lable at ScienceDirect

Vacuum 84 (2010) 1389–1392

Contents lists avai

Vacuum

journal homepage: www.elsevier .com/locate/vacuum

Measurement for dissociation ratio of source gases in plasmas usinga quartz sensor

Atsushi Suzuki*, Hidehiko NonakaNational Institute of Advanced Industrial Science and Technology (AIST), Tsukuba Central 2, 1-1-1 Umezono, Tsukuba, Ibaraki 3058568, Japan

a r t i c l e i n f o

Article history:Received 7 July 2009Received in revised form15 December 2009Accepted 5 January 2010

Keywords:Plasma dissociationDissociation ratioQuartz sensorPlasma diagnosisGas compositionAmmonia plasmaSpatial distributionGas viscosity

* Corresponding author. Tel.: þ81 29 861 5461; faxE-mail address: a-suzuki@aist.go.jp (A. Suzuki).

0042-207X/$ – see front matter � 2010 Elsevier Ltd.doi:10.1016/j.vacuum.2010.01.004

a b s t r a c t

The dissociation ratio of source gas molecules is measured using a quartz sensor, for which the outputdepends on the average molecular weight and viscosity of gases. The change in the pressure andtemperature-normalized quartz sensor output (NQO) by discharge correlates with changes of the signalintensity of the source gas of ammonia (NH3), as measured by gas analysis using a quadrupole massspectrometer. Thereby, the dissociation ratio is obtainable from the change in NQO by the discharge. Thespatial distributions of the change in NQO in the plasma chamber show a larger change in NQO near theplasma electrode, indicating that the change in NQO correlates with the dissociation ratio of NH3. Finally,the dissociation ratios of NH3 near and between the plasma electrodes were obtained from the spatialdistribution of the dissociation ratio of NH3. Results show that the Q-sensor measurement is simple anduseful to derive the dissociation ratio of the source gases for the plasmas of reactive gases.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Dissociation of gas molecules occurring via collision with elec-trons is the initial reaction in the gas phase of plasmas. Thesedissociation reactions determine the initially generated activechemical species such as radicals and ions, and lead to secondarygas phase reactions in the plasmas.

Several reports have described dissociation ratios of gas mole-cules in environmental studies for which the dissociation ratio iscrucial. Environmental studies typically use gas chromatography(GC) because measurements are done at atmospheric pressure[1,2]. Other techniques such as infrared absorption spectroscopy,ion chromatography, and emission spectroscopy are also useful tostudy the decomposition of toxic or gas molecules of key speciesrelated to global warming [3–5].

Few studies have addressed the dissociation of the source gasesin plasmas, particularly for industrial plasma processes. Dissocia-tion of hydrogen has been studied using laser-induced fluorescencefor hydrogen atoms [6–9]. Coherent anti-Stokes Raman scatteringmethod has been used to measure the dissociation of silanemolecules used for preparation of amorphous silicon films [10].

: þ81 29 861 5301.

All rights reserved.

However, the techniques in the works described above mainlyrequire expensive lasers. For that reason, they are unsuitable for usewith industrial plasma processes. Moreover, GC is only applicable toplasmas at high pressures because of the necessity for gassampling. In contrast, gas analysis using quadrupole mass spec-trometry (QMS) alone can measure the gas phase of the plasmas atlow pressures; it is unsuitable for use at higher pressures [3,11].Therefore, gas analysis is not applicable in high-pressure conditionslike those used for plasma processing for thin silicon films.Furthermore, it is complicated to obtain information on the disso-ciation ratio of the source gas molecules.

Information related to the dissociation of the source gas mole-cules should be important because the degree of the dissociationratio affects properties of the products processed by the plasma.Moreover, because one can know the consumption rate of thesource gas molecules from information on the dissociation of thesource gas molecules, information about the dissociation ofthe source gas molecules is expected to be important from aneconomic perspective. Therefore, simpler measurements of thedissociation ratio of the source gases are expected to be useful forindustrial plasma processes such as those for silicon films: animportant material used for thin film solar cells [12].

We have taken measurements using a quartz sensor, for whichthe output depends on pressure, molecular weight, and viscosity ofgases measured [13–20]. Because the pressure and temperature-

20

21

22

23

5.25

5.30

5.35

5.40

Time (min)0 5 10 15 20

Plasma on Plasma off

Q s

enso

r out

put (

arb.

uni

t)

Pres

sure

(Pa)

Fig. 1. Temporal changes in the total pressure (B), the Q-sensor output (C) and thepressure and temperature-normalized output (,) with and without plasmasmeasured at Z¼ 70 mm.

0.000.020.040.060.080.100.120.140.160.18

0 50 100 150 200 250Rf power (W)

Cha

nge

in N

QO

(V)

Fig. 2. Rf power dependence of the change in the pressure and temperature-normalized output measured at Z¼ 70 mm.

A. Suzuki, H. Nonaka / Vacuum 84 (2010) 1389–13921390

normalized quartz sensor output (NQO) depends only on themolecular weight and viscosity of the gases measured, ourQ-sensor measurement can detect changes in chemical species inthe gas phase of the plasmas [19]. This Q-sensor measurement canmeasure the dissociation ratio of the source gas molecules in theplasmas if this change in NQO induced by the plasma correlates tothe dissociation ratio of the source gas molecules.

The size of quartz in the Q-sensor is less than 1 cm: it can befixed easily anywhere in the plasma chamber. Its small size is alsoadvantageous for measuring the spatial distributions of the disso-ciation ratio of the source gas molecules by sliding the Q-sensor inthe plasma chamber. Application of the Q-sensor measurement forobtaining the dissociation ratio of the source gases is convenientbecause it is like a simple pressure measurement without samplingof the gas from the chamber, which would reduce the pressure.Moreover, it is applicable to highly reactive gases.

For this study, we attempted to use Q-sensor measurement tomeasure the dissociation ratio of the source gas in low-pressureradio frequency (rf) NH3 plasmas. We used NH3 to avoid chemicalinfluences on the quartz in the Q-sensor such as deposition andetching because this is the first attempt of Q-sensor measurementof plasmas with reactive gases. The NH3 plasma is used practicallyas a coating to reduce ashing damage, nitridation, and deposition[21–23]. Comparison by gas analysis using QMS revealed that thedissociation ratio of NH3 by the plasmas can be derived from thechange of the Q-sensor output.

2. Experimental

Details of the experimental setup are shown elsewhere [19,20].Briefly, the rf 13.56 MHz glow discharge is generated betweenparallel plasma electrode plates. The diameter and spacing of theplasma electrodes are, respectively, 112 and 20 mm. The respectiveflow rate and pressure of NH3 are 10 sccm and 20 Pa. The rf powerwas 20–250 W. The anode electrode temperature was set to 100 �C.The Q-sensor, installed near the plasma electrodes, can be slid andclose to the plasma electrodes. Here, we define Z as the distancebetween the Q-sensor and the edge of the plasma electrodes [20].

The electric impedance of the quartz oscillator has been analyzedtheoretically using a simple model in which the fork-shape quartzoscillator vibrates with a single frequency [13]. The Q-sensor outputis the voltage output which is converted from the impedance of thequartz oscillator, depending on molecular weight and viscosity ofthe gas measured [13]. However, the total pressure and temperaturealso affect the Q-sensor output [13,24]. Thereby, we used the pres-sure and temperature-normalized Q-sensor output (NQO) to obtaininformation about the gas composition in the plasmas because NQOonly depends on viscosity and molecular weight of the gases [19].Details of the derivation for NQO from the Q-sensor output aredescribed in Ref. [19].

For comparison to the Q-sensor measurement, the gas phase inthe plasma was analyzed using QMS. In this study, etchingor nitridation were not observed to affect the results from theQ-sensor measurement.

3. Results and discussion

3.1. Gas composition change measured using a Q-sensor

The Q-sensor measurement can measure the gas compositionchange in plasmas because the Q-sensor output depends onmolecular weight and viscosity of the gases, which depends on thegas composition. Fig. 1 portrays temporal changes in the totalpressure, the Q-sensor output, and NQO with and without plasmas.With the discharge turned on, the pressure and the Q-sensor

output changed and reached constant outputs. After the dischargewas turned off, they reverted to the output levels before thedischarge. Then NQO was calculated from the pressure-normalizedQ-sensor (PQO) output and its temperature dependence; PQO wascalculated from the Q-sensor output and its pressure dependence[19]. Because NQO depends only on the molecular weight andviscosity of the gas measured, the change in NQO between with andwithout the discharge depicted in Fig. 1 means that the gascomposition changes as a result of the discharge.

The rf power dependence of the change in NQO in Fig. 2 alsosupports that the change in NQO correlates to the gas compositionchange in the gas phase by the plasmas. As expected, the change inNQO increases with the rf power, indicating that the change in NQOreflects the gas composition change in the gas phase of the plasmasbecause they increase with rf power. These gas compositionchanges were attributed to dissociation of the source gas: NH3.Therefore, the change in NQO is expected to correlate with thedissociation ratio of NH3.

3.2. Working curve for the dissociation ratio of NH3

Gas analysis using QMS was used to evaluate the dissociationratio of NH3. Fig. 3 depicts the mass spectrum measured using QMS

10 20 301E-7

1E-6

1E-5

1E-4

Sign

al in

tens

ity (a

rb. u

nit)

m/z

Fig. 3. Mass spectrum measured by QMS with (open) and without (close) plasmas.

A. Suzuki, H. Nonaka / Vacuum 84 (2010) 1389–1392 1391

with and without the plasma at rf power of 100 W. In NH3 plasma,a decrease in m/e¼ 16 and 17 by the dissociation of NH3 and anincrease in m/e¼ 2 and 28 by the production of H2 and N2 wereobserved. By calibrating the relative ion signal intensity in the massspectrum, the relative densities of the gases can be compared.These results show that the overall reaction in the plasma is

2NH3 / 3H2 D N2:

The reaction presented above agrees with that presented in anearlier report [25]. The Q-sensor measurement also supports thatthe product of the dissociation reaction of NH3 is a mixture of H2

and N2 with a large ratio of H2 [18].The dissociation ratio of NH3 (Rdis) was calculated from the

relative ion signal intensity of m/e¼ 17 as

Rdis ¼Igas � Ipla

Igas: (1)

Here, Igas and Ipla, respectively, represent the signal intensities ofm/e¼ 17 before and after the plasma. Fig. 4 portrays the calculateddissociation ratio of NH3 as a function of the change in NQO. FromFig. 4, the change in NQO is found to correlate to Rdis when varying

Dis

soci

atio

n ra

tio (%

)

Change in NQO

05

101520253035404550

0.00 0.05 0.10 0.15 0.20

Fig. 4. Dissociation ratio of NH3 calculated from the results of ion signal of m/e¼ 17plotted against the change in NQO at Z¼ 70 mm.

the rf power. This correlation between Rdis and the change in NQOare attributed to the constant relative production ratio for eachchemical species including H2 and N2 in the plasma. This isprobably because of the unique plasma character of a-regimeunder conditions of a constant flow rate and total pressure of NH3

[26,27].Fig. 4 can be used to measure Rdis under the present flow rate

and pressure. One can measure Rdis by applying the change in NQOfrom the Q-sensor measurement to the working curve of Fig. 4.Once the working curve for dissociation of the source gas isobtained, one can easily obtain the Q-sensor measurement valueusing the curve.

3.3. Spatial distribution of the dissociation ratio

The Rdis obtained in the previous subsection is at thechamber inner wall because the edge of the gas sampling tubefor the QMS is positioned there. Because this is slightly distantfrom the plasma glow, the information related to the dissocia-tion of NH3 is expected to be more convenient to understandhow much the source gases dissociate in the plasma glow. Toobserve this, we measured the spatial distribution of Rdis in theplasma chamber. This is done by sliding the Q-sensor fromthe inside of the pipe to the plasma electrodes. Fig. 5 presentsthe spatial distribution of Rdis plotted against the distance (Z)between the Q-sensor head and the plasma electrodes with rfpower maintained at 100 W. Here, results obtained by themeasurement at Z< 20 mm were excluded because they includelarge inaccuracies, probably because Z is smaller than thespacing of the plasma electrodes and a part of the dischargegenerated from the Q-sensor.

As with the change in NQO, Rdis calculated using Fig. 4 as theworking curve increases with the decrease in Z in Fig. 5. To estimatethe dissociation ratio of NH3 at Z< 20 mm, we assumed that thespatial distribution of Rdis depends on the spatial distribution ofthe gas composition in the gas phase of the plasmas. The deviationof Rdis in the spatial distribution can be expressed as proportional toRdis as shown below if the dissociation of NH3 only occurs atZ¼ 0 mm.

�dRdis

dZ¼ kRdis (2)

29

30

31

32

33

34

35

36

0 50 100 150

Dis

soci

atio

n ra

tio (%

)

Distance between Q-sensorand plasma electrode; Z (mm)

Fig. 5. Spatial distribution of dissociation ratio of NH3 as a function of the distancebetween the Q-sensor and the plasma electrode. The line is the fitting curve based onEq. (3).

A. Suzuki, H. Nonaka / Vacuum 84 (2010) 1389–13921392

Therein, k is a constant related to the diffusion constant. Thesolution of Eq. (2) is

Rdis ¼ Rdis;0expð�kZÞ; (3)

where Rdis,0 is the dissociation ratio of NH3 at Z¼ 0 mm. Eq. (3) iswell fitted to the experimental data, as shown by the line in Fig. 5.By the fitting, Rdis,0 is derived as 35% at the edge of the plasmaelectrodes. The position of the centre of the plasma electrodescorresponds to Z¼�52 mm, thereby, the Rdis at the centre of theplasma shows the maximum of 37%.

These dissociation ratios of 30–40% mostly agreed with theprevious studies [25,28]. Consequently, it can be concluded thatthe Q-sensor measurement can measure the dissociation ratio nearthe plasma electrodes. Because the present Q-sensor measurementis simple and does not affect the plasma if it is located far from theplasma electrodes, it is expected to be convenient to obtain thedissociation ratio of NH3.

4. Summary

Results show that the dissociation ratio of source gases inplasmas can be measured using the Q-sensor measurementbecause the change in the normalized Q-sensor output (NQO) iscorrelated to the dissociation ratio of the source gases at constantflow rate of 10 sccm and pressure of 20 Pa. The working curves forthe dissociation ratio to other plasma conditions are under inves-tigation. This method is simple, practical, and applicable to indus-trial plasma processes for any source gas.

References

[1] Sun R-B, Xi Z-G, Chao F-H, Zhang W, Zhang H-S, Yang D-F. Atmos Environ2007;41:6853–9.

[2] Kim DY, Park DW. Surf Coat Technol 2008;202:5280–3.[3] Mista W, Kacprzyk R. Catal Today 2008;137:345–9.[4] Mok YS, Demidyuk V, Whitehead JC. J Phys Chem A 2008;112:6586–91.[5] Okuda I, Takahashi E, Kato S, Matsumoto Y. Jpn J Appl Phys 2008;47:8054–6.[6] Cheshire RC, Graham WG, Morrow T, Kornas V, Dobele HF, Donnelly K, et al.

Appl Phys Lett 1995;66:3152–4.[7] SchulzvonderGathen V, Dobele HF. Pla Chem Pla Proc 1996;16:461–86.[8] Mozetic M. Heat and mass transfer under plasma conditions 1999;891:

325–32.[9] Mozetic M, Vesel A, Monna V, Ricard A. Vacuum 2003;71:201–5.

[10] Hata N, Matsuda A, Tanaka K, Kajiyama K, Moro N, Sajiki K. Jpn J Appl Phys1983;22:L1–3.

[11] Jauberteau JL, Jauberteau I, Aubreton J. Int J Mass Spec 2007;266:15–24.[12] Hamakawa Y. Thin-film solar cells. Berlin: Springer; 2004.[13] Kokubun K, Hirata M, Murakami H, Toda Y, Ono M. Vacuum 1984;34:731–5.[14] Suzuki A, Nonaka H. Shinku 2005;48:139–41.[15] Suzuki A, Kurokawa A, Nonaka H. In: Proceedings of 18th international

symposium on plasma chemistry. 2007. p. 395.[16] Suzuki A, Kurokawa A, Nonaka H. Shinku 2007;50:184–6.[17] Suzuki A, Nonaka H. J Vac Soc Jpn 2008;51:155–7.[18] Suzuki A, Nonaka H. J Vac Soc Jpn 2009;52:121–3.[19] Suzuki A, Nonaka H. Rev Sci Instrum 2009;80:095109.[20] Suzuki A, Nonaka H. Vacuum 2009;84:554–8.[21] Chang KM, Tseng MH, Deng IC, Tsai YP, Yeh SJ. Jpn J Appl Phys

2001;40:6663–7.[22] Bellakhal N. Mater Res Bull 2002;37:2539–46.[23] Choi JH, Lee TY, Choi SH, Han J-H, Yoo J-B, Park C-Y, et al. Thin Solid Films

2003;435:318–23.[24] Kawashima H. Jpn J Appl Phys 1995;34:2596–600.[25] Miller GP, Baird JK. J Phys Chem 1993;97:10984–8.[26] Pennebaker WB. IBM J Res Develop 1979;23:16–23.[27] Godyak VA, Khanneh AS. IEEE Trans Plasma Sci 1986;14:112–23.[28] Carbaugh DC, Munno FJ, Marchello JM. J Chem Phys 1967;47:5211–4.