Neutral gas temperatures in a multipolar electron cyclotron resonance plasma
`J. Hopwood and J. Asmussen
`
`Citation: Applied Physics Letters 58, 2473 (1991); doi: 10.1063/1.105232
`View online: http://dx.doi.org/10.1063/1.105232
`View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/58/22?ver=pdfcov
`Published by the AIP Publishing
`
`Articles you may be interested in
`Oxidation of silicon in an oxygen plasma generated by a multipolar electron cyclotron resonance source
`J. Vac. Sci. Technol. B 10, 2211 (1992); 10.1116/1.586191
`
`Ion and neutral energies in a multipolar electron cyclotron resonance plasma source
`J. Vac. Sci. Technol. A 10, 1265 (1992); 10.1116/1.578237
`
`Ion and neutral temperatures in electron cyclotron resonance plasma reactors
`Appl. Phys. Lett. 58, 458 (1991); 10.1063/1.104606
`
`Plasma parameter and etch measurements in a multipolar confined electron cyclotron resonance
`discharge
`J. Vac. Sci. Technol. B 9, 29 (1991); 10.1116/1.585786
`
`Microwave multipolar plasmas excited by distributed electron cyclotron resonance: Concept and
`performance
`Rev. Sci. Instrum. 59, 1072 (1988); 10.1063/1.1139728
`
`
`
` This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
`134.84.0.128 On: Sat, 12 Apr 2014 22:19:59
`
`TSMC-1110
`TSMC v. Zond, Inc.
`Page 1 of 4
`
`

`
`in a multipolar
`
`electron
`
`cyclotron
`
`Neutral gas temperatures
`plasma
`J. Hopwood and J. Asmussen
`Department of Electrical Engineering, Michigan State University, East Lansing, Michigan 48824
`(Received 5 November 1990; accepted for publication 8 March 1991)
`Optical emission measurements of the Doppler broadening of argon (549.6 nm) and helium
`(501.6 nm) neutral
`lines in the unmagnetized
`regions of an electron cyclotron
`resonance
`plasma show that
`the gas temperature
`ranges from 300 to 900 K. After compensation
`Zeeman splitting, Doppler widths are found
`to be constant across the radius of the
`plasma. Plasma heating of the argon gas (0.77 mTorr)
`is shown to increase from 300 to 500
`K as microwave power absorption
`increases from 80 to 330 W. Long neutral residence
`times are observed to increase the argon gas temperature
`to ~900 K. Helium and argon
`neutral
`temperatures decrease as the neutral mean free path increases indicating
`that
`the gas may be heated by ion-neutral collisions
`including charge exchange.
`
`resonance
`
`for
`
`resonance (ECR) plasmas gain
`As electron cyclotron
`acceptance with
`the plasma processing industry,
`it is im-
`portant
`that
`the plasma physics within
`these devices be
`better understood.
`Ions and electrons typically have ener-
`gies on the order of 0.1-20 eV in ECR plasmaslA due to
`interactions with strong electric and magnetic
`fields, Neu-
`trals in the plasma may then gain considerable kinetic en-
`ergy from elastic collisions or charge exchange with
`these
`more energetic charged particles. Although ECR plasma
`processes such as dry etching are often “ion assisted”
`in
`nature, neutrals and free radicals in the discharge also con-
`tribute
`to plasma-surface
`interactions
`(e. g., spontaneous
`etching of trench sidewalls).
`In addition, neutral heating
`may ultimately
`limit
`the ion density attainable
`in a plasma
`due to rarefaction of the source gas. In a preliminary effort
`to quantify
`the role of neutral and free radical energies in
`multipolar-ECR
`plasma processing, we have determined
`the temperature of argon and helium neutrals by measur-
`ing optical emission
`line profiles as a function of micro-
`wave power absorption, gas pressure, and mass flow rate.
`The ECR plasma source used in this work
`is shown
`schematically
`in Fig.
`1, and
`described
`in
`detail
`elsewhere.a Microwave
`radiation
`(2.45 GHz)
`is intro-
`duced into a variable length resonant cavity which focuses
`the microwaves in a 12.5 cm i.d. quartz discharge chamber.
`The ECR magnetic
`field is produced by an octapole array
`of Nd-Fe-B magnets which surround
`the discharge. The B
`field radially decreases from > 1000 G at the inside wall of
`the chamber
`to ~50 G in the center and passes through
`the ECR field strength of 875 G at z 1 cm from the cham-
`ber wall. Since there is no axial B field like that in diuer-
`gent-jield ECR sources,1-3 the plasma diffuses into the pro-
`cessing chamber rather
`than streaming.
`Emission, primarily
`from the active discharge region of
`the plasma, was collected by an optical fiber bundle with an
`acceptance cone of 15” (see Fig. 1). The light emerging
`from
`the opposite end of the fiber was collimated and
`passed through a Fabry-Perot
`interferometer, a monochro-
`mator, and detected by a cooled photomultiplier
`tube
`(PMT). The instrumental broadening of this optical sys-
`tem was determined by measuring
`the 480.7 nm emission
`
`from the unmagnetized volume of a 150 W, 0.67 mTorr, 10
`seem xenon ECR discharge. After deconvolution
`of the
`(7&1)~10-~
`nm Doppler width
`(600*150
`K) of the
`xenon line, the instrument
`response was found to be Gauss-
`ian with a full width at half maximum
`(FWHM)
`resolu-
`tion of (M/A)
`= (3.5 10.1) x 10 - ‘. This corresponds to
`a spectral
`resolution of ( 1,92=l=O.O6) x 10 - 3 nm for
`the
`549.6 nm Ar
`line and
`(1.76rtO.05) ~10~~
`nm
`for
`the
`501.6 nm He line. All subsequently measured line profiles
`were deconvolved
`from the instrument
`response. Detail
`in
`the argon spectral line shapes was obscured by instrument
`broadening. Helium
`lines, however, were considerably
`wider
`than
`the
`instrument
`resolution
`(typically
`0.0045
`nm) and appeared to be symmetric and Gaussian in shape.
`The emission line profiles observed in plasmas may be
`broadened by many factors. Natural and collisional broad-
`ening are insignificant’
`compared
`to the Doppler width
`(FWHM)
`which
`is given by
`A;l,=7.16~10-~&[T/M]~‘*,
`where T is in Kelvin and M
`is the mass of the emitting
`species in amu. Typical charged particle densities in this
`discharge are on the order of 10”/cm3 which
`is roughly
`two orders of magnitude
`less than densities which produce
`significant Stark broadening.’ Stark broadening due to the
`microwave
`fields in the plasma [ - 7 kV/m at 250 W (Ref.
`9)]
`is estimated to be on the order of lo-”
`nm. It is not
`possible, however, to neglect Zeeman splitting of lines in an
`
`FIG. 1. Schematic view of the microwave plasma disk reactor (MPDR)
`and associated spectroscopic equipment.
`
`2473
`
`Appl. Phys. Lett. 58 (22), 3 June 1991
`
`0003-6951
`
`I91 /222473-03$02.00
`
`@ 1991 American
`
`Institute of Physics
`
`2473
`
` This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
`134.84.0.128 On: Sat, 12 Apr 2014 22:19:59
`
`TSMC-1110 / Page 2 of 4
`
`

`
`z~~j.......::,9~m
`
`E 2 e L E ?OO
`F
`P s
`d
`E-
`2 500
`P
`E 2
`i
`
`300
`
`I f-mi
`
`(cm)
`
`Argon Presson
`
`(mTorr)
`
`10 %cm, hsllum
`350 w0ff.a
`
`Neutral Mean Free Path
`
`(cmj
`
`-2
`Radial Position
`
`1 I
`
`I
`
`.
`
`I
`
`.
`
`I
`
`I
`
`in
`
`the Dischorgs
`
`Chamber
`
`I )
`2-
`
`FIG. 2. Comparison of measured line widths and Zeeman effect imply a
`uniform temperature distribution
`in the radial direction. Note the mea-
`sured peak magnetic held strengths are listed next to each data point.
`
`FIG. 4. Variation of helium and argon neutral temperature with neutral
`mean free path. Note that
`the upper axis applies to argon only (the
`helium pressure ranges from 7.5 to 23 m?brr).
`
`the plasmas are
`in
`(Electrons
`recombinations.
`ion wall
`heated by electron cyclotron resonance and ions are accel-
`erated by VB forces and/or plasma potential gradients.)
`Extrapolation of these data indicates that the neutral tem-
`perature may exceed 1000 K for absorbed powers~ of l-2
`kW as previously reported by differential pressure mea-
`surements.” The error bars were determined by repeated
`temperature measurements. The uncertainty in the temper-
`ature measurement is dominated by background noise
`rather than instrument resolution calibration.
`Gas temperatures as a function of process chamber
`pressure for a constant absorbed power and mass How rate
`are presented in Fig. 4. When plotted as a function of
`neutral mean free path (mfp),
`the temperatures are ob-
`served to-decrease nearly linearly with mfp, thus support-
`ing the model that
`the neutrals are-collisionally heated.
`Under conditions of long mfp the neutrals are also in better
`contact with
`the chamber- walls and are therefore cooled
`more effectively. Finally, note that helium neutrals
`(,%a
`
`ECR discharge due to the strong static magnetic fields
`(875 G). The measured 549.6 nm line widths in a 20 seem,
`0.77 mTorr argon discharge are plotted as a function of
`radial position (and B field)
`in the plasma in Fig. 2. As-
`suming a constant Doppler profile of 0.002 nm (i.e., a
`constant temperature),
`the solid curve represents the com-
`puted Zeeman broadeninga”’ after convolution with
`the
`instrument response. The close agreement between exper-
`iment and calculation
`implies that the temperature is uni-
`form across the discharge as expected when the mean free
`path is of the order of the plasma chamber characteristic
`length. Subsequent data were measured along the vertical
`axis in the central volume (us= 0) of the discharge cham-
`ber where the Zeeman splitting
`is negligible.
`The argon (/z = 549.6 nm) neutral temperature at low
`input power ( < 100 WI
`is nearly room temperature as
`shown in Fig. 3. increasing the absorbed power in an ECR
`discharge increases not only the ion density,g but also the
`neutral gas temperature. This suggests that energetic neu-
`tral particles are produced by collisions with charged par-
`ticles, charge exchange with ions, or generated by electron-
`
`
`
`jroo l---------
`
`c
`8
`2
`
`-.
`
`.^^I_____
`
`Argon Flow Rats
`
`(seem)
`
`1300~--L41ul.~-
`
`.- 1 100
`:s
`3
`% wo
`e r"
`
`700
`
`f
`s
`
`B
`
`sot’
`
`
`
`4 .OO
`
`Residence Time
`
`(psoc.)
`
`I 10al....,..,..,l,,..,,,.r.~I,r.,ll,.,.I.,,,F
`
`200
`Power
`
`(w)
`
`JO0
`
`0
`
`100
`
`FIG. 3. Argon neutral temperature measured aIong the central axis of the
`MPDR
`increases as microwave absorbed power increases.-
`
`FIG. 5. Argon neutral temperature variation as a function of gas resi-
`dence time.
`
`2474
`
`Appl. Phys. Lett., Vol. 58, No. 22, 3 June 1991
`
`J. Hopwood and J. Asmussen
`
`2474
`
` This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
`134.84.0.128 On: Sat, 12 Apr 2014 22:19:59
`
`TSMC-1110 / Page 3 of 4
`
`

`
`less energetic than argon
`= 501.6 nm) are considerably
`neutrals due, at least in part, to higher neutral gas thermal
`conductivity and lower ion densities .in helium.
`temper-
`The effect of residence time on argon neutral
`ature was measured by varying the flow rate at a constant
`pressure of 2 mTorr
`(see Fig. 5). The rapid increase in gas
`temperature observed for long residence time plasmas in-
`dicates that as flow rate increases a significant portion of
`the neutral gas energy is transported downstream from the
`ECR heating zones. These data also show that variation of
`pumping speeds and flow
`rates can be used to control
`plasma temperatures, perhaps most critically under high-
`power absorption conditions.
`further elucidate
`to
`Additional
`research
`is planned
`plasma heating and cooling mechanisms, including
`the ef-
`fect of plasma chamber and substrate temperature control.
`Free radical and ion temperatures in reactive process gases
`using absorbed powers up to 3 kW
`in both
`larger and
`smaller discharge geometries should also add insight
`into
`neutral gas heating mechanisms.
`
`The authors wish to thank P. Mak for design improve-
`ments to the ECR plasma source and T. Grotjohn
`for help-
`ful discussions concerning optics. This work
`is supported
`in part by Wavemat,
`Inc.
`
`‘E. A. Den Hartog, H. Persing, and R. C. Woods, Appl. Phys. Lett. 57,
`661 (1990).
`‘Y. H. Lee, J. E. Heidenreich III, and G. Fortuno, J. Vat. Sci. Technol.
`A 7, 903 (1989).
`3D. J. Trevor, N. Sadeghi, T. Nakano, J. Derouard, R. Gottscho, P. Foo,
`and J. Cook, Appl. Phys. Lett. 57, 1188 (1990).
`4J. Hopwood, D. K. Reinhard, and J. Asmussen, J. Vat. Sci. Technol. A
`8, 3103 (1990).
`5 J. Asmussen and M. Dahimene, J. Vat. Sci. Technol. B 5, 328 ( 1987).
`6J. Asmussen, D. K. Reinhard, and M. Dahimene, U. S. Patent No.
`4,727,293, Feb. 23, 1988.
`‘B. W. Shore and D. H. Menzel Principles of‘rltomic Spectra, (Wiley,
`New York, 1968),p. 38.
`New York, 1964).
`*H. R. Griem, Plasma Spectroscopy, (McGraw-Hill,
`‘J. Hopwood, R. Wagner, D. K. Reinhard, and J. Asmussen, J. Vat. Sci.
`Technol. A 8, 2904 ( 1990).
`“G. V. Marr, Plasma Spectroscopy (Elsevier, Amsterdam, 1968), p. 190.
`“S. M. Rossnagel, S. J. Whitehair, C. R. Guamieri, and J. J. Cuomo, J.
`Vat. Sci. Technol. A 8, 3113 ( 1990).
`
`2475
`
`Appl. Phys. Lett., Vol. 58, No. 22, 3 June 1991
`
`J. Hopwood and J. Asmussen
`
`2475
`
` This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
`134.84.0.128 On: Sat, 12 Apr 2014 22:19:59
`
`TSMC-1110 / Page 4 of 4