Microfluidic technology: An economical and versatile approach for the synthesis of O-(2-[18F]fluoroethyl)-l-tyrosine ([18F]FET)

Article abstract of DOI:10.1016/j.bmcl.2012.01.083

A new synthesis of O-(2-[¹⁸F]fluoroethyl)-l-tyrosine [ ¹⁸F]FET was developed using a NanoTek microfluidic synthesis system (Advion BioSciences, Inc.). Optimal reaction conditions were studied through screening different reaction parameters like temperature, flow rate, reaction time, concentration of the labeling precursor, and the applied volume ratio between the labeling precursor and [¹⁸F]fluoride. [ ¹⁸F]FET was obtained after HPLC purification with 50% decay-corrected radiochemical yield starting from as little as 40 μg of labeling precursor. Small animal PET studies in EMT-6 tumor bearing mice showed radioactivity accumulation in the tumor (SUV_60min 1.21 ± 0.2) resulting in an slightly increasing tumor-to-muscle ratio over time.

Full text of DOI:10.1016/j.bmcl.2012.01.083

Bioorganic & Medicinal Chemistry Letters 22 (2012) 2291–2295
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Microfluidic technology: An economical and versatile approach for the
synthesis of O-(2-[18F]fluoroethyl)-L-tyrosine ([18F]FET)
⇑
Vincent Bouvet, Melinda Wuest, Pui-Hang Tam, Monica Wang, Frank Wuest
Department of Oncology, University of Alberta, Edmonton, Canada T6G 1Z2
a r t i c l e i n f o
a b s t r a c t
A new synthesis of O-(2-[18F]fluoroethyl)- -tyrosine [18F]FET was developed using a NanoTekÒ microflu-
L
Article history:
Received 16 December 2011
Revised 17 January 2012
Accepted 20 January 2012
Available online 2 February 2012
idic synthesis system (Advion BioSciences, Inc.). Optimal reaction conditions were studied through
screening different reaction parameters like temperature, flow rate, reaction time, concentration of the
labeling precursor, and the applied volume ratio between the labeling precursor and [18F]fluoride.
[
18F]FET was obtained after HPLC purification with 50% decay-corrected radiochemical yield starting from
as little as 40 lg of labeling precursor. Small animal PET studies in EMT-6 tumor bearing mice showed
radioactivity accumulation in the tumor (SUV60min 1.21 0.2) resulting in an slightly increasing tumor-
to-muscle ratio over time.
Keywords:
[
Microfluidic technology
Positron emission tomography (PET)
Small animal PET imaging
18F]Fluoroethyltyrosine ([18F]FET)
Ó 2012 Elsevier Ltd. All rights reserved.
The development of radiolabled molecular probes for molecular
imaging with positron emission tomography (PET) has evolved into
a complex chemical science. Besides significant advances in radio-
nuclide production and the development of novel radiolabeling
chemistry, the application of novel technologies has further stimu-
lated new direction in current radiotracer development. Over the
last years, various micro-reactor and microfluidic devices have
been introduced as novel and highly innovative technology to the
synthesis of radiolabeled compounds. Application of micro-reactor
and microfluidic technology offers major advantages like higher
yields, shorter reaction times and more selective product forma-
tion compared to conventional batch reactions. Miniaturization
of radiosynthesis devices based on microreactor and microfluidic
technology is especially attractive for the synthesis of radiotracers
labeled with the short-lived positron emitter fluorine-18 (18F,
t1/2 = 109.8 min). Micro-reactor and microfluidic technology have
successfully been applied to the synthesis of important PET radio-
for the radiosynthesis of [18F]FET. The first reaction route involves
formation of 2-[18F]fluoroethyl tosylate starting from 1,2-bis(tosyl-
oxy)ethane and subsequent 18F-fluoroethyl alkylation reaction
with the disodium salt of L
-tyrosine to afford [18F]FET (route 1,
Fig. 1).10,11 A second synthesis approach commences with an
aliphatic nucleophilic radiofluorination reaction using no-carrier-
added [18F]fluoride on fully protected tosyloxyethyl-
L-tyrosine
followed by an acidic deprotection step using trifluoroacetic acid
(TFA) (route 2, Fig. 1).12 Figure 1 summarizes the two different
synthesis routes for the preparation of [18F]FET.
[
18F]FET syntheses were performed using conventional heating
or microwave activation. Multiple optimizations of such methodol-
ogy have been published and the most commonly used are sum-
marized in Table 1.12–14
Although all these methodology afford reasonable radiochemi-
cal yields, they also require a significant amount of labeling precur-
sor ranging from 5 mg up to 30 mg per synthesis. In the last
decade, various micro-fluidic technologies have emerged for the
synthesis of radiolabeled compounds.
Recently we have applied microfluidic technology for the radio-
synthesis of the hypoxia imaging agent [18F]FAZA.3 By investigat-
ing the influence of different reaction parameters (concentration,
flow rate, temperature and residency time) of the NanotekÒ micro-
fluidic synthesis system, we demonstrated the potential of micro-
fluidic technology in radiopharmaceutical chemistry as a fast and
reliable synthesis approach for 18F-labeled radiotracers while using
only very small amounts of labeling precursor.
tracers [18F]FDG,1 18F]FLT,2 and [18F]FAZA.3
[
Various 18F-labeled amino acids have also found many applica-
tions in clinical and pre-clinical molecular imaging of tumors.4
Several studies have described the use of 18F-labeled amino acids
or analogs for measurement of protein synthesis rate or solely
amino acid transport in tumors.5–7
A widely used 18F-labeled amino acid is O-(20-[18F]fluoro-ethyl)-
L
-tyrosine ([18F]FET).8 18F]FET is a frequently used clinical PET
[
radiotracer for the detection of hyperbolic brain lesions such as
cerebral gliomas.9 Two different reaction routes are currently used
In the present study, we have carried out a comparative study to
optimize radiosynthesis of [18F]FET using microfluidic technology.
We have analyzed the complex array of different reaction
⇑
Corresponding author.
E-mail address: wuest@ualberta.ca (F. Wuest).
0960-894X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2012.01.083

2292
V. Bouvet et al. / Bioorg. Med. Chem. Lett. 22 (2012) 2291–2295
Synthesis route 1:
OTs
TsO
CO2Na
[
18F]F-
NH2
NaO
OTs
18F
CO2H
NH2
18F
O
Synthesis route 2:
2
t
18F]FET
CO2 Bu
[
1. [18F]F-
2. TFA
TsO
O
NHTrt
1
Figure 1. Conventional [18F]FET syntheses.12–14
conditions linked to the microfluidic synthesis system set-up.
Microfluidic technology-prepared [18F]FET was used for cell uptake
studies in murine breast cancer cell line EMT6. Dynamic small ani-
mal PET studies were carried out in EMT-6 tumor-bearing mice.
The synthesis of [18F]FET (2) involves a two-step process con-
sisting of (1) incorporation of [18F]fluoride into tosylate labeling
precursor 1 via nucleophilic radiofluorination followed by (2)
hydrolysis of the protecting using acidic conditions 2.5 N HCl at
100 °C for 7 min. The reaction sequence for the radiosynthesis of
The second parameter that we investigated was the volume ra-
tio (V[precursor]/V[18F]) between labeling precursor
and
18F]fluoride injected into the reactor. The influence of the volume
1
[
ratio on the radiochemical yield is given in Table 2. The volume ra-
tios in this set of reactions were 1 (Table 2, entry 3), 2 (Table 2, en-
try 5) and 3 (Table 2, entry 6). Best radiochemical yields were
obtained by using a volume ratio of 2 (46 4) or 3 (45 6%). Appli-
cation of a volume ratio of 1 gave lower radiochemical yields of
21 3% (Table 2, entry 3). The different volume ratios were
achieved by injecting the respective volume equivalents of the
labeling precursor 1 into the mixing chamber of the reactor. To
achieve a volume ratio of 2 (Table 2, entry 5), two volume equiva-
[
[
18F]FET is depicted in Figure 1 (route 2). Cyclotron-produced
18F]fluoride was converted into powerful nucleophilic radiofluori-
nation agent [18F]KF by means of kryptofix K222 and K2CO3 in ace-
tonitrile/water and subsequent azeotropic distillation using the
Advion Nano-Tek concentrator as displayed in Scheme 1.
lents (20
L/min for each volume equivalent (10
min) of [18F]fluoride into the mixing chamber of the reactor. This
resulted in an overall flow rate of 12 L/min. Increased volume ra-
lL) of labeling precursor 1 were injected at a flow rate of
8
l
l
L at a flow rate of 4 L/
l
In our experimental setup involving microfluidic reactor system
(Scheme 1), labeling precursor 1 and redissolved [18F]KF were both
solubilized in DMSO and introduced into the system through their
dedicated injection loops. Once the reactants were transferred onto
the injection loops, the solvent (DMSO) was pushed through Port E
to deliver the reactants into the microreactor through Port C (right
panel of Scheme 1). Each reaction was performed using 5–200 MBq
l
tio from 2 to 3 seems not to result in significantly higher radio-
chemical yields. Consequently, further increase of the volume
ratio from 2 to 3 is not useful and would only result in using more
precursors. The most efficient volume ratio V[precursor]/V[18F] for
the incorporation of [18F]fluoride into labeling precursor 1 was 2.
These results seem consistent with the incorporation of [18F]fluo-
ride into [18F]FAZA that we previously described.
Another parameter that needs to be explored is the flow rate at
which the reaction mixtures are entering the reactor. The flow rate
is directly linked with two additional reaction parameters like de-
scribed in our previous study: The mixing of the reactants and the
residence time of the aliquot within the reactor. In order to change
the flow rate of the reactions while keeping all the other parame-
ters constant, especially the residence time within the reactor, it
was necessary to perform the reactions in different reactors sizes.
Table 3 summarizes the influence of different flow rates on the
radiochemical yield. Data of entries 1, 2 and 3 in Table 3 clearly
demonstrate the positive effect of an increased flow rate on the
radiochemical yield while maintaining the reactor residency time
of [18F]KF solubilized in 10–20
lL of DMSO.
In the first set of reactions, we tested the influence of the reac-
tion temperature and the volume ratio on the radiochemical yield
as reflected by the incorporation of [18F]fluoride into our labeling
precursor. The results are summarized in Table 2 (entry 1–5).
Results show that the optimum reaction temperature for the incor-
poration of [18F]fluoride into labeling precursor 1 via microfluidic
nucleophilic radiofluorination is 140 °C (Table 2, entry 3 and 5).
At lower temperatures, incorporation yield of [18F]fluoride drops
significantly to 9 2% at 120 °C for a volume ratio of 2 (Table 2,
entry 4) and no incorporation occurs at 120 °C for a volume ratio
of 1 (Table 2, entry 2). At higher reaction temperature of 160 °C,
substantial degradation occurred which resulted in the formation
of various radioactive by-products.
Table 1
Comparison of recent conventional and microwave-assisted [18F]FET synthesis
Conventional synthesis I [12]
Conventional synthesis II [13]
Microwave synthesis [14]
Precursor amount (mg)
Reaction time (min)a
HPLC purification
10
80
Yes
85
50
5
5
60
No
90
35
30
Yes
N/A
37
Reaction temperature (°C)
Radiochemical yieldb (%)
a
Reaction time of radiofluorination step.
Radiochemical yields are decay-corrected.
b

V. Bouvet et al. / Bioorg. Med. Chem. Lett. 22 (2012) 2291–2295
2293
Scheme 1. Advion NanoTek set-up.
Table 2
Influence of the temperature and the volume ratio on the [18F]fluoride incorporationa
Entry
Temp. (°C)
Volume ratiob
Overall flow rate (
l
L/min)
Residency timec (sec)
Radiochemical yield (n = 3) [%] SDd
1
2
3
4
5
6
100
120
140
120
140
140
1
1
1
2
2
3
12
12
12
12
12
12
320
320
320
320
320
320
0
0
21
9
46
45
3
2
4
6
a
All reactions were carried out in 64
10 to 20 L.
Volume ratio=V[precursor]/V[18F].
lL reactors (100 l
m ID) using DMSO as solvent and a labeling precursor 1 concentration of 1 mg/mL. [18F]Fluoride bolus varied from
l
b
c
Residency time reflects the time the solution resided inside the reactor.
d
The radiochemical yields were determined by radio-HPLC referring to the percentage of the radioactivity area of 18F-labeled intermediate to the total radioactivity area.
Table 3
Influence of the flow rate and residency time on [18F]fluoride incorporationa
Entry
Reactor size (
lL)
Volume ratio
Overall flow rate (lL/min)
Residency time (sec)
Radiochemical yieldb (n = 3) [%] SD
1
2
3
4
5
32
64
128
32
2
2
2
2
2
9
18
36
18
18
213
213
213
107
426
33
66
72
40
62
6
8
5
7
5
128
a
All reactions were carried out at 140 °C using DMSO as solvent and a labeling precursor 1 concentration of 2 mg/mL. [18F]Fluoride bolus varied from 10 to 20
lL. For 10 lL
[
18F]Fluoride bolus, 20
lL (40
lg) of precursor 1 are used.
b
The radiochemical yields were determined by radio-HPLC referring to the percentage of the radioactivity area of 18F-labeled intermediate to the total radioactivity area.
constant at 213 s. High radiochemical yields of up to 66 8% and
72 5% were obtained in a 64 and 128- L reactor (two 64- l reac-
tors in series) at an overall flow rate of 18 and 36 L/min, respec-
observed while optimizing the synthesis of [18F]FAZA and was
attributed to the chaotic mixing within the reactors due to the
viscosity of DMSO.3,15 In this particular case flow rates above
l
l
l
tively (entry 2 and 3). Lower flow rates (Table 3, entry 1) seem to
18
lL/min do not seem to significantly improve the radiochemical
provide lower radiochemical yields. Similar phenomenon was
yield. Furthermore, higher flow rate, especially in the range of

2294
V. Bouvet et al. / Bioorg. Med. Chem. Lett. 22 (2012) 2291–2295
Table 4
Influence of the precursor 1 concentration on the [18F]fluoride incorporationa
Entry
Conc. (mg/mL)
Volume ratio
Overall flow rate (
l
L/min)
Residency time (sec)
Radiochemical yieldb (n = 2) [%] SDb
1
2
3
4
0.5
1
2
2
2
2
2
12
12
12
12
320
320
320
320
22
46
65 10
8
5
4
69
2
a
All reactions were carried out at 140 °C using DMSO as solvent and a 64 l l
L reactor. [18F]Fluoride bolus varied from 10 to 20
L.
b
The radiochemical yields were determined by radio-HPLC referring to the percentage of the radioactivity area of 18F-labeled intermediate to the total radioactivity area.
36
l
L/min, results in higher back pressure which can cause over-
but also implies some limitations. Unless working with excessive
amounts of [18F]fluoride, it is manageable to resolubilize about
pressure complications.
The influence of the reactor size, which is directly linked with
the residency time, was studied in another set of reactions while
all other reaction conditions remained constant (Table 3, entry 2,
4 and 5). In this case, the flow rate needs to stay constant
(18
change the residency time. The residency maximum radiochemical
yields of 66 8% and 72 4% were obtained when 64 and 128
reactors were used (Table 3, entry 2 and 5). Due to the fact that
there are no significant improvements while increasing the resi-
dency time from 213 to 426 s, we decided to continue our optimi-
100–200 MBq of [ lL of solvent (especially
18F]fluoride in 10
DMSO). Therefore, the system is perfectly suited for preclinical
studies like small animal PET, and single dose patient studies.
Higher radioactivity levels (>200 MBq) can be achieved by increas-
ing the injected volume. However, this would lead to the need to
increase the overall reaction time and the amount of precursor.
Microfluidic technology-prepared [18F]FET was used for cell up-
take studies in murine breast cancer cells EMT-6. The murine
mammary breast cancer model EMT-6 was used in previous radio-
pharmacological studies to analyze the uptake of radiotracers for
measuring tumor hypoxia or tumor metabolism.3,16 The obtained
radiotracer accumulated over time in EMT-6 cells reaching a max-
imum cell uptake of 105 7% ID/mg protein (n = 3) after 15 min
incubation time. Over the following time course the radioactivity
amount in the cells decreased to 60 4% ID/mg protein (n = 3) indi-
cating no trapping mechanism for this tracer in the EMT-6 cells
(Fig. 2). Decreasing intracellular radiotracer levels over time and
tracer washout was also found studying in vitro uptake of
lL/min) while the size of the reactor is modified in order to
lL
zation using 64 lL reactors as it shortens the reaction time by half.
Multiple examples in literature have demonstrated that the
radiochemical yield for direct or indirect radiofluorination directly
correlates to the amount of labeling precursor. In the last set of
reaction, we studied the influence of different labeling precursor
concentrations on the radiochemical yield of [18F]FET. The results
displayed in Table 4 clearly demonstrate the importance of the
amount of labeling precursor on the radiochemical yield. Only a lit-
tle amount of product (22%) was formed when low concentration
(0.5 mg/mL) of labeling precursor 1 was used (Table 4, entry 1).
Significantly higher radiochemical yields of 65% and 69% were ob-
tained at higher concentrations of 2 mg/mL (Table 4, entry 3) and
4 mg/ml (Table 4, entry 4). This finding is consistent with results
reported in the literature emphasizing the importance of labeling
precursor amount upon the radiochemical yield.
[
18F]FET into other tumor cell lines such as SW707 (human colon
carcinoma cell line)17 and EL4 (mouse lymphoma cell line).18
Dynamic small animal PET imaging showed [18F]FET uptake in
EMT-6 tumors in vivo (Fig. 3). The tumor starts to become clearly
visible at 10 min pi, with increasing tracer accumulation over time
reaching SUV values of 1.28 0.2 at 30 min and 1.21 0.2 at
60 min, respectively. Based on its missing trapping mechanism
and because [18F]FET is not metabolized in vivo, tumor uptake of
the radiotracer reaches its maximum after 30–60 min depending
on the tumor model studied.17,19 Muscle uptake of [18F]FET in
BALB/c mice remained fairly constant after 10 min pi, and tended
to decrease after 30 min pi towards 60 min pi. The tumor/muscle
ratio slightly increased reaching a value of 4.04 0.8 (n = 3) at
the end of the experiment.
Optimal parameter obtained for [18F]FET synthesis were compa-
rable to those observed for our previously reported [18F]FAZA syn-
thesis.3 Optimal temperatures for microfluidic syntheses are
usually significantly higher than for conventional syntheses
(140 °C (Table 2) vs 90 °C (Table 1)) and can be attributed to the
short residency time of the precursor within the reactor. Also, opti-
mal volume ratio of 2–3 (Table 2) and higher flow rates (Table 3)
result in significantly improved yields and can be attributed to
an optimized mixing.3,15
Various biodistribution studies in different mouse tumor mod-
els revealed constant tumor/muscle ratios up to 2 h pi17 or
decreasing values from 2.7 to 1.7, respectively.19 Based on its
Best results were obtained using the following reaction condi-
tions: 40 lg of labeling precursor 1 in 20 lL of DMSO (2 mg/ml),
overall flow rate = 18 lL/min, volume ratio (V[precursor]/
V[18F]) = 2, reaction temperature = 140 °C, residence time = 213 s.
The microfluidic radiofluorination was carried out in 7 min (104000
for the injection of 30 lL reaction mixture at 18 l
L/min, 303300 res-
idence time in the reactor, 104700 to transfer the reaction mixture
from the reactor into the hydrolysis vial). These parameters affor-
ded radiochemical yields of 66% 8 for the first step of the reaction
sequence within the synthesis sequence of [18F]FET. This repre-
sents a significant improvement compared to conventional radiola-
beling method especially with respect to the amount of labeling
precursor used (40 lg vs 5–20 mg). Application of optimized reac-
tion conditions in combination with subsequent acid hydrolysis of
our radiolabeled intermediate with 2.5 N HCl at 100 °C for 7 min
and HPLC purification afforded [18F]FET in 50% radiochemical yield
(decay-corrected) in less than 45 min.
It is important to mention that drying and resolubilization of
[
18F]fluoride using the NanotekÒ microfluidic system is convenient
Figure 2. Cell uptake of [18F]FET in EMT-6 cells.

V. Bouvet et al. / Bioorg. Med. Chem. Lett. 22 (2012) 2291–2295
2295
Figure 3. Dynamic small animal PET image in EMT-6 tumor bearing BALB/c mice.
favorable radiopharmacokinetic parameters, [18F]FET has been
developed and frequently applied as a clinical PET radiotracer in
recent years.8
2. Akula, M.; Collier, T.; Kabalka, G.; Wall, J.; Kennel, S.; Stuckey, A.; LeBlanc, A. J.
Nucl. Med. 2010, 51, 1473.
3. Bouvet, V. R.; Wuest, M.; Wiebe, L. I.; Wuest, F. Nucl. Med. Biol. 2011, 2, 235.
4. McConathy, j.; Goodman, M. M. Cancer Metastasis Rev. 2008, 27, 555.
5. Hawkins, R. A.; Huang, S. C.; Barrio, J. R.; Keen, R. E.; Feng, D.; Mazziotta, J. C.,
et al J. Cereb. Blood Flow Metab. 1989, 9, 446.
6. Jager, P. L.; Vaalburg, W.; Pruim, J.; de Vries, E. G.; Langen, K. J.; Piers, D. A. J.
Nucl. Med. 2001, 42, 432.
7. Wiesel, F. A.; Blomquist, G.; Halldin, C., et al J. Nucl. Med. 1991, 32, 2041.
8. Langen, K. J.; Hamacher, K.; Weckesser, M.; Floeth, F.; Stoffels, G.; Bauer, D.,
et al Nucl. Med. Biol. 2006, 33, 287.
9. Lahoutte, T.; Caveliers, V.; Camargo, S. M.; Franca, R.; Ramadan, T.; Veljkovic, E.,
et al J. Nucl. Med. 2004, 45, 1591.
10. Wester, H. J.; Herz, M.; Weber, W.; Heiss, P.; Senekowitsch-Schmidtke, R.;
Schwaiger, M.; Stocklin, G. J. Nucl. Med. 1999, 40, 205.
11. Zuhayra, M.; Alfteimi, A.; Von Forstner, C.; Lutzen, u.; Meller, B.; Henze, E.
Bioorg Med. Chem. 2009, 17, 7441.
In summary, we have optimized and investigated the influence
of different reaction parameters of the NanoTek Microfluidic Syn-
thesis System by the optimization of the radiosynthesis of
[
18F]FET. Careful screening of various reaction parameters afforded
a significant improvement on the amount of precursor consumed
for one synthesis. Less than 1% of precursor amount used in con-
ventional methodology is actually used in our optimized microflu-
idic methodology. This procedure allows the convenient
preparation of [18F]FET suitable for preclinical studies.
12. Hamacher, K.; Coenen, H. H. Appl. Radiat. Isot. 2002, 57, 853.
13. Wang, H.-E.; Wua, S.-Y.; Chang, C.-W.; Liu, R.-S.; Hwang, L.-C.; Lee, T.-W.; Chen,
J.-C.; Hwan, J.-J. Nucl. Med. Biol. 2005, 32, 367.
Acknowledgments
14. Wang, M.; Yin, D.; Cheng, D.; Li, G.; Wang, Y. J. Radioanal. Nucl. Chem. 2006, 270,
439.
15. Bringer, M. R.; Gerdts, C. J.; Song, H.; Joshua, D.; Ismagilov, T.; Ismagilov, R. F.
Philos. Trans. R. Soc. London, Ser. A 2004, 362, 1087.
16. Wuest, M.; Trayner, B. J.; Grant, T. N.; Jans, H. S.; Mercer, J. R.; Murray, D.; West,
F. G.; McEwan, A. J.; Wuest, F.; Cheeseman, C. I. Nucl. Med. Biol. 2011, 38, 461.
17. Heiss, P.; Mayer, S.; Herz, M.; Wester, H. J.; Schwaiger, M.; Senekowitsch-
Schmidtke, R. J. Nucl. Med. 1999, 40, 1367.
The authors would like to thank John Wilson, David Clendening
and Jayden Sader from the Edmonton PET Center for radionuclide
production and excellent technical support. F.W. thanks the Dianne
and Irving Kipnes Foundation for supporting this work.
References and notes
18. Abe, K.; Hayashi, K.; Sasaki, M.; Koga, H.; Kaneko, K.; Sawamoto, H.; Himuro, K.;
Honda, H. Acta Radiol. 2006, 47, 1042.
1. Lee, C. C.; Sui, G. D.; Elizarov, A.; Shu, C. Y. J.; Shin, Y. S.; Dooley, A. N., et al
Science 2005, 310, 1793.
19. Tang, G.; Wang, M.; Tang, X.; Luo, L.; Gan, M. Appl. Radiat. Isot. 2003, 58, 219.