An intense pulsed electrical discharge source for OH molecular beams

Article abstract of DOI:10.1016/S0009-2614(01)00223-8

In this Letter we describe and characterize a pulsed DC discharge source for a molecular beam of OH radicals. The absolute line-integrated OH density has been measured by cavity ring-down spectroscopy, while the off-axis distribution of the radicals has been determined by 1-dimensional laser-induced fluorescence spectroscopy. Combining both measurements the total OH centerline flux at the maximum of the pulse was determined to be (2.2±0.1)×10¹⁷ molecules/sr s. No anomalous population distribution was found for the Λ-doublet components of the rotational ground state.

Full text of DOI:10.1016/S0009-2614(01)00223-8

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
6 April 2001
Chemical Physics Letters 337 92001) 237±242
www.elsevier.nl/locate/cplett
An intense pulsed electrical discharge source
for OH molecular beams
M.C. van Beek *, J.J. ter Meulen
Department of Molecular and Laser Physics, University of Nijmegen, P.O. Box 9010, NL-6500 GL Nijmegen, Netherlands
Received 23 November 2000; in ®nal form 5 February 2001
Abstract
In this Letter we describe and characterize a pulsed DC discharge source for a molecular beam of OH radicals. The
absolute line-integrated OH density has been measured by cavity ring-down spectroscopy, while the o€-axis distribution
of the radicals has been determined by 1-dimensional laser-induced ¯uorescence spectroscopy. Combining both mea-
surements the total OH centerline ¯ux at the maximum of the pulse was determined to be ꢀ2:2 Æ 0:1†  1017 molecules/
sr s. No anomalous population distribution was found for the K-doublet components of the rotational ground
state. Ó 2001 Elsevier Science B.V. All rights reserved.
1. Introduction
H ‡ NO2 ! OH ‡ NO [2,3]. This source was es-
timated to yield a centerline ¯ux of 2 Â 1016 mol-
ecules/sr s [4]. Later, continuous OH molecular
beams have been generated by radiofrequency [5]
or corona [6,7] discharges of H2O, while pulsed
beams have been produced by photolysis of HNO3
with 193 nm light [8,9]. Unfortunately no reliable
estimates of the OH ¯ux of these sources has been
reported. In our laboratory an OH beam source
has been developed based on a pulsed DC dis-
charge in a H2O/Ar jet at the beginning of the
expansion into the vacuum chamber [10]. Com-
pared to the other production methods, a pulsed
discharge provides a simple, cheap, intense and
stable OH source. To our knowledge the only
other OH pulsed discharge sources have been re-
ported by Bramble and Hamilton [11] and by
Anderson et al. [12]. These sources have been used
for spectroscopic studies of OH and OH±Ar in the
free-jet expansion.
Because of their high reactivity, free radicals
play a key role in the dynamics of many chemical
reactions and numerous experiments have been
reported on 9half-)collisions involving these mole-
cules. However, their high reactivity also causes
diculties in the generation of free radicals for
laboratory studies. In particular, the radicals have
to be produced in a low-pressure environment to
prevent their immediate loss by unwanted reac-
tions. The methods which have been used for the
production of free radicals can be grouped into
four categories: photolysis, pyrolysis, gas dis-
charges and chemical reactions [1].
One of the most frequently studied radicals in
molecular beam experiments is the OH molecule.
The ®rst intense continuous OH beam source re-
ported was based on the chemical reaction
In this Letter we describe and characterize our
OH source. There were several reasons for per-
forming this experiment. First, the absolute ¯ux of
* Corresponding author. Fax: +31-24-3653311.
E-mail address: mvbeek@sci.kun.nl 9M.C. van Beek).
0009-2614/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 0 0 9 - 26 1 4 9 0 1 ) 0 0 223 - 8

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
238
M.C. van Beek, J.J. ter Meulen / Chemical Physics Letters 337 /2001) 237±242
the molecular beams has to be known in order to
determine absolute cross sections in crossed beam
collision experiments. Secondly, from the obtain-
able ¯ux the feasibility of speci®c OH beam exper-
iments can be estimated. Thirdly, in previous
experiments a remarkable di€erence in population
between the OH K-doublet components of the
lowest rotational states has been found when OH
was produced in a pulsed discharge [10]. Since the
energy spacing between these K-doublet states is
only 0.05 cmÀ1 the population of these states is
expected to be almost equal after the supersonic
expansion. A large population di€erence in the
beam can only be explained if there is a very strong
propensity for one of the K-doublet states in the
production of OH. Such a propensity has been ob-
served for the photodissociation of H2O by 157 nm
light [13]. It should be noted that for the OH source
described in this Letter a di€erent pulsed valve has
been used than in previous experiments [10]. The
properties of the discharge, however, are similar.
mechanism. It has a repetition rate of 10 Hz and a
pulse duration of about 40 ls. On top of the stan-
dard nozzle with a diameter of 0.5 mm a stainless
steel nozzle with an ori®ce of 0.2mm and a conical
shape is mounted. A 0.5 mm thick stainless steel ring
with a diameter of 4 mm located on-axis 2.5 mm
from the ori®ce is kept at a voltage of À2.4 kV. The
OH radicals are produced by the dissociation of
H2O at the beginning of the expansion by the elec-
trical discharge between the ring and the grounded
valve body. This discharge is initiated by the rise of
the local pressure when the valve opens; it auto-
matically stops when the valve closes. The conical
shape of the nozzle con®nes the discharge to the
ori®ce. The current between the ring and the valve
has a pulse length of 55 ls 9FWHM) and a pulse
maximum of 300 lA. During operation the average
pressure in the vacuum chamber is 2 Â 10À6 mbar.
After production the OH radicals are cooled down
to the lowest rotational state 9X2P3=2; v ˆ 0;
J ˆ 3=2) during adiabatic expansion into the vac-
uum chamber. This state is split into two K-doublet
components which are denoted by e 9lower) and f
9upper).
2. Experimental
To investigate the characteristics of the OH
beam, a laser beam crosses the molecular beam 2.3
cm downstream of the ori®ce. The relative spatial
distribution as a function of the distance from the
molecular beam axis was determined by saturated
1-dimensional laser-induced ¯uorescence 91D-
LIF), and the absolute line-integrated OH density
was measured by cavity ring-down spectroscopy
9CRDS). For both experiments the A2Rꢀv ˆ 0†
X2Pꢀv ˆ 0† transition around 308 nm was
used. Since 94% of the OH radicals are in one of
the K-doublet states of the rotational ground state,
only these states are considered in this Letter [10].
The population of the e state is probed by the P1ꢀ1†
transition, whereas the Q1ꢀ1† ‡ Q21ꢀ1† transitions
were used to measure the population of the f state
[14]. The laser light was produced by a dye-laser
9Continuum TDL-60) operating with Sulforhod-
amine 640 dye and pumped by a 10 Hz Nd:YAG
laser 9Quantel 681C), and was frequency doubled
in a KDP crystal. The resulting 308 nm light has
an energy of 4 mJ/pulse, a bandwidth of 0.4 cmÀ1
and a pulse length of 5 ns. The diameter of the
laser beam is about 5 mm.
The discharge geometry is depicted schematically
in Fig. 1. The OH beam is produced by expanding a
3% H2O-in-Ar mixture at a total pressure of 800
mbar into a vacuum chamber. This mixture is ob-
tained by ¯owing Ar through liquid water at room
temperature 9vapor pressure of 24 mbar) in a bub-
bler. For the expansion, a commercially available
pulsed supersonic valve 9R.M. Jordan) is used. The
operation of this valve is based on the current-loop
current loop
800 mbar 3% H2O in Ar
φ0.5 mm
4.6 mm
φ0.2 mm
1.0 mm
2.5 mm
valve body (0 kV)
ring electrode
(-2.4 kV)
vacuum
4.0 mm
Fig. 1. Geometry of pulsed nozzle and discharge.

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
M.C. van Beek, J.J. ter Meulen / Chemical Physics Letters 337 /2001) 237±242
239
2.1. 1-Dimensional LIF
completely saturate these transitions. The ¯uores-
cence signal detected by the CCD chip was inte-
grated over a period of 1.4 ls. Each ®nal image
consisted of 640 single shot images. The back-
ground 9caused by the discharge, scattered laser
light, stray light and dark current) was measured
by tuning the laser out of resonance, and sub-
tracted from the ¯uorescence images. The resulting
¯uorescence signal was integrated in the direction
perpendicular to the laser beam to give the OH
distribution along the laser beam.
The setup for the 1D-LIF measurements is de-
picted schematically in Fig. 2A. The ¯uorescence
was measured spatially resolved by an intensi®ed
CCD camera 9Princeton instruments ICCD-512-T)
placed at a distance of 60 cm from the OH beam.
With a Nikon UV 105 mm f/4.5 objective an area
of 4 Â 4 cm2 could be imaged onto the 512 Â 512
pixels of the CCD chip. Two Schott colored ®lters
9UG11 and WG295) were mounted in front of the
camera to suppress stray light from the laboratory
and scattered light from the discharge. After the
laser beam exits the vacuum chamber, the intensity
is attenuated and the relative power is measured
with a photomultiplier.
2.2. Cavity ring-down
The CRD setup is represented schematically in
Fig. 2B. The cavity is formed by two highly re¯ective
convex mirrors with a radius R ˆ 1 m separated by
27 cm. The re¯ection coecient of the mirrors was
about 99.6%, which is rather low for CRDS exper-
iments. We preferred to use older mirrors with a
lower re¯ectivity since the ions produced in the
discharge slightly damage the mirrors during the
experiment. The ring-down time of the empty cavity
was around 215 ns. The light leaking out of the
cavity was detected by a fast photomultiplier tube
9EMI 9235 Q) and visualized on a 10 bit 300 MHz
oscilloscope 9Yokogawa DL-42000). For each data
point 8 pulses were averaged on the oscilloscope
before a single exponential decay function was ®tted
to the ring-down signal. The decay time s was stored
on a PC as a function of the laser wavelength. In
order to control the alignment of the cavity during
the experiment a fraction of the light leaking out of
the cavity was imaged onto an intensi®ed CCD
camera.
The laser frequency was scanned around the
P1ꢀ1† and the Q1ꢀ1† ‡ Q21ꢀ1† transitions to deter-
mine the population of both the e and f K-doublet
components of the rotational ground state at the
maximum pulse intensity. The population of the
excited rotational states was too low to obtain a
reasonable signal-to-noise ratio. To prevent satu-
ration e€ects the laser power was attenuated by a
factor of 3 before entering the cavity. From the
spatial distribution obtained in the 1D-LIF mea-
surements the velocity distribution along the laser
beam can be determined. This velocity distribution
corresponds to a Doppler width of the absorption
The images were measured with the laser fre-
quency ®xed at the center of the P1ꢀ1† or Q1ꢀ1†
transition. A 3 mm diaphragm was used to cut o€
the low-energy wings of the laser beam. The power
in the center of the laser beam was high enough to
window
(A)
PMT
attenuator
OH beam
dye
window
filters
spatial OH distribution
laser YAG
CCD
(B)
alignment cavity
mirror
CCD
PMT
OH beam
window
dye
spectrum
laser YAG
Fig. 2. Experimental setup for the saturated 1D-LIF mea-
surements 9A) and for the CRDS measurements 9B). The di-
rection of the OH beam, represented by the concentric circles, is
perpendicular to the paper.

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
240
M.C. van Beek, J.J. ter Meulen / Chemical Physics Letters 337 /2001) 237±242
lines of 0.048 cmÀ1. This is much larger than the
mode spacing of the cavity 90.0044 cmÀ1) but
smaller than the bandwidth of the laser 9ꢁ0.4
cmÀ1). Zalicki and Zare [15] have shown that if the
laser bandwidth is larger than the absorption
linewidth, the absolute number of molecules can
be determined within an accuracy of 10% if the
area of the absorption line is measured. For the
analysis of the CRD spectra the laser bandwidth is
approximated by a Gaussian frequency pro®le.
The central position, width and area were ®tted to
the measured spectrum. Since the Q1ꢀ1† and Q21ꢀ1†
lines partially overlap 9the separation between
these lines is 0.336 cmÀ1), these lines cannot be
analyzed independently. In the analysis we assume
that the Q1ꢀ1† and Q21ꢀ1† lines have the same
linewidth, and that the relative area is propor-
tional to the relative absorption coecient since
both transitions probe the same ground state.
ratio is very sensitive to variations in the back-
ground signal. Averaged over all distances the
¯uorescence ratio is equal to 0:665 Æ 0:027, which
is in perfect agreement with the ratio of 0.68 re-
ported by Schreel et al. [10]. Since the spatial ¯u-
orescence distribution is similar for the detected
states, the ¯uorescence from the di€erent states is
summed to provide the relative density distribu-
tion of the OH beam. Using the values for the total
number of molecules along the detection path,
determined with CRDS as is described in the fol-
lowing paragraphs, the relative density has been
scaled to an absolute number density. This density
distribution is also plotted in Fig. 3.
A typical CRDS spectrum is depicted in Fig. 4.
From the integrated absorption of the P1ꢀ1† and
the Q1ꢀ1† ‡ Q21ꢀ1† spectra the line-integrated
number densities of the e and f states were deter-
mined. Both spectra were recorded 12times. The
standard deviation was about 5%; however, be-
cause the bandwidth of the laser is larger than the
Doppler width a systematic error of the order of
10% can be made [15]. Therefore, the accuracy of
the number densities is set to 10%. The line-inte-
grated OH densities were found to be Nlineꢀe† ˆ
ꢀ3:32 Æ 0:34†  1011 and Nlineꢀf† ˆ ꢀ3:68 Æ 0:37† Â
1011 molecules/cm3 for the rotational ground state.
The ratio between the populations of the e and f
states is 0:90 Æ 0:14. This population ratio seems
not to be in accordance with the ¯uorescence ratio
of 0.665. However, in the interpretation of satu-
rated LIF signals the degeneracy of the ground
and excited states has to be taken into account by
3. Results
In Fig. 3 the ratio of the ¯uorescence from the
P1ꢀ1† to that of the Q1ꢀ1† transition is plotted as a
function of the distance to the molecular beam
axis. From the ®gure it can be seen that this ratio is
nearly constant. The ratio slightly increases far
away from the beam axis; however, this is also the
region where the OH signals are very small and the
Fig. 3. Ratio of the ¯uorescence from the P1ꢀ1† and Q1ꢀ1†
transition 9r), distribution of the OH radicals perpendicular to
the molecular beam axis 9ꢂ) and analytical ®t to the distribution
9gray line).
Fig. 4. Typical CRD spectrum of the P1ꢀ1† transition.