A simple radionuclide-driven single-ion source
ABSTRACT We describe a source capable of producing single barium ions through nuclear recoils in radioactive decay. The source is fabricated by electroplating 148 G d onto a silicon α -particle detector and vapor depositing a layer of BaF 2 over it. 144 S m recoils from the alpha decay of 148 G d are used to dislodge Ba + ions from the BaF 2 layer and emit them in the surrounding environment. The simultaneous detection of an α particle in the substrate detector allows for tagging of the nuclear decay and of the Ba + emission. The source is simple, durable, and can be manipulated and used in different environments. We discuss the fabrication process, which can be easily adapted to emit most other chemical species, and the performance of the source.
A simple radionuclide-driven single-ion source
M. Montero D´ ıez,1K. Twelker,1W. Fairbank Jr.,2G. Gratta,1P.S. Barbeau,1K. Barry,1R. DeVoe,1
M.J. Dolinski,1M. Green,1F. LePort,1A.R. M¨ uller,1R. Neilson,1K. O’Sullivan,1N. Ackerman,3B. Aharmin,4
M. Auger,5C. Benitez-Medina,2M. Breidenbach,3A. Burenkov,6S. Cook,2T. Daniels,7K. Donato,4J. Farine,4
G. Giroux,5R. Gornea,5K. Graham,8C. Hagemann,8C. Hall,9K. Hall,2D. Hallman,4C. Hargrove,8S. Herrin,3
A. Karelin,6L.J. Kaufman,9, a)A. Kuchenkov,6K. Kumar,7J. Lacey,8D.S. Leonard,9, b)D. Mackay,3
R. MacLellan,10B. Mong,2E. Niner,10A. Odian,3A. Piepke,10A. Pocar,7C.Y. Prescott,3K. Pushkin,10
E. Rollin,8P.C. Rowson,3D. Sinclair,11S. Slutsky,9V. Stekhanov,6J.-L. Vuilleumier,5U. Wichoski,4J. Wodin,3
L. Yang,3and Y.-R. Yen9
1)Physics Department, Stanford University, Stanford CA, USA
2)Physics Department, Colorado State University, Fort Collins CO, USA
3)SLAC National Accelerator Laboratory, Menlo Park CA, USA
4)Physics Department, Laurentian University, Sudbury ON, Canada
5)LHEP, Physikalisches Institut, University of Bern, Bern, Switzerland
6)Institute for Theoretical and Experimental Physics, Moscow, Russia
7)Physics Department, University of Massachusetts, Amherst MA, USA
8)Physics Department, Carleton University, Ottawa ON, Canada
9)Physics Department, University of Maryland, College Park MD, USA
10)Dept. of Physics and Astronomy,University of Alabama, Tuscaloosa AL,USA
11)Physics Dept., Carleton University, Ottawa and TRIUMF, Vancouver, Canada
(Dated: 23 August 2010)
We describe a source capable of producing single barium ions through nuclear recoils in radioactive decay.
The source is fabricated by electroplating148Gd onto a silicon α-particle detector and vapor depositing a
layer of BaF2over it.
BaF2layer and emit them in the surrounding environment. The simultaneous detection of an α particle in
the substrate detector allows for tagging of the nuclear decay and of the Ba+emission. The source is simple,
durable, and can be manipulated and used in different environments. We discuss the fabrication process,
which can be easily adapted to emit most other chemical species, and the performance of the source.
144Sm recoils from the alpha decay of148Gd are used to dislodge Ba+ions from the
PACS numbers: 29.25.Ni, 34.50.-s, 81.15.Pq
It is often necessary in experimental settings to pro-
duce specific ionic species in a controlled manner. In-
deed a large number of techniques to produce ions have
been documented, from the early Penning sources to laser
and electron-beam produced plasmas to thermal ioniza-
tion1,2. These sources, however, can only operate under
vacuum, and in many cases require expensive and com-
plex infrastructure to allow for ion separation and deliv-
The work discussed here arises from the need to de-
velop a simple source of monoatomic, singly-ionized bar-
ium that can be used in a cryogenic liquid, a gaseous
environment, and in vacuum. Such a source is necessary
for the R&D efforts of the Enriched Xenon Observatory
(EXO). The EXO Collaboration is planning a series of
experiments designed to determine the mass of the neu-
trino by searching for the neutrinoless double-beta decay3
of136Xe. To suppress possible radioactive backgrounds,
EXO plans to add to the standard low background tech-
niques that are customary in these experiments the abil-
a)Now at Indiana University, Bloomington IN, USA
b)Now at the University of Seoul, Seoul, Korea
ity to tag the barium daughter of the double-beta de-
cay using resonant fluorescence4–6. Since possible detec-
tor technologies would include xenon Time Projection
Chambers (TPCs) in high pressure gas4or liquid7phase,
the R&D work on Ba-tagging requires a method of releas-
ing Ba+in such media. The injection through a thin win-
dow of Ba ions produced in an accelerator would clearly
be possible but it would require expensive equipment and
would result in very highly ionized states, owing to the
electron-stripping in the entrance window and in the high
The source presented here makes use of the Sm daugh-
ter of the decay
62Sm + α(1)
to dislodge a Ba+ion from a BaF2 layer coated over
the Gd source. The α recoiling against the Sm is then
detected in a semiconductor detector that is used as a
substrate for the assembly, providing a tag for the ion
half-life (74.6 y) provides a good compromise between
high specific activity and durability.
decay of148Gd is directly in the ground state of144Sm,
so that no other radioactivity is produced. Finally, Gd
can be reliably electroplated in thin, uniform layers, as
will be discussed.
148Gd is a convenient α emitter because its
In addition, the
arXiv:1008.3422v1 [physics.atom-ph] 20 Aug 2010
The Gd source is deposited onto a PIPSR
tor8and a special technique is used to confine the ac-
tivity to a small central region. A thin layer of BaF2
is then vapor-deposited over the entire detector. The as-
sembly is mounted in a custom metal-ceramic holder that
is suitable for ultra-clean operations and moderate vac-
uum bakeouts. The ion source obtained is inexpensive,
compact and suitable for a variety of environments. The
selection of coatings other than BaF2can provide similar
sources for most elements.
II. SOURCE FABRICATION
148Gd is obtained in the form of GdCl3in HCl aque-
ous solution. Natural Gd is present as carrier in the solu-
tion in the approximate ratio9 natGd:148Gd=1:6.6×10−4.
Gd electroplating works best in isopropanol (IPA) from
gadolinium nitrate10. So the chloride is transformed into
GdCl3(aq)+ 3HNO3(aq)→ Gd(NO3)3(aq)+ 3HCl(aq)
The reaction is achieved by drying a 500µL (0.1µCi)
148GdCl3batch on a hot plate and adding 500µL of 1M
HNO311. The water is then evaporated again and more
HNO3is added, repeating the process three times. Af-
ter the last drying cycle, the148Gd(NO3)3is dissolved in
500µL of IPA12. 25µL of this solution are calculated to
have an activity of 185 Bq. A measurement of the activ-
ity, performed by dissolving one 25µL batch in a liquid
scintillation counter13, confirms a carry-over efficiency of
∼100%. Since the Gd(NO3)3concentration is extremely
low, a drop of 0.05M HNO3is added in 10L of the IPA in
order to increase the conductivity of the plating solution
to an acceptable level.
The use of a conventional plating cell is excluded be-
cause of the requirement to deposit the148Gd only in a
specific and small region of the detector. Therefore, a
plating circuit is established by mounting a PIPSR
tector on a horizontal support and lowering an anode to
a position about 0.5 mm above the center of the detector
using a micrometer, as shown in Figure 1. Using a cali-
brated pipette a small drop of ∼ 5µL of solution is then
deposited between the anode and the detector, which
serves as the cathode in the circuit. The drop remains
in place because of surface tension. A 200 V potential is
then applied to the circuit via a 10 MΩ current-limiting
resistor, establishing a current that is measured to be
18 µA. Depending on the activity required more 5µL
batches are added, as the solvent evaporates, for a total
of 25 to 50µL. Finally, pure IPA aliquots are added, again
to counter evaporation. A total plating time of 15 min
was found to be adequate by trial and error by measuring
the activity on the detector (since the charge transport
in the cell is completely dominated by the HNO3added
to the IPA and has no relation with the amount of Gd
deposited). Because of the non-radioactive Gd carrier, a
FIG. 1. Schematic view of the radioactive Gd plating setup.
The micrometers allow for an accurate and stable positioning
of the anode ? 0.5 mm above the center of the detector being
plated. A drawing of the Canberra PIPSR
provided in the inset.
?detector used is
source of ?200 Bq activity, obtained from 25µL plating
solution, roughly covers a surface of 8 × 10−2cm2with
∼10 monolayers of Gd.
Initial tests using Pt anodes were found to produce
large contaminations in the Gd coatings. This was evi-
dent by optical inspection of the plated detector. Such
contaminations are unacceptable because they reduce the
energy of the Sm recoils, as confirmed by the analysis of
the α energy spectrum from the source measured in an
external surface barrier detector. Anodes obtained using
4 mm - wide slices of a B-doped silicon wafer (resistivity
≈ 50 Ωm) were found to be sufficiently clean and were
used for the work presented here.
The α detector used as a substrate has surface con-
tacts formed by shallow ion implantation and it is espe-
cially appropriate because of its ruggedness. Its hexag-
onal shape (also shown in Figure 1) is of no particular
significance here. The functionality of the detector af-
ter plating and the quality of the plated source are ver-
ified by α-counting both with the source-substrate de-
tector (“self-counting”) and, in some cases, with an ex-
ternal surface-barrier detector14in a vacuum chamber.
In addition, an external241Am source is used to verify
the detector performance and resolution after plating. A
typical241Am spectrum recorded by the PIPSR
tor with a 50 V bias and 2 µs shaping time, before and
after Gd plating is shown in Figure 2. The three peaks
at 5485.6 keV, 5442.8 keV and 5388.2 keV are clearly
visible before plating, with a 22 keV FWHM resolution
for the most prominent peak. This is somewhat worse
than the 14 keV FWHM stated by the manufacturer.
The spectrum after Gd plating is substantially broader
(∼ 130 keV FWHM), presumably due to some deterio-
ration of the detector and some energy loss in the Gd
and other impurities left on the surface by the plating
53005350540054505500 5550 5600
FIG. 2. α spectrum of an external241Am source recorded by
cles) the Gd plating. The three main α peaks at 5485.6 keV,
5442.8 keV and 5388.2 keV are clearly distinguishable in the
first data set. In this case the resolution of the 5485.6 keV
peak is of 22 keV FWHM. After plating, the energy resolution
degrades substantially and the new resolution of ∼ 130 keV
FWHM makes it impossible to resolve the three241Am lines.
?detector before (full circles) and after (empty cir-
process. This degradation, however, appears to be stable
in time and is not important for the simple α detection
required in tagging the ion emission. The spectrum from
the single 3182.7 keV α emitted by the148Gd plated on
a detector shows (Figure 3(a)) a resolution of ∼ 60 keV
The activity of the plated148Gd is measured by read-
ing out the signal from the α detector, as will be de-
scribed in the next section. For typical sources this ac-
tivity is ?200 Bq and matches, within experimental un-
certainties, the activity measured in the plating solution,
implying that the plating efficiency is rather high.
A coating of the appropriate chemical species and
thickness will generally produce free ions and neutral
atoms dislodged by the collision of the 89 keV recoiling
144Sm. This is expected to be true for most elements,
although with varying efficiencies, mainly due to the ra-
tio of masses between the Sm and the “target” species.
Different (in particular, heavier) α emitters (e.g.
or209Po) could replace the148Gd, although this was not
attempted in our work.
BaF2 was chosen as the chemical species containing
Ba because of its relatively good stability in air15and
the fact that the other atomic species, fluorine, has a
mass that is very different from that of Ba. The optimal
layer thickness is estimated using the SRIM16simulation
package with full damage cascades simulation. Figure 4
shows the predicted Ba yields as a function of thickness.
BaF2was simulated by simply providing SRIM with the
appropriate number densities of Ba and F. Yields greater
than one, like those reported in Figure 4, are not unusual
in sputtering processes.
While the Ba yield can be larger than 1, as in the
2600 2800 3000320034003600
260028003000 32003400 3600
the spectrum of the
resolution is ∼60 keV. Panel (b) shows the spectrum after
40 nm of BaF2 are deposited over the source. The FWHM
energy resolution is in this case ∼170 keV.
148Gd “self-counted” α spectra. Panel (a) shows
148Gd layer only. The FWHM energy
case of sputtering, it should be kept in mind that SRIM
does not distinguish between neutral atoms and ions in
different charge states, so that the yield of Ba+is not
known from the simulation. The energy spectrum for
Ba is predicted by SRIM to be peaked at zero with a
long tail extending above 100 eV. 15 nm was chosen as
the optimal layer thickness for the BaF2, although some
early sources were fabricated with thicknesses from 8 nm
to 40 nm.
As already mentioned, electroplating was chosen as the
technique to deposit the α emitter because of the layer
quality and the possibility of obtaining deposits that are
easily limited to a specific area. The technology for the
overcoating by the target species was then chosen so as
not to chemically interfere with the radionuclide and, at
Layer Thickness (nm)
0 1020 3040 50607080
Atomic yield per incident Sm
FIG. 4. Yield for Ba atoms as a function of layer thicknesses,
as calculated by a SRIM simulation. Since SRIM does not
distinguish between different charge states the yield from the
graph includes neutral atoms, singly charged ions as well as
ions with a higher ionization state.
the same time, provide a uniform layer over the entire
front surface of the detector. Vacuum evaporation is suit-
able for this since the energy of the ions is low (compared
to sputtering) and no wet chemistry is involved. In our
system, tantalum boats are ohmically heated to an ap-
propriate temperature and the thickness of the deposited
layer is monitored by a microbalance located next to the
source being plated. Most materials can be deposited
in this way or using electron-beam heating for refrac-
tory materials. As shown in Figure 3(b), the α energy
resolution from the148Gd source is further deteriorated
after the BaF2coating. While the origin of the deterio-
ration is not understood, the detector is still functional
and completely adequate to tag the decay and the sub-
sequent emission. In order to ensure the best possible
purity for the BaF2 coating, a small scintillation-grade
BaF2crystal17was used in the evaporation process.
Sources prepared as described above are then mounted
in the special holders shown in Figure 5. All metallic
parts are made of gold-plated stainless steel, and the insu-
lating ring separating the two diode contacts is made out
sulation to maintain a constant pressure on the contacts
is replaced by a gold-plated wavy washer. The holders
allow for cooldown to 77 K and are compatible with the
extreme cleanliness required for ultra-high vacuum oper-
ations as well as for operation in liquefied noble gases (in
particular liquid xenon).
?. The elastomer used in the standard encap-
The ion emission from the source is analyzed with a
simple time of flight (TOF) spectrometer comprising two
Einzel lenses and designed to maximize the transport ef-
ceramic UHV mount. The “wavy washer” provides a uniform
force on the detector to ensure proper contact. All metallic
parts are made out of gold-plated stainless steel. The ring
to the right of the drawing is made out of MacorR
Photograph of a completed source.
Top: Exploded view of an ion source in its metal-
ficiency of ions from the rather large emittance produced
by the source to a channel electron multiplier18(CEM).
The spectrometer is designed with a central drift region
and electrodes at the two ends that allow for the indepen-
dent adjustment of the electric field at the surface of the
source (Esource, 350 V/cm < Esource< 750 V/cm) and
the ion impact energy on the CEM (K 2400 eV < K <
3800 eV). The overall length of ? 17 cm produces a TOF
for Ba+of ∼5 µs, depending on the exact configuration
of potentials. The “start” signal for the TOF measure-
ment is provided by the PIPSR
source is built and the “stop” is provided by the CEM
pulse. In this case the 45 V bias of the PIPSR
is provided by a battery, so that the entire source can
be floated to high voltage as required by the different
configurations of the spectrometer. Both signals from
the source and the CEM are digitized with flash ADCs
(FADCs) at 250 MS/s. The signal from the source is
used to trigger an acquisition. While a fast preamplifier
is used for the CEM pulse, the signal from the source
is fed into an Ortec model 142 preamplifier followed by
a model 474 timing filter amplifier (TFA) with 100 ns
shaping. Timing thresholds are applied offline.
?detector on which the
A typical TOF spectrum obtained for the intermedi-
ate values of Esource= 550 V/cm and K = 2800 eV is
shown in Figure 6. The prominent peak around 5 µs has
a position and width consistent with those predicted by
a SIMION19simulation for Ba+ions. Such simulation
includes the correct natural isotopic mix for the atomic
FIG. 6. Time-of-flight spectrum for a typical Ba source. The
source has a measured148Gd activity of 173 Bq and a 15 nm
thick BaF2 coating. The Ba+yield, measured as the integral
of the main peak, is of 2.7 Hz.
mass of Ba and an initial energy distribution derived by
the SRIM output, as explained below. A fit of this peak
to five Gaussian functions representing the main Ba iso-
tope masses provides an estimate for the spectrometer
FWHM resolution of 5.8 amu at ?137 amu. The assign-
ment of the peak to Ba+is corroborated by the observa-
tion that no such peak is present in a source where the
BaF2coating is replaced by Al. In addition, and maybe
most importantly, Ba deposited from a source of this type
onto a clean silicon substrate was then desorbed and res-
onantly ionized with two lasers. From this very specific
result it is clear that positive Ba ions are indeed emitted
by the source.
According to SRIM the energy spectrum of the Ba
emitted by the source peaks at zero and decreases mono-
tonically, with 17% of the spectrum having energies
greater than 300 eV. Since, as already mentioned, the
SRIM simulation is unable to distinguish between neu-
tral and charged states, a different method is required
to estimate the expected ion fraction, α = n+/ntotal.
Ideally this ratio would be known as a function of Ba en-
ergy. Unfortunately, a general understanding of the ion
fraction in sputtering is still an open question, even for
pure metals, and data is completely lacking for ionic com-
pounds, such as BaF220. Some models and experimental
data describe α by a decaying exponential in the inverse
velocity plus a constant term, with the transition tak-
ing place around a critical velocity of ∼2 cm/µs20. The
net result of this function combined with the sputtering
energy distribution is that the fraction of ions is low at
low energy and approaches unity at higher energy. The
transition takes place around 2 cm/µs, or ∼300 eV in the
case of Ba.
The actual energy distribution of Ba+ions affects
the transport efficiency of the spectrometer significantly.
While such efficiency is calculated to be 85% for the en-
tire energy distribution produced by SRIM, it is only 8%
for the part of such energy distribution exceeding 300 eV.
Also the TOF peak position and width are to some extent
predicted to vary for different regions of the emittance of
the source. A simple consistency check for the Ba+yield
can be made by assuming in the simulation that only ions
(neutrals) are produced above (below) a certain energy
cutoff. The yield, TOF peak position and width are then
compared with data for a variety of values of Esourceand
K. A reasonably good description of the data occurs for
a 300 eV cutoff. The simulated rate reproduces the data
within a factor of two when Esourceand K are varied from
550 V/cm and 2800 eV to the extremes mentioned above.
With the same cutoff value the simulation also provides
a good prediction for the Ba+TOF peak position (bet-
ter than 10%) and width (better than 25%), again when
scanning Esourceand K from their central values to the
For the spectrum in Figure 6 the Ba+yield is measured
to be 2.7 Hz. Using the α activity of 173 Bq, SRIM
predicts a rate of Ba emission of 430 Hz for the 15 nm
thick BaF2 coating (see Figure 4). Assuming that the
ions comprise only those Ba atoms ejected with energy
above 300 eV, or, according to SRIM 17% of the total, the
Ba+emission rate is calculated to be 73 Hz. Using the
spectrometer efficiency of 8% appropriate for this energy
range, a rate of 5.8 Hz is predicted to arrive at the CEM.
The ratio between this figure and the measured rate of
2.7 Hz is consistent with typical CEM efficiencies for ions
with the velocities expected here.
TABLE I. Possible assignment of the TOF peaks for the
source in Fig. 6. The associations of H+and Ba+(in ital-
ics) are assumed and used to compute the mass values for the
Time Measured mass Possible assignment
(known mass in [ ])
The TOF spectrum for light masses is similar to the
one observed for the source coated with Al. Its interpre-
tation nevertheless poses some challenges because of the
limited resolution of the spectrometer, the large emit-
tance of the source and the difficulty in predicting the
charge to neutral ratio as a function of the species and
initial energy. A possible assignment, obtained using as
reference the Ba+peak and the peak at 0.31 µs, assum-
ing to be due to H+is shown in Table I. The presence of a
prominent Na+peak is not surprising, given the common
presence of sodium contamination on surfaces. Emission
of some Si from the substrate is not unexpected, while
C2H5O+and C3H8O+are commonly observed in mass
spectra of the IPA that is used in the Gd plating pro-
cess. No compelling assignment is available for the peak
at 2.23 µs. Finally, the satellite peak at a mass larger
than that of Ba is not present in the Al coated source
and it may be attributed to BaF+.
We have built a simple, inexpensive and compact
source of single Ba+ions using the recoils of an α emit-
ting radionuclide to dislodge Ba atoms from a BaF2coat-
ing. The source uses a surface barrier detector as a sub-
strate, so that a tag is provided for the ion emission. Its
simplicity and the possibility of delivering tagged ions
in various environments make it attractive in applica-
tions where a low rate of individual ions is desirable. In
addition the emission of neutral Ba, not useful for our
purposes, may find other applications. The replacement
of the BaF2with other materials provides the means of
emitting ions from a large number of different species.
We would like to thank K. Moody (LLNL) and
L. Moretto (UC Berkeley and LBNL) for their radio-
chemistry advice. The support from the staff of the
Stanford Health Physics group and, in particular, of
D. Menke, is also gratefully acknowledged. We are also
indebted to H. Newman and R-Y. Zhu (Caltech) for pro-
viding a high quality crystal of barium fluoride. Finally,
we would like to thank Canberra Semiconductor NV and,
in particular, M. Morelle for advice and for providing us
with un-encapsulated samples of their detectors used for
practicing the different processing steps. This work was
supported in part by NSF grant PHY-0652416.
1L. Valyi, Atom and Ion Sources (John Wiley & Sons, London,
2B. Wolf, ed., Handbook of Ion Sources (CRC, Boca Raton, 1995).
3F.T. Avignone III, S.R. Elliott, J. Engel, Rev. Mod. Phys. 80
4M. Danilov et al., Phys. Lett. B 480 12 (2000).
5B. Flatt et al., Nucl. Inst. Meth. A 578 399 (2007).
6M. Green et al., Phys. Rev. A 76 023404 (2007).
7E. Conti et al., Phys. Rev. B 68 054201 (2003).
8Model PD50 50 mm2active area unmounted wafer, Canberra,
9Eckert & Ziegler Isotope Products, Private Communication.
10K. Moody, Lawrence Livermore National Lab, Private Commu-
11TraceMetal grade, Fisher Chemical, Pittsburgh, PA.
13Measurement performed by the Stanford Health Physics Depart-
ment, Stanford, CA.
14Model BU-014-050-100, Ortec, Oak Ridge, TN.
15R. Weast, ed., CRC Handbook of Chemistry and Physics, 56th
Ed. (CRC, Boca Raton, 1975) p. B-74-76.
16J.F. Ziegler, J.P. Biersack and M.D. Ziegler SRIM, The Stopping
and Range of Ions in Matter (Lulu Press, Morrisville, 2008).
17Produced by the Beijing Glass Research Institute, Beijing, China.
18Model 2403, DeTech, Palmer, MA.
20A. Wucher, Applied Surface Science 255 1144 (2008).
?>99.9% grade, Sigma-Aldrich, St. Louis,