Iliescu et al. report HF nitride at about 100 $nm/min$ at 100 $\mathrm{W}$, rising to 250 to 320 $nm/min$ at 600 $\mathrm{W}$. That is a 2.5 to 3.2 times rise for a 6 times rise in power (computed). Rasmussen et al. confirm that RF power is the main factor in rate. The CNS values are more likely typical for a standard low-power, low-stress user recipe, and Iliescu’s for a high-power optimised process (my reading). A PECVD rate must therefore always be quoted with its power and frequency (my reading). Reference: Harvard University Center for Nanoscale Systems, ‘STS PECVD’ tool page (NNCI): HF 13.56 MHz top electrode and LF 380 kHz supply, substrate from room temperature to 300 C (recipes at 300 C platen, 250 C lid), gases SiH4, NH3, N2O, N2, O2, Ar, PH3/Ar and B2H6/H2, films SiO2, Si3N4, low-stress Si3N4, a-Si and doped versions, rates SiO2 79 (LF) and 56 (HF) nm/min, Si3N4 45 (LF) and 10 (HF) nm/min, indices SiO2 1.48 (LF) and 1.485 (HF), Si3N4 1.977 (LF) and 2.01 (HF), mixed-frequency stress tailoring by the HF/LF power-on ratio (fetch-tool summary); C. Iliescu, M. Avram, B. Chen, A. Popescu, V. Dumitrescu, D. P. Poenar, A. Sterian, D. Vrtacnik, S. Amon and P. Sterian, ‘Residual stress in thin films PECVD depositions: a review’, Journal of Optoelectronics and Advanced Materials 13(4), 387-394 (2011): SiNx at 20 to 80 W with HF 13.56 MHz giving 120 to 150 MPa tensile and LF 380 kHz (400 kHz in the abstract) giving 600 to 700 MPa compressive, alternating or simultaneous HF and LF reaching zero stress, HF power 100 to 600 W lowering tensile stress from 100 to 150 MPa to 4 to 60 MPa and raising rate from about 100 to 250 to 320 nm/min, about 4 MPa at 600 W, 700 to 1100 mTorr studied with about 900 mTorr optimal and higher stress at 1100 mTorr, higher SiH4 fraction raising rate and lowering stress, a-Si:H LF compressive 310 to 450 MPa, a-SiC refractive index 2.5 to 2.6, recommended SiNx 600 W alternating HF/LF at 800 mTorr with N2/NH3/SiH4 1200/30/60 sccm (fetch-tool summary); F. E. Rasmussen, B. Geilman, M. Heschel, O. Hansen and A. M. Jorgensen (Mikroelektronik Centret, DTU, and Hymite), ‘Development and characterisation of KOH resistant PECVD silicon nitride for microsystems applications’, ECS meeting abstract: all films at 300 C, 380 kHz, 13.56 MHz and mixed excitation, hydrogen content strongly affecting stress, index and etch rate, nitrogen-rich films with N-H peaks at 3350 and 1150 cm-1 and Si-H near 2200 cm-1 mostly undetectable, index rising with SiH4/NH3 ratio, 1.86 to 1.93 for LF and 1.93 to 2.25 for mixed frequency, BOE rate falling as index rises, stress less compressive with higher power and pressure, KOH etch rate rising with power and pressure, LF and mixed films more KOH resistant, power the main factor in rate (fetch-tool summary).

Ques: Is the low frequency 380 kHz or 400 kHz?

380 $\mathrm{kHz}$ is more likely correct. Iliescu et al. give the low frequency as 380 $\mathrm{kHz}$ in the body of their review but as 400 $\mathrm{kHz}$ in the abstract. Harvard CNS specifies a 380 $\mathrm{kHz}$ low-frequency supply alongside its 13.56 $\mathrm{MHz}$ supply. Rasmussen et al. also used 380 $\mathrm{kHz}$ and 13.56 $\mathrm{MHz}$ with mixed excitation. Two independent sources and the body of the review therefore agree on 380 $\mathrm{kHz}$, so the 400 $\mathrm{kHz}$ in the abstract is most likely a rounded value (my reading). The practical point is unaffected: both describe a frequency low enough to give compressive films, 600 to 700 $\mathrm{MPa}$, against tensile films at 13.56 $\mathrm{MHz}$ (Iliescu et al.). Reference: C. Iliescu, M. Avram, B. Chen, A. Popescu, V. Dumitrescu, D. P. Poenar, A. Sterian, D. Vrtacnik, S. Amon and P. Sterian, ‘Residual stress in thin films PECVD depositions: a review’, Journal of Optoelectronics and Advanced Materials 13(4), 387-394 (2011): SiNx at 20 to 80 W with HF 13.56 MHz giving 120 to 150 MPa tensile and LF 380 kHz (400 kHz in the abstract) giving 600 to 700 MPa compressive, alternating or simultaneous HF and LF reaching zero stress, HF power 100 to 600 W lowering tensile stress from 100 to 150 MPa to 4 to 60 MPa and raising rate from about 100 to 250 to 320 nm/min, about 4 MPa at 600 W, 700 to 1100 mTorr studied with about 900 mTorr optimal and higher stress at 1100 mTorr, higher SiH4 fraction raising rate and lowering stress, a-Si:H LF compressive 310 to 450 MPa, a-SiC refractive index 2.5 to 2.6, recommended SiNx 600 W alternating HF/LF at 800 mTorr with N2/NH3/SiH4 1200/30/60 sccm (fetch-tool summary); Harvard University Center for Nanoscale Systems, ‘STS PECVD’ tool page (NNCI): HF 13.56 MHz top electrode and LF 380 kHz supply, substrate from room temperature to 300 C (recipes at 300 C platen, 250 C lid), gases SiH4, NH3, N2O, N2, O2, Ar, PH3/Ar and B2H6/H2, films SiO2, Si3N4, low-stress Si3N4, a-Si and doped versions, rates SiO2 79 (LF) and 56 (HF) nm/min, Si3N4 45 (LF) and 10 (HF) nm/min, indices SiO2 1.48 (LF) and 1.485 (HF), Si3N4 1.977 (LF) and 2.01 (HF), mixed-frequency stress tailoring by the HF/LF power-on ratio (fetch-tool summary); F. E. Rasmussen, B. Geilman, M. Heschel, O. Hansen and A. M. Jorgensen (Mikroelektronik Centret, DTU, and Hymite), ‘Development and characterisation of KOH resistant PECVD silicon nitride for microsystems applications’, ECS meeting abstract: all films at 300 C, 380 kHz, 13.56 MHz and mixed excitation, hydrogen content strongly affecting stress, index and etch rate, nitrogen-rich films with N-H peaks at 3350 and 1150 cm-1 and Si-H near 2200 cm-1 mostly undetectable, index rising with SiH4/NH3 ratio, 1.86 to 1.93 for LF and 1.93 to 2.25 for mixed frequency, BOE rate falling as index rises, stress less compressive with higher power and pressure, KOH etch rate rising with power and pressure, LF and mixed films more KOH resistant, power the main factor in rate (fetch-tool summary).

Ques: How do pressure and plasma design affect PECVD films?

They set stress, uniformity and step coverage. Iliescu et al. studied SiN${}_x$ from 700 to 1100 $\mathrm{mTorr}$ at 600 $\mathrm{W}$. About 900 $\mathrm{mTorr}$ was optimal for a stable plasma and low stress, and stress was higher at 1100 $\mathrm{mTorr}$. Their recommended recipe is 600 $\mathrm{W}$ alternating HF and LF at 800 $\mathrm{mTorr}$, with N${}_2$/NH${}_3$/SiH${}_4$ at 1200/30/60 $\mathrm{sccm}$. Rasmussen et al. found that higher pressure and power make films less compressive but raise the KOH etch rate, a trade-off between stress and mask quality. Georgia Tech notes that remote-plasma PECVD near 0.1 $\mathrm{Torr}$ deposits directionally with poorer step coverage than older systems above 1 $\mathrm{Torr}$. In my reading, conformal PECVD needs the higher pressures. Reference: C. Iliescu, M. Avram, B. Chen, A. Popescu, V. Dumitrescu, D. P. Poenar, A. Sterian, D. Vrtacnik, S. Amon and P. Sterian, ‘Residual stress in thin films PECVD depositions: a review’, Journal of Optoelectronics and Advanced Materials 13(4), 387-394 (2011): SiNx at 20 to 80 W with HF 13.56 MHz giving 120 to 150 MPa tensile and LF 380 kHz (400 kHz in the abstract) giving 600 to 700 MPa compressive, alternating or simultaneous HF and LF reaching zero stress, HF power 100 to 600 W lowering tensile stress from 100 to 150 MPa to 4 to 60 MPa and raising rate from about 100 to 250 to 320 nm/min, about 4 MPa at 600 W, 700 to 1100 mTorr studied with about 900 mTorr optimal and higher stress at 1100 mTorr, higher SiH4 fraction raising rate and lowering stress, a-Si:H LF compressive 310 to 450 MPa, a-SiC refractive index 2.5 to 2.6, recommended SiNx 600 W alternating HF/LF at 800 mTorr with N2/NH3/SiH4 1200/30/60 sccm (fetch-tool summary); F. E. Rasmussen, B. Geilman, M. Heschel, O. Hansen and A. M. Jorgensen (Mikroelektronik Centret, DTU, and Hymite), ‘Development and characterisation of KOH resistant PECVD silicon nitride for microsystems applications’, ECS meeting abstract: all films at 300 C, 380 kHz, 13.56 MHz and mixed excitation, hydrogen content strongly affecting stress, index and etch rate, nitrogen-rich films with N-H peaks at 3350 and 1150 cm-1 and Si-H near 2200 cm-1 mostly undetectable, index rising with SiH4/NH3 ratio, 1.86 to 1.93 for LF and 1.93 to 2.25 for mixed frequency, BOE rate falling as index rises, stress less compressive with higher power and pressure, KOH etch rate rising with power and pressure, LF and mixed films more KOH resistant, power the main factor in rate (fetch-tool summary); W. A. Doolittle, ECE 6450 Lectures 13 and 14 ‘CVD and Epitaxy’, Georgia Institute of Technology: remote-plasma PECVD at about 0.1 torr giving directional deposition and poorer step coverage, older systems above 1 torr giving better step coverage, film stress limiting thickness (fetch-tool summary).

References

1. C. Iliescu, M. Avram, B. Chen, A. Popescu, V. Dumitrescu, D. P. Poenar, A. Sterian, D. Vrtacnik, S. Amon and P. Sterian, “Residual stress in thin films PECVD depositions: a review,” Journal of Optoelectronics and Advanced Materials, vol. 13, no. 4, pp. 387-394, 2011. Available: https://joam.inoe.ro/articles/residual-stress-in-thin-films-pecvd-depositions-a-review/fulltext
2. Harvard University Center for Nanoscale Systems, “STS PECVD,” NNCI tool page. Available: https://nnci.net/tools/sts-pecvd
3. F. E. Rasmussen, B. Geilman, M. Heschel, O. Hansen and A. M. Jorgensen, “Development and characterisation of KOH resistant PECVD silicon nitride for microsystems applications,” ECS meeting abstract 367. Available: https://www.electrochem.org/dl/ma/203/pdfs/0367.pdf
4. KTH Royal Institute of Technology, “Lecture 8: CVD and ALD,” course slides. Available: https://www.kth.se/social/upload/510bc6e8f276547a4fe3c222/Lecture%208.pdf
5. W. A. Doolittle, “CVD and epitaxy,” ECE 6450 Lectures 13 and 14, Georgia Institute of Technology. Available: https://alan.ece.gatech.edu/ECE6450/Lectures/ECE6450L13and14-CVD%20and%20Epitaxy.pdf
6. LPCVD Furnace page of this vault.
7. Wafer Curvature Stress Gauge page of this vault.
8. Spectroscopic Ellipsometer page of this vault.

Related pages in this vault: LPCVD Furnace, Wafer Curvature Stress Gauge, Spectroscopic Ellipsometer, Langmuir Probe, Optical Emission Spectroscopy Endpoint (OES), Atomic Layer Deposition (ALD) Reactor.