- Email: [email protected]
Modeling surface effects in negative-ion volume sources
6 September 199 I
CHEMICAL PHYSICS LETTERS
Volume 183, number 5
Modeling surface effects in negative-ion volume sources P. Berlemont, D.A. Skinner and M. Bacal Laboratoire de Physiquedes MilieuxIonis& Laboratoire du CNRS,Ecole Polytechnique,91128Palaiseau Cedex, France Received 23 April 199 I ; in final form I9 June 199 1
In modeling negative-ion volume sources, surface processes may play an important role in the performance to be expected from the source. We ran our simulation code and showed that increasing the surface/volume ratio, S/i’, can lead to a lower level of gas dissociation, and thus to an enhancement of the vibrational spectrum at high power.Whensomenegative-ionsurfaceproduction is added, an increase of the H- density with S/ Vis observed.
They are only lost by recombination i.e.
1. Introduction In high-power magnetic multicusp ion sources, the large amount of H atoms, which is due to the high level of dissociation, causes saturation in the production of negative ions, as was demonstrated experimentally [ I] and numerically [2]. In volume negative-ion sources, most negative ions are thought to be produced through dissociative attachment of electrons to vibrationally excited molecules e+H,(v”)+H-+H.
(DA)
It has been shown that hydrogen atoms are an important cause for de-excitation of vibrationally excited molecules HZ( II” ) in the range 6 < v” d 8 (which is the most efficient for H- production), through the so-called “V-T” process (vibrational-translational energy transfer). Besides, a high degree of dissociation leads also to a poor density of molecules which are susceptible to be vibrationally excited. In hydrogen discharges, the H atoms are mainly produced by the following processes: e+H,(v”)+e+Ht(b3C:,c311,,a3C:) +e+2H,
(ra)
No single production process is dominant. Therefore, the most effective way to reduce the atomic density is to increase the recombination rate of atoms on the wall. This can be done by finding a wall material with a high recombination factor yH, or by increasing the ratio surface/volume, S/K In the calculations presented in this Letter, we take the value YH= 0.1, which usually gives good simulation results and which is close to the value determined by Wise and Wood [3]. As stated above, dissociative attachment is believed to be responsible for H- production in volume sources. However, positive ions colliding with low workfunction surfaces (covered with Cs or Ba) can also lead to the formation of negative ions with a non-negligible probability, even at low energies (a few electronvolt). In the same way, thermal atoms reflecting on such surfaces also yield negative ions. We thus add in our code the following processes (for a review, see for example ref. [ 41):
(DS)
H,+ +e(wall)+H+... Hi +e(volume)+H+... H: +Hz(u”)-+H+H: , with n= I, 2 or 3. 0009-2614/91/S
H+H(wall)+H2(v”).
on the walls,
H,+ +wall+H-+...
,
(0.01; 0.05) ,
(si)
(0.0001; 0.0005) .
(sa)
(h) ,
H+wall+HWV) (IM)
The numbers on the right of the reactions indicate the probabilities of H- formation used in the code for the two sets of calculations shown below. Ac-
03.50 0 1991 Elsevier Science Publishers B.V. All rights reserved.
397
Volume 183.number 5
6 September I99 1
CHEMICALPHYSICSLETTERS
cording to van OS et al. [ 51, when approaching the
surface, positive ions can be neutralized (and resultant atoms are scattered or implanted) and also sputter adsorbed atoms from the surface. They calculated a negative-ion-formation probability of 0.1 at IOeV for hydrogen atoms leaving normal to a Bacovered W surface. They also pointed out that the conversion efficiency, i.e. the ratio between the extracted negative-ion current and the positive-ion current, depends on the discharge current and how the surface is contaminated by evaporated material from the filaments [ 61. Seidl et al. [ 71 show that 60% of H atoms with an energy greater than 1.5 eV (for a temperature of 0.22 eV) will produce negative ions when colliding on a surface covered with cesium oxide. For this temperature, it would give a probability of 0.005. The values taken in our calculations for the processes labeled “si” and “sa” were chosen because, as we will see below, they give a sensible but not dominant contribution to the negative-ion production. According to Hiskes and Karo, vibrationally excited molecules can be generated by molecular ions incident upon metal surfaces [8]. However, to our knowledge, no experiments in the scale of energy involved in our kind of plasmas (less than z 10 eV) confirm these theoretical calculations. Besides, our simulation results usually show higher H,( v” ) populations than measurements, and adding a new production process would increase this difference. Nevertheless, we include this process in further calculations. The zero-dimensional time-dependent model used for the simulation is based on the code developed by Gorse et al. [9-l 11. It solves simultaneously the
Boltzmann equation for the electron-energy distribution function (EEDF), the vibrational master equation for HZ(zY) and the kinetics for the atomic density, positive- and negative-ion densities. For the calculations presented below, a new version of this code, faster and more flexible, has been used [ 121. The original source chosen for the simulations is a one-chamber cylindrical vessel, which was studied experimentally at Lawrence Berkeley Laboratory [ 13 1. This source was a cylinder 20 cm high with a radius of 10 cm, operating in pure hydrogen. Magnets placed on the whole surface of the source provide a multicusp configuration which strongly confines the electrons. Comparisons between simulation results for this source and measurements give a fair agreement and are reported in refs. [ 11,141. We will now present results obtained from the simulations for differently sized sources, keeping constant the input power (discharge voltage V,= 120 V, discharge current I,=25 A), or keeping constant the ratio power/volume (at two values: 3 W cme3 and 0.6 W cmm3). The effect of changing the source size will be different according to the discharge conditions used.
2. Results 2.1. Pure negative-ion volume production Table 1 gives the various dimensions of the source for which the simulations were made. Subscript “gas” refers to the gas total volume and bounding surface, while subscript “plas” refers to the unmagnetized plasmas region and its surface [ 111. A first set of calculations was performed for a 120 V 8 mTorr 25 A
Table I Variousdimensions of the source studied (the original one is printed in italics) h (cm)
r (cm1 S,, (cm2I v,, (cdl V,,lS,, (cm ) SW,/V,. (cm-’ I S,,.,! VP,,,(cm-’ ) VplsrlSplas (cm) S,,., (cm?
v,,.*(cm’)
398
40 20 7540 50265 6.66 0.15 0.156 6.41 6985 44820
20 10 1880
6300 3.33 0.3 0.32 3.12 1600 5000
15 1.5
1060 2650 2.5 0.4 0.446 2.24 860 1930
10
5 471 785 1.66 0.6 0.707 1.41 341 482
6
3 170 170
1.33 0.75 95.4 71.6
Volume
183,number 5
CHEMICAL PHYSICS LETTERS
6 September 1991
0.6
0.4 0.2 0 0.3
eWY
1.3 S,#,,,
(cm-‘)
Fig. I. Normalized production rates of H atoms for 2 values of the S,.,/ V,., ratio. Labels refer to reactions defined in the text.
discharge, which were the original conditions for which most of the experimental and previous computing work was done. Keeping the discharge current constant at 1,=25 A did not lead to a decrease of the H density with increasing Splas/ VPlps.When the input power is maintained constant, the plasma density rises as the source volume is reduced. In fact, the positive-ion recombination on the wail is the main cause for the increasing atomic density (ri process, see fig. 1): the positive-ion flux to the wall is more important because of the S,,,/ VP,,,increase, and also because of the larger plasma density. We proceed with the same kind of calculations, but keeping the power/volume ratio constant at 3 W cme3. This represents an increase by a factor of 5 (I,= 125 A for the “original” source, instead of 25 A for the previous series of calculations). In these series of calculations, we attempt to have roughly the same EEDF in the high-energy range (the tail, corresponding to fast electrons). Fast electrons are important for vibrationally exciting the molecules on
Fig. 2. JZEDFsobtained for different values of the ratio S,/ Vm: (-) 0.15; (- - -)0.32; (---)0.44; (...) 0.7 and (- - -) 1.3 (cm-‘).
levels which are of interest for H(u” > 6), via the so-called EV process: e+H*(v=O)+e+Ht(B
‘Z:,
production
C'II,) ,
+e+Hz(v”)+hv.
(W
In the results presented here, at least 50% of the H,( v” = 7) are produced by the EV process. Therefore, keeping the fast-electron density constant would roughly keep the Hz (v” > 6) production rate constant. In a magnetic multipole discharge, electrons with energies larger than the plasma potential are lost in the cusps and in the non-magnetized surfaces. In our calculations, we kept constant the ratio Se& Vplap, where Seleeis the total loss surface. In our case, this value is very low, so the effect of this modification of the loss area will not be that important: electrons are principally lost in volume reactions. Run conditions are summarized in table 2. Fig. 2 represents the EEDFs obtained for the various SplJ VP,,,ratios for which calculations have been made. They are ap-
Table 2 Run conditions. Vd=120 V
Id (A) &,, (cm*) &kc/S,, V&lV,,(Wcm-‘) &Jv,, (cm-‘)
1125 108 0.015 3 0.0024
125 12 0.0075 3 0.0024
50 4.63 0.0054 3 0.0024
12.5 1.16 0.0034 3 0.0024
1.8 0.17 0.0018 3 0.0024
399
Volume 183, number 5
0.0 16
I
0
I
0.2
0.4
I
I
I
0.6
0.8
1
SPflPh
I
1.2
I
1.4
m-9
Fig. 3. Densities of the various species as function of .S,,,./ V,I_, withZ,V,/V,,,=3Wcm-3. (O)N,/lOO: (A) lON,-;(O)N,/ 1000; (+)ivu* (u”77).
proximately the same: the differences in the tail of the EEDF come from the different degree of gas dissociation, as the fast-electrons energy-loss rate depends on the gas composition through ionization and dissociation processes. In fig. 3, we have plotted the densities of electrons N,, of vibrationally excited molecules on v” =7, NH*(v” = 7)) of H atoms, NH, and of negative ions NH-. The atomic density is drastically reduced: the ratio NH/ ( NH+NH2) varies from 0.84 to 0.38 as SPlas/VPlpsvaries from 0.15 to 1.3 cm- ‘. We can see that the density of H, ( U”= 7) is increased by a factor of more than 3 when S,,,aJ/VPlrsincreases from 0.15 to 1.33 cm-‘. The main cause of destruction of the
0
la1
L.
0.15
0.3
0:4
- il.7
’ ‘1.3
sp,Jvs/v,, (cm?
Fig. 4. Normalized negative-ion production rates as function of&JV,,.,: refer to reactions defined in the text.
400
6 September 199 1
CHEMICAL PHYSICS LETTERS
vibrationally excited molecules remains the deactivation by the VT process. By increasing Splas/Vplasr we have lowered this deactivation. With increasing S,,,,l v,,,,, the negative-ion density is only increased by a factor of less than 2, but this is due to the simultaneous reduction of the electron density. The variation of N, is related to the existence of two regimes. For S,,,,/V,,,,=O.15 to 0.4 cm-‘, H+ remains the dominant positive-ion species (due to the high level of dissociation). The only way those ions are lost is by recombination on the walls: increasing S,,,,/ Vplasthen enhances the loss of those ions. For Splas/VPlaslarger than 0.4, H: becomes the dominant species (due to the higher density of Hz molecules); since Hz can also recombine in volume, the decrease in the plasma density is less pronounced (see NCin fig. 3). We performed another set of calculations with the original (lower) input power per volume (0.6 W crne3, kept constant, which corresponds to I,=25 A for the “original” source). We observe an enhancement of the vibrational population in the range Splal/ VPlas<0.4 cm- ‘, followed by saturation, although the atomic density was reduced in the whole Splas/ Vplas range. For this case, the wall relaxation of molecules on the surface is important#’ and is favoured for high values of Splas/vPlas. “’ We assume in our calculations a deactivation probability y( v” ) for H2(u”)+HZ(u) y(u”>7)=1.
0
lb1
varying
. _, .,. .
.
0.15
0.3
0.4
s&V,,
from
y(v”=l)=O.S
to
.
0.7 -
1.3‘-.’
(cm-‘)
(a) low; (b) high probabilities for surface production. Labels
CHEMICAL PHYSICS LETTERS
Volume 183, number 5
1091”“*,.‘,~~~~~‘~~~~~,~,.~,~~~ 0
0.2
0.4
0.6 Spflpb
0.8
1
1.2
1.4
(cd
Fig. 5. Negative-ion density as function of S&V,,.. with and without surface production. (0 ) low,(A ) high probabilities for surface production and ( x ) nosurface production.
6 September 1991
ing the input power per volume constant under purevolume-production conditions leads to a lower fraction of atomic hydrogen for high-power discharge, and, thus, to an enhancement of the vibrational population. In order to obtain this effect, it was necessary to reduce the fast-electron losses, as the device size decreases (meaning stronger confining magnetic field). However, we found a modest enhancement (by a factor of 2) of the negative-ion density at high S/ V, due to reduced plasma density as positive-ion loss rate on the wall surface rises. Introducing additional surface production leads to a significant increase of the H- density when the S/V is enhanced.
Acknowledgement 2.2. Addition of negative ion surface production
The additional effect of negative-ion surface production is shown in figs. 4 and 5. All calculations were made for a 120 V 8 mTorr 25 A discharge (i.e. Id kept constant). In fig. 4, we have plotted the production rates of negative ions. For the first set of calculations, dissociative attachment is still the main production mechanism for all S,,,,/ Vplarratios. This is no longer the case when we increase the probabilities of H- surface production by a factor of 5. In fig. 5, the negative-ion density is plotted versus the S,,,,/ V,,,, ratio for 3 sets of calculations, corresponding to low, high and no surface production. Note that for relatively high probabilities of H- surface production, the negative-ion density is enhanced by a factor of 6 when Splas/T/plasis increased from 0.15 to 1.33 cm-‘. When increasing Splas/V,l,s for the same discharge current, the plasma density becomes higher, and, thus, the positive-ion flux to wall is more important. When no surface production is added, the H- density saturates: even if the electron density increases, which would give a higher dissociative attachment production rate, the higher temperature and slightly higher fast-electron fraction lead to an increase of destruction of negative ions by electron detachment. 3. Conclusion It has been shown that increasing S/ V when keep-
This work was supported in part by the US Air Force Office of Scientific Research (Grant No. AFOSR 89-0538). Computing means were granted by Conseil Scientifique du Centre de Calcul Vectoriel pour la Recherche, Palaiseau, France.
References [ 1IA.J.T. Holmes, L.M. Lea, A.F. Newman and M.P.S. Nightingale, Rev. Sci. Instr. 58 (1987) 223. [2] M. Bacal, P. Berlemont and D.A. Skinner, Proc. SPIE 1061 (1989) 528. [ 31 H. Wise and B.J. Wood, in: Advan. At. Mol. Phys. 3 ( 1976) 291. [4] A.W. Kleyn, Proceedings of the 5th International Symposium on Production and Neutralization of Negative Ions and Beams, Brookhaven National Laboratory, 1989, AIP Conf. Proceedings 2 10 ( 1990) 3. [S] C.F.A. van OS, P.W. van Amenfoort and J. Los, J. Appl. Phys. 64 (1988) 3863. [6]C.F.A. vanOs, K.N. Leung,A.F. Lietzke, J.W. Steamsand W.B. Kunkel, Proceedings of the 5th International Symposium on Production and Neutralization of Negative Ions and Beams, Brookhaven National Laboratory, 1989, AIP Conf. Proceedings 210 (1990) 17. [7]M. Seidl, S.T. Melnychuk, S.W. Lee and W.E. Carr, Proceedings of the 5th International Symposium on Production and Neutralization of Negative Ions and Beams, Brookhaven National Laboratory, AIP Conf. Proceedings 210 (1990) 30. [S] J.R. Hiskes andA.M. Karo, J. Appl. Phys. 67 (1990) 6621. [9] C. Gorse, M. Capitelli, J. Bretagne and M. Bacal, Chem. Phys. 93 (1985) I.
401
Volume 183, number 5
CHEMICALPHYSICS LETTERS
[ IO] C. Gorse, M. Capitelli, M. Bacal,J. Bretagneand A. Lagana, Chem. Phys. 117 (1987) 177. [ II] D.A. Skinner, P. Berlemont and M. Bacal, Proceedings of the 5th International Symposium on Production and Neutralization of Negative Ions and Beams, Brookhaven National Laboratory, 1989, AJP Conf. Proceedings 210 (1990) 557.
402
6 September 1991
[ 121D.A. Skinner, P. Bcrlemont and M. Bacal, Proceedings of the 4th European Workshop on the Production and Application of Light Negative Ions, Belfast, 26-28 March 1991,to be published. [ 131G.C. Stutzin, A.T. Young,AS. Schlachter, K.N. Leung and W.B. Kunkel, Chem. Phys. Letters 155 (1989) 475. ( 141M. Capitelli, C. Gorse, P. Berlemont, D.A. Skinner and M. Bacal,Chem. Phys. Letters 179 (1991) 48.
CHEMICAL PHYSICS LETTERS
Volume 183, number 5
Modeling surface effects in negative-ion volume sources P. Berlemont, D.A. Skinner and M. Bacal Laboratoire de Physiquedes MilieuxIonis& Laboratoire du CNRS,Ecole Polytechnique,91128Palaiseau Cedex, France Received 23 April 199 I ; in final form I9 June 199 1
In modeling negative-ion volume sources, surface processes may play an important role in the performance to be expected from the source. We ran our simulation code and showed that increasing the surface/volume ratio, S/i’, can lead to a lower level of gas dissociation, and thus to an enhancement of the vibrational spectrum at high power.Whensomenegative-ionsurfaceproduction is added, an increase of the H- density with S/ Vis observed.
They are only lost by recombination i.e.
1. Introduction In high-power magnetic multicusp ion sources, the large amount of H atoms, which is due to the high level of dissociation, causes saturation in the production of negative ions, as was demonstrated experimentally [ I] and numerically [2]. In volume negative-ion sources, most negative ions are thought to be produced through dissociative attachment of electrons to vibrationally excited molecules e+H,(v”)+H-+H.
(DA)
It has been shown that hydrogen atoms are an important cause for de-excitation of vibrationally excited molecules HZ( II” ) in the range 6 < v” d 8 (which is the most efficient for H- production), through the so-called “V-T” process (vibrational-translational energy transfer). Besides, a high degree of dissociation leads also to a poor density of molecules which are susceptible to be vibrationally excited. In hydrogen discharges, the H atoms are mainly produced by the following processes: e+H,(v”)+e+Ht(b3C:,c311,,a3C:) +e+2H,
(ra)
No single production process is dominant. Therefore, the most effective way to reduce the atomic density is to increase the recombination rate of atoms on the wall. This can be done by finding a wall material with a high recombination factor yH, or by increasing the ratio surface/volume, S/K In the calculations presented in this Letter, we take the value YH= 0.1, which usually gives good simulation results and which is close to the value determined by Wise and Wood [3]. As stated above, dissociative attachment is believed to be responsible for H- production in volume sources. However, positive ions colliding with low workfunction surfaces (covered with Cs or Ba) can also lead to the formation of negative ions with a non-negligible probability, even at low energies (a few electronvolt). In the same way, thermal atoms reflecting on such surfaces also yield negative ions. We thus add in our code the following processes (for a review, see for example ref. [ 41):
(DS)
H,+ +e(wall)+H+... Hi +e(volume)+H+... H: +Hz(u”)-+H+H: , with n= I, 2 or 3. 0009-2614/91/S
H+H(wall)+H2(v”).
on the walls,
H,+ +wall+H-+...
,
(0.01; 0.05) ,
(si)
(0.0001; 0.0005) .
(sa)
(h) ,
H+wall+HWV) (IM)
The numbers on the right of the reactions indicate the probabilities of H- formation used in the code for the two sets of calculations shown below. Ac-
03.50 0 1991 Elsevier Science Publishers B.V. All rights reserved.
397
Volume 183.number 5
6 September I99 1
CHEMICALPHYSICSLETTERS
cording to van OS et al. [ 51, when approaching the
surface, positive ions can be neutralized (and resultant atoms are scattered or implanted) and also sputter adsorbed atoms from the surface. They calculated a negative-ion-formation probability of 0.1 at IOeV for hydrogen atoms leaving normal to a Bacovered W surface. They also pointed out that the conversion efficiency, i.e. the ratio between the extracted negative-ion current and the positive-ion current, depends on the discharge current and how the surface is contaminated by evaporated material from the filaments [ 61. Seidl et al. [ 71 show that 60% of H atoms with an energy greater than 1.5 eV (for a temperature of 0.22 eV) will produce negative ions when colliding on a surface covered with cesium oxide. For this temperature, it would give a probability of 0.005. The values taken in our calculations for the processes labeled “si” and “sa” were chosen because, as we will see below, they give a sensible but not dominant contribution to the negative-ion production. According to Hiskes and Karo, vibrationally excited molecules can be generated by molecular ions incident upon metal surfaces [8]. However, to our knowledge, no experiments in the scale of energy involved in our kind of plasmas (less than z 10 eV) confirm these theoretical calculations. Besides, our simulation results usually show higher H,( v” ) populations than measurements, and adding a new production process would increase this difference. Nevertheless, we include this process in further calculations. The zero-dimensional time-dependent model used for the simulation is based on the code developed by Gorse et al. [9-l 11. It solves simultaneously the
Boltzmann equation for the electron-energy distribution function (EEDF), the vibrational master equation for HZ(zY) and the kinetics for the atomic density, positive- and negative-ion densities. For the calculations presented below, a new version of this code, faster and more flexible, has been used [ 121. The original source chosen for the simulations is a one-chamber cylindrical vessel, which was studied experimentally at Lawrence Berkeley Laboratory [ 13 1. This source was a cylinder 20 cm high with a radius of 10 cm, operating in pure hydrogen. Magnets placed on the whole surface of the source provide a multicusp configuration which strongly confines the electrons. Comparisons between simulation results for this source and measurements give a fair agreement and are reported in refs. [ 11,141. We will now present results obtained from the simulations for differently sized sources, keeping constant the input power (discharge voltage V,= 120 V, discharge current I,=25 A), or keeping constant the ratio power/volume (at two values: 3 W cme3 and 0.6 W cmm3). The effect of changing the source size will be different according to the discharge conditions used.
2. Results 2.1. Pure negative-ion volume production Table 1 gives the various dimensions of the source for which the simulations were made. Subscript “gas” refers to the gas total volume and bounding surface, while subscript “plas” refers to the unmagnetized plasmas region and its surface [ 111. A first set of calculations was performed for a 120 V 8 mTorr 25 A
Table I Variousdimensions of the source studied (the original one is printed in italics) h (cm)
r (cm1 S,, (cm2I v,, (cdl V,,lS,, (cm ) SW,/V,. (cm-’ I S,,.,! VP,,,(cm-’ ) VplsrlSplas (cm) S,,., (cm?
v,,.*(cm’)
398
40 20 7540 50265 6.66 0.15 0.156 6.41 6985 44820
20 10 1880
6300 3.33 0.3 0.32 3.12 1600 5000
15 1.5
1060 2650 2.5 0.4 0.446 2.24 860 1930
10
5 471 785 1.66 0.6 0.707 1.41 341 482
6
3 170 170
1.33 0.75 95.4 71.6
Volume
183,number 5
CHEMICAL PHYSICS LETTERS
6 September 1991
0.6
0.4 0.2 0 0.3
eWY
1.3 S,#,,,
(cm-‘)
Fig. I. Normalized production rates of H atoms for 2 values of the S,.,/ V,., ratio. Labels refer to reactions defined in the text.
discharge, which were the original conditions for which most of the experimental and previous computing work was done. Keeping the discharge current constant at 1,=25 A did not lead to a decrease of the H density with increasing Splas/ VPlps.When the input power is maintained constant, the plasma density rises as the source volume is reduced. In fact, the positive-ion recombination on the wail is the main cause for the increasing atomic density (ri process, see fig. 1): the positive-ion flux to the wall is more important because of the S,,,/ VP,,,increase, and also because of the larger plasma density. We proceed with the same kind of calculations, but keeping the power/volume ratio constant at 3 W cme3. This represents an increase by a factor of 5 (I,= 125 A for the “original” source, instead of 25 A for the previous series of calculations). In these series of calculations, we attempt to have roughly the same EEDF in the high-energy range (the tail, corresponding to fast electrons). Fast electrons are important for vibrationally exciting the molecules on
Fig. 2. JZEDFsobtained for different values of the ratio S,/ Vm: (-) 0.15; (- - -)0.32; (---)0.44; (...) 0.7 and (- - -) 1.3 (cm-‘).
levels which are of interest for H(u” > 6), via the so-called EV process: e+H*(v=O)+e+Ht(B
‘Z:,
production
C'II,) ,
+e+Hz(v”)+hv.
(W
In the results presented here, at least 50% of the H,( v” = 7) are produced by the EV process. Therefore, keeping the fast-electron density constant would roughly keep the Hz (v” > 6) production rate constant. In a magnetic multipole discharge, electrons with energies larger than the plasma potential are lost in the cusps and in the non-magnetized surfaces. In our calculations, we kept constant the ratio Se& Vplap, where Seleeis the total loss surface. In our case, this value is very low, so the effect of this modification of the loss area will not be that important: electrons are principally lost in volume reactions. Run conditions are summarized in table 2. Fig. 2 represents the EEDFs obtained for the various SplJ VP,,,ratios for which calculations have been made. They are ap-
Table 2 Run conditions. Vd=120 V
Id (A) &,, (cm*) &kc/S,, V&lV,,(Wcm-‘) &Jv,, (cm-‘)
1125 108 0.015 3 0.0024
125 12 0.0075 3 0.0024
50 4.63 0.0054 3 0.0024
12.5 1.16 0.0034 3 0.0024
1.8 0.17 0.0018 3 0.0024
399
Volume 183, number 5
0.0 16
I
0
I
0.2
0.4
I
I
I
0.6
0.8
1
SPflPh
I
1.2
I
1.4
m-9
Fig. 3. Densities of the various species as function of .S,,,./ V,I_, withZ,V,/V,,,=3Wcm-3. (O)N,/lOO: (A) lON,-;(O)N,/ 1000; (+)ivu* (u”77).
proximately the same: the differences in the tail of the EEDF come from the different degree of gas dissociation, as the fast-electrons energy-loss rate depends on the gas composition through ionization and dissociation processes. In fig. 3, we have plotted the densities of electrons N,, of vibrationally excited molecules on v” =7, NH*(v” = 7)) of H atoms, NH, and of negative ions NH-. The atomic density is drastically reduced: the ratio NH/ ( NH+NH2) varies from 0.84 to 0.38 as SPlas/VPlpsvaries from 0.15 to 1.3 cm- ‘. We can see that the density of H, ( U”= 7) is increased by a factor of more than 3 when S,,,aJ/VPlrsincreases from 0.15 to 1.33 cm-‘. The main cause of destruction of the
0
la1
L.
0.15
0.3
0:4
- il.7
’ ‘1.3
sp,Jvs/v,, (cm?
Fig. 4. Normalized negative-ion production rates as function of&JV,,.,: refer to reactions defined in the text.
400
6 September 199 1
CHEMICAL PHYSICS LETTERS
vibrationally excited molecules remains the deactivation by the VT process. By increasing Splas/Vplasr we have lowered this deactivation. With increasing S,,,,l v,,,,, the negative-ion density is only increased by a factor of less than 2, but this is due to the simultaneous reduction of the electron density. The variation of N, is related to the existence of two regimes. For S,,,,/V,,,,=O.15 to 0.4 cm-‘, H+ remains the dominant positive-ion species (due to the high level of dissociation). The only way those ions are lost is by recombination on the walls: increasing S,,,,/ Vplasthen enhances the loss of those ions. For Splas/VPlaslarger than 0.4, H: becomes the dominant species (due to the higher density of Hz molecules); since Hz can also recombine in volume, the decrease in the plasma density is less pronounced (see NCin fig. 3). We performed another set of calculations with the original (lower) input power per volume (0.6 W crne3, kept constant, which corresponds to I,=25 A for the “original” source). We observe an enhancement of the vibrational population in the range Splal/ VPlas<0.4 cm- ‘, followed by saturation, although the atomic density was reduced in the whole Splas/ Vplas range. For this case, the wall relaxation of molecules on the surface is important#’ and is favoured for high values of Splas/vPlas. “’ We assume in our calculations a deactivation probability y( v” ) for H2(u”)+HZ(u) y(u”>7)=1.
0
lb1
varying
. _, .,. .
.
0.15
0.3
0.4
s&V,,
from
y(v”=l)=O.S
to
.
0.7 -
1.3‘-.’
(cm-‘)
(a) low; (b) high probabilities for surface production. Labels
CHEMICAL PHYSICS LETTERS
Volume 183, number 5
1091”“*,.‘,~~~~~‘~~~~~,~,.~,~~~ 0
0.2
0.4
0.6 Spflpb
0.8
1
1.2
1.4
(cd
Fig. 5. Negative-ion density as function of S&V,,.. with and without surface production. (0 ) low,(A ) high probabilities for surface production and ( x ) nosurface production.
6 September 1991
ing the input power per volume constant under purevolume-production conditions leads to a lower fraction of atomic hydrogen for high-power discharge, and, thus, to an enhancement of the vibrational population. In order to obtain this effect, it was necessary to reduce the fast-electron losses, as the device size decreases (meaning stronger confining magnetic field). However, we found a modest enhancement (by a factor of 2) of the negative-ion density at high S/ V, due to reduced plasma density as positive-ion loss rate on the wall surface rises. Introducing additional surface production leads to a significant increase of the H- density when the S/V is enhanced.
Acknowledgement 2.2. Addition of negative ion surface production
The additional effect of negative-ion surface production is shown in figs. 4 and 5. All calculations were made for a 120 V 8 mTorr 25 A discharge (i.e. Id kept constant). In fig. 4, we have plotted the production rates of negative ions. For the first set of calculations, dissociative attachment is still the main production mechanism for all S,,,,/ Vplarratios. This is no longer the case when we increase the probabilities of H- surface production by a factor of 5. In fig. 5, the negative-ion density is plotted versus the S,,,,/ V,,,, ratio for 3 sets of calculations, corresponding to low, high and no surface production. Note that for relatively high probabilities of H- surface production, the negative-ion density is enhanced by a factor of 6 when Splas/T/plasis increased from 0.15 to 1.33 cm-‘. When increasing Splas/V,l,s for the same discharge current, the plasma density becomes higher, and, thus, the positive-ion flux to wall is more important. When no surface production is added, the H- density saturates: even if the electron density increases, which would give a higher dissociative attachment production rate, the higher temperature and slightly higher fast-electron fraction lead to an increase of destruction of negative ions by electron detachment. 3. Conclusion It has been shown that increasing S/ V when keep-
This work was supported in part by the US Air Force Office of Scientific Research (Grant No. AFOSR 89-0538). Computing means were granted by Conseil Scientifique du Centre de Calcul Vectoriel pour la Recherche, Palaiseau, France.
References [ 1IA.J.T. Holmes, L.M. Lea, A.F. Newman and M.P.S. Nightingale, Rev. Sci. Instr. 58 (1987) 223. [2] M. Bacal, P. Berlemont and D.A. Skinner, Proc. SPIE 1061 (1989) 528. [ 31 H. Wise and B.J. Wood, in: Advan. At. Mol. Phys. 3 ( 1976) 291. [4] A.W. Kleyn, Proceedings of the 5th International Symposium on Production and Neutralization of Negative Ions and Beams, Brookhaven National Laboratory, 1989, AIP Conf. Proceedings 2 10 ( 1990) 3. [S] C.F.A. van OS, P.W. van Amenfoort and J. Los, J. Appl. Phys. 64 (1988) 3863. [6]C.F.A. vanOs, K.N. Leung,A.F. Lietzke, J.W. Steamsand W.B. Kunkel, Proceedings of the 5th International Symposium on Production and Neutralization of Negative Ions and Beams, Brookhaven National Laboratory, 1989, AIP Conf. Proceedings 210 (1990) 17. [7]M. Seidl, S.T. Melnychuk, S.W. Lee and W.E. Carr, Proceedings of the 5th International Symposium on Production and Neutralization of Negative Ions and Beams, Brookhaven National Laboratory, AIP Conf. Proceedings 210 (1990) 30. [S] J.R. Hiskes andA.M. Karo, J. Appl. Phys. 67 (1990) 6621. [9] C. Gorse, M. Capitelli, J. Bretagne and M. Bacal, Chem. Phys. 93 (1985) I.
401
Volume 183, number 5
CHEMICALPHYSICS LETTERS
[ IO] C. Gorse, M. Capitelli, M. Bacal,J. Bretagneand A. Lagana, Chem. Phys. 117 (1987) 177. [ II] D.A. Skinner, P. Berlemont and M. Bacal, Proceedings of the 5th International Symposium on Production and Neutralization of Negative Ions and Beams, Brookhaven National Laboratory, 1989, AJP Conf. Proceedings 210 (1990) 557.
402
6 September 1991
[ 121D.A. Skinner, P. Bcrlemont and M. Bacal, Proceedings of the 4th European Workshop on the Production and Application of Light Negative Ions, Belfast, 26-28 March 1991,to be published. [ 131G.C. Stutzin, A.T. Young,AS. Schlachter, K.N. Leung and W.B. Kunkel, Chem. Phys. Letters 155 (1989) 475. ( 141M. Capitelli, C. Gorse, P. Berlemont, D.A. Skinner and M. Bacal,Chem. Phys. Letters 179 (1991) 48.