68
GE-68GA PRODUCTION: EXCITATION FUNCTIONS, TARGET
PREPARATION AND PC-CONTROLLED RADIOCHEMISTRY SYSTEM
R. Adam Rebeles, A. Hermanne, P. Van den Winkel, L. De Vis, R. Waegeneer, Cyclotron
Laboratory, Vrije Universiteit Brussel (VUB), Belgium
F. Tarkanyi, S. Takacs, Institute of Nuclear Research of the Hungarian Academy of Sciences
(Atomki), Debrecen, Hungary
M.P. Takacs, Institute of Physics, University of Debrecen, Hungary
J-M. Geets, Ion Beam Applications (IBA), Louvain la Neuve , Belgium
Abstract
Activation cross sections by protons (up to 36 MeV)
and deuterons (up to 50 MeV) induced reactions were
measured in a stacked foil irradiation using thin Ga-Ni
(70/30%) targets. Excitation functions for 68,69Ge are
discussed and the results are compared with the literature
values and with the theoretical predictions of TALYS
nuclear reaction code.
High beam power withstanding targets for production
of 68Ge by proton irradiation of Ga were developed using
electrodeposition of Ga-Ni alloy (70/30%) on thick Cu
backings. A PC-controlled radiochemistry system for the
production of nca 68Ge was developed.
INTRODUCTION
The interest for use of 68Ga (T1/2 = 68 minutes, decay
β : 89.1 % , EC 10.9 % ) for clinical PET imaging is
rising fast in recent years. Evidence for this are the 50 or
so talks where application of 68Ga labelled tracers were
discussed in the dedicated conference in 2011 in Bad
Birka (abstracts in [1]), several presentations in the NRC8
conference held in September 2012 or the discussion in
the recent overview article by Qaim [2].
68
Ga has the advantage that it is shorter lived compared
to 18F allowing in principle easier multiple imaging of a
patient and reduction of radiation burden. The main
advantage comes form the fact that 68Ga can be obtained
from a generator system, being the decay product of the
long lived 68Ge (T1/2 = 277 d). An overview and detailed
information required for production and chemical
processing of, among others, 68Ge is discussed in an
IAEA document published in 2010 [3]. The preferred
production route of 68Ge is through the 69Ga(p,2n)68Ge
and 71Ga(p,4n)68Ge reactions, the latter reaction taking
place only at higher proton energies (Q = 28.2 MeV). As
the melting point of Ga is 29.7°C, centers like Los
Alamos National Lab - IPF facility (USA), Brookhaven
National Lab - BLIP facility (USA) and Ithemba - Faure
(South Africa) use liquid Ga encapsulated in a Nb
container as target while in the Russian Federation at
Cyclotron Co. Lt. (Obninsk), a Ga-Ni alloy pressed on Cu
block serves as target.
In collaboration with Ion Beam Applications in
Louvain la Neuve (Belgium) the idea is studied to use
their commercially available 30 MeV proton cyclotrons
for optimized production of parent 68Ge. These
+
accelerators can deliver high beam intensity (up to 500
µA) and standard 11.7 cm2 Cu block targets with adapted
water cooling, irradiated at a 6° beam/target angle, have
proven to withstand the high power in prolonged
irradiations.
The present study involves the following steps needed
to set-up a reliable production chain:
Optimize reliable high power Ga-targets by
electroplating a Ga-Ni alloy with high melting point on
Cu block targets.
Optimize a computer controlled automated
chemistry with high chemical yield and high purity,
delivering a nca 68Ge solution acceptable for loading a
generator system;
Complete the cross section database for reactions
of middle energy protons on Ga to identify optimal energy
ranges for production of 68Ge and minimization of
contaminants;
In the frame of our systematic comparison of
proton and deuteron induced routes for radionuclide
production: study of the excitation function of deuterons
on Ga up to 50 MeV;
TARGET PREPARATION
The set-up and methodology developed earlier for
plating of Tl (for 201Tl production); Zn (for 67Ga), and Cd
(for 111In) production targets used at IBA Cyclone 30
accelerators and for different type of targets used in
research [4, 5, 6] served as basis. The system consist of a
cylindrical plating vessel, allowing insertion of 4 targets
carriers, power supplies and control units, custom made at
VUB [7].
Pre-treatment of the copper target carriers
To ensure good adhesion to the Cu carrier special care
must be taken in preparation of the copper substrate
(brushed with fine abrasive wool, polished with fine
abrasive papers and diamond pastes, removal of metallic
particle or traces of grease by ultrasonic cleaning).
Next the Woods nickel strike is applied. The surface of
the Ni strike is larger (22.1 cm2) compared to the actual
surface of the Ni-Ga alloy (11.7 cm2). The exact
composition of the nickel strike bath as well as the
operation conditions are described in the ASTM standard
B281-88 (reapproved 2001). Activation of that strike
immediately before starting the Ni-Ga plating is done in
diluted sulphuric acid (10 mA/cm2, 3 min). More detailed
information about the activation method can be found in
ASTM standard B343-92a (reapproved 2004).
Electrodeposition of the gallium-nickel target
layer
The Ni-Ga target layer was deposited, simultaneously
on 4 Cu carriers, by using the AC Constant Current
Electrolysis technique (AC-CCE) applying a cathodic
chopped sine wave form. The net cathodic plating current
density was 34.2 mA/cm2 (plating area 11.7 cm2). In the
acidic plating bath the Ni and Ga species exist as NiSO4
and Ga2(SO4)3 respectively. The bath was prepared by
dissolving 31.6 g Ga2O3 in 150 ml (27%) H2SO4. Under
heating (200ºC) and vigorous stirring about 99% of the
Ga2O3 was dissolved in 14 hours. The solution was
filtered, de-ionized water was added until a volume of
450 ml was obtained and the pH is stepwise adjusted to
1.5 by addition of concentrated ammonia. Under gently
heating to 45°C, 17.7 g of NiSO4*6H2O were added and
the solution transferred to the cylindrical electroplating
vessel. During the electrolysis the pH drop is
compensated for by addition of diluted NH4OH by means
of a peristaltic pump coupled with associated feedback
electronics and a pH meter. Up to 51 mg/cm2 of Ga-Ni
alloy were deposited within 6.5 h hours, resulting in a
thickness of the target layer of 104 µm. In a 6º
beam/target angle geometry this results in an energy
degradation of about 10 MeV for incident protons with
energy in the 26-28 MeV range. As the depletion ratio of
the plating bath is limited (to 24.1% for Ni and to 9.5%
for Ga) the composition of the deposited alloy varies only
slightly (± 3%) with the thickness of the layer in the 100
mg to 800 mg Ga-Ni/target range. The alloy composition
was determined by anodic stripping voltammetry to be
30% Ni(±) - 70% (±3%) Ga by weight.
EXCITATION FUNCTIONS AND THICK
TARGET YIELDS
Experimental conditions
The cross sections were measured by using stacked foil
irradiation technique followed by non destructive γspectrometry with general experimental procedures as
described in several of our recent publications [8, 9]. The
targets were Ga-Ni alloy (70/30%) prepared by using the
electroplating technique previously described on Au (25
µm, for deuteron experiment) or Cu (12.5 µm, proton
experiment) foils as backing material. The nominal
thickness of the Ga-Ni layer was 17.3 µm (deuteron
experiment) or 13.2 µm (proton experiment).
In the stack irradiated with 50 MeV deuterons at the
CGR930 cyclotron in Louvain la Neuve (LLN), 22 Ga-Ni
targets on Au backing were interleaved with 55 Al (98 µm
and 49.6 µm) and 15 Ni (23.9 µm) monitor/degrader foils.
Second stack consisting of 20 Ga-Ni alloy on Cu backing
targets, interleaved with Tl/Cu “degraders” was irradiated
with a 36 MeV proton beam of the CGR560 cyclotron at
Brussels (VUB). In this experiment the Cu backings were
used for monitoring. All base foils used were of high
purity (> 99.5%) and were provided by Goodfellow (UK).
The targets were mounted in Faraday cup like target
holders provided with a long collimator. The beam current
was kept constant at about 100 nA (± 5 %) during the 4060 min irradiation.
After dismantling of the targets and a 1 to 4 h cooling
period, the activity of the induced radionuclides was
assessed by HPGe γ-spectrometry, without chemical
separation.
Optimal and accurate identification and quantification
of full energy peaks of short and longer lived activation
products was performed in several measurements at
different times after EOB with varying sample – detector
distance. Typical dead-times were between 5 and 0.5 %.
Energy dependent detector efficiency was determined
by repeated measurements using calibrated standard
sources and a fit allowing knowledge of the efficiency at
any γ-ray energy (see [8]).
Data Analysis
The initial irradiation characteristics (incident energy
and particle flux) were derived from the accelerator
settings and from the integrated charge measured on the
Faraday cup. These parameters were further adapted
taking into account the analysis of the monitor reactions
nat
Cu(p,x)62Zn, respectively natAl(d,x)24,22Na.
A very good overlap of measured and recommended
excitation functions was obtained after changing the
estimated incident proton energy from 36.5 MeV to 36.7
MeV and increasing the current by 25%, while for the
deuteron beam no change in energy was needed and the
beam current was increased by 5%. The SRIM 2006 code
was used for estimation of the energy degradation in each
foil of the stacks [9]. The incident energy was estimated
with an uncertainty of ± 0.2 MeV. Due to the variations in
foils thickness and energy spread, the uncertainty in the
median energy estimation is increasing along the stack,
reaching a maximum of 1.2 MeV for protons and 2 MeV
for deuterons respectively in the last foils.
Decay and spectrometric characteristics of the
investigated radionuclides were taken from Nudat2 [10].
The uncertainties regarding the cross section values were
estimated by extracting the positive square root from the
sum all individual contributions into quadrature.
The following individual uncertainties are taken into
account: absolute abundance of the used gamma rays (1-5
%), determination of the peak areas including statistical
errors (1-5 %), the number of target nuclei including nonuniformity (5 %), detector efficiency (5%) and incident
particle intensity (5 %). The total uncertainty is evaluated
at 8-13 %.
Theoretical calculations
The experimental results were compared with data
retrieved from the online TENDL2011 library, based on
calculations with the 1.4 version of the TALYS code
family [11, 12]. A priori calculations, without any
optimization of the input parameters, were performed for
each stable target isotope separately and where
appropriate, a weighted sum of the contributing reactions
was made and represented on the figures.
Results
Although more than 20 γ-emitting radionuclides were
identified as activation products, only results for 68,69Ge
will be discussed here together with earlier published
experimental data and values from the theoretical
calculation.
Proton induced reactions
In Fig. 1 our new results for the natGa(p,xn)68Ge
excitation function are compared to the values of
Levkovskii [13] (values corrected by 20%) and Porile et
al. [14]. The agreement up to 35 MeV between these 3
data sets is fair, be it that the Levkovski data are at the
low side and some outlying points of Porile et al. [14] are
noted . A fit made to our new values results in a maximal
cross section that is slightly higher than what is published
in the online version of TECDOC-1211 [15].
Figure 2: Excitation function for natGa(p,xn)69Ge,
comparison with literature values and theoretical values
Deuteron induced reactions
In Fig.3 we compare the results of this work for 68Ge
production by deuteron irradiation of natGa to the data
points published by Karpeles [17]. These literature values
are about 25% higher than ours but show overall the same
dependence on energy. Compared to protons the
maximum for the deuteron reaction on 69Ga occurs at
about 6 MeV higher energy and the cross section value is
about 25% lower. In the energy range studied no clear
contribution from the reaction on 71Ga can be seen. The
TENDL2011 prediction is, like for protons, shifted to
lower energy and underestimates the maximal value. A
slight increase in the theoretical excitation function, due
to 71Ga(d,5n)68Ge, is seen between 40 and 50 MeV.
Figure 1: Excitation function for natGa(p,xn)68Ge,
comparison with literature values and theoretical values
Although it concerns simple (p,2n) and (p,4n) reactions
the TENDL2011 prediction shows a shift to lower energy
and too low maximal values.
For production of 69Ge some discrepancies exist
between our new measurements and the literature data
(see Fig. 2). In the 10-15 MeV region our data are energy
shifted compared to the values of Levkovski [13]
(corrected values by 20%) but the agreement in the higher
energy region is excellent. The new data are also lower
than what was published by Porile et al. [14] obtained by
weighted summation of results on enriched 69Ga and 71Ga
targets. In the low energy region (up to 5 MeV) the new
data overlap with the values of Johnson et al. [16].
The results of TENDL2011 for the (p,n) and (p,3n)
reactions agree here very well with our experimental
values.
Figure 3: Excitation function for natGa(d,xn)68,69Ge,
comparison with literature values and theoretical values
The excitation function for production of 69Ge is very
similar for deuterons and protons although, as expected, a
shift towards higher energy occurs (Fig. 3). The
TENDL2011 values represent well the shape of the
function but the predicted maximal value for the
69
Ga(d,2n)69Ge reaction is 50% higher than the measured
one.
Thick targets yields
Liquid/liquid extraction of nca 68Ge into toluene
From fits to our experimental results thick target yields
were calculated and presented in Fig. 4. For the
nat
Ga(p,x)68Ge the present results confirm the IAEA
TECDOC-1211 recommended yields. The difference of
10 percents is within experimental uncertainties.
From Fig. 4 it is clear that deuteron induced production
of 68Ge on Ga targets is not interesting as the expected
yield is lower than for proton irradiation, thicker targets
are needed and higher deuteron energy are required.
At the end of the stripping process the solution is
transferred into the extractor by means of a high-speed
peristaltic pump. The PVDF Extractor is fitted with a
magnetic Teflon-coated stirring bar and a microconductivity phase sensor at the outlet. Two
electromagnets surrounding the extractor body in an
appropriate geometry generate a magnetic rotating field
causing the magnetic stirring bar to rotate when 50Hz
current is applied through the coils of the magnets. An
inlet tube mounted in the cover plate allows introduction
of reagents by means of a peristaltic pump, a second tube
in the cover plate is connected to a gas-trap to collect any
radioactive component escaping from the extractor and
nitrogen pressure can be applied through a third tube
connected to a nitrogen canister via a valve.
Upon addition of 16 ml toluene, 3.3 ml HCl 10 N is
added to the extractor followed by 16 more ml of toluene.
Mixing of both phases is achieved by vigorous stirring (3
min). During this mixing more than 97% of the 68Ge is
selectively transferred into the toluene phase as 68GeCl4.
Next, separation of both phases under gravity force is
allowed for 15 min whereupon the aqueous phase is
transferred to the waste applying 50 mb nitrogen pressure
and opening the extractor outlet valve. Back-extraction of
68
Ge into the aqueous phase (20 ml 0.05 N HCl) is done
in a similar way. The aqueous phase containing the 68Ge is
transferred to a Temporary Storage Vial (TSV).
Figure 4: Thick target yields for production of 68Ge by
proton and deuteron irradiation of natural gallium
PC-CONTROLLED RADIOCHEMISTRY
SYSTEM FOR THE PRODUCTION OF
68
GE
The set-up, described in the following sections, is
similar to what was developed in collaboration with IBA
earlier for production of 64Cu [18].
Target dissolution
Upon bombardment and appropriate cooling down,
the irradiated target is mounted in the heated flow-through
stripper (fitted with a polycarbonate window closed by
means of compressed air). As stripping solution 27 ml
mixture of H2SO4 (11.45M ) and HNO3 (1.04N) acid is
re-circulated 30 min through a loop built up from a
recycling vial, a peristaltic pump and the flow through
stripper connected by Teflon tubing through an
appropriate combinations of valves. The stripper is heated
up to 100°C.
As during the next separation step (extraction of 68Ge
into toluene) NO gas bubbles may disturb the proper
separation of organic and aqueous phases in the microconductivity sensor cell at the outlet of the extractor, the
NO is reduced to N2 gas showing a limited solubility in
aqueous solutions. Therefore after 30 min of dissolution,
4 ml of 10% sulfamic acid is slowly (1ml.min-1) added
into the recycling vial and the recirculation continued for
30 more min. The quantity of copper dissolved during the
entire stripping process is always much less than 250 mg.
Final purification by ion exchange
chromatography
The removal of possible metallic trace impurities (Ga,
Ni, Cu …) is done by means of cation exchange
chromatography using Dowex-50 WX4 200-400 mesh
resin in the H+ form. The polycarbonate column used for
this purpose has a bed volume of 5.2 ml and is fitted with
a conductivity phase sensor/ 3/2 way valve at the inlet
while the outlet is directly connected to the Volume
Measuring Unit (VMU). The solution from the TSV is
pumped to the column at a flow-rate of 0.77 ml.min-1. In
0.05 N HCl, metallic impurities are exchanged for protons
as the solution moves through the column.
Volume measuring and sterilization
The neutral 68GeCl4 passes into the eluate that is
directly collected in the VMU. The latter is fitted with a
moving needle conductivity sensor mounted in the top
plate and reads the volume (after appropriate calibration)
with a resolution of 100 μl.
Samples for chemical and gamma-spectrometric
analysis are taken from the VMU by means of a peristaltic
pump. After sampling, the bulk solution is transferred to a
penicillin vial using a peristaltic pump/0.22 μm filter
combination.
Yield and purity of the bulk solution
Tracer experiments using the shorter lived 69Ge (T1/2 =
39 hr) tracer revealed a mean yield of 90% (Table 1). The
total chemistry time amounts to 2.5 hr. Mean bulk
impurities measured by Stripping Voltammetric Analysis
are quite acceptable: Cu: 0.15ppm; Ni: 0.06ppm and Ga:
0.14ppm (Table 2).
Table 1: Overview of 69Ge losses during radiochemistry
Target Plate Dissolution window Recycling vial Acid waste Organic waste Extractor
%
%
%
%
%
%
0.19
0.51
0.11
4.67
0.45
1.10
0.10
0.52
0.22
2.57
0.50
1.15
0.13
0.71
0.30
2.16
0.27
1.57
0.09
0.61
0.19
1.28
0.57
1.09
0.13
0.59
0.21
2.67
0.45
1.23
Test Nr
1
2
3
4
Mean
Table 2: Chemical purity of the bulk solution
Test Nr
1
2
3
4
Mean
Cu
ppm
0.161
0.140
0.213
0.101
0.15
Ni
ppm
0.064
0.083
0.045
<0.04
0.06
Ga
ppm
0.114
0.110
0.245
0.074
0.14
CONCLUSIONS
In order to make optimized production of nca 68Ge,
parent for a 68Ge-68Ga generator, possible using
commercially available 30 MeV proton cyclotrons, we
developed high quality Ga-Ni alloy (70/30%) targets. The
targets can withstand 350°C and are electroplated in 6.5 h
from an acidic bath in the standard set-up developed at
VUB. Four targets can simultaneous be prepared and
good quality targets with thicknesses ranging from 100 to
130 µm alloy are currently obtained. This is sufficient to
degrade incident beams of 28-26 MeV protons by about
10 MeV in the 6° beam/target angle geometry mostly
used around the IBA Cyclone cyclotrons. This ideal
energy range was derived from the upgraded data set for
cross sections of the natGa(p,xn)68Ge reaction where our
new results confirm the earlier experimental results and
the thick target yields published in the IAEA TECDOC1211. For an irradiation of 21 days at 300 µA on target a
batch yield of 2.9 Ci can be obtained. Deuteron induced
production of 68Ge on Ga targets is not interesting as the
expected yield is 30% lower than for proton irradiation,
thicker targets are needed and higher deuteron energy are
required.
For handling of the irradiated targets an automated, PC
controlled separation-purification chemistry has been
developed. In four steps (dissolution in H2SO4/HNO3,
extraction in toluene, back extraction in diluted HCl and
final purification over a cation exchange chromatographic
column), nca 68Ge is obtained in 25 ml 0.05 M HCl. The
full procedure was performed several times using 69Ge as
tracer and showed a reproducible overall yield of about
90%.Mean bulk impurities measured are quite acceptable:
Cu: 0.15 ppm; Ni: 0.06 ppm and Ga: 0.14 ppm.
REFERENCES
[1] F. Rösch, World Journal of Nuclear Medicine, 10
(2011) 26
[2] S.M. Qaim, Radiochimica Acta, 100 (2012) 635
[3] IAEA Radioisotopes and radiopharmaceuticals
series, N° 2: Production of long lived parent nuclides
Volatilized 69Ge
%
0.21
0.04
0.11
0.00
0.09
Chromatographic unit
%
<0.1
0.04
0.07
0.04
0.06
Bulk yield 69Ge vs reference solution
%
92.41
90.07
93.42
91.37
91.82
for generators: 68Ge, 82Sr, 90Sr and 188W. IAEA
Vienna (2010)
[4] R. Adam-Rebeles, P. Van den Winkel, L. De Vis,
Appl. Radiat. Isotopes 65 (2007) 995
[5] R. Adam-Rebeles, “Optimization of accelerator based
radionuclide production”, PhD thesis presented at the
Vrije Universiteit Brussel (2011)
[6] A. Hermanne, L. Daraban, R. Adam Rebeles, A.
Ignatyuk, F. Tarkanyi, S. Takacs, Nucl. Instr.
Methods B 268 (2010) 1376
[7] A. Hermanne, R. Adam Rebeles, F. Tárkányi, S.
Takacs, M. Takacs, A. V. Ignatyuk, Nucl.
Instr.Methods B 269 (2011) 2563
[8] A. Hermanne, S. Takacs, F. Tarkanyi, R. AdamRebeles, A. V. Ignatyuk, Appl. Radiat. Isotopes 70
(2012) 2763
[9] J.F. Ziegler, M.D. Ziegler, J.P. Biersack, SRIM2006,
http://www.srim.org
[10] National
Nuclear
Data
Center,
USA,
http://www.nndc.bnl.gov
[11] A.J. Koning, S. Hilaire, S. Goriely, TALYS-1.4, A
nuclear reaction program, User manual, Consultancy
Group (NRG), Petten, The Netherlands, (2011)
[12] A.J. Koning, D. Rochman, TENDL-2011 TALYSbased Evaluated Nuclear Data Library, Nuclear
Research and Consultancy Group (NRG), Petten, The
Netherlands, Release date December 8, 2011,
http://www.talys.eu/tendl-2011
[13] V.N Levkovskii, Activation cross sections for the
nuclides of medium mass region (A = 40-100) with
protons and a particles at medium (E = 10-50 MeV)
energies (Experiments and systematics). Inter-Vesi
Moscow, (1991) – EXFOR A0510
[14] N.T. Porile, H. Tanaka, H. Amano, Nuclear Physics
43 (1963) 500
[15] F. Tarkanyi, S. Takacs, K. Gul, A. Hermanne, M.G.
Mustafa, M. Nortier, P. Oblozinsky, S.M. Qaim, B.
Scholten, Yu. N. Shubin, Z. Youxiang, Charged
particle cross section database for medical
radioisotope production: diagnostic radioisotopes and
monitor reactions. IAEA TECDOC-1211, IAEA,
Vienna (2001), http://www-nds.iaea.org/medical/.
[16] C.H. Johnson, C.C. Trail, A. Galonsky, Physical
Review B, 136 (1964) 1719
[17] A. Karpeles, Radiochimica Acta 12 (1969) 212
[18] R. Adam Rebeles, P. Van den Winkel, A. Hermanne,
L. De Vis, R. Waegeneer, J. Radioanalytical and
Nuclear Chemistry 286 (2010) 665