Increased atomic hydrogen flux from a cascaded arc plasma source by changing the nozzle geometry

Article abstract of DOI:10.1063/1.1879112

A very high flux of hydrogen atoms with energies in the eV range has been obtained by using a thermal plasma source and by optimization of the nozzle exit geometry. It proves that the flux of hydrogen atoms emerging from a cascaded arc plasma source depends strongly on the geometry of the nozzle. By decreasing the nozzle length by a factor 2, the atomic hydrogen flux is increased by a factor of 13, and a further increase of a factor of 2.5 can be obtained by increasing the nozzle diameter. The resulting atomic hydrogen flux is 1.2×1021 s-1, corresponding to a dissociation degree of over 30%. It is argued that the main loss channel for atomic hydrogen is surface recombination, and that by using nozzle geometries that reduce the surface loss, the atomic flux is increased.

Full text of DOI:10.1063/1.1879112

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Increased atomic hydrogen flux from a cascaded arc plasma source by changing the
nozzle geometry
P. Vankan, R. Engeln, and D. C. Schram
Citation: Applied Physics Letters 86, 101501 (2005); doi: 10.1063/1.1879112
View online: http://dx.doi.org/10.1063/1.1879112
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/86/10?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Production of high transient heat and particle fluxes in a linear plasma device
Appl. Phys. Lett. 97, 081502 (2010); 10.1063/1.3484961
Charged particle dynamics and molecular kinetics in the hydrogen postdischarge plasma
Phys. Plasmas 13, 113505 (2006); 10.1063/1.2395929
Study of the Ar metastable atom population in a hollow cathode discharge by means of a hybrid model and
spectrometric measurements
J. Appl. Phys. 97, 123305 (2005); 10.1063/1.1929857
Numerical studies of collisionless and collisional sheath evolution in plasma source ion implantations
Phys. Plasmas 12, 043501 (2005); 10.1063/1.1861057
Transition from the constant ion mobility regime to the ion-atom charge-exchange regime for bounded collisional
plasmas
Phys. Plasmas 12, 023502 (2005); 10.1063/1.1844512
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:
129.49.251.30 On: Wed, 10 Dec 2014 21:48:00

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
APPLIED PHYSICS LETTERS 86, 101501 ͑2005͒
Increased atomic hydrogen flux from a cascaded arc plasma source
by changing the nozzle geometry
P. Vankan, R. Engeln,a͒ and D. C. Schram
Department of Applied Physics, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven,
The Netherlands
͑Received 30 September 2004; accepted 17 January 2005; published online 1 March 2005͒
A very high flux of hydrogen atoms with energies in the eV range has been obtained by using a
thermal plasma source and by optimization of the nozzle exit geometry. It proves that the flux of
hydrogen atoms emerging from a cascaded arc plasma source depends strongly on the geometry of
the nozzle. By decreasing the nozzle length by a factor 2, the atomic hydrogen flux is increased by
a factor of 13, and a further increase of a factor of 2.5 can be obtained by increasing the nozzle
diameter. The resulting atomic hydrogen flux is 1.2ϫ1021 s−1, corresponding to a dissociation
degree of over 30%. It is argued that the main loss channel for atomic hydrogen is surface
recombination, and that by using nozzle geometries that reduce the surface loss, the atomic flux is
increased. © 2005 American Institute of Physics. ͓DOI: 10.1063/1.1879112͔
Atomic hydrogen has a strong technological relevance,
since H atoms often serve as primary reactive particles. Typi-
cal applications can be found in areas like surface modifica-
tion and thin film deposition, for example, the deposition of
nanocrystalline silicon layers and diamond.1–5 In this type of
application, hydrogen atoms play an important role in the
crystallization process and in the formation of the radicals
from the precursors. Larger fluxes of H atoms allow for
higher deposition rates and therefore more cost-efficient pro-
duction.
Another type of application where large fluxes of atomic
hydrogen can make a difference is in the area of molecule
formation via plasma-activated catalysis.6 In such applica-
tions the injected gases are fragmented and subsequently
converted at the catalytic surface into different stable mol-
ecules. An example is the production of ammonia in N2–H2
discharges.6,7 The initial gases are partly dissociated and re-
act on the nearby catalytic surfaces to form ammonia. Other
examples are the catalytic formation of methanol and hydra-
zine. The production efficiency in all of these examples de-
pends strongly on atomic fluxes.
been performed at axial positions close to the source.
The hydrogen atoms emerging from the plasma source
are spatially probed using the two-photon absorption laser
induced fluorescence ͑TALIF͒ technique.9,10 A Nd:YAG laser
pumped dye laser is operated at a wavelength of around 615
nm. The laser beam is frequency tripled using a KDP and a
BBO crystal, resulting in tunable, 8 ns pulses at a wave-
length of around 205 nm. The 205 nm photons are used to
excite H atoms from the 1s2S ground state to the 3d2D and
3s2S excited states, via a two-photon transition. The resulting
Balmer-␣ fluorescence is detected by a photomultiplier tube.
From a spectral scan over the two-photon transition the H
atom density, temperature and velocity along the laser beam
are obtained. Since the laser is directed into the plasma ex-
pansion parallel to the expansion axis, axial velocities, and
therefore, via the multiplication with the density, axial flux
densities ␾H=nH are determined. By moving the plasma
vz
source radially or axially, spatial profiles of the axial atomic
hydrogen flux density can now be measured with a spatial
resolution of less than 1 mm. In all of the measurements the
dye laser wavelength is calibrated by the simultaneous re-
cording of the well known absorption spectrum of molecular
iodine. The measurements are furthermore corrected for all
of the experimental parameters ͑laser power, optical trans-
mission, and detector efficiency͒ and the linearity is checked.
The densities are calibrated using a two-photon transition in
a known amount of krypton.9 Since the ratio of the two-
photon excitation cross sections of krypton and hydrogen is
known,11 the relative hydrogen densities can be made abso-
lute.
We have studied the H atom fluxes emerging from the
plasma source using three different nozzles and as function
of the hydrogen flow, while all other experimental param-
eters like arc current and background pressure are kept con-
stant. The first nozzle has a length of L=14 mm and a con-
stant diameter of 4 mm. This nozzle will be referred to as the
long nozzle. The second nozzle has the same diameter, but its
length is L=7 mm instead of 14 mm. This nozzle will be
referred to as the short nozzle. In Fig. 1 the axial H atom flux
density as a function of the radial position at z=2 mm from
the source using the long and the short nozzle are shown.
We present measurements of the atomic hydrogen flux
emerging from a cascaded arc plasma source.8 The cascaded
arc has the advantage of operating at high pressure, high
power and small volume, leading to a high power density.
The arc channel with a diameter of 4 mm and a length of 50
mm is formed by the central holes in four insulated cascade
plates. A current of 60 A generates a thermal plasma in flow-
ing hydrogen gas at a typical power of 8 kW. The molecular
hydrogen flow through the arc is ⌽0H =1.25ϫ1021 s−1 unless
stated otherwise. Due to the power d2issipation, a thin plasma
channel is formed in the middle of the arc channel mainly
consisting of atomic hydrogen. The plasma source is con-
nected to the reactor vessel using a grounded cylindrical
nozzle, which also serves as the anode. The plasma expands
supersonically from the source, where the pressure is 90
mbar, into the reactor vessel, which is kept at a pressure of 1
mbar. All of the measurements presented in this letter have
a͒
Author to whom correspondence should be addressed; electronic mail:
r.engeln@phys.tue.nl
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:
0003-6951/2005/86͑10͒/101501/3/$22.50 86, 101501-1 © 2005 American Institute of Physics
129.49.251.30 On: Wed, 10 Dec 2014 21:48:00

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
101501-2 Vankan, Engeln, and Schram
Appl. Phys. Lett. 86, 101501 ͑2005͒
FIG. 2. Axial atomic hydrogen flux as function of the flow of H2 through
the cascaded arc for three different nozzles. The flux using the short and the
long nozzle has been measured at an axial position of z=2 mm and the flux
using the large diameter nozzle at z=8 mm.
FIG. 1. Axial atomic hydrogen flux density at 2 mm from the source as
function of the radial position, using the short and the long nozzle. Note that
the flux densities using the long nozzle have been multiplied by 5 for a
better graphical comparison. A schematic representation of both nozzles has
been added. The cross indicates the measurement position.
time also varies with the radial position via the convection
velocity, but on average 1.4 ␮s is a reasonable value. Assum-
ing a recombination coefficient of ␥=0.3 to Ϫ0.5 of H atoms
on copper13,14 this crude estimation predicts a loss in atomic
flux of around one order of magnitude over ⌬L, which agrees
well with the observed factor of 13. It readily follows that,
contrary to volume recombination, surface recombination
does give a good explanation for the loss of atomic flux.
The atomic hydrogen flux is furthermore measured as
function of the H2 flow through the arc for both the long and
the short nozzle, as shown in Fig. 2. In both cases, the flux
increased more than linear with increasing flow. So the dis-
sociation degree increased with increasing flow. This reduc-
tion in the loss is explained by less efficient surface recom-
bination. The H atom and H2 molecule densities in the arc
increase linearly with the H2 flow, leading to an increase of
the diffusion time of H atoms towards the surface. Therefore,
the probability of an atom to reach the surface sufficiently
fast decreases and with that the loss probability via surface
recombination. Hence more atoms survive the transit through
the nozzle.
Note that the flux density using the long nozzle is multiplied
by a factor of 5 to be able to better compare it graphically. A
schematic representation of both nozzles has been added to
the figure. In the case of the short nozzle a maximum flux
density of 4.0ϫ1025 m−2 s−1 is reached on axis. From the
measurements the total axial H atom flux ⌽H can be deter-
mined. For the short nozzle it is ⌽short=4.0ϫ1020 s−1 and for
the long nozzle the H atom flux equals ⌽long=3.1
ϫ1019 s−1. Using ⌽H, the dissociation degree ␤ can be de-
termined via
⌽
2⌽0H
H
␤ =
.
͑1͒
2
For the short nozzle a dissociation degree of ␤=0.16 is
reached at z=2 mm, whereas it is 0.012 using the long
nozzle. This difference is rather remarkable, since the only
difference is the nozzle length. In other words, the extra 7
mm length of the long nozzle induces a loss of the H atom
flux of a factor of 13.
Several improvements can now be implemented using
the knowledge that surface recombination is the main loss
channel for hydrogen atoms. First, the wall should be as far
from the plasma as possible and second, also in the nozzle
region power should be dissipated, to better confine the
plasma. Both these improvements are accomplished using a
nozzle with a larger diameter. Using this nozzle, there is a
sudden increase in the radius from 2 mm to 6 mm after the
last cascade plate, enlarging the distance between surface
and plasma. Furthermore, the current cannot attach immedi-
ately at the beginning of the nozzle, but attaches further
down on the nozzle, dissipating power further downstream
compared to the situations using the short and long nozzle. A
schematic representation of the nozzle is given in Fig. 3.
The results for the atomic hydrogen flux are shown in
Fig. 3, where the radial profile of the flux density at z
=8 mm in the large diameter case is compared to that of the
short nozzle at 8 mm. Note that the distance z is the distance
from the last narrow opening, as indicated in the figure.
Equal z now means equally far expanded. Since in the case
of the large diameter nozzle no velocities could be deter-
mined because of a too low signal-to-noise ratio, the veloci-
This enormous loss cannot be accounted for by volume
recombination of the H atoms. At the highest density ͑near
the wall͒ of 1023 m−3 and using a three-body recombination
rate for H atoms of k3b=5ϫ10−44 m6 s−1,12 a characteristic
time for the volume loss of H atoms of ␶3b=2 ms can be
calculated. The characteristic convection time ␶conv=⌬L/wz,
the time a particle needs to transit the extra length ⌬L of the
long nozzle ͑7 mm͒ at the convection velocity wz ͑around
7000 m/s͒, is around 1 ␮s. From a comparison of these char-
acteristic times it follows that the volume recombination is
three orders of magnitude too slow to account for the loss in
H atom flux.
The only other possibility, recombination of hydrogen
atoms on the nozzle surface can explain the loss of hydrogen
atoms, as shown by the following calculation. Since the
nozzle is grounded it serves as the anode. Therefore, the
current attaches to the nozzle in the first millimeter. The
power dissipation into the plasma ends, and the atomic hy-
drogen diffuses towards the surface of the nozzle. This dif-
fusion time varies with the radial position in the arc channel,
but a good estimation is ␶diff=0.5 ␮s. This should now be
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:
compared to the characteristic time for convection ␶conv. This
ties using the short nozzle at the corresponding position have
129.49.251.30 On: Wed, 10 Dec 2014 21:48:00

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
101501-3 Vankan, Engeln, and Schram
Appl. Phys. Lett. 86, 101501 ͑2005͒
The flow dependence of the H atom flux using the large
diameter nozzle is much less strong than with the other
nozzles. It showed a slightly less than linear increase with
the H2 flow, for a slightly decreasing dissociation degree, see
Fig. 2. This would point to near total dissociation of the
plasma emerging from the plasma source. The fact that the
measurements at z=8 mm point to a dissociation degree of
0.33 rather than a value close to 1 can partly be explained by
the estimated error in the measurements of 20–30%. How-
ever, another, more probable, explanation is that the H atoms
diffuse out of the expansion during the first few millimeters,
where the wall, and thus the loss channel for H atoms, is not
too far from the plasma.
We conclude that the nozzle has a determining influence
on the flux of atomic hydrogen emerging from the plasma
source. The main loss channel for the atomic hydrogen is
surface recombination on the nozzle surface. By implement-
ing some relatively small changes in the nozzle shape, the
dissociation has already increased from around 1% to more
than 33% and H atom fluxes of more than 1021 s−1 are
reached. Besides an effect on the radical flux, surface pro-
cesses are also believed to influence the ionic fluxes, which
is confirmed by the variation in plasma emission of several
orders of magnitude between the different nozzles.
FIG. 3. Axial atomic hydrogen flux density at 8 mm from the source as
function of the radial position, using the short nozzle and the large diameter
nozzle. A schematic representation of both nozzles has been added. The
cross indicates the measurement position.
been used. This procedure is supported by the fact that a
freely expanding plasma cannot reach any surface of the
nozzle/source in the large diameter case, so the expansion is
in terms of velocity identical to the expansion using the short
nozzle. The induced error is estimated to be within 25%. The
atomic flux at z=8 mm using the large diameter nozzle is
1S. Sriraman, S. Agarwal, E. S. Aydil, and D. Maroudas, Nature ͑London͒
418, 62 ͑2002͒.
⌽
largeD=8.1ϫ1020 s−1, whereas it is ⌽short=3.3ϫ1020 s−1
2J. C. Angus and C. C. Hayman, Science 241, 913 ͑1988͒.
3K. Hassouni, X. Duten, A. Rousseau, and A. Gicquel, Plasma Sources Sci.
Technol. 10, 61 ͑2001͒.
using the short nozzle. So due to the change in geometry, the
dissociation degree has increased to a value of ␤=0.33 at z
=8 mm in the expansion. This corresponds to equal fluxes of
molecular and atomic hydrogen.
4A. von Keudel, T. Schwarz-Selinger, M. Meier, and W. Jacob, Appl. Phys.
Lett. 76, 676 ͑2000͒.
5M. C. M. van de Sanden, R. J. Severens, W. M. M. Kessels, R. F. G.
Meulenbroeks, and D. C. Schram, J. Appl. Phys. 84, 2426 ͑1988͒.
6P. Vankan, T. Rutten, S. Mazouffre, D. C. Schram, and R. Engeln, Appl.
Phys. Lett. 81, 418 ͑2002͒.
The H atom flux using the large diameter nozzle may
even be higher than observed at 8 mm. Compare for example
the decrease in atomic flux using the short nozzle when mov-
ing downstream. It decreases by 17% when moving from z
=2 mm to 8 mm. This effect has been extensively studied,
both in a pure H2 expansion and in Ar–H2 expansions10,15
and is explained by the diffusion of H atoms out of the ex-
pansion. This diffusion is induced by H atom density gradi-
ents between the background gas in the vessel and the
plasma. The H atom background density is low due to sur-
face recombination of the hydrogen atoms on the vessel
walls. Assuming the loss of H due to diffusion in the large
diameter nozzle case is the same as for the short nozzle case,
a total flux at 2 mm in the large diameter case of 9.8
ϫ1020 s−1 can be calculated, which indicates a dissociation
degree of 0.4.
7J. Amorim, G. Bavarian, and G. Sultan, Appl. Phys. Lett. 68, 1915 ͑1996͒.
8M. C. M. van de Sanden, J. M. de Regt, G. M. Janssen, J. A. M. van der
Mullen, D. C. Schram, and B. van der Sijde, Rev. Sci. Instrum. 63, 3369
͑1992͒.
9M. G. H. Boogaarts, S. Mazouffre, G. J. Brinkman, H. W. P. van der
Heijden, P. Vankan, J. A. M. van der Mullen, D. C. Schram, and H. F.
Döbele, Rev. Sci. Instrum. 73, 73 ͑2002͒.
10S. Mazouffre, P. Vankan, R. Engeln, and D. C. Schram, Phys. Plasmas 8,
3824 ͑2001͒.
11K. Niemi, T. Mosbach, and H. F. Döbele, Chem. Phys. Lett. 367, 549
͑2003͒.
12N. Cohen and K. R. Westerberg, J. Phys. Chem. Ref. Data 2, 531 ͑1983͒.
13B. Jackson and M. Pearson, J. Chem. Phys. 96, 2378 ͑1992͒.
14C. T. Rettner, Phys. Rev. Lett. 69, 383 ͑1992͒.
15S. Mazouffre, M. G. H. Boogaarts, J. A. M. van der Mullen, and D. C.
Schram, Phys. Rev. Lett. 84, 2622 ͑2000͒.
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:
129.49.251.30 On: Wed, 10 Dec 2014 21:48:00