`Comparison of Optical Emission Spectrometric Measurements of the
`Concentration and Energy of Species in Low-pressure Microwave and
`Radiofrequency Plasma Sources*
`
`803
`
`Jiirgen Ropcke, Andreas Oh1 and Martin Schmidt
`Institute for Low- Temperature Plasma Physics, Robert-Blum-Str. 8- 10, D- 17489 Greifswald, Germany
`
`Understanding of specific plasmachemical reactions occurring in plasma sources with the addition of
`complicated molecules requires knowledge of the particular plasma conditions. The subject of the present paper
`is a comparison of the spectrometrically measured relative concentration distributions and energies of species
`in low-pressure argon discharges containing an organosilicon compound, hexamethyldisiloxane (HMDSO),
`which has a large molecular size. The optical diagnostics were performed with three different plasma sources: a
`special planar microwave plasma source (v=2.45 GHz), an r.f. planar reactor (v=13.56 MHz) and a capacitively
`coupled r.f. model discharge tube (v=460 kHz). The energy distribution functions of each plasma are not the
`same, but their general forms are similar to a Maxwell-Boltzmann distribution. The results of comprehensive,
`spatially resolved measurements of the relative concentrations of atoms and radicals (H, Si, CH and C2), the
`neutral species gas temperature, rotational temperature and optical excitation temperature are reported. The
`use of the same gas mixture in three plasma sources, each of distinct construction, excited by different
`frequencies, once again clearly indicates that the comparison and interpretation of optical diagnostic results has
`to be done very carefully, taking into consideration the discharge conditions. Results obtained by actinometry
`show considerably different particle density gradients in the plasmas. Gradients of the excitation temperatures
`( Te,,=0.45-0.7 eV) are less, but the neutral gas temperatures also exhibit large spatial gradients (T,=ambient
`temperature, about 2000 K). This clearly indicates the absolute necessity of spatially resolved optical emission
`measurements for the purpose of comparisons between different plasma sources.
`Keywords: Optical emission spectrometry; plasma diagnostics and characterization; low-pressure microwave
`and r. f. discharges; species concentration and energy
`
`The parameter set needed to describe the properties of non-
`equilibrium, low-pressure gas discharge plasmas is very
`large. Owing to specific excitation mechanisms many
`different particle ensembles exist. Since energy is trans-
`ferred step by step from the external electric field to plasma
`electrons and plasma ions, and then to activated neutral
`species, and as the transfer efficiencies of these processes
`differ by orders of magnitude, particle ensembles with
`different thermodynamic properties exist simultaneously.
`The energy of single particles of each of these ensembles is
`distributed within a certain characteristic range. Normally,
`the energy distribution functions are not equal to each other
`but their general form is similar to a Maxwell-Boltzmann
`distribution.
`Therefore,
`following
`thermodynamic
`methods, each of these particle ensembles can be character-
`ized by generalized thermodynamic variables, such as
`temperature and the number densities of the particles. In
`this way, it is possible, as a first approach, to describe the
`properties of low-pressure, non-equilibrium plasmas by a
`set of temperatures and number densities. When consider-
`ing systematic investigations of the fundamental properties
`and application of plasmas, knowledge of this set of
`parameters must be available. As the largest differences
`exist between the electron ensemble on one side and all
`other, heavier particle ensembles on the other side, the
`parameter set can be reduced to two ensembles in the
`simplest case. The heavy particle ensemble includes ions as
`well as radicals and non-activated neutral species. If, as is
`usual, the degree of ionization is low, this heavy particle
`ensemble can be described by the properties of the neutral
`species. This two-temperature model is the simplest ap-
`proach to characterize low-pressure, non-equilibrium plas-
`mas, and is well established in the characterization of pure
`rare gas discharges.
`
`* Presented in part at the 1993 Winter Conference on Plasma
`Spectrochemistry, Granada, Spain, January 10- 15, 1993.
`
`The addition of molecules, especially of complex mole-
`cules, to rare gas discharges leads to an enormous increase
`of the number of species ensembles. Nevertheless, the
`differences between electrons and neutral species remain
`high. Hence, the characterization of these plasmas can also
`begin with the determination of electron temperature,
`neutral species gas temperature and densities of the most
`relevant species. However, many of these molecular gas
`plasmas, which are very relevant in practical applications,
`are chemically active. Under these circumstances a complex
`investigation of plasmas including probe measurements,
`mass spectrometry and optical methods, is difficult, and
`measuring methods which do not disturb the plasma
`become very important. Optical emission spectrometry
`(OES) is one of these methods of measurement. It can
`detect a certain number of different species and the
`determination of characteristic temperatures
`is also
`possible.
`The subject of the present paper is OES measurements
`in low-pressure argon discharges containing small additions
`of an organosilicon compound, hexamethyldisiloxane
`(HMDSO). In some cases the carrier gas also contains a few
`per cent of helium or nitrogen for measuring purposes.
`Three different plasma sources are compared: a special
`planar microwave plasma source (v=2.45 GHz), an r.f.
`planar reactor (v= 13.56 MHz) and a capacitively coupled
`r.f. model discharge tube (v=460 kHz). The results con-
`cerning the spatial distribution of hydrogen and silicon
`atoms and of CH and C2 radicals, neutral species gas
`temperature, rotational temperature and optical excita-
`tion temperature, which is closely related to the energy of
`the electron gas, are reported. Knowledge of these para-
`meters is vital for the detailed analysis of the plasmachemi-
`cal activity of these plasmas, which results in the growth of
`so called plasma polymer films on the plasma vessel walls
`and on the electrodes. Although it is known, especially with
`plasmapolymerization of HMDSO, that large differences
`between processes in r.f. excited plasmas and microwave
`
`Published on 01 January 1993. Downloaded by University of Minnesota - Twin Cities on 12/04/2014 23:29:01.
`
`View Article Online
`
` / Journal Homepage
`
` / Table of Contents for this issue
`
`TSMC-1109
`TSMC v. Zond, Inc.
`Page 1 of 6
`
`
`
`804
`
`JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1993, VOL. 8
`
`the microphysical explanation of these
`plasmas exist,'
`differences is still a matter of discussion. The latter is a
`result of the complex character of these plasmas. Hence, a
`complex comparison of these would be interesting.
`The analysis of the emission measurements includes a
`number of assumptions, which are described as follows.
`As a matter of principle, it is assumed that the Corona
`model is valid, describing the emission of spectral lines or
`molecular bands corresponding to the population of excited
`states, as determined by the balance between collisional
`excitation and spontaneous radiative decay only. The
`validity of this assumption is mainly based on the low
`degree of ionization in these plasmas, which is a maximum
`of 1 x
`in all cases.
`If the population of the excited atomic states corresponds
`to a Boltzmann distribution, the so called electronic
`excitation temperature or excitation temperature T,,, of the
`emitting atoms can be derived by means of relative line
`intensity measurements. In this case the value log(Iil/gA) is
`a linear function of the excitation energy. Using the
`Boltzmann plot method and plotting log(Iil/gA) as a
`dependent of the excitation energy, the slope of the plot is
`related
`to the excitation temperature. It
`is equal to
`-0.625/Te,, when E is in cm-I, I is the line intensity, ;I the
`line wavelength, A the transition probability and g=gk/gj is
`the quotient of the statistical weights of the two atomic
`levels (k>j]. In the present paper the excitation tempera-
`ture is measured by using the argon line intensities. The
`reason for the use of argon lines is the existence of a large
`number of good detectable lines in the spectral range
`available. The transition probabilities were taken from the
`literature.2 In this instance the argon lines with wavelengths
`603.2 nm (transition probability A=0.246 x lo8 s-l), 614.5
`nm (A=0.079x lo8 s-l), 667.7 nm (A=0.0241 x lo8 S - I )
`and 696.5 nm (A =0.674 x lo8 s-I) were used.
`The spatially resolved relative concentrations of atoms
`and molecules are determined by actinometry. This method
`compares the emission of the species of interest with the
`emission of, for example, an inert gas, the so called
`actinomer, which
`is added in small amounts to the
`discharge gas m i x t ~ r e . ~ The preconditions ensuring validity
`of actinometric results are as follows. The addition of the
`actinomer should not essentially influence the discharge
`characteristics, the electron energy distribution has to be
`spatially constant and the emitting states are solely excited
`by electron impact upon the ground state. The last
`condition means that dissociative excitation should be
`negligible.
`Often these preconditions are not completely fulfilled,
`therefore the plasmas under investigation are always
`checked for changes of electron energy distribution. For this
`purpose the intensity ratio of the H a to the HP line is
`assumed to be a useful measure of relative changes, as has
`been done previ~usly.~
`For the determination of translational or kinetic temper-
`ature of heavy particles from Doppler broadening of
`spectral lines, the velocity distribution of these particles is
`assumed to be Maxwellian. This is self-evident, since
`spectral linewidth measurements are only performed for
`neutral species, and since the degree of ionization of the
`plasmas is low, the kinetic temperature is assumed to agree
`with the neutral gas temperature. The influence of mi-
`crofield Stark effect is neglected owing to low absolute
`electron densities, not exceeding 1 x 1 0-12 ~ m - ~ . Because in
`most cases natural linewidths also can be negligible, the line
`profile is pure Gaussian. The full width of half maximum
`(FWHM) of the Gaussian profile directly represents the
`Doppler broadening:
`Ad = (8 x ln2 x k/mc2)'12 x Tg112 x Tg112 x Am
`where T,=neutral gas temperature (K); m=particle mass
`
`(nm) in the middle of the profile;
`(kg); A,=wavelength
`Ic=the Boltzmann constant; and c=the velocity of light
`(m s - ' ) . ~
`The H a line is an extensively studied spectral line.
`Owing to the fine structure splitting it consists of seven
`closely related components. The relative intensities are
`lfound to depend on the discharge conditions. At neutral
`species gas temperatures lower than about 600 K, this fine
`structure line splitting must be taken into account by
`deconvolution. In this temperature region the H a line
`,appears as a doublet, each component being a blend of
`different fine structure lines. An experimental separation
`of the two main components of 0.014 nm, which only
`requires a relatively
`low instrumental error of about
`nm, is sufficient for determination of the neutral
`1 x
`species gas temperature.
`Another method for the determination of the gas temper-
`ature used in this work is the measurement of the rotational
`intensity distribution of excited molecular bands. In order
`to use this method, close correspondence of the rotational
`state population to the gas temperature must be assumed.
`The energy separation between the rotational levels of the
`measured molecules is very small. This makes the rota-
`tional excitation via interaction with the translation degrees
`of freedom of the molecules very efficient. Excitation by
`electron collisions or other processes, which can result in
`non-Boltzmann distributions of the excited rotational
`levels, is therefore neglected.
`The rotational temperature is derived by measuring the
`relative intensities of rotational lines within a vibrational
`band. It is assumed that all rotational lines in the vibra-
`tional band have the same transition probability. The
`sensitivity of the apparatus is taken into account. Under
`these conditions the following relationship can be written:
`+ = R-branch
`Log(I/p)=const- [p(pk 1)B,hc]/2.3026kTro,
`- = P-branch
`where T,,, is the rotational temperature, B,, the rotational
`constant of the upper electronic level, I the rotational line
`intensity, p is the rotational quantum number and k the
`Boltzmann constant. In a Boltzmann plot diagram
`log(l/p)=flpO,-t l)] the slope of the straight line gives the
`rotational temperature.
`In this paper results of rotational temperature measure-
`ments made by means of nitrogen molecule emission are
`reported. A small amount of nitrogen is added to the
`plasma. When considering the nitrogen emission, the
`molecular
`band of
`the
`second positive
`system
`C311,(0)-B311,(O) with its bandhead at 337.1 nm is used. To
`minimize the influence of branch overlapping, the R-branch
`rotational line distribution from quantum numbers greater
`than 20 is measured. For these large quantum numbers the
`rotational lines were clearly identified. In this case the
`deviations from the linear relationship in the Boltzmann
`plot were less than 5%.
`
`Experimental
`The measurements were performed using a special planar
`microwave plasma source (v= 2.45 GHz), a capacitively
`coupled r.f. model discharge tube (v=460 kHz) and an r.f.
`planar reactor (v= 13.56 MHz). In each case argon was used
`as the carrier gas. Hexamethyldisiloxane was added in
`concentrations of 0.2-10%.
`The optical diagnostic system, a schematic diagram of
`which is shown in Fig. 1, in combination with the
`microwave plasma source, consists of a computer con-
`trolled 2 m grating spectrometer and a high sensitivity
`optical multichannel analyser system, OMA-VISION
`(PAR).
`
`Published on 01 January 1993. Downloaded by University of Minnesota - Twin Cities on 12/04/2014 23:29:01.
`
`View Article Online
`
`TSMC-1109 / Page 2 of 6
`
`
`
`JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1993, VOL. 8
`
`805
`
`Gas handling unit
`
`Ar -+Q
`
`HMDSO
`
`I--+} 7
`{-G
`>
`
`Plasma source
`
`Microwave supply
`
`Vacuum vessel
`
`c
`
`Mechanical
`translation
`system
`
`I
`
`'
`
`I
`
`El,
`
`Fig. 1 Schematic diagram of the experimental set-up including the planar microwave plasma source
`
`The 2 m grating spectrometer with quartz collimation
`optics and a grating with 1302 grooves mm-' achieves a
`wavelength resolution of about 1 x
`nm. This allows
`Doppler line broadening measurements of the red Ha line
`for determination of neutral species gas temperature in the
`plasma down to ambient temperature. With respect to the
`plasma conditions, which are characterized by a variety of
`low intensity atomic lines and molecular bands and by
`relatively high noise levels, lock-in signal detection is
`used. The computer controlled data collection system used
`allows further treatment of the stored spectra. Fit and
`deconvolution procedures are very important, but smooth-
`ing, spreading out or averaging procedures were also
`used.
`The OMA-VISION system is equipped with a 0.5 m
`grating monochromator and a Peltier cooled highly
`sensitive UV intensified charge coupled device (CCD)
`matrix detector. The maximum sensitivity of this detector
`is 4-10 photons per count. This system is
`used for
`
`molecular band analysis for determination of the rotational
`temperature and for atomic line intensity measurements.
`Molecular rotational band analysis is performed using a
`grating with 2400 grooves mrn-l. This provides a wave-
`length resolution of about 0.1 nm, which is sufficient for
`most cases. Atomic line intensities were measured using a
`grating with 600 grooves mrn-l. For spatially resolved
`measurements, a quartz telescope and a rectangular slit-
`like aperture are mounted on a mechanical translation
`system combined with a quartz light guide cable used for
`light transfer to the monochromators.
`The planar microwave plasma source investigated works
`on the principle of distributed coupling of microwave
`power into the vaccum vessel. In this way the difficulties
`concerning the generation of large area microwave plasmas
`are overcome, which arise mainly because of the short
`wavelength of microwaves and their low penetration depth
`into the plasma. For a more detailed description of this type
`of microwave plasma source see ref. 6.
`
`Telescope
`
`I Opticall window
`
`Molybder
`
`im boat
`
`Substrate
`Microwave window
`
`X
`
`n\\\\\U
`
`ry
`
`- 7 Substrate
`
`Vessel wall
`
`I
`
`Thermocouples
`
`II
`
`holder
`
`Fig. 2 Details of the plasma region and substrate arrangement (x, distance to the microwave window)
`
`Published on 01 January 1993. Downloaded by University of Minnesota - Twin Cities on 12/04/2014 23:29:01.
`
`View Article Online
`
`TSMC-1109 / Page 3 of 6
`
`
`
`806
`
`,
`
`4;,
`
`kHz
`
`A r + H e HMDSO
`
`JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1993, VOL. 8
`
`\$; ,
`
`Electrode
`
`-
`
`Observation
`
`,--+--+
`
`Observation
`
`Pump
`
`Fig. 3 Schematic diagram of the r.f. model discharge tube (I, side-on observation position)
`
`The test source used generates a long, extended, laterally
`homogeneous, planar plasma for low-pressure plasma pro-
`cessing with lateral dimensions of 14 x 4 cm. Sources with
`up to 50 x 10 cm lateral dimensions have been achieved. In
`Fig. 2 details of the microwave plasma region and substrate
`arrangement are shown. The microwaves which penetrate
`into the reaction vessel through the quartz microwave
`window are strongly absorbed. Therefore, the plasma is
`strongly decaying in the direction normal to the window.
`The thickness of the active microwave power absorbing
`plasma region, with electron densities above the so called
`cut-off density of about 7 x 1 0lo ~ m - ~ ,
`does not exceed a few
`centimeters at low pressures, and increasing the pressure
`compresses this region. The active plasma region is fol-
`lowed by another decaying, diffusive afterglow plasma
`region. Usually, the extended planar substrates for plasma
`polymerization are placed in this afterglow region.
`The optical emission of optically thin plasmas was
`spatially resolved, being observed in an end-on configura-
`tion by means of a special quartz telescope, resulting in
`laterally integrated information. Plasma homogeneity along
`the line of sight was ensured. Therefore, margin effects of
`the plasma edge regions, which significantly influence
`linewidth measurements, can be neglected.
`Fig. 3 shows a schematic diagram of the r.f. model
`discharge tube used. To ensure well defined, stable dis-
`charge conditions a gas flow regime with a separate supply
`I
`
`.-
`0
`
`C
`
`'5 0.2
`C
`-
`
`10
`0
`
`5
`
`10
`15
`xlmm
`Fig. 4
`(a) Dependence of the intensity ratios: A, C2 516 nm
`system:Ar 750.4 nm; B, CH 430 nm system:Ar 750.4 nm; C, Si
`288.2 nm:Ar 750.4 nm; and D, HmAr 750.4 nm ( x 10) on the
`distance to the microwave window. (b) Dependence of the intensity
`ratio Ha:HP on the distance to the microwave window. Ar+0.2%
`HMDSO+2.5Oh He; p=230 Pa; and substrate position, s=25 mm
`
`20
`
`25
`
`(a)
`
`1.6
`
`1.4 -
`1.2 -
`1.0 -
`0.8 -
`0.6 -
`
`A
`
`A
`
`D
`
`1
`
`I
`
`of the monomer downstream of the argon supply is
`necessary. The r.f. energy (v=460 kHz) with maximum 40
`'W power is capacitively coupled to the plasma. The over-all
`(discharge volume is about 50 cm3. In addition to the end-on
`observation, side-on observation of the spectral emission is
`also possible.
`The 13.56 MHz r.f. planar reactor is of the usual type. It
`consists of two circular electrodes, one of which is grounded
`and the other powered by capacitive coupling. Both
`electrodes have the same diameter of 13 cm. The distance
`between the electrodes is 4 cm. The r.f. power is supplied
`via a matching network. The optical emission is observed
`side on. Additionally, Langmuir probe and mass spectrome-
`try measurements are possible. The gas supply is organized
`in a flow regime using a flow controller gas inlet system.
`
`Results and Discussion
`In Fig. 4(a) measurements of the relative spatial concentra-
`tion distributions of Si atoms, H atoms, CH radicals and C2
`radicals between the microwave quartz window and the
`substrate in the microwave plasma source are shown with
`addition of 0.2% HMDSO. The species exhibit pronounced
`local concentration maxima, which are shifted toward the
`active plasma region near the window. Under these dis-
`charge conditions, atomic line emission of the helium
`actinomer is undetectable, hence an argon line (750.4 nm)
`is used for actinometric estimation. The low helium
`emission is believed to be due to the relatively low electron
`temperature. In contrast, between the electrodes of the r.f.
`25
`
`1 1/11
`
`(a)
`
`15
`
`l o t
`
`I
`
`I
`
`30
`
`40
`
`0
`
`10
`
`20
`dlm m
`Fig. 5
`(a) Dependence of the intensity ratio of A, C2 516 nm
`system:He 587.6 nm and B, CH 430 nm system:He 587.6 nm on
`the position between the electrodes, d, of the r.f. planar reactor.
`(b) Dependence of the intensity ratio Har:H/? on the position
`between the electrodes, d, of the r.f. planar reactor. Ar+ 10%
`HMDSO+ 5% He, p = 70 Pa, d=O mm grounded electrode
`
`Published on 01 January 1993. Downloaded by University of Minnesota - Twin Cities on 12/04/2014 23:29:01.
`
`View Article Online
`
`TSMC-1109 / Page 4 of 6
`
`
`
`JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, SEPTEMBER 1993, VOL. 8
`
`807
`
`100
`
`I
`
`
`
`1
`0
`
`I
`I
`0.50
`0.25
`Relative wavelengthIA
`
`I
`0.75
`
`Fig. 8 Experimental H a line profile fitted by Gaussian profiles
`
`than in the r.f. plasma sources [Figs. 5(b) and 6(b)]. This
`indicates the expected lower electron energy of microwave
`plasmas. A certain flat maximum between the microwave
`window and the substrate must be recognized. In accor-
`dance to the H a to HP intensity rdtio, the spatial distribu-
`tion of the excitation temperature T,,,, determined by argon
`line intensity relationships, shows a flat minimum between
`the microwave window and the substrate (Fig. 7). The
`tendency to decrease with increasing distance from the
`microwave window agrees well with former Langmuir
`probe measurements.' In the vicinity of the substrate the
`excitation temperature seems to be influenced by this
`additional wall.
`In contrast to the excitation temperature, the neutral
`species gas temperature exhibits larger spatial variations.
`In the r.f. model discharge tube, the neutral species gas
`temperature is derived from Doppler line broadening of the
`red H a line (656.3 nm) and rotational band analysis of the
`nitrogen molecule. The influence of fine structure line
`splitting on Doppler broadening is taken into account by
`using numerical fit procedures. This is recommended
`particularly in the medium temperature range exhibiting
`incomplete line splitting. So the error in temperature
`determination is reduced to 5-10%. Fig. 8 shows an
`example of such an experimental H a line profile fitted by
`Gaussian profiles. In Fig. 9 the dependence of the neutral
`species gas temperature, Tg, on the supplied power, P, is
`shown. The value of Tg increases slightly with increasing
`power. In contrast to this, the analysis of the nitrogen
`molecular emission resulted in rotational temperatures of
`
`600
`
`A
`
`/
`
`550
`
`500
`
`I
`
`20
`
`1
`30
`
`I
`40
`
`50
`
`PMI
`Fig. 9 Dependence of the neutral gas temperature, Tg, on the
`power, P, in the r.f. model discharge tube ( 0 , Ar 5 5 Pa: A, Ar 87
`Ar+ 10% HMDSO 60 Pa, derived from Ha Doppler
`Pa; and
`line broadening)
`
`50
`
`0 .-
`c. h
`r
`c. .-
`fn
`
`; o - l l b i l
`
`5
`0
`
`100
`
`200
`Ilm m
`
`300
`
`400
`
`(a) Dependence of the intensity ratio of: A, C2 516 nm
`Fig. 6
`system:He 587.6 nm ( x loT2); B, CH 430 nm system:He 587.6 nm
`and C, Si 288.2 nm: He 587.6 nm on the side-on
`( x
`observation position, 1, in the r.f. model discharge tube. (b)
`Dependence of the intensity ratio Ha:HP on the side-on observa-
`tion position, 1, in the r.f. model discharge tube Ar+lO%
`HMDSO+2% He, p=70 Pa
`planar reactor such a concentration maximum of the CH
`and C2 molecules is not observable [Fig. 5(a)].
`In the r.f. model discharge tube, maxima of the relative
`Si, CH and C2 concentrations appear directly downstream
`from the position where the HMDSO feed gas comes into
`the tube. For this discharge, spatial distributions of the
`polymer deposition rate were observed. The optical emis-
`sion maxima [Fig. 6(a)] correlate with the spatial depen-
`dence of the deposition rate.
`In all of the plasmas, the validity of actinometry was
`ensured. The ratios of the hydrogen line intensities indicate
`an almost constant electron temperature over the plasma
`region investigated [Figs. 4(b), 5(b) and 6(b)]. No order of
`magnitude changes were observed. One possible explana-
`tion for this effect is the relatively low level of HMDSO,
`resulting in an argon-hydrogen gas mixture of constant
`composition as the dominant plasma gas component over
`the entire discharge volume. The HMDSO is known to
`produce large amounts of atomic hydrogen under the
`influence of gas discharges. The ratio of H a to HP in the
`microwave plasma source [Fig. 4(b)] is essentially higher
`
`0.6
`
`I-.- 0.5
`
`0.4
`0
`
`I
`
`5
`
`I
`
`10
`x/mm
`
`I
`
`15
`
`I
`
`20
`
`25
`
`Fig. 7 Dependence of the excitation temperature T,,, on the
`distance to the microwave window, x, (Ar+0.2% HMDSO; p=230
`Pa; substrate position: s=25 mm, derived from argon line
`intensities)
`
`Y '
`.,
`
`450 1
`I
`
`Ann 1
`7"" 0
`
`0
`
`1
`
`10
`
`Published on 01 January 1993. Downloaded by University of Minnesota - Twin Cities on 12/04/2014 23:29:01.
`
`View Article Online
`
`TSMC-1109 / Page 5 of 6
`
`
`
`808
`
`JOURNAL O F ANALYTIICAL ATOMIC SPECTROMETRY, SEPTEMBER 1993, VOL. 8
`
`2ooo :
`-
`
`iaoo
`5 1600
`
`G
`
`1400
`
`-
`
`-
`
`transfer are different. In the r.f. plasma source the plasma
`sheath, which transfers most of the energy, is localized at
`the powered electrode. This is due to capacitative coupling
`of this electrode. In the microwave discharge the localiza-
`tion of power absorption is due to the low penetration depth
`of microwaves into the plasma.
`A comparison of all measurements, especially the re-
`ported considerable spatial gradients of neutral species gas
`temperature measurements as well as particle density
`measurements, clearly demonstrates the complexity of the
`problem under investigation. If different optical tempera-
`ture measuring methods are used to obtain information
`concerning the energy of species in the plasmas, possible
`error sources have to be taken into account. The validity of
`the measured excitation temperatures depends on the
`uncertainty of the transition probabilities used, which can
`be estimated to be typically 20-30O/0.~~ In contrast, the error
`in neutral gas temperature determination by Ha line
`broadening is about 5- 1 O%, and dissociative excitation can
`be neglected. The error in the rotational temperature
`measurements is also about this value.
`The rotational temperatures of nitrogen molecules exhibit
`strong deviations in comparison with the neutral species gas
`temperature. The addition of nitrogen disturbs the plasmas
`being considered. Therefore, measurements of the rotational
`temperature of hydrogen molecules will be more useful for
`comparison purposes in further investigations.
`It is clear that comprehensive, complex emission spectro-
`metric measurements are not sufficient to compare r.f. and
`microwave plasma sources satisfactorily. In particular, the
`absolute necessity of spatially resolved measurements has
`been demonstrated, i.e., at present only a comparison of
`locally and energetically well defined plasma source volume
`elements seems to be promising.
`
`The authors would like to thank Urte Kellner and Dietmar
`Gott for carrying out measurements and for technical
`assistance and Mathias Schaller for software development.
`This work was supported by BMFT (Federal Ministry of
`Research and Technology) of Germany, Contract No.
`13N5939.
`
`1
`
`2
`
`3
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`11
`
`12
`13
`
`References
`Wertheimer, M. R., and Moisan, M., J. Vac. Sci. Technol.,
`1985, A3, 2643.
`in Inductively Coupled Plasma Emission
`Mermet, J.-M.,
`Spectroscopy, Part 11: Applications and Fundamentals, ed.
`Boumans, P. W. J. M., Wiley-Interscience, New York, 1987.
`Coburn, J. W., and Chen, M., J. Vac. Sci. Technol., 198 1,18,353.
`Lopata, E., and Countrywood, J., J. Vac. Sci. Technol., 1988,
`A6, 2949.
`Griem, H. R., Plasma Spectroscopy, McGraw-Hill, New York,
`1964.
`Ohl, A., in Microwave Discharges: Fundamentals and Applica-
`tions, ed. Ferreira, C. M., NATO AS1 Series, Plenum, New
`York, in the press.
`Ropcke, J., and Ohl, A., Contrib. Plasma Phys., submitted for
`publication.
`Krysmanski, K. H., and Walter, W., Beitr. Plasmaphys., 1978,
`18, 367.
`Marawi, I., Bielski, B. A., Caruso, J. A., and Meeks, F. R., J.
`Anal. At. Spectrom., 1992, 7 , 899.
`Wojaczek, K., Beitr. Plasmaphys., 1984, 24, 551.
`Neumann, W., Ergebnisse der Plasmaphysik und Gaselektro-
`nik, eds. Rompe, R., and Steenbeck, M., Akademie-Verlag,
`Berlin, 1970.
`Ropcke, J., and Ohl, A., Contrib. Plasma Phys., 199 1, 31, 669.
`Herzberg, R. G., Jaffe, S. M., Larjo, J., Saari, J., and
`Vattulainen, J., ISPC- 10, Bochum, Germany, 199 1, conference
`abstract 1.2-6.
`Paper 3 /O I 06 OK
`Received February 22, 1993
`Accepted May 4, 1993
`
`1
`
`400
`
`1200
`0
`
`I
`
`100
`
`0
`
`I
`
`1
`300
`
`200
`Mm m
`Fig. 10 Dependence of the rotational temperature, T,,, on the
`distance, 1 along the tube (Ar+ 5% N2, p = 54 Pa, P=45 W, derived
`from nitrogen band analysis)
`more than 1000 K. Fig. 10 shows the distribution of these
`rotational temperatures along the tube. A minimum value
`of the rotational temperature in the region between the two
`electrodes can be observed. The phenomenon that nitrogen
`band analysis leads to too high rotational temperatures has
`also been reported by other worker^.^.^ In ref. 8, gas
`temperatures measured by interferometry were compared
`with rotational temperatures. The latter were considerably
`higher, although rotational Boltzmann distribution was
`ensured. Deviations from a linear relationship were less
`than 5%. However, it has also to be taken into account, that
`addition of nitrogen can influence the discharge character-
`istics. Obviously, longer relaxation times of the excited
`nitrogen molecules disturb the local equilibrium between
`molecular vibration and translation. An accumulation of
`vibrational energy in the discharge could be caused by
`differences between excitation and relaxation times. lo In
`ref. 11 correspondence between rotational and gas tempera-
`ture was predicted for higher pressures.
`In Fig. 11 the position dependence of the neutral species
`gas temperature, Tg, between the electrodes of the r.f. planar
`reactor is reported. The value of Tg is derived from Ha line
`Doppler broadening. A pronounced maximum of the
`neutral species gas temperature of about 2000 K near the
`powered electrode is detectable. This temperature distribu-
`tion is similar to the spatial distribution of the neutral gas
`temperature in the microwave discharge. The latter also
`exhibits a local maximum in the region near the microwave
`window.I2 This similarity is explained by similarities in the
`absorption energy. In both plasmas, most of the energy
`transfer from the electric field to the plasma is localized
`near one of the walls and the electrodes, respectively. Only
`the conditions of localization and the mechanisms of energy
`
`0
`
`1
`
`I
`-
`2000
`-
`
`1500
`
`-
`Y
`Za 1000
`-
`
`500
`
`0
`
`10
`
`20
`dlm rn
`
`30
`
`40
`
`Fig. 11 Dependence of the neutral gas temperature, T,, on the
`position, d, between the electrodes of the r.f. planar reactor
`(Ar+lO% HMDSO+5% He, p = 7 0 Pa, d=O mm, grounded
`electrode, derived from H a line Doppler broadening)
`
`Published on 01 January 1993. Downloaded by University of Minnesota - Twin Cities on 12/04/2014 23:29:01.
`
`View Article Online
`
`TSMC-1109 / Page 6 of 6
`
`