Plasma Enhanced Atomic Layer Deposition of SiN-AlN Nanolaminates ...
Supporting information for
Plasma Enhanced Atomic Layer Deposition of SiN-AlN Nanolaminates for Ultra Low Wet Etch Rates in Hydrofluoric Acid Yongmin Kim†,#, J Provine†,#*, Stephen P. Walch†, Joonsuk Park§, Witchukorn Phuthong§, Anup L. Dadlani*, Hyo-Jin Kim†, Peter Schindler†, Kihyun Kim∥ and Fritz B. Prinz†,§
†
Department of Mechanical Engineering,
*Department of Chemistry, §
Department of Materials Science and Engineering, Stanford University, Stanford, California 94305,
United States ∥
Manufacturing Technology Center, Samsung Electronics, Suwon, Gyeonggi-Do, South Korea
Keywords: plasma enhanced atomic layer deposition (PEALD), silicon nitride, aluminum incorporation, nanolaminate, thin film deposition, wet etch, composite films
#These authors contributed equally to this work *Corresponding author: [email protected]
S1
Reflectance (a.u.)
Reflectance (a.u.)
AlN:SiN=1:19 AlN:SiN=1:24 AlN:SiN=1:49 AlN:SiN=1:99
2300
3000
2200
2100
2000
-1
Wavenumber (cm )
2500
2000
1500
1000
-1
Wavenumber (cm )
Figure S1. FTIR spectra of PEALD intermixed SiN-AlN films with different AlN:SiN ratios. The inset figure shows a magnified plot of the spectra in the spectral range, 2300-2000 cm-1, associated with the SiH stretching mode. Figure S1 shows FTIR spectra for various SiN-AlN nanolaminates. The spectra represent the surface bonding states due to the use of an Attenuate Total Reflection (ATR) surface probe during the scans. At differing concentrations of Al in the nanolaminates, FTIR spectra showed an Si-H stretching mode peak at ~2180 cm-1 with similar peak intensity regardless of the amount of aluminum, from which we can conclude the H content of the films are similar. No N-H stretching mode peak at ~3345 cm-1 was observed (Figure S1). With increasing Al content the Si-O-Si stretching frequency decreases and can possibly be used as a marker for determining Al concentration. The decrease in stretching frequency (lower wavenumber) observed in FTIR spectrum likely indicates a weaker Si-O bond.
S2
Si-O
Si-N
(b)
Si-C
103.3 102.3
Intensity (cps)
100.1
Measured Fitted (Si-N) Fitted (Si-C) Fitted (Si-O)
106
104
102
100
98
Si-N
Si-C
103.3
101.5
99.2
96
Intensity (cps)
104
102
100
106
104
(d)
98
96
Si2p Binding energy (eV)
Intensity (cps)
(e)
Si-N
Si-C
101.3
100.0
99.5
102
100
98
96
Si2p Binding energy (eV)
Measured Fitted (Si-N) Fitted (Si-C) Fitted (Si-O)
106
Si-C
102.0
Si-O
Si-N
Si-C
103.3
101.9
99.7
Intensity (cps)
Si-O
Si-N
103.3
Measured Fitted (Si-N) Fitted (Si-C) Fitted (Si-O)
Si2p Binding energy (eV)
(c)
Si-O
Intensity (cps)
(a)
Measured Fitted (Si-O) Fitted (Si-C) Fitted (Si-N)
106
104
102
100
98
96
Si2p Binding energy (eV)
Measured Fitted (Si-N) Fitted (Si-C)
106
104
102
100
98
96
Si2p Binding energy (eV)
Figure S2. Si2p XPS spectra of (a) AlN:SiN=1:19, (b) AlN:SiN=1:24, (c) AlN:SiN=1:49, (d) AlN:SiN=1:99, and (e) SiN. Peaks at 103.3 eV are assigned to Si-O and peaks at 99.1-100.1 eV are assigned to Si-C and silicide.1 S3
The contribution of Al energy loss peaks at 99.8 eV and 104.3 eV to the binding energy (BE) shift of Si2p is negligible. The Si2p BE range of interest is 99-105 eV. According to the XPS handbook,1 two Al energy loss peaks at 99.8 and 104.3 eV possibly overlap with the Si2p BE. These Al energy loss peaks, however, are negligible compared to Si2p de-convoluted peaks at 100.1 and 103.3 eV. Firstly, the aforementioned Al energy loss peaks are relatively smaller than Al2p peak (75.3 eV), i.e., peak heights are only 4% (99.8 eV) and 9% (104.3 eV) with respect to Al2p. Furthermore, the Al2p peak height is largely smaller than Si2p; in case of AlN:SiN=1:19 with the highest Al content compared to other samples, the peak ratio of Al2p to Si2p is only ~30%. Therefore, the heights of Al energy loss peaks are only 1% (99.8 eV) and 3% (104.3 eV) with respect to Si2p. In addition, from deconvolution, we confirmed the contribution of Al energy loss peak at 104.3 eV to the BE shifts of Si2p is less likely and another Al energy loss peak at 99.8 eV is relatively smaller than Si-C peak at 100.1 eV. Supposing that there is some contribution, since the main focus of deconvolution of Si2p peak is to separate the contribution of Si-N and investigate the shift of this peak, we can conclude Al energy loss peaks have nothing to do with Si-N peak shift.
S4
(a)
(b) as-deposited 0.1 min etching 0.2 min etching 0.3 min etching
692
690
688
as-deposited 0.1 min etching
686
Binding energy (eV)
684
682
692
690
688
686
684
682
Binding energy (eV)
Figure S3. XPS depth profiles for fluorine (F 1s, 685 eV) in (a) the SiN-AlN nanolaminate film (AlN:SiN = 1:49, Al/Si atomic ratio = 0.18) and (b) the pure SiN. The film depth profiling for atomic composition was carried out by X-ray photoelectron spectroscopy (XPS, PHI VersaProbe Scanning XPS Microscope), utilizing Ar+ ion sputtering (Sputter 2 kV, 1 uA, 2x2 µm2, Neutralizer 10 eV).
Figure S3 provides the XPS depth profiles for a representative AlN-SiN nanolaminate and a pure SiN PEALD film after the WER test has been performed. Fluorine incorporation is more clearly present in the nanolaminate, which is attributable to the formation of Al-F bonds which are substantially more difficult to dissociate in aqueous HF than Si-F bonds as shown by our ab initio calculations
S5
7.43 (1/nm)
Figure S4. The FFT image of the TEM image of the pure AlN, presented in Figure 4a, demonstrating the formation of hexagonal AlN polycrystalline, as revealed by the brightest sharp diffraction ring corresponding to (100) plane with d(100) of 0.26 nm (Hexagonal AlN, ICDD reference code: 00-025-1133), superimposed on Si {200} spot.2
S6
Figure S5. AFM topography images of (a) SiN, (b) AlN:SiN = 1:49, (c) AlN:SiN = 1:24, (d) AlN:SiN = 1:19, and (e) AlN. Window sizes are 1 µm x 1 µm. The detailed information for the samples, such as thickness and chemical composition, is listed in Table 2. Figure S5 provides the AFM surface scans of PEALD films of pure SiN, pure AlN, and various nanolaminate compositions. The nanolaminate films are slightly rougher than the pure SiN (2 Å RMS roughness compared to 1 Å for pure SiN), and the pure AlN is significantly rougher (RMS roughness of 4 Å). This increase in film roughness is further evidence of the crystalline component of the pure AlN film that is not present in the amorphous SiN film and nanolaminates.
S7
References (1)
(2)
S8
Briggs, D. Handbook of X-Ray Photoelectron Spectroscopy C. D. Wanger, W. M. Riggs, L. E. Davis, J. F. Moulder and G. E.Muilenberg Perkin-Elmer Corp., Physical Electronics Division, Eden Prairie, Minnesota, USA, 1979. 190 Pp. $195. Surf. Interface Anal. 1981, 3 (4), v – v. Ozgit-Akgun, C.; Kayaci, F.; Donmez, I.; Uyar, T.; Biyikli, N. Template-Based Synthesis of Aluminum Nitride Hollow Nanofibers Via Plasma-Enhanced Atomic Layer Deposition. J. Am. Ceram. Soc. 2013, 96 (3), 916–922.
Plasma Enhanced Atomic Layer Deposition of SiN-AlN Nanolaminates for Ultra Low Wet Etch Rates in Hydrofluoric Acid Yongmin Kim†,#, J Provine†,#*, Stephen P. Walch†, Joonsuk Park§, Witchukorn Phuthong§, Anup L. Dadlani*, Hyo-Jin Kim†, Peter Schindler†, Kihyun Kim∥ and Fritz B. Prinz†,§
†
Department of Mechanical Engineering,
*Department of Chemistry, §
Department of Materials Science and Engineering, Stanford University, Stanford, California 94305,
United States ∥
Manufacturing Technology Center, Samsung Electronics, Suwon, Gyeonggi-Do, South Korea
Keywords: plasma enhanced atomic layer deposition (PEALD), silicon nitride, aluminum incorporation, nanolaminate, thin film deposition, wet etch, composite films
#These authors contributed equally to this work *Corresponding author: [email protected]
S1
Reflectance (a.u.)
Reflectance (a.u.)
AlN:SiN=1:19 AlN:SiN=1:24 AlN:SiN=1:49 AlN:SiN=1:99
2300
3000
2200
2100
2000
-1
Wavenumber (cm )
2500
2000
1500
1000
-1
Wavenumber (cm )
Figure S1. FTIR spectra of PEALD intermixed SiN-AlN films with different AlN:SiN ratios. The inset figure shows a magnified plot of the spectra in the spectral range, 2300-2000 cm-1, associated with the SiH stretching mode. Figure S1 shows FTIR spectra for various SiN-AlN nanolaminates. The spectra represent the surface bonding states due to the use of an Attenuate Total Reflection (ATR) surface probe during the scans. At differing concentrations of Al in the nanolaminates, FTIR spectra showed an Si-H stretching mode peak at ~2180 cm-1 with similar peak intensity regardless of the amount of aluminum, from which we can conclude the H content of the films are similar. No N-H stretching mode peak at ~3345 cm-1 was observed (Figure S1). With increasing Al content the Si-O-Si stretching frequency decreases and can possibly be used as a marker for determining Al concentration. The decrease in stretching frequency (lower wavenumber) observed in FTIR spectrum likely indicates a weaker Si-O bond.
S2
Si-O
Si-N
(b)
Si-C
103.3 102.3
Intensity (cps)
100.1
Measured Fitted (Si-N) Fitted (Si-C) Fitted (Si-O)
106
104
102
100
98
Si-N
Si-C
103.3
101.5
99.2
96
Intensity (cps)
104
102
100
106
104
(d)
98
96
Si2p Binding energy (eV)
Intensity (cps)
(e)
Si-N
Si-C
101.3
100.0
99.5
102
100
98
96
Si2p Binding energy (eV)
Measured Fitted (Si-N) Fitted (Si-C) Fitted (Si-O)
106
Si-C
102.0
Si-O
Si-N
Si-C
103.3
101.9
99.7
Intensity (cps)
Si-O
Si-N
103.3
Measured Fitted (Si-N) Fitted (Si-C) Fitted (Si-O)
Si2p Binding energy (eV)
(c)
Si-O
Intensity (cps)
(a)
Measured Fitted (Si-O) Fitted (Si-C) Fitted (Si-N)
106
104
102
100
98
96
Si2p Binding energy (eV)
Measured Fitted (Si-N) Fitted (Si-C)
106
104
102
100
98
96
Si2p Binding energy (eV)
Figure S2. Si2p XPS spectra of (a) AlN:SiN=1:19, (b) AlN:SiN=1:24, (c) AlN:SiN=1:49, (d) AlN:SiN=1:99, and (e) SiN. Peaks at 103.3 eV are assigned to Si-O and peaks at 99.1-100.1 eV are assigned to Si-C and silicide.1 S3
The contribution of Al energy loss peaks at 99.8 eV and 104.3 eV to the binding energy (BE) shift of Si2p is negligible. The Si2p BE range of interest is 99-105 eV. According to the XPS handbook,1 two Al energy loss peaks at 99.8 and 104.3 eV possibly overlap with the Si2p BE. These Al energy loss peaks, however, are negligible compared to Si2p de-convoluted peaks at 100.1 and 103.3 eV. Firstly, the aforementioned Al energy loss peaks are relatively smaller than Al2p peak (75.3 eV), i.e., peak heights are only 4% (99.8 eV) and 9% (104.3 eV) with respect to Al2p. Furthermore, the Al2p peak height is largely smaller than Si2p; in case of AlN:SiN=1:19 with the highest Al content compared to other samples, the peak ratio of Al2p to Si2p is only ~30%. Therefore, the heights of Al energy loss peaks are only 1% (99.8 eV) and 3% (104.3 eV) with respect to Si2p. In addition, from deconvolution, we confirmed the contribution of Al energy loss peak at 104.3 eV to the BE shifts of Si2p is less likely and another Al energy loss peak at 99.8 eV is relatively smaller than Si-C peak at 100.1 eV. Supposing that there is some contribution, since the main focus of deconvolution of Si2p peak is to separate the contribution of Si-N and investigate the shift of this peak, we can conclude Al energy loss peaks have nothing to do with Si-N peak shift.
S4
(a)
(b) as-deposited 0.1 min etching 0.2 min etching 0.3 min etching
692
690
688
as-deposited 0.1 min etching
686
Binding energy (eV)
684
682
692
690
688
686
684
682
Binding energy (eV)
Figure S3. XPS depth profiles for fluorine (F 1s, 685 eV) in (a) the SiN-AlN nanolaminate film (AlN:SiN = 1:49, Al/Si atomic ratio = 0.18) and (b) the pure SiN. The film depth profiling for atomic composition was carried out by X-ray photoelectron spectroscopy (XPS, PHI VersaProbe Scanning XPS Microscope), utilizing Ar+ ion sputtering (Sputter 2 kV, 1 uA, 2x2 µm2, Neutralizer 10 eV).
Figure S3 provides the XPS depth profiles for a representative AlN-SiN nanolaminate and a pure SiN PEALD film after the WER test has been performed. Fluorine incorporation is more clearly present in the nanolaminate, which is attributable to the formation of Al-F bonds which are substantially more difficult to dissociate in aqueous HF than Si-F bonds as shown by our ab initio calculations
S5
7.43 (1/nm)
Figure S4. The FFT image of the TEM image of the pure AlN, presented in Figure 4a, demonstrating the formation of hexagonal AlN polycrystalline, as revealed by the brightest sharp diffraction ring corresponding to (100) plane with d(100) of 0.26 nm (Hexagonal AlN, ICDD reference code: 00-025-1133), superimposed on Si {200} spot.2
S6
Figure S5. AFM topography images of (a) SiN, (b) AlN:SiN = 1:49, (c) AlN:SiN = 1:24, (d) AlN:SiN = 1:19, and (e) AlN. Window sizes are 1 µm x 1 µm. The detailed information for the samples, such as thickness and chemical composition, is listed in Table 2. Figure S5 provides the AFM surface scans of PEALD films of pure SiN, pure AlN, and various nanolaminate compositions. The nanolaminate films are slightly rougher than the pure SiN (2 Å RMS roughness compared to 1 Å for pure SiN), and the pure AlN is significantly rougher (RMS roughness of 4 Å). This increase in film roughness is further evidence of the crystalline component of the pure AlN film that is not present in the amorphous SiN film and nanolaminates.
S7
References (1)
(2)
S8
Briggs, D. Handbook of X-Ray Photoelectron Spectroscopy C. D. Wanger, W. M. Riggs, L. E. Davis, J. F. Moulder and G. E.Muilenberg Perkin-Elmer Corp., Physical Electronics Division, Eden Prairie, Minnesota, USA, 1979. 190 Pp. $195. Surf. Interface Anal. 1981, 3 (4), v – v. Ozgit-Akgun, C.; Kayaci, F.; Donmez, I.; Uyar, T.; Biyikli, N. Template-Based Synthesis of Aluminum Nitride Hollow Nanofibers Via Plasma-Enhanced Atomic Layer Deposition. J. Am. Ceram. Soc. 2013, 96 (3), 916–922.
