Synthesis of [18F]xenon difluoride as a radiolabeling reagent from [18F]fluoride ion in a micro-reactor and at production scale

Article abstract of DOI:10.1016/j.jfluchem.2010.07.009

[¹⁸F]Xenon difluoride ([¹⁸F]XeF₂), was produced by treating xenon difluoride with cyclotron-produced [ ¹⁸F]fluoride ion to provide a potentially useful agent for labeling novel radiotracers with fluorine-18 (t_1/2 = 109.7 min) for imaging applications with positron emission tomography. Firstly, the effects of various reaction parameters, for example, vessel material, solvent, cation and base on this process were studied at room temperature. Glass vials facilitated the reaction more readily than polypropylene vials. The reaction was less efficient in acetonitrile than in dichloromethane. Cs⁺ or K⁺ with or without the cryptand, K 2.2.2, was acceptable as counter cation. The production of [¹⁸F]XeF₂ was retarded by K₂CO₃, suggesting that generation of hydrogen fluoride in the reaction milieu promoted the incorporation of fluorine-18 into xenon difluoride. Secondly, the effect of temperature was studied using a microfluidic platform in which [ ¹⁸F]XeF₂ was produced in acetonitrile at elevated temperature (≥85 °C) over 94 s. These results enabled us to develop a method for obtaining [¹⁸F]XeF₂ on a production scale (up to 25 mCi) through reaction of [¹⁸F]fluoride ion with xenon difluoride in acetonitrile at 90 °C for 10 min. [¹⁸F]XeF ₂ was separated from the reaction mixture by distillation at 110 °C. Furthermore, [¹⁸F]XeF₂ was shown to be reactive towards substrates, such as 1-((trimethylsilyl)oxy)cyclohexene and fluorene.

Full text of DOI:10.1016/j.jfluchem.2010.07.009

Journal of Fluorine Chemistry 131 (2010) 1032–1038
Contents lists available at ScienceDirect
Journal of Fluorine Chemistry
journal homepage: www.elsevier.com/locate/fluor
Synthesis of [18F]xenon difluoride as a radiolabeling reagent from [18F]fluoride ion
in a micro-reactor and at production scale
*
Shuiyu Lu , Victor W. Pike
PET Radiopharmaceutical Sciences Section, Molecular Imaging Branch, National Institute of Mental Health, National Institutes of Health, 10 Center Drive, Room B3 C346,
Bethesda, MD 20892-1003, USA
A R T I C L E I N F O
A B S T R A C T
18F]Xenon difluoride ([18F]XeF2), was produced by treating xenon difluoride with cyclotron-produced
18F]fluoride iontoprovide a potentially usefulagentforlabelingnovel radiotracerswith fluorine-18(t1/2
Article history:
[
[
Received 27 May 2010
Received in revised form 14 July 2010
Accepted 19 July 2010
=
109.7 min) for imaging applications with positron emission tomography. Firstly, the effects of various
reaction parameters, for example, vessel material, solvent, cation and base on this process were studied at
room temperature. Glass vials facilitated the reaction more readily than polypropylene vials. The reaction
was less efficient in acetonitrile than in dichloromethane. Cs+ or K+ with or without the cryptand, K 2.2.2,
was acceptable as counter cation. The production of [18F]XeF2 was retarded by K2CO3, suggesting that
generation of hydrogen fluoride in the reaction milieu promoted the incorporation of fluorine-18 into
xenon difluoride. Secondly, the effect of temperature was studied using a microfluidic platform in which
Available online 27 July 2010
Keywords:
[
[
18F]Xenon difluoride
18F]Fluoride ion
Labeling agent
Micro-reactor
Radiotracers
[
18F]XeF2 was produced in acetonitrile at elevated temperature (ꢀ85 8C) over 94 s. These results enabled us
to develop a method for obtaining [18F]XeF2 on a production scale (up to 25 mCi) through reaction of
[
18F]fluoride ion with xenon difluoride in acetonitrile at 90 8C for 10 min. [18F]XeF2 was separated from the
reaction mixture by distillation at 110 8C. Furthermore, [18F]XeF2 was shown to be reactive towards
substrates, such as 1-((trimethylsilyl)oxy)cyclohexene and fluorene.
Published by Elsevier B.V.
1. Introduction
Xenon difluoride has a rich chemistry for fluorinating organic
molecules, including arenes, alkenes, enols, thioethers, aryl
Positron emission tomography (PET) is increasingly applied to
drug development [1] and clinical research [2] as a molecular
imaging modality in animal and human subjects. PET can measure
biochemical targets and processes provided that specific radio-
tracers labeled with a positron-emitting radionuclide are available.
The development of such radiotracers is therefore an active field of
research [3]. Fluorine-18 (t1/2 = 109.7 min) is particularly attractive
as a label because it can mimic hydrogen or hydroxyl in organic
molecules. Furthermore, fluorine-18 can be produced in very high
activities (multi-Curies) with moderate energy cyclotrons according
to the 18O(p,n)18F nuclear reaction from 18O-enriched water [4]. This
process produces [18F]fluoride ion, which can be used directly in
nucleophilic substitution reactions to introduce fluorine-18 at
aliphatic and aryl carbons bearing suitable leaving groups [3].
However, [18F]fluoride ion cannot be used directly to replace
hydrogen at aryl or alkyl C–H groups. Electrophilic reagents, such as
aldehydes and aryl ketones under relatively mild conditions [7].
Xenon difluoride also performs fluorodeiodination [8], fluorode-
silylation [9] and fluorodecarboxylation [10,11] reactions. Hence,
18F-labeled xenon difluoride ([18F]XeF2) has long been recognized
as a potentially useful ‘electrophilic’ radiofluorination agent.
However, this potential has not been realized, mainly because of
the lack of a practical means for producing [18F]XeF2 rapidly in
useful activities (i.e., multi-mCi). In fact, only a few reports have
described the use of [18F]XeF2 to prepare a PET radiotracer [12].
[
18F]XeF2 was earlier prepared by the reaction of [18F]fluorine
with xenon [13], by the 19F(p,pn)18F nuclear reaction on XeF2 [12b],
and by ‘fluorine exchange’ of xenon difluoride with either
[
18F]hydrogen fluoride [14] or [18F]fluoride ion [15]. The exchange
with [18F]fluoride ion is attractive because of its direct use of readily
available cyclotron-produced [18F]fluoride ion. Thus, [18F]XeF2 was
obtained from
[ [
18F]fluoride ion (as 18F]FÀ–Cs+–K 2.2.2) in
[
18F]fluorine gas [5] and [18F]acetyl hypofluorite [6], have been used
dichloromethane in a ‘glassy carbon’ vessel at room temperature
[15]. However, reaction of xenon difluoride with [18F]FÀ–K+–K 2.2.2
in dichloromethane in either a glass or a Teflon vessel at room
temperature did not produce [18F]XeF2 [16]. These seemingly
for this purpose, but these labeling reagents cannot be prepared in a
single step from cyclotron-produced [18F]fluoride ion.
dissonant results prompted us to pursue
a more extensive
investigation of the effects of reaction parameters, for example,
vessel material, solvent, cation and base on the production of
[
* Corresponding author. Tel.: +1 301 451 3904; fax: +1 301 480 5112.
18F]XeF2. Moreover, interest in applying micro-reactor technology
E-mail address: Shuiyu.Lu@mail.nih.gov (S. Lu).
0022-1139/$ – see front matter. Published by Elsevier B.V.
doi:10.1016/j.jfluchem.2010.07.009

S. Lu, V.W. Pike / Journal of Fluorine Chemistry 131 (2010) 1032–1038
1033
to radiochemistry with fluorine-18 has recently increased, since it
has been demonstrated that the use of this technology is
advantageous for radiochemistry optimization and radiotracer
production [17]. Here we also took advantage of a microfluidic
device to study the synthesis of [18F]XeF2 from [18F]fluoride ion at
elevated temperatures. Finally, results from these studies in glass or
plastic vials and in a micro-reactor facilitated our development of a
new method for producing [18F]XeF2 on a multi-mCi scale.
When the reaction vial material was changed from glass to
polypropylene, with Cs+–K 2.2.2 as cation and dichloromethane as
solvent, the RCY of [18F]xenon difluoride was very low (3%) when
using a low amount (64
mmol) of xenon difluoride in a 90 min
reaction, but became appreciable (31%) when the amount of xenon
difluoride was nearly doubled.
In essence, the results obtained here in borosilicate glass vials in
the presence of Cs+–K 2.2.2 and NCA [18F]fluoride ion accord with
those obtained previously in glassy carbon reaction vials [15].
2. Results and discussion
2.1.2. In the presence of K+–K 2.2.2 and NCA [18F]fluoride ion
The decay-corrected radiochemical yields (RCYs) of [18F]XeF2
from reactions of xenon difluoride with NCA [18F]fluoride ion in the
presence of K+–K 2.2.2 under various conditions are listed in Table 1.
When K+–K 2.2.2 was used as counter cation, the trend in RCYs
was similar to that with Cs+–K 2.2.2. A combination of low xenon
This study began with the replication of published results [15]
and culminated in a new method for producing [18F]XeF2 on a
multi-mCi scale from cyclotron-produced [18F]fluoride ion. Essen-
tially, experiments in closed reaction vials, at room temperature,
showed that [18F]XeF2 production was low or slow in polypropyl-
ene vials and also in acetonitrile, relative to reactions in
borosilicate glass vials or dichloromethane, respectively, whereas
experiments in a silica glass micro-reactor showed that [18F]XeF2
could be produced in acetonitrile at elevated temperature. These
key findings were exploited to devise the method for producing
difluoride amount (56
mmol), dichloromethane as solvent and
glass as vial material gave [18F]XeF2 in 22% RCY after 120 min
(Table 1, entries 1–3), while an approximate doubling of the
amount of xenon difluoride resulted in 79% RCY after only 60 min
(Table 1, entries 4 and 5). In contrast to the results obtained with
Cs+ as cation, omission of K 2.2.2 (Table 1, entries 6 and 7) almost
halved the RCY (c.f. entries 7 vs. 5). This finding appears consistent
with the shorter charge separation and therefore lower nucleo-
philicity of the K+/18FÀ ion pair compared to that of the K+
–K 2.2.2/18FÀ ion pair. Only very low RCYs were obtained in a
polypropylene vessel, even with a higher amount of xenon
difluoride (Table 1, entries 8–11).
[
18F]XeF2 on a multi-mCi production scale.
2.1. Synthesis of [18F]xenon difluoride in conventional reaction vials at
room temperature
Reactions of xenon difluoride are well-known to be influenced
by the reaction vessel material [11,18,19]. Whether Pyrex
(borosilicate) glass might be compatible with [18F]XeF2 production
was a question of primary interest, since our original report had
described [18F]XeF2 production in a ‘glassy carbon’ vessel [15].
Glassy carbon is not a glass, but a vitreous carbon [20]; its inert
surface therefore clearly differs from that of borosilicate glass,
which shows protolytic behaviour [21]. For comparison we also
tested readily available plastic (polypropylene) vials.
Reactions of xenon difluoride may also depend on solvent [19]. In
our original report [15], dichloromethane supported [18F]XeF2
production in glassy carbon but acetonitrile failed to do so. We
thereforetested reactions inboth solvents. Ithasbeen suggested that
the production of [18F]XeF2 is promoted by hydrogen fluoride [16]
thatmaybegeneratedfromthereactionofexcessxenondifluorideor
fluoride ion [22] with organic reaction components, such as the
solvent (dichloromethane) or the kryptand (K 2.2.2). Reactions were
thereforeconductedwithandwithoutK2.2.2, and alsowithdifferent
amounts of base (carbonate) that might be expected to consume any
initially generated hydrogen fluoride. All reactions in this part of the
study were carried out at room temperature.
With the reaction still performed in a glass vial with K+–K 2.2.2 as
cation and with a low amount of xenon difluoride, change of solvent
to acetonitrile resulted in no [18F]XeF2 after 120 min (Table 1, entry
12). However, when the amount of xenon difluoride was almost
doubled, a high RCY of [18F]XeF2 was rapidly obtained (Table 1,
entries 13 and 14). No yield was obtained in a polypropylene vial
(Table 1, entries 15). By contrast, when K2CO3 was used in larger
amounts than usual (Table 1, entries 16 and 17), production of
[
18F]XeF2 was almost completely inhibited.
2.1.3. In the presence of K+–K 2.2.2 and CA [18F]fluoride ion
The RCYs of [18F]XeF2 from reactions of xenon difluoride with
carrier-added (CA) [18F]fluoride ion in the presence of K+–K 2.2.2
under various conditions are listed in Table 2. In glass vials, when
the base, K2CO3, was replaced with KF (3
mmol), the RCYs followed
the general trend observed in Table 1. The early rate of [18F]XeF2
production was however noticeably increased, since RCYs at
10 min were already substantial (Table 2, entries 1, 5 and 9). In
polypropylene vials with KF replacing K2CO3, RCYs were generally
low or very low (Table 2, entries 3, 7 and 11).
2.1.1. In the presence of Cs+–K 2.2.2 and NCA [18F]fluoride ion
In glass vials, a reaction of no-carrier-added (NCA) [18F]fluoride
2.1.4. General trends and possible mechanisms
ion with a low amount of xenon difluoride (11
presence of Cs+–K 2.2.2 in dichloromethane (300
produced no [18F]XeF2 but an increase in amount of xenon
difluoride to 27 mol increased RCY to 26% over 90 min. By use of
m
mol) in the
The data presented in Tables 1 and 2 revealed some strong
general trends. First, reactions conducted in glass vessels occurred
more readily than those in polypropylene vessels. Second,
reactions conducted in dichloromethane gave higher RCYs than
those in acetonitrile. Third, reactions conducted with low amounts
m
L) for 90 min
m
an even higher amount of xenon difluoride (106
higher RCY (78%) was obtained after 90 min.
mmol), a much
of xenon difluoride (<70
mmol) gave low RCYs and those with high
The presence of K 2.2.2 proved unnecessary, since comparable
RCYs (76% vs. 78%) were obtained initsabsenceand somewhat more
rapidly. This finding is consistent with the high reactivity of Cs18F in
organic solvent [23], and also the weaker capacity of K 2.2.2 to bind
the Cs+ cation relative to its ability to bind the smaller K+ cation [24].
However, the inclusion of K 2.2.2 was expedient, since it greatly
reduced deposition of [18F]fluoride ion onto glass vessel walls [25]
and consequently improved the efficiency of radioactivity transfer
from the glass vessel that was initially used for drying of the reagent
to the vessel subsequently used for its reaction.
amounts (100–130 mol) high RCYs. Fourth, the amount of added
m
carbonate base (K2CO3) had a dramatic effect on reaction outcome;
reactions proceeded readily in the absence of carbonate but were
suppressed by higher amounts of carbonate. These observations
with regard to the effect of various parameters on RCY are
generally consistent with a major role for hydrogen fluoride in
promoting the formation of [18F]XeF2, as proposed previously [16].
Hydrogen fluoride might be generated through reaction of
xenon difluoride with the reaction vessel wall, trace water [26],
solvent [27] or other susceptible material (e.g., K 2.2.2) [16] in the

1034
S. Lu, V.W. Pike / Journal of Fluorine Chemistry 131 (2010) 1032–1038
Table 1
RCYs of [18F]XeF2 from the treatment of xenon difluoride with no-carrier-added (NCA) [18F]fluoride ion in the presence of K+ or K+–K 2.2.2 as counter ion at RT, using different
vial materials, amounts of xenon difluoride, solvents, and reaction times.
Vessel
XeF2
(
m
mol)a
K2CO3
(
mmol)
K 2.2.2 (
m
mol)
Solvent
Time (min)
RCY of [18F]XeF2 (%)b
c
1
2
3
Glass
Glass
Glass
56
56
56
3.6
3.6
3.6
11
11
11
CH2Cl2
10
60
n.d.
8
CH2Cl2
CH2Cl2
120
22
4
5
Glass
Glass
122
122
3.6
3.6
11
11
CH2Cl2
CH2Cl2
10
60
66
79
6
7
Glass
Glass
125
125
3.0
3.0
0
0
CH2Cl2
CH2Cl2
10
60
8
40
8
9
Polypropylene
Polypropylene
62
62
3.6
3.6
11
11
CH2Cl2
CH2Cl2
10
60
n.d.
n.d.
10
11
Polypropylene
Polypropylene
121
121
3.6
3.6
11
11
CH2Cl2
CH2Cl2
50
90
1
3
12
Glass
58
3.6
11
MeCNd
120
n.d.
13
14
Glass
Glass
118
118
3.6
3.6
11
11
MeCN
MeCN
10
60
67
67
15
Polypropylene
125
3.6
11
MeCN
120
n.d.
16
17
Glass
Glass
95
13.9
23.1
41.7
69.3
MeCN
MeCN
10–180
10–180
2–3
n.d.
122
Data within horizontal lines are from the time-course of a single reaction mixture.
n.d. = not detected (<1%).
a
Reaction volume was 300
m
L.
b
c
Calculated as % peak area in the HPLC radio-chromatogram of the reaction mixture.
Efficiency of [18F]FÀ–K+–K 2.2.2 transfer from preparation vessel to reaction vessel was ꢁ20%.
Efficiency of [18F]FÀ–K+–K 2.2.2 transfer from preparation vessel to reaction vessel was ꢁ80%.
d
reaction medium. Thus, xenon difluoride is less stable in Pyrex
glass vessels than in Teflon-FEP vessels [19]. The instability in glass
is attributed to the acidity of the surface, which may therefore
generate some hydrogen fluoride from fluoride ion. This may
explain why [18F]XeF2 production was generally greater in glass
vessels. Xenon difluoride is reduced slowly with water at room
temperature and produces hydrogen fluoride [26]. Xenon difluor-
ide is known to react readily with dichloromethane at RT, but morÀe
stubbornly with acetonitrile, in each case to generate the HF2
anion, presumably via the generation of hydrogen fluoride [26]. In
turn, this may explain why the RCYs of [18F]XeF2 were generally
lower in acetonitrile than in dichloromethane. Xenon difluoride
has been shown to react slowly with K 2.2.2, also with the likely
production of hydrogen fluoride [16]. It seems likely that
incorporation of fluorine-18 into xenon difluoride only occurs
readily once any generated hydrogen fluoride has been produced
in excess of the available base (CO32À).
Originally, it was proposed that [18F]XeF2 production was
catalyzed in glassy carbon vessels by Cs+–K 2.2.2 complex [15], and
subsequently this proposal was questioned [16]. However ‘naked’
fluoride ion does exchange with xenon difluoride under rigorously
anhydrous and HF-free conditions [16]. Also, remarkably, aqueous
hydrogen/potassium [18F]fluoride slowly exchanges with xenon
difluoride at low temperature (ꢁ0.8% after 2 h at 0 8C) [26]. The
[
18F]fluoride ion used in our reactions was ‘dried’ by azeotropic
evaporation of the 18O-water with acetonitrile, and would
therefore not be completely ‘naked’ but hydrated to a low and
unknown extent [28]. Nevertheless, the reactivity of this reagent
would be much greater than that of aqueous fluoride ion.
Therefore, it is conceivable that a direct exchange mechanism
Table 2
RCYs of [18F]xenon difluoride from the treatment of xenon difluoride with carrier-added (CA) [18F]fluoride ion in the presence of KF or KF–K 2.2.2 at RT.
Vessel
XeF2
(
mmol)
KF (
m
mol)
K 2.2.2 (
m
mol)
Solvent
Time (min)
RCY of [18F]XeF2 (%)a
1
2
Glass
Glass
56
56
3
3
0
0
MeCNb
MeCN
10
40
20
19
3
4
Polypropylene
Polypropylene
60
60
3
3
0
0
MeCN
MeCN
10
40
15
21
c
5
6
Glass
Glass
90
90
3
3
0
0
CH2Cl2
10
40
72
72
CH2Cl2
7
8
Polypropylene
Polypropylene
102
102
3
3
0
0
CH2Cl2
CH2Cl2
10
40
n.d.
2
9
Glass
Glass
118
118
3
3
3.3
3.3
MeCN
MeCN
10
40
43
71
10
11
12
Polypropylene
Polypropylene
107
107
3
3
3.3
3.3
MeCN
MeCN
10
40
n.d.
n.d.
Data within horizontal lines were obtained by following the time-course of a single reaction mixture.
n.d. = not detected (<1%).
a
Calculated as % peak area of the HPLC radio-chromatogram of the reaction mixture.
b
Efficiency of K+[18F]FÀ transfer from preparation vessel to reaction vessel was ꢁ80% in the presence of K 2.2.2 and ꢁ20% in its absence.
c
Efficiency of K+[18F]FÀ transfer from preparation vessel to reaction vessel was ꢁ20%.

S. Lu, V.W. Pike / Journal of Fluorine Ch
[
emistry 131 (2010) 1032–1038
1035
may be operational under some circumstances, such as in the
reaction conducted in a glass vessel with Cs18F in dichloromethane,
which rapidly gave high RCYs of [18F]XeF2.
2.2. Synthesis of [18F]xenon difluoride in a micro-reactor
In experiments in glass or polymer vials, acetonitrile provided
several clear advantages over dichloromethane as reaction solvent.
First, it was the solvent of choice for obtaining dry [18F]FÀ–K+–K
2.2.2 from cyclotron-irradiated 18O-enriched water. Treatment of
aqueous [18F]fluoride ion in the presence of potassium carbonate
and K 2.2.2 to cycles of acetonitrile addition and evaporation
rapidly provided this reagent. Moreover this radiofluorination
reagent was more soluble in acetonitrile than in dichloromethane.
Thus, typically, only about 20% of the dried [18F]FÀ–K+–K 2.2.2
could be transferred efficiently out of a glass preparation vessel in
dichloromethane solution, while 80% or more could be transferred
in acetonitrile solution. Furthermore, [18F]XeF2 generated in
acetonitrile solution could be stored in a polypropylene vial for
several hours without decomposition. Finally, any unreacted
Fig. 1. Stacked plot of radio-HPLC chromatograms showing synthesis of [18F]xenon
difluoride in a micro-reactor using [18F]FÀ–K+–K 2.2.2 and xenon difluoride as
reagents in acetonitrile at different temperatures.
present in the micro-reactor (internal volume 31.4
mL) would have
been about 700 mol). Therefore, the micro-reactor
m
g
(ꢁ4
m
[
18F]fluoride ion could be readily removed from acetonitrile
enables low amounts of xenon difluoride to be used relative to
that used in a sealed reactor. Thus, simultaneous use of a high
activity of [18F]fluoride ion (e.g., 1 Ci in the storage loop) would
have potential to produce a specific radioactivity greater than
solutions of [18F]XeF2 by filtration of the reaction mixture through
a small plug of silica gel. Thus, we were interested to test whether
[
18F]XeF2 could be produced in acetonitrile at elevated tempera-
ture, and proposed to test this on a small reaction scale in a
commercially available microfluidic device.
The microfluidic platform used in this study allowed radio-
fluorination reactions to be performed under controlled conditions
of reaction time, temperature, reagent concentrations and reagent
stoichiometry, as has been previously described [29]. The micro-
30 mCi/mmol, which is comparable to or better than that normally
obtained for other electrophilic radiofluorination agents. For
example, the specific radioactivity of cyclotron-produced
[
18F]fluorine gas is typically in the 1–10 mCi/
mmol range [30].
2.3. Large-scale production of [18F]xenon difluoride
reactor itself is composed of narrow-bore (100
mm i.d.) fused-silica
glass tubing. We established that the silica glass micro-reactor was
resistant to dilute xenon difluoride solutions and could be exposed
repetitively without obvious damage or incident, even at raised
temperatures.
Reaction (or residence) times in the micro-reactor were deter-
mined by the total rate of influx of reagents. In initial experiments,
The micro-reactor experiments showed that a high RCY of
[
18F]XeF2 could be produced rapidly in acetonitrile by
heating the reaction mixture to around 125 8C. In view of this
finding, the reaction was carried out at production scale in
conventional Pyrex glass apparatus (Table 3). Reaction was
performed at 90 8C, because in a crimp-sealed vial it was found
difficult to contain xenon difluoride in acetonitrile above this
temperature. No more than 115 mCi of [18F]fluoride ion were
used in order to avoid excessive personal radiation burden in the
absence of fully automated equipment. However, the results are
expected to be fully scalable to the highest activities of
the total rate of influx was 20
and [18F]FÀ–K+–K 2.2.2 solution each infused at 10
reaction time of about 94 s. For reactions conducted with
mL/min, withxenon difluoridesolution
mL/min, giving a
[
18F]fluoride ion in the presence of K+–K 2.2.2 in acetonitrile, no
reaction occurred below 65 8C. At 85 8C, [18F]XeF2 was detected
(Fig. 1). At 125 8C, nearly half the radioactivity was incorporated into
[
18F]fluoride ion that can be produced from a single cyclotron
[
18F]XeF2. Whenthecounterionwaschanged toCs+ (K2.2.2present),
irradiation (multi-Ci). It was found that [18F]XeF2 could be
separated from inorganic material and K 2.2.2 in the reaction
mixture by filtration through a plug of silica gel or by distillation
at 110 8C through a double-tipped needle into a collection vial.
The overall RCY of [18F]XeF2 varied between 13 and 43%, but
was highest when a high amount of xenon difluoride was used
(Table 3).
exchange was clearly observable, even below 65 8C (chromatogram
not shown). Cs+ was not however the cation of choice because the
efficiency for transfer of 18FÀ–Cs+–K 2.2.2 from the drying vessel into
the storage loop of the micro-reactor system was low (19%).
The concentration of xenon difluoride within the micro-reactor
was 135 mM. At any instant the amount of xenon difluoride
Table 3
Multi-mCi production of [18F]xenon difluoride from [18F]fluoride ion.
XeF2
(
m
mol)a
[
18F]FÀ ion
Activity collected (mCi)
Decay-correction of
[
18F]XeF2 as % of
RCY of [18F]XeF2 (%)b
reagent (mCi)
collected activity (mCi)
collected activity
1
2
34
56
19.7
16.8
25.0
115
1.74
6.30
7.20
28.7
5.85
9.88
43.1
5.40
5.00
7.89
2.01
7.38
8.33
33.4
12.0
11.6
51.8
6.24
6.77
13.3
0
44
55
45
58
71
58
40
52
76
0
19
18
13
22
31
28
16
20
43
3
102
109
114
137
143
172
179
253
4
5
31.3
26.8
109
6
7
8
15.5
17.4
23.4
9
10
a
Reaction volume, 300
m
L.
b
Decay-corrected to the start of reaction from [18F]fluoride ion reagent.

1036
[(Schme_1)TD$FIG]
S. Lu, V.W. Pike / Journal of Fluorine Ch
)
emistry 131 (2010) 1032–1038
Scheme 1. Synthesis of [18F]xenon difluoride from [18F]fluoride ion and use of
18F]xenon difluoride as a radiofluorination agent.
[
2.4. Reactivity of [18F]xenon difluoride
The reactivity of [18F]xenon difluoride, produced in either the
micro-reactor or on a larger scale in a glass vial, was tested with the
known substrates 1-((trimethylsilyl)oxy))cyclohexene [19] and
fluorene [31] (Scheme 1).
Fig. 3. Stacked plot of radio-HPLC chromatograms showing reaction of [18F]xenon
difluoride with 1-((trimethylsilyl)oxy))cyclohexene in a micro-reactor at different
temperatures. Note [18F]fluorotrimethylsilane (not shown) eluted at 20.8 min.
The synthesis of [18F]XeF2 and its subsequent reaction with 1-
((trimethylsilyl)oxy))cyclohexene in acetonitrile were achieved at
micro-reactor scale by utilizing the two micro-reactors and three
available delivery pumps of the apparatus. The first micro-reactor
was used for [18F]XeF2 production and the second for its reaction
with a substrate (Fig. 2). When the solvent, xenon difluoride
concentration and infusion speeds were kept constant, the RCY of
reactor 2 to give [18F]2-fluorocyclohexanone. When the tempera-
ture of reactor
production also increased, albeit still to a low RCY (Fig. 3). The
identity of the [18F]2-fluorocyclohexanone was confirmed by GC
analysis.
When the radiofluorination of 1-((trimethylsilyl)oxy))cyclo-
hexene was carried out in acetonitrile in a glass vessel with
2 was increased, [
18F]2-fluorocyclohexanone
[
18F]XeF2 increased with reaction temperature, and consequently
the RCY of the product [18F]2-fluorocyclohexanone also increased.
When both micro-reactors were maintained at room temperature
no product was observed. When reactor 1 was heated to 100 8C,
[
18F]XeF2 that had been distilled out of a Pyrex glass reaction vial,
the RCY of [18F]2-fluorocyclohexanone was 38%. The remainder of
the radioactivity was comprised of [18F]fluorotrimethylsilane
(40%) and unreacted [18F]fluoride ion (22%). Hence, the reaction
[
18F]XeF2 was produced. The resulting [18F]XeF2 reacted with 1-
((trimethylsilyl)oxy))-cyclohexene at room temperature in micro-
[(ig._2)TD$FIG]
Fig. 2. A schematic illustration of micro-reactors, reagent storage and delivery, product collection valves, and flow directions for reaction of [18F]xenon diluoride (produced in
micro-reactor 1) with TMSO-cyclohexene in micro-reactor 2.

S. Lu, V.W. Pike / Journal of Fluorine Chemistry 131 (2010) 1032–1038
1037
had proceeded to 78% of the theoretical RCY. In this type of
procedure, acetonitrile distils over with the [18F]XeF2 reagent,
restricting the use of the reagents to reactions in acetonitrile or
acetonitrile mixtures.
5
[
m
m, 250 mm  4.6 mm; Phenomenex). This method separated
18F]XeF2 (tR = 3.9–4.1 min) from 18F]fluoride ion (tR = 2.6–
[
2.7 min), although not with complete resolution, since no-carrier-
added [18F]fluoride ion tends to exhibit broad tailing. By this HPLC
method, we verified that an acetonitrile solution of [18F]XeF2 was
stable for up to 3 h at room temperature in a polypropylene vial.
Hence, all analytes were injected onto HPLC columns in solvent
composed predominantly of acetonitrile. Column and elution
conditions are described later for each individual procedure.
Reactions of xenon difluoride are commonly performed in other
solvents, besides acetonitrile, such as dichloromethane [7]. We
therefore tested whether [18F]XeF2 produced in dichloromethane
could be used as a labeling agent in the same solvent. [18F]XeF2
produced in dichloromethane in a glass vessel was treated with
fluorene in the same pot. [18F]Fluorofluorenes were obtained in low
RCYs (5–10%). Only one radioactive peak was observed in the HPLC
analysis of product. Further analysis of this product by GC however
showed that two radiofluorinated products had been formed,
consistent with the previously observed production of 2-fluoro- and
4-fluoro-fluorene in 23% yield from the reaction of xenon difluoride
with fluorene in dichloromethane for 3 h at room temperature [31].
4.2. Synthesis of [18F]xenon difluoride in Pyrex glass or polypropylene
vials
Anhydrous [18F]FÀ–K+–K 2.2.2 solution (2–10 mCi) in acetoni-
trile (ꢁ300
ride ion, potassium carbonate (3.6
and added to a clear Pyrex glass vial (2.0 mL, Waters) containing
xenon difluoride (2–30 mg; 11.8–177 mol). The reaction mixture
was allowed to stand at room temperature for a set period. An
aliquot of the reaction mixture (10 L) was taken into a plastic vial
(PP LV, 500 L, Grace) and quenched with acetonitrile (300 L). A
sample (10–30 L) was analyzed on a reverse phase radio-HPLC
column (C18, 5
m, 250 mm  4.6 mm, Phenomenex) eluted at
m
L) was prepared from cyclotron-produced [18F]fluo-
m
mol) and K 2.2.2 (11 mmol)
3. Conclusion
m
In conclusion, [18F]XeF2 was produced by treating xenon
difluoride with cyclotron-produced [18F]fluoride ion in dichlor-
omethane at room temperature or in acetonitrile at elevated
temperature. [18F]XeF2 was obtained on a multi-mCi scale through
reaction of [18F]fluoride ion with xenon difluoride in acetonitrile at
90 8C for 10 min followed by separation from reaction mixture by
distillation at 110 8C. Studies are now in progress to explore further
the utility of this labeling agent and to improve the specific
radioactivity of this labeling agent over that currently attainable.
m
m
m
m
m
1 mL/min with aqueous HCOONH4 (25 mM)-MeCN at 55: 45 (v/v)
(HPLC method 1). The retention time of [18F]XeF2 was 3.9–4.1 min.
Reactions under modified conditions (e.g. Cs+, K+ or Cs+–K 2.2.2
as the cation, dichloromethane as solvent and PP LV vial as reaction
vessel) were performed similarly.
4. Experimental
4.3. Synthesis of [18F]xenon difluoride in a micro-reactor
4.1. General procedures
For this experiment two syringe pumps and one micro-reactor
of the micro-fluidic apparatus (Nanotek; Advion; cf. Fig. 2) were
used. Acetonitrile solutions of [18F]FÀ–K+–K 2.2.2 (10–50 mCi; K+–
K 2.2.2; 26 mM) and xenon difluoride (270 mM), were loaded into
Xenon difluoride (99.99%), kryptofix 2.2.2 (4,7,13,16,21,24-
hexaoxa-1,10-diazabicyclo[8.8.8]hexacosane, K 2.2.2; 98%), am-
monium formate (99.995%), potassium carbonate (99%), cesium
carbonate (99%), potassium fluoride (99.99%), dichloromethane
(anhydrous, ꢀ 99.8%), acetonitrile (anhydrous, 99.8%), 1-((tri-
methylsilyl)oxy)cyclohexene (99%), and fluorene (98%) were
purchased from Sigma–Aldrich (Milwaukee, WI) and used as
received. Acetonitrile (high purity solvent, Burdick & Jackson;
Morristown, NJ) for HPLC mobile phase was also used without
further treatment. NCA [18F]fluoride ion was obtained through the
18O(p,n)18F nuclear reaction by irradiating [18O]water (95 atom%)
separate storage loops (capacity 255
each loop were infused simultaneously at 10
coiled silica glass micro-reactor (100 m i.d., 4-m long; internal
volume 31.4 L). Reactions were quenched by diluting the reactor
effluent with acetonitrile (400 L) in a plastic vial. An aliquot
(20 L) of the reaction mixture was analyzed with HPLC method 1.
m
L). Solutions (20
mL) from
m
L/min into the
m
m
m
m
The procedure was repeated with the micro-reactor at different set
temperatures.
for 90–120 min with a proton beam (14.1 MeV; 20
m
A) produced
4.4. Synthesis of [18F]xenon difluoride in glass vials at a multi-mCi
level
by a PETrace cyclotron (GE, Milwaukee, WI). Cyclotron-produced
aqueous [18F]fluoride ion was dried azeotropically in the presence
of appropriate base (Cs2CO3, K2CO3 or KF) with or without K 2.2.2
(in 1.5 molar excess of metal ion if used), under microwave-
enhanced conditions [32].
Anhydrous [18F]FÀ–K+–K 2.2.2 solution (15–115 mCi) in aceto-
nitrile (ꢁ300
m
L) was added to a clear Pyrex glass V-vial (1.0 mL,
Alltech) containing xenon difluoride (5.7–42.8 mg; 34–253
mmol).
Radioactivity was measured with a calibrated dose calibrator
(Atomlab 300; Biodex Medical Systems Inc., Shirley, NY). Radio-
syntheses were performed in lead-shielded hot-cells. RCYs were
estimated by integrating the radio-peaks on HPLC chromatograms
or by measuring isolated product. GC was performed on a 6850
Network GC system (Agilent Technologies, Foster City, CA) using a
DB-WAX capillary column (30 m length, 0.25 mm ID).
The vial was closed with a crimp seal having a PTFE-coated septum,
and then heated at 90 8C for 10 min. The vial was connected via a
double-tipped needle to another similarly sealed empty V-vial
sitting on dry ice. The reaction vessel was heated at 110 8C so that
any volatile [18F]XeF2, together with acetonitrile, was transferred to
the collection vial. An aliquot of the collected product (10
quenched with acetonitrile (300 L) in a plastic vial (PP LV, 500
Grace) and a sample (10–20 L) analyzed with HPLC method 1.
m
L) was
m
mL,
The production of [18F]XeF2 was monitored by radio-HPLC on a
system comprising a solvent module (System Gold 126; Beckman
m
Coulter, CA) coupled to a UV absorbance detector (
l
= 250 nm [33];
4.5. Reaction of [18F]XeF2 with 1-((trimethylsilyl)oxy)cyclohexene in a
micro-reactor
Model 168; Beckman Coulter) and a radioactivity detector (PMT,
Flow-count; Bioscan, Washington, DC). Xenon difluoride is quite
stable in aqueous media [34] and is also stable for hours in
Acetonitrile solutions of [18F]FÀ–K+–K 2.2.2 (40 mCi; K+–K 2.2.2;
26 mM), xenon difluoride (370 mM) and 1-((trimethylsilyl)oxy)cy-
clohexene (223 mM), were separately loaded into designated
acetonitrile.
A mixture of acetonitrile and 25 mM aqueous
ammonium formate solution was therefore used as mobile phase
for the HPLC of xenon difluoride on a reversed phase column (C18,
storage loop (255
mL capacity) of the microfluidic apparatus

1038
S. Lu, V.W. Pike / Journal of Fluorine Chemistry 131 (2010) 1032–1038
(Fig. 2). [18F]Fluoride ion solution (20
m
L) and xenon difluoride
L/min, into coiled
m i.d., 4-m long; internal volume
L). The reaction mixture from this reactor was then infused
L/min) simultaneously with the 1-((trimethylsilyl)oxy)cyclo-
hexene solution at a rate of 5 L/min into the second micro-reactor
(100 m i.d., 4-m long; internal volume 31.4 L). The effluent from
this reactor was quenched by dilution with acetonitrile (400 L) in a
plastic vial. An aliquot (20 L) of the reaction mixture was analyzed
with HPLC method 1 ([18F]2-fluorocyclohexanone, tR = 5.3 min;
18F]Me3SiF tR = 20.8 min). The radioactive fraction was also
[3] (a) L. Cai, S.Y. Lu, V.W. Pike, Eur. J. Org. Chem. 17 (2008) 2853–2873;
(b) P.W. Miller, N.J. Long, R. Vilar, A.D. Gee, Angew. Chem. Int. Ed. 47 (2008) 8998–
9033;
(c) S.M. Ametamey, M. Honer, P.A. Schubiger, Chem. Rev. 108 (2008) 1501–1506.
[4] (a) T.J. Ruth, A.P. Wolf, Radiochim. Acta 26 (1979) 21–24;
(b) M. Guillaume, A. Luxen, B. Nebeling, M. Argentini, J.C. Clark, V.W. Pike, Appl.
Radiat. Isot. 42 (1991) 749–762.
[5] V. Casella, T. Ido, A.P. Wolf, J.S. Fowler, R.R. Macgregor, T.J. Ruth, J. Nucl. Med. 21
(1980) 750–757.
[6] (a) D.M. Jewett, J.F. Potocki, R.E. Ehrenkaufer, J. Fluorine Chem. 24 (1984)
477–484;
solution were infused simultaneously, each at 5
silica glass micro-reactor 1 (100
31.4
(10
m
m
m
m
m
m
m
m
m
(b) R. Chirakal, G. Firnau, E.S. Garnett, Appl. Radiat. Isot. 39 (1988) 1099–
1101.
[7] (a) R. Filler, Isr. J. Chem. 17 (1977) 71–79;
[
(b) M. Zupan, Xenon halide halogenations, in: S. Patai, Z. Rappaport (Eds.), The
Chemistry of Functional Groups, Suppl. D, John Wiley and Sons Ltd., Chichester,
1983, pp. 657–679;
analyzed by GC (2-fluorocyclohexanone tR = 6.18 min). The
procedure was repeated with the micro-reactors set at different
temperatures.
(c) M.A. Tius, Tetrahedron 51 (1995) 6605–6634;
ˇ
ˇ
(d) M. Tramsek, B. Zemva, Acta Chim. Slov. 53 (2006) 105–116.
[8] E.W. Della, N.J. Head, J. Org. Chem. 57 (1992) 2850–2855.
[9] A.P. Lothian, C.A. Ramsden, Synlett 10 (1993) 753–755.
[10] (a) T.B. Patrick, K.K. Johri, D.H. White, W.S. Betrand, R. Mokhtar, M.R. Kilbourn,
M.J. Welch, Can. J. Chem. 64 (1986) 138–141;
4.6. Reaction of distilled [18F]xenon difluoride with 1-
((trimethylsilyl)oxy)cyclohexene
(b) T.B. Patrick, S. Khazaeli, S. Nadji, K. Hering-Smith, D. Reif, J. Org. Chem. 58
(1993) 705–708.
Anhydrous [18F]FÀ–K+–K 2.2.2 solution (5–20 mCi) in acetoni-
[11] C.A. Ramsden, M.M. Shaw, Tetrahedron Lett. 50 (2009) 3321–3324.
[12] (a) S. Sood, G. Firnau, E.S. Garnett, Int. J. Appl. Radiat. Isot. 34 (1983) 743–
745;
(b) C.-Y. Shiue, K.-C. To, A.P. Wolf, J. Labelled Compd. Radiopharm. 20 (1983)
157–162.
[13] R. Chirakal, G. Firnau, G.J. Schrobilgen, J. Mckay, E.S. Garnett, Int. J. Appl. Radiat.
Isot. 35 (1984) 401–404.
[14] G. Schrobilgen, G. Firnau, R. Chirakal, E.S. Garnett, J. Chem. Soc. Chem. Commun.
(1981) 198–199.
[15] H. Constantinou, F.I. Aigbirhio, R.G. Smith, C.A. Ramsden, V.W. Pike, J. Am. Chem.
Soc. 123 (2001) 1780–1781.
[16] N. Vasdev, B.E. Pointner, R. Chirakal, G.J. Schrobilgen, J. Am. Chem. Soc. 124 (2002)
12863–12868.
trile (ꢁ300
mL) was added to a clear Pyrex glass V-vial (1.0 mL,
Alltech) containing xenon difluoride (2–30 mg; 11.8–177
mmol).
Thereactionmixturewaskeptat90 8Cfor10 min. [18F]XeF2 together
with acetonitrile was distilled over 10 min into another V-vial
containing 1-((trimethylsilyl)oxy)cyclohexene (2–10
tion was allowed to stand at room temperature for 20 min or heated
to 80 8C for5–10 min. Analiquotof the reaction mixture(10 L) was
quenched with acetonitrile (400 L) in a plastic vial (PP LV, 500 L,
Grace). A sample (20 L) was then analyzed with HPLC method 1.
mL). The reac-
m
m
m
m
4.7. Reaction of [18F]xenon difluoride with fluorene in a glass vial in
situ
[17] (a) S.Y. Lu, V.W. Pike, in: P.A. Schubiger, L. Lehmann, M. Friebe (Eds.), PET
Chemistry—The Driving Force in Molecular Imaging, Springer-Verlag, Heidelberg,
2007, pp. 271–287;
(b) H. Audrain, Angew. Chem. Int. Ed. 46 (2007) 1772–1775;
(c) A.M. Elizarov, Lab Chip 9 (2009) 1326–1333;
(d) P.W. Miller, J. Chem. Technol. Biotechnol. 84 (2009) 309–315.
[18] (a) P. Nongkunsarn, C.A. Ramsden, J. Chem. Soc. Perkin Trans. 1 (2) (1996)
121–122;
(b) C.A. Ramsden, R.G. Smith, Org. Lett. 1 (1999) 1591–1594;
(c) M.M. Shaw, R.G. Smith, C.A. Ramsden, J. Fluorine Chem. 116 (2002) 71–73.
[19] C.A. Ramsden, R.G. Smith, J. Am. Chem. Soc. 120 (1998) 6842–6843.
[20] (a) F.C. Cowlard, J.C. Lewis, J. Mater. Sci. 2 (1967) 507–512;
(b) P.J.F. Harris, Philos. Mag. 84 (2004) 3159–3167.
Anhydrous [18F]FÀ–Cs+–K 2.2.2 solution (5–10 mCi) in dichlor-
omethane (200
Alltech) containing xenon difluoride (10–20 mg, 59–118
The reaction mixture was kept at room temperature for 30 min.
Fluorene (1.6 mg, 10 mol) in dichloromethane (50 L) and
BF3.Et2O (1 M, 3 L) were added to the vial. The reaction mixture
m
L) was added to a clear Pyrex glass vial (1.0 mL,
m
mol).
m
m
m
immediately turned from blue to brown and was left standing at
room temperature for 30–60 min. An aliquot of the reaction
[21] M. Iuliano, L. Ciavatta, G. De Tomasso, J. Colloid Interface Sci. 310 (2007) 402–
410.
mixture (10
plastic vial (PP LV, 500
analyzed on a reverse phase HPLC column (Phenomenex, C18, 5
250 mm  4.6 mm) eluted at 1 mL/min with aqueous HCOONH4
(25 mM)-MeCN at 35:65 (v/v). The retention time of [18F]fluoro-
fluorene(s) was 15.1 min. The radioactive product fraction was
collected and analyzed by GC [fluorofluorene(s) tR = 10.4 min].
m
L) was quenched with acetonitrile (300
mL) in a
[22] K.O. Christe, W.W. Wilson, J. Fluorine Chem. 47 (1990) 117–120.
[23] J.R. Ballinger, B.M. Bowen, G. Firnau, E.S. Garnett, F.W. Teare, Int. J. Appl. Radiat.
Isot. 35 (1984) 1125–1128.
[24] J.M. Lehn, J.P. Sauvage, J. Am. Chem. Soc. 97 (1975) 6700–6707.
[25] (a) J.W. Brodack, M.R. Kilbourn, M.J. Welch, J.A. Katzenellenbogen, Appl. Radiat.
Isot. 37 (1986) 217–221;
(b) R.J. Nickles, S.J. Gatley, J.R. Votaw, M.L. Kornguth, Appl. Radiat. Isot. 37 (1986)
649–650.
[26] E.H. Appelman, Inorg. Chem. 6 (1967) 1268–1269.
[27] (a) W.W. Dukat, J.H. Holloway, E.G. Hope, P.J. Townson, R.L. Powell, J. Fluorine
Chem. 62 (1993) 293–296;
m
L, Grace) and a sample (20 L) was then
m
m
,
Acknowledgements
(b) H. Meinert, S.Z. Ru¨diger, Z. Chem. 7 (1967) 239–1239;
(c) L.P. Ingman, J. Jokisaari, K. Oikarinen, R. Seydoux, J. Magn. Reson. 111 (1994)
155–160.
This research was supported by the Intramural Research
Program (project # Z01-MH-002793) of the National Institutes
of Health (National Institute of Mental Health). We are grateful to
the NIH Clinical PET Department for the production of fluorine-18.
We also thank Professor C. A. Ramsden (Keele University, UK) and
Dr. F. I. Aigbirhio (Cambridge University, UK) for their comments
on the draft manuscript.
[28] M.L. Korguth, T.R. DeGrado, J.E. Holden, S.J. Gatley, J. Labelled Compd. Radio-
pharm. 25 (1988) 369–381.
[29] (a) S. Lu, A.M. Giamis, V.W. Pike, Curr. Radiopharm. 2 (2009) 49–55;
(b) J.H. Chun, S. Lu, Y.S. Lee, V.W. Pike, J. Org. Chem. 75 (2010) 3332–3338.
[30] S.M. Qaim, J.C. Clark, C. Crouzel, M. Guillaume, H.J. Helmeke, B. Nebeling, V.W.
Pike, G. Sto¨cklin, in: G. Sto¨cklin, V.W. Pike (Eds.), Radiopharmaceuticals for
Positron Emission Tomography, Kluwer Academic Publishers, Dordrect/London,
Dev. Nucl. Med., vol. 24, 1993, pp. 1–33.
[31] (a) M. Zupan, J. Iskra, S. Stavber, J. Org. Chem. 63 (1998) 878–880;
(b) J. Iskra, S. Stavber, M. Zupan, Collect. Czech. Chem. Commun. 73 (2008)
1671–1680.
References
´
[32] N. Lazarova, F.G. Simeon, J.L. Musachio, S.Y. Lu, V.W. Pike, J. Labelled Compd.
[1] M. Rudin, R. Weissleder, Nat. Rev. Drug Discov. 2 (2003) 123–131.
[2] (a) G. Sedvall, Triangle 30 (1991) 11–20;
Radiopharm. 50 (2007) 463–465.
[33] (a) E.H. Appelman, J.G. Malm, J. Am. Chem. Soc. 86 (1964) 2297–2298;
(b) H. Kunkely, A. Vogler, Inorg. Chim. Acta 357 (2004) 2407–2409.
[34] J.L. Weeks, M.S. Matheson, Inorg. Syn. 8 (1966) 260–264.
(b) T. Jones, Eur. J. Nucl. Med. 23 (1996) 807–813;
(c) J.M. Miller, D. Kumar, J.J. Mann, R.V. Parsey, Curr. Radiopharm. 1 (2008) 12–16.