Full Paper
Deposition of ‘‘Polysiloxane’’ Thin Films Containing Silver Particles by an RF Asymmetrical Discharge Bernard Despax,* Patrice Raynaud
The aim of the present work is to produce silver-containing SiCxOyHz thin films by using simultaneous sputtering of silver and plasma polymerization in an HMDSO plasma. The ratio of plasma polymerization to sputtering was adjusted by using a pulsed flow rate of the hexamethyldisiloxane precursor, which permitted an accurate control of the co-deposition process. The presence of silver in the gas phase was detected by optical emission spectroscopy. The silver volume fraction was controlled over a wide range by superimposing physical sputtering onto the polymerization process taking place simultaneously within a very low and narrow range of HMDSO plasma pressure. The polymer matrix structures were analyzed by several techniques. FTIR spectra revealed the presence of Si–O–Si, Si–C–Si, SiCH3 and C – C groups whose amount varied with the silver content. TEM analysis showed Ag nanoparticles with sizes varying from 4 to 70 nm depending on the silver volume fraction. Introduction Composite thin films consisting of granular mixtures of insulating and metallic particles have been extensively studied during the last decades. A convenient method for preparation of metal-containing plasma polymerized hydrocarbon films has been described in several previous papers[1–4] and referenced therein. This deposition technique allows one to obtain very thin films ranging from 10 to 1 000 nm. The modifications of the electrical and optical properties of such metal-containing materials have been studied using percolation models and/or effectivemedium theories.[5–11] Moreover, laser irradiation can induce specific modification of the film properties leading to novel applications such as the laser writing of conducting path or the optical storage.[12] In the case of a nonB. Despax, P. Raynaud Laboratoire de Ge´nie Electrique Universite´ Paul Sabatier (UMR 5003), 118 route de Narbonne, F-31062 Toulouse cedex 09, France E-mail:
[email protected] Plasma Process. Polym. 2007, 4, 127–134 ß 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
reactive metal, the carbon matrices in which the metal clusters are embedded are easily scratched and their ageing under oxygen or humid atmosphere can lead to film damages with delamination and appearance of cracks. In fact, the presence of numerous carbon dangling bonds favors the matrix oxidation which renders the polymer network more or less unstable.[13] Then, this material is not really relevant for the above applications. In order to avoid these problems, organosiloxane precursors such as hexamethyldisiloxane (HMDSO) could be used for the plasma deposition process in order to produce an oxisilicon carbide network which is harder and less sensitive to the atmospheric oxidation. Moreover, the scratch resistance and the chemical stability of the matrix in these composites make this material promising for other possible applications, such as anticorrosion and antifouling layers to prevent the biofilm formation. This sort of silver-containing polysiloxane was first obtained by Salz et al.[14] Their deposition process relied on the simultaneous metal evaporation and plasma polymerization. The metal evaporation was performed outside the plasma chamber. The
DOI: 10.1002/ppap.200600083
127
B. Despax, P. Raynaud
difference of pressure between the two locations ensured an argon flow, leading to silver atom injection into the plasma chamber through a pipe. Under our conditions, the deposition process involves simultaneous silver sputtering and HMDSO plasma polymerization in an asymmetrical radiofrequency discharge. We will pay particular attention to the variation of silver volume fraction in the film as a function of the operating conditions. Moreover, the variation of physicochemical properties of the SiCOH matrix and the silver-containing SiCOH matrix will be carefully analyzed.
Experimental Part The 13.56 MHz RF plasma deposition system (Figure 1) consists of a 30 cm diameter stainless steel chamber fitted with stainless steel end plates with observation and pumping ports. All the walls of the reactor are grounded. The RF powered electrode is a 10 cm diameter, 0.6 cm thick silver disc capacitively coupled to a RF generator by means of a LC matching network. The RF electrode is mounted 3.5 cm above a 12 cm diameter stainless steel electrode which is electrically grounded. An input power of 100 W, corresponding to a self-bias voltage of 800 V was applied to the silver target through the impedancematching network, while the other electrode held the samples (silicon, glass). Hexamethyldisiloxane was obtained from Sigma Aldrich with a purity >99.5% and 99.9995% pure argon from Air Liquid. The argon flow rate was 3.1 sccm and its partial pressure, measured using an MKS baratron gauge, was 5.32 Pa. The HDMSO amount was monitored by a pulsation of the HMDSO mass flow rate with a period (T ¼ TON þ TOFF) of 5 s and a variable duty cycle (TON) using a OMICRON mass flow controller equipped with an AGILENT impulse generator. The maximum flow rate used was 0.4 sccm, the average flow rates can be estimated as 0.4 TON/T. HDMSO and argon flow were mixed in a buffer chamber before being introduced into the plasma chamber by means of a ring gas injector located in the upper part of the reactor at the periphery of the RF electrode. A weak pressure fluctuation was observed during the HMDSO injection in the plasma. In this paper, the difference between the maximum of plasma pressure with HMDSO and the pressure without HMDSO will be called HMDSO pressure. Let us
remark that this procedure seems to be the only way to get simultaneous silver sputtering and plasma polymerization with very low controllable flow rates. The reactor temperature was maintained at 40 8C.
Film Deposition The method for depositing metal-containing hydrocarbon films has already been described in several previous papers.[1–3] In our case, the plasma decomposition of HMDSO produced SiCOH fragments which polymerized on the walls. Owing to the asymmetrical design of RF discharge, the silver RF-powered electrode was subjected to an energetic argon ion bombardment. Consequently, silver and plasma polymer deposits were sputtered from the powered electrode. Simultaneous deposition of polymer and silver on the samples resulted in the growth of an Ag-composite in which the silver volume fraction was controlled by adjusting the argon and HMDSO flow rates. With a power of 100 W, the deposition rate varied from 15 to 12 nm min1 with HMDSO corresponding to the pressure between 0.27 and 1.06 Pa. Note that the samples were initially protected by a shutter which was removed when the plasma steady state conditions were reached. The plasma optical emission was measured in the range 300–650 nm through a glass window with an optical fiber located in front of the gas discharge. The optical fiber was mounted on a monochromator, closely aligned with the entrance slit. A 0.25 m Chromex monochromator was used with a 600 or a 150 lines mm1 grating coupled with a Princeton Instrument OMA with a silicon-intensified detector in order to record optical spectra. SiCxOyHz with or without silver films were characterized by Fourier transform infrared (FTIR) spectroscopy at room temperature. More quantitative information about the film composition was deduced from Rutherford backscattering spectrometry (RBS). The silicon, oxygen, carbon and silver atomic composition were extracted from RBS spectra under a 4Heþ beam, Ec ¼ 2.2 MeV, I ¼ 10 nA, Area ¼ 1 mm2. Electron recoil detection analysis (ERDA) was used to measure hydrogen composition of deposits under a 4Heþ beam, 2.3 MeV, 2 nA, 3 mm2. Scanning electron microscopy (SEM) was carried out by a JEOL JSM 6700F microscope on films deposited on silicon substrate to obtain a general overview of the bulk cross section microstructure. Two kinds of substrates were used: 25 36 mm2 glass plate for the density determination, 10 10 mm2 intrinsic silicon plate for FTIR, SEM, RBS and ERDA. Transmission electron micrographs (TEM) were obtained with a JEOL JEM 1011 microscope. TEM samples with different silver volume fraction were deposited during 50 s on silicon grids with silicon nitride support. The silver volume fraction ( fv) in each film was calculated from its density (r) by using the relation:
fv ¼
Figure 1. Experimental setup.
128
Plasma Process. Polym. 2007, 4, 127–134 ß 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
rx rCSi rAg rCSi
(1)
where rAg and rCSi are the bulk silver and polymer densities. rx was determined by weighing the sample before and after deposition, and by calculating the film volume (thickness was measured by profilometry). The limiting values, i.e., rAg and rCSi,
DOI: 10.1002/ppap.200600083
Deposition of ‘‘Polysiloxane’’ Thin Films Containing Silver Particles . . .
were respectively measured on pure silver and pure polymer films by the same method in order to account for the systematic instrumental error in the measurements. The overall precision of r ˚ thick films. is estimated to be 0.05 for 3 000 A
Results and Discussion Investigation of the Plasma Phase and the Films’ Metal Volume Fractions by Emission Spectroscopy Plasma was investigated by optical emission spectroscopy for different silver-SiCxOyHz composite films which were conditioned by the hexamethyldisiloxane-argon mixtures. As the pulsed gas injection of HMDSO led to a fluctuation of the plasma light emission, the acquisition time for the spectra (10 s) was kept higher than the HMDSO injection period (5 s) in order to obtain average values of the peak intensities. The principal features of the emission spectra in the range between 300 to 650 nm are depicted in Figure 2a. In order to control the process deposition and particularly the silver sputtering against the plasma polymerization, we focused our attention in the range between 480 and 550 nm (Figure 2b). The silver lines (546.55 and 520.91 nm), the Hb transition (486 nm) and the Ar transition (549.59 nm) were easily detected and identified (Figure 2). The Ag lines and Hb transition are assumed to be related to the silver atoms presence and to the HMDSO dissociation in the plasma phase, respectively. Spectra were recorded as a function of the partial pressure of hexamethyldisiloxane. Figure 3 compares Ar, Ag and Hb peak intensities as a function of the HMDSO partial pressure ranging from 0.2 to 0.47 Pa. Silver lines exhibit a drop with increasing the hexamethyldisiloxane partial pressure. In contrast, the Hb emission line increases up to 0.38 Pa, decreasing slowly thereafter. Argon emission decrease is mainly related to the increase of HMDSO amount, which modifies the gas electron interaction. The cathode coverage appears at 0.21 Pa. The argon plasma modification coupled with the partial cathode coverage due to the SiCxOy deposition explains the silver emission decrease. Above 0.66 Pa of HDMSO at 100 W, the silver lines disappear completely and, consequently, the deposited layers do not contain any silver. Figure 4 shows the variation of the silver volume fraction with the HMDSO partial pressure from 0.25 to 0.47 Pascal. This region corresponds to a transition zone from metal-rich to polymer-rich composites. This transition zone in fact represents a critical zone where the balance between the partial cathode coverage by SiCxOy deposit and the target sputtering takes time (30 min to reach the steady state conditions). Above 0.4 Pa, we observe an abrupt variation of the Ag volume fraction due to the complete coverage of target center (Figure 4). Indeed, only a very small part of the silver electrode at the Plasma Process. Polym. 2007, 4, 127–134 ß 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Figure 2. Typical emission lines in an HMDSO-argon plasma obtained during the deposition of Ag-containing polysiloxane films: a) in the 300–670 nm range, and b) in the 480–553 nm range.
Figure 3. Evolution of the main emission lines as a function of the HDMSO partial pressure.
www.plasma-polymers.org
129
B. Despax, P. Raynaud
Figure 4. Variation of the silver volume fraction as a function of the HMDSO partial pressure.
periphery close to the shield remains uncovered. All the process and diagnostics are realized when the steadystate conditions are reached. After the shutter removal, the deposition process takes place on the substrate and the plasma composition is monitored by keeping the ratio between the specific lines constant. The evolution of Ag (546.55 nm)/Ar (549.59 nm) line ratio gives an image of the amount of Ag in the plasma (Figure 4). Then, the Ag/Ar ratio can be related to the Ag volume fraction in the deposit (Figure 5). This kind of chart constitutes useful data to obtain reproducible results.
Physicochemical Properties SiCxOyH Matrix without Silver Films free of silver were obtained at HDMSO partial pressure of 1.07 Pa under various power conditions. Structural characterization of amorphous films is particularly complex. FTIR spectroscopy has already given lots of information concerning bonding arrangements in the SiCxOyH films. The assignment of the peaks as found in the literature[15–20] is summarized in Table 1. The band between 1 034 and 990 cm1 is classically assigned to the Si –O –Si asymmetric stretching in silicon suboxide – Si –O –Si – environment, such as – – XyO3y (with X ¼ C or Si and 1 y 3). Actually, the film spectra are different when increasing the power (Figure 6, spectra normalized in respect to the thickness). At a low power (5 W), the film spectrum exhibits a main band at 1 034 cm1 (Table 1, Figure 6), whereas the film obtained at a high power (100 W) shows a broad band at around 1 000 cm1 (Table 1, Figure 6). Although the presence of a –(Si –O –Si) – bridge is on the whole undeniable, the occurrence of a few Si –CH2 – Si bridges (bending mode) related to the peaks at
130
Plasma Process. Polym. 2007, 4, 127–134 ß 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Figure 5. Variation of the silver volume fraction as a function of the ratio between the silver (546.54 nm) and argon (549.58 nm) emission lines intensities.
1 358 cm1 indicates also a probable contribution of Si –CH2 –Si bridge (wagging mode v)[12–14] at around 1 020 cm1 in the band between 1 034 and 1 000 cm1. The Si –CH2 –Si bridge (1 358 cm1) is better brought out in the films obtained at 20 and 40 W where a CH2 band as a shoulder at 2 876 cm1appears in the IR spectra. Another feature in the film spectra is the absence of Si –H bond at low power (5 W), whereas for films obtained at higher powers (>10 W, Figure 6, Table 1), the Si –H groups appear. Moreover, when increasing power up to 100 W, the CH3 band decreases until complete disappearance, while the aromatic C – – C bands (1 539 cm1) appear. The atomic ratios of C/Si, O/Si and H/Si obtained in RBS and ERDA are plotted in Figure 7 as a function of the RF power. Under 5 W, the film composition is SiO0.7C1.9 H2.4. Since the HMDSO elemental composition is SiO0.5C3H9, it can be remarked that at this low power the O/Si ratio (0.8) is greater than the one in HMDSO (0.5) whereas the C/Si ratio (1.9) and H/Si ratio (2.35) are lower. This specific composition with a strong decrease in hydrogen concentration indicates the creation of new bonds such as Si –O, Si –C, C – – C and Si –Si in the film network in order to satisfy the valence bonding of each atom. This statement seems to be in agreement with infrared measurements, which detected some Si –CH2 –Si at 1 358 cm1. With increasing plasma power from 5 W to 100 W, the carbon content remains nearly constant, whereas the oxygen content decreases slightly from 0.8 to 0.6, and the hydrogen content is significantly decreased from 2.4 to 1.2. The latter is in good agreement with the methyl disappearance observed in the infrared measurements. Besides, with increasing the RF power, the variation of density from 1.45 at 5 W to 1.85 at 100 W indicates a film densification
DOI: 10.1002/ppap.200600083
Deposition of ‘‘Polysiloxane’’ Thin Films Containing Silver Particles . . .
Table 1. Comparison between the infrared bands (in cm1) of the HMDSO plasma deposits obtained under different power conditions (vs: very strong, s: strong, qs: quite strong, w: weak, vw: very weak, sh: shoulder).
5W
20 W
40 W
100 W
Mode
2 906 (w)
2 960 (s)
2 960 (qs)
2 954 (vw)
2 906 (w)
2 906 (w)
2 910 (w)
2 876 (w-sh)
2 876 (w-sh) 2 875 (vw)
2 131 (w)
2 140 (w)
Ref.
H–C – –C
3 024 (vw) 2 960 (s)
Chemical groups
a
sp3 CH3
s
n C–H3
sp3 CH3
ns C–H2
sp3 CH2
n C–H3
[15–17] [15],[16]
2 131 (qs)
[15–18]
1 870 (vw) 1 721 (vw) 1 718 (vw)
ns C – –O
1 716 (w)
1 575 (vw) 1 570 (vw)
1 556 (w)
1 410 (vw) 1 410 (vw)
1 410 (vw)
[15],[17] X–C – –C aromatic (X ¼ H or Si)
1 539 (qs) da C–H3 a
SiMex
[15],[20]
1 358 (vw) 1 359 (vw)
1359 (vw)
d C–H2
Si–CH2–Si
[15],[20]
1 255 (vs)
1 257 (s)
ds C–H3
SiMex
[15–17],[20]
Si–O–Si–X
[15],[21]
1 034 (vs)
1 256 (vs) 1 031 (vs)
1 026 (vs)
1 000 (vs)
a
n Si–O–Si v Si–CH2–Si
839 (vs)
837 (vs)
835 (w-sh)
835 (vw-sh) n Si–C
796 (vs)
756 (w-sh)
754 (vw-sh)
794 (vs)
802 (vs)
[15] Si(Me)3
rCH 796 (s)
[19]
H–SiO
[15],[20]
nSi–C or d Si–O SiCx
[15–17]
r CH
Si(Me), Si(Me)2
[20]
SiC3 in Si(CH3)3
[15–17]
686 (vw)
?
613 (w)
?
Figure 6. FTIR spectra of SiCxOyHz deposits obtained under different RF power conditions. Plasma Process. Polym. 2007, 4, 127–134 ß 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[17],[20]
Figure 7. Evolution of C/Si, O/Si and H/Si ratios as a function of the RF power.
www.plasma-polymers.org
131
B. Despax, P. Raynaud
Table 2. Comparison between the infrared bands (in cm1) of the HMDSO plasma deposits obtained with different silver volume fractions (vs: very strong, s: strong, qs: quite strong, w: weak, vw: very weak, sh: shoulder).
0 £ fv(Ag) £ 0.1
0.3 2 958 (qs)
2 902 (w) 2 130 (qs)
0.5
Chemical groups
2 958 (qs)
na C–H
sp3 CH3
2 914 (vw)
ns C–H
sp3 CH3
2 872 (w-sh)
2 873 (qs)
2 140 (w)
2 133 (w)
1 880 (vw)
Mode
s
n C–H
sp3 CH2
1 880 (vw) ns C – –O
1 540 (qs)
1 000 (vs)
1 566 (qs) 1 456
1 444 (vw)
1 400 (w)
1 404 (w)
da C–H3
1 340 (vw-sh)
1 332 (vw-sh)
daC–H2
Si–CH2–Si
1 255 (qs)
1 253 (qs)
ds CH3
in Si(CH3)
1 010 (vs)
1 024 (vs)
a
n Si–O
Si–O–Si
842 (s)
r CH or/and d H–SiO
Si (CH3)3
798 (w-sh)
ns Si–C
642 (w-sh)
?
950 (s-sh)
941 (qs-sh)
835 (vw-sh)
833 (s-sh)
800 (s)
802 (w-sh)
The IR peak assignments of Ag-SiCxOyHz are summarized in Table 2. From a qualitative point of view, the AgSiCxOyHz composite films exhibit the same IR spectra (Figure 8, Table 2) as the SiCxOyHz films. The only difference concerns the relatively marginal changes in composition which are revealed by the shift or/and the increase of some
bands. Indeed, the bands at around 1 000 cm1 shift to greater energy while the methyl peaks (1 410, 1 256, 835 cm1) and Si –CH2 –Si peak reappear with the increase in silver content. These results indicate that the presence of silver seems to favor the incorporation of oxygen (1 024 cm1) and methyl incorporations in the film network. The spectra exhibit a significant amount of aromatic C – – C bonds in contrast with the films obtained without any silver between 5 and 40 W. It suggests that the C – – C bonds come from carbon sputtered fragments.
Figure 8. FTIR spectra of SiCxOyHz films containing different silver volume fractions.
Figure 9. Evolution of the film composition as a function of the Ag/Si ratio.
related to the methyl disappearance, and the formation of C– – C aromatic bonds. SiCOH Matrix with Silver
132
C– –C
1 566 (qs)
Plasma Process. Polym. 2007, 4, 127–134 ß 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/ppap.200600083
Deposition of ‘‘Polysiloxane’’ Thin Films Containing Silver Particles . . .
The C/Si, O/Si and H/Si ratios obtained by RBS and ERDA measurements are plotted in Figure 9 as a function of the Ag/Si atomic ratio. These results show that the H, C and O percentages rise in respect to the Ag/Si ratio present in the layers. The latter results are in perfect agreement with the presence of methyl groups, and with the shift of Si –O –Si stretching mode to the greater energy in the IR spectra.
Microscopic Analysis of the Film Structure Chemical analyses and physical properties indicated that our films were silver-containing SiCxOyHz materials under certain operating conditions. Mass density analysis revealed the presence of different silver volume fractions related to the process parameters. The last point concerns the distribution in shape and size of the silver particles inside the SiCxOyHz matrix. The film microstructure was examined by TEM and SEM. Typical TEM micrographs obtained are shown in Figure 10. The
Figure 11. SEM micrographs of the cross section of the bulk material: a) at low silver volume fraction (0.6).
relatively good homogeneity of particle distribution is quite remarkable and shows the efficiency of the HMDSO pulsed injection mode for the control of the deposition process. The two micrographs show the cluster size and shape for two different metal volume fractions. At a low volume fraction below 0.25, the picture (Figure 10a) shows homogeneous and spherical metal grains immersed in matrix. Their diameter varies between 2 and 10 nm, 3 nm been the highest probability. However, the picture shown in Figure 10a does not give a good estimation of the volume fraction because of the stacking of two or three cluster attributed to a too long deposition time (50 s). At a high volume fraction (Figure 10b), the material can be described as having very large non spherical silver clusters separated by thin and elongated polymer branches. Their dispersion in size and shape is important with sizes varying between 30 and 100 nm. The matrix network indicates a fast diffusion of silver atoms or particles, and their easy coalescence during the film growth. SEM micrographs of the bulk material cross section at two volume fractions are shown in Figure 11. The microstructure at low volume fraction (Figure 11a) shows a granite-like aspect of the film and a homogeneous dispersion of the silver nanoparticles (
