Applied Radiation and Isotopes 57 (2002) 303–308

High-yield, low-pressure [18O] water targets of titanium and

niobium for F-18 production on MC-17 cyclotrons
Marc S. Berridge*, Ken W. Voelker, Bonnie Bennington
Departments of Radiology and Chemistry, Case Western Reserve University and University Hospitals of Cleveland,
11100 Euclid Ave, Cleveland, OH 44106, USA
Received 29 May 2001; received in revised form 4 September 2001; accepted 19 September 2001

Abstract

Large volume low-pressure targets were designed for use on the wide-beam Scanditronix MC-17 cyclotron. The
design was optimized experimentally by construction of aluminum test targets. Final production targets were
constructed of titanium and niobium. The targets tripled the routine production yield of the previous target, giving up
to 90 GBq (2.5 Ci) of fluoride and essentially eliminated the need for periodic target maintenance. The final targets
represent a considerable improvement. However, they have not performed as well as predicted, indicating that further
improvement is possible. r 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Positron tomography; Target; Fluorine-18; Titanium; Niobium; Low pressure

1. Introduction water chamber wide enough to accept the full width of

The F-18 target supplied with Scanditronix MC-17
cyclotrons (Berridge and Kjellstrom, 1999) is a silver
dual-foil target with a thin water chamber and ‘keyhole’
design. It was designed for moderate production needs
and conservation of O-18 water. Since the MC-17 beam
is wide (3 cm), the 2 cm wide target chamber conserved
water by wasting nearly half of the proton beam. The
target can produce up to 40 GBq (1 Ci), though 30 GBq
(750 mCi) is a more common yield. High-beam currents
or long irradiations are not well tolerated by the shallow
(2.5 mm) water chamber. The original target functioned
well for over 10 years (Berridge and Kjellstrom, 1999),
but production demand has increased and O-18 water
recovery methods have improved (Schlyer et al., 1990;
Jewett et al., 1990).
We have now constructed targets with increased
production capacity. They are single-foil targets with a
*Corresponding author. Department of Radiology, Univer-
sity Hospitals of Cleveland, 11100 Euclid Ave, Cleveland, OH
44106, USA. Tel.: +1-216-844-7135; fax: +1-216-844-7351.
E-mail address: msb5@po.cwru.edu ( S. Berridge).
the MC-17 beam, and have provisions for efficient
cooling. The water chamber is also thicker than the
previous design and incorporates a large head space to
accept high-beam current without loss of yield to
cavitation (Tewson et al., 1988). Vigorous boiling and
cavitation was expected in the thicker target without
high pressurization (which over-stresses the large-area
beam entrance foil), so the head space was important.
Another goal was to improve maintenance intervals.
The silver target required cleaning every few weeks to
remove a white adhering layer in the target chamber.
Though it was not rigorously identified, a similar deposit
was formed by exposure of silver to dilute nitric acid for
several weeks. Given that nitrate is produced in carrier
quantities in ammonia targets that are chemically
identical to these targets (Berridge and Landmeier,
1993; Gersberg et al., 1976; Parks and Krohn, 1978;
Krizek et al., 1973), the deposit could be silver nitrate or
nitrite, or silver chloride formed from their anion
exchange with ambient chloride, which could cause the
observed effects. The new targets were made of titanium
and niobium and are less reactive to acid than silver so
would not be expected to give a similar result. The

0969-8043/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 9 6 9 - 8 0 4 3 ( 0 1 ) 0 0 2 4 5 - 7
304 M.S. Berridge et al. / Applied Radiation and Isotopes 57 (2002) 303–308
hardness and inertness of these metals was expected to
increase maintenance intervals and service lifetime. In
several published reports (Mazza and Yoshizumi, 1994;
Erdman, 2000; van der Jagt et al., 2000; Helmeke et al.,
2001), and in other cases that remain unpublished,
titanium has been found to be acceptable for fluorine-18
production targets. Niobium has recently been reported
(Zeisler et al., 2000) as having desirable properties and
being compatible with fluorine-18 production and use.
A series of experimental targets were constructed from
aluminum with various dimensions and tested using the
proton bombardment of natural abundance water to
produce N-13 (Berridge and Landmeier, 1993). These
established the minimum necessary dimensions of the
target water chamber to achieve the production goals.
Targets for F-18 production were then constructed with
those dimensions.
2. Methods
The general target construction, mounting of the
targets, materials used, and methods for target irradia-
tion and product analysis have been previously de-
scribed (Berridge and Kjellstrom, 1999; Berridge and
Landmeier, 1993; Krasikova et al., 1999). Briefly, the
targets are designed for a Scanditronix MC-17 cyclotron
which has no external beam optics and places a
3.5
0.6 cm2 (w
h) beam at the target entrance. Yields
from natural abundance water were measured as N-13.
Yields from oxygen-18 enriched water were measured as
fluoride ion after collection and subsequent elution from
Bio-Rad AG1
8 ion exchange resin (Schlyer et al.,
1990; Jewett et al., 1990). Reactivity of fluoride was
tested by the synthesis (Hamacher et al., 1986) of [18F]2-
deoxy-2-fluoroglucose (FDG). Target yields, as in earlier
reports, are expressed as saturation yield per micro-
ampere (S) calculated from the measured target yield
and irradiation parameters. Reaction yields are ex-
pressed as chemical yields (radiochemical yields to which
decay-correction was applied) relative to starting fluor-
ide ion. The titanium target was treated with 1 N HCl by
filling with the acid (Fisher Scientific, diluted to 1 N with
deionized water) at room temperature for 30 min then
rinsing with deionized water until pH paper indicated a
neutral pH. The procedure was repeated 3 times.
3. Target designs
All of the targets were constructed according to
Figs. 1 and 2. Features that remain unchanged from
previous MC-17 targets (Berridge and Kjellstrom, 1999)
are not shown here. The new helium cooling chamber
(Fig. 1) provided a jet of helium directed from below and
aimed low on the foil through a thin slit across the full
width of the beam. An unrestricted collection area at the
top of the chamber minimized the resistance to gas flow.
The outline of the water chamber (Fig. 2, solid line)
was determined by beam size and location and by
mounting and construction constraints. Water was
introduced through flat-bottom 1/4-28 fittings compa-
tible with several commercial laboratory tubing and
fitting systems. Helicoflex O-rings (Helicoflex Co.,
Columbia, SC) were used. The chamber shape achieved
maximum head space volume within the O-ring (groove
visible on Fig 1). The cooling water (Fig. 2B, right side)
was separated from the target chamber by 2 mm in the
aluminum test targets, and by 0.8 mm in the titanium
(Grade 5) and niobium (Grade 395) production targets.
Cooling water was introduced through holes in a simple
flat back plate of aluminum with a flat gasket (not
shown), entering at lower right and leaving at upper left
of the cooling water chamber as shown in Fig. 2A. Fins
were used in the cooling water chamber to direct the
water flow and provide mechanical support. (Fig. 2) A
series of test targets were constructed of aluminum with
a variety of water chamber dimensions. Width (Fig. 2,
w) varied from 21.6 to 40 mm. Depth, or target thickness
(Fig. 2, d) varied from 5.4 to 10 mm. The targets were
operated as described previously (Berridge and Land-
meier, 1993) but with a 5 bar methane overpressure
(Krasikova et al., 1999) so that ammonia was always the
major product generated from natural abundance water,
regardless of beam dose. The same methane gas and
pressure was used during F-18 production. This was
done to reduce the potential of complications from the
production of nitric acid carrier by identical processes
that take place in the natural-abundance water targets,
though we have no evidence that such complications can
occur when using titanium or niobium targets. It was
previously established and discussed (Krasikova et al.,
Fig. 1. Helium cooling section showing helium jets, mounting,
and O-ring groove. O-ring groove faces the water chamber
(Fig. 2).
 M.S. Berridge et al. / Applied Radiation and Isotopes 57 (2002) 303–308 305

Fig. 2. Target water and water cooling chambers: (A) Front view showing the target water chamber (solid lines, front side), and in
dotted lines the cooling water chamber and fins and O-18 water tube fittings (reverse side), and internal O-18 water chamber. (B) Side
view, O-18 water chamber on left, cooling water on right. (C) Top view. O-18 water chamber dimensions discussed in the text (h, w, d)
are shown on views B and C.

1999) and we reconfirm in this work that there is no
observed tar formation or other adverse consequences of
irradiation of a methane-pressurized water target.
Results from the aluminum test targets determined the
dimensions used for the fluoride production targets,
constructed of titanium and niobium. The specified
chamber height, width, and depth of the production
targets were 31, 32 and 6 mm, respectively. The actual
depth of the niobium target was 5.8 mm. Test irradia-
tions were performed on natural abundance water and
on 75% and 95% O-18 enriched water. Chemical
reactivity of the fluoride was tested using the synthesis
of [18F]fluorodeoxyglucose, FDG, as in our previous
work (Berridge and Kjellstrom, 1999).
4. Results and discussion
The effect of target widths was clear. Saturation yields
for ammonia production were 30 (mCi/mA) at widths of
40 and 32 mm, but dropped to 25, 23, 20 and 18 at
widths of 27, 24, 23 and 21 mm, respectively. The low
yields in the narrower targets resulted from exclusion of
the outer ends of the 3 cm wide beam. The beam dose
rate and current were kept under 10 mA and 10 min,
respectively. The results demonstrate the horizontal
beam density profile of the MC-17 and permit a width
to be chosen in compromise between O-18 water
conservation and yield maximization. The choice
depends on the relative priority of those two items, a
variable in recent years. Our choice was a width of
32 mm.
The effect of depth (d; Fig. 1) was then investigated
only at the chosen width. Results of individual experi-
ments are shown in Fig. 3. A trend of decreasing
saturation yield can be seen with increasing beam dose.
The general decrease is probably due to factors other
than the depth of the target, and seems to be a general
feature of cyclotron targets. At low-beam dose, there
was little if any difference as target depth changed, as
would be expected from previous results with targets
that had depths as low as 3 mm. At higher-beam doses, a
5.4 mm depth clearly gave lower yields than 6.0 or
7.5 mm. The general dependence on dose rate that was
observed in our previous target investigations (Berridge
and Kjellstrom, 1999; Berridge and Landmeier, 1993)
was also seen here. Degradation of performance is seen
at lower total dose when a high-dose rate is used, as is
typical of most targets. However, since the major
interest in this work was in the region of high beam
dose rate and high total beam dose, we will not repeat
that discussion. There was no large difference between
the other two depths, though the number of data points
was insufficient to rule out a small difference. Because
the potential for a small difference did not justify an
increase in depth, the final target depth was chosen to be
6.0 mm.
The final 6.0 mm deep and 3.2 cm wide aluminum
prototype target produced an average S value for
N-13 of 28 mCi/mA at beam doses under 400 mA min.
Performance was maintained to 1000 mA min, but by
5000 mA min the average S value had dropped to 25. The
original prototype, 10 mm deep and 4 cm wide, produced
virtually identical values. At low beam dose S averaged
29 and dropped to 25 by 4000 mA min. It is clear that
306 M.S. Berridge et al. / Applied Radiation and Isotopes 57 (2002) 303–308
Fig. 3. Prototype yields as a function of target depth. Yields of N-13 vs. beam dose are shown from individual experiments.
there is no practical difference in performance between
the two targets, though their internal volumes differ by
more than a factor of two. Since the relative perfor-
mance of target chambers smaller than the final
prototype is degraded, we concluded that the prototype
was near the optimum size for an MC17 target, at least
for one made of aluminum. Water loaded into the
targets during the experiments varied throughout
because the size of the target chambers varied. The
approach used is as we have previously reported
(Berridge and Kjellstrom, 1999; Berridge and Land-
meier, 1993). The target was not filled to the top of the
chamber, rather a head space was allowed to remain.
Preliminary experiments at both high- and low-beam
dose were done to determine the minimum effective load
volume. In the final prototype and the production
targets, the total load capacity is 3.9 ml, while the
amount of water loaded is between 2.5 and 3.0 ml. A
load volume of 2.5 is sufficient at low total beam dose,
but not sufficient for high production demands. The
minimum necessary load volume at high dose rate is
between 2.8 and 3.0 ml, depending on how one interprets
experimental variations. Load volumes >3.5 ml reduce
performance at high beam dose when water is ejected
from the target due to insufficient head space volume.
The production target chambers were constructed
according to the optimum prototype design. The
thickness of metal between the target water and the
cooling water was reduced to 0.8 mm. Titanium is up to
four times stronger, with thermal conductivity 10 times
poorer, than aluminum. Niobium is stronger than
titanium and has twice its thermal conductivity. Simple
two-dimensional heat transfer calculations were per-
formed using the thermal conductivity and thickness of
the metal and the temperature of the cooling water to
calculate an equilibrium temperature in the target water
chamber:
Dt ¼ HT=ðAkÞ
where Dt is the temperature rise in the target chamber
over cooling water temperature, H is the heat load
(850 W maximum; 50 mA at 17 MeV), T is the thickness
of the metal, A is the metal area in contact with the
target water, and k is the thermal conductivity. Other
routes of cooling including helium cooling were
neglected. Efficient mixing was assumed. The calcula-
tions indicated that the O-18 water would remain under
851C even with a 1.5 mm thickness of titanium. Clearly,
to assume ideal convection is invalid and actual water
temperature will be both higher and non-uniform.
However, the calculation does indicate that the metal
thickness should not limit the target performance
(Steinbach et al., 1990). The production tests with
natural abundance water were consistent with prototype
results beyond 40 mA for 1 h. Results of the first 6
months of fluorine-18 production with each target after
break-in are shown in Fig. 4. These represent beam
currents to 40 mA and F-18 yields up to 90 GBq (2.5 Ci).
The performance was similar to that of the prototype.
The previous target’s yield fell rapidly at beam doses
above 1000 mA min (Berridge and Kjellstrom, 1999). The
new targets produce similar saturation yield, but
continue to higher-beam currents. The new design was,
therefore, successful in raising the beam handling
capacity of the target to better match the beam
production capacity of the cyclotron. Yield of 2-
fluoro-2-deoxyglucose obtained from fluoride produced
 M.S. Berridge et al. / Applied Radiation and Isotopes 57 (2002) 303–308
Fig. 4. Fluorine-18 production yields of the two production targets, of titanium (open circles) and niobium (closed triangles) vs. beam
dose.
with these targets was 50–60% (uncorr.) at EOS
(45 min), indicating reasonable reactivity of the fluoride.
There were, however, problems with both production
targets. The yield of the titanium target (Fig. 4), even at
low beam dose, is below normal. The expected value, in
practical use, is about 180–200 mCi/mA for a 17 MeV
proton irradiation, depending on O-18 water enrich-
ment. An error in procurement of the titanium was
discovered in hindsight. The titanium alloy contained
6% aluminum, which is known to retain fluoride in a
target, to complex with it, and to reduce its reactivity
(Tewson et al., 1988). Initially very low fluoride yields
and low fluorination yields improved with 1 N HCl
treatment and with continued use to give the results
shown. We fully expect that the same target made of
pure titanium would produce appropriate reactive
fluoride yields (Mazza and Yoshizumi, 1994; Erdman,
2000; van der Jagt et al., 2000; Helmeke et al., 2001).
The niobium target also required a break-in period of
several irradiations presumably to remove surface
impurities. Due to machining problems the final
niobium target water chamber depth was 5.8 mm rather
than the intended 6.0 mm. We suspect that the error
contributes to the inability so far of this target to accept
50 mA of beam current without reduction in fluoride
yield. If these problems are avoided in future target
constructions we would expect results equal to or better
than those reported here. Though both targets produce
well, we have preferred the niobium target. Radiation
from the titanium target on dismounting after several
months of use was 250 mSv/h (25 R/h) at 5 cm and
decayed with a 16 day half-life, consistent with the
expected V-48. The niobium target after similar use
307
produces o6 mSv/h (0.6 R/h) on the first day of
removal, and the primary activation product is 6.9 h
Mo-93 m (Zeisler et al., 2000).
5. Conclusion
The previous Scanditronix production target (Ber-
ridge and Kjellstrom, 1999), produced a maximum of
48 GBq (1300 mCi) using 120 min of 15 mA (or 60 min of
20 mA) beam on the narrow water chamber (=25–35 mA
on the Faraday cup). But yields dropped at higher-beam
doses, probably due to thinning of the target by
cavitation and boiling. Therefore, the maximum yield
was obtained with 30 mA on the target assembly, but
only 15–20 mA actually in the target water.
The new targets have provided a considerable
improvement. Over twice the previous maximum fluor-
ide yield is routinely produced. The most common
irradiation has been 30 mA, 75 min to produce 78 GBq
(2.1 Ci). Up to 90 GBq (2.5 Ci) has been produced to
date. Further, both production targets have so far been
used for periods of 6 months to 1 year without
maintenance. Production results and chemical reactivity
of the product from the two targets seem satisfactory
and equivalent.
References
Berridge, M.S., Kjellstrom, R., 1999. Designs and use of silver
[O-18]water targets for [F-18]fluoride production. Appl.
Radiat. Isot. 50, 699–705.
308 M.S. Berridge et al. / Applied Radiation and Isotopes 57 (2002) 303–308
Berridge, M.S., Landmeier, B.J., 1993. Target production of
[13N]ammonia: target design, products and operating
parameters. Appl. Radiat. Isot. 44 (1993), 1433–1441.
Erdman, K. (2000) Features of the design of the titanium target.
Proceedings of the Eighth Workshop on Targetry and
Target Chemistry, St. Louis, MO, pp. 53–56.
Gersberg, R., Krohn, K.A., Peek, N., Goldman, C.R., 1976.
Denitrification studies with N-labeled nitrate. Science 192,
1229–1231.
Hamacher, K., Coenen, H.H., Stocklin, G., 1986. Efficient
stereospecific synthesis of no-carrier-added 2-[18F]fluoro-2-
deoxy-D-glucose using aminopolyether supported nucleo-
philic substitution. J. Nucl. Med. 27, 235–238.
Helmeke, H.J., Harms, T., Knapp, W.H., 2001. A water target
with beam sweep for routine fluorine-18 production. Appl.
Radiat. Isot. 54, 753–759.
Jewett, D.M., Toorongian, S.A., Bachelor, M.A., Kilbourn,
M.R., 1990. Extraction of [F-18]fluoride from [O-18]water
by a fast fibrous anion exchange resin. Appl. Radiat. Isot.
41, 583–586.
Krasikova, R.N., Fedorova, O.S., Korsakov, M.V., Benning-
ton, B.L., Berridge, M.S., 1999. Improved [N-13]ammonia
yield from the proton irradiation of water using methane
gas. Appl. Radiat. Isot. 51, 395–401.
Krizek, H., Lembares, N., Dinwoodie, R., Gloria, I., Lathrop,
K.A., Harper, P.V., 1973. Production of radio-
chemically pure [N-13]NH3 for biomedical studies using
the [O-16](p,Alpha)[N-13] reaction. J. Nucl. Med. 14,
629–630.
Mazza, S.M., Yoshizumi, T., 1994. 48V radionuclidic impurity
in product streams from liquid cyclotron targets with
titanium windows. Nucl. Med. Biol. 21, 677–21679.
Parks, N.J., Krohn, K.A., 1978. The synthesis of N-13 labeled
ammonia, dinitrogen, nitrite, and nitrate using a single
cyclotron target system. Int. J. Appl. Radiat. Isot. 29,
754–756.
Schlyer, D.J., Bastos, M.A., Alexoff, D., Wolf, A.P., 1990.
Separation of [F-18]fluoride from [O-18]water using anion
exchange resin. Appl. Radiat. Isot. 41, 531–533.
Steinbach, J., Guenther, K., Loesel, E., Grunwald, G., Mikecz,
P., Ando, L., Szelecsenyi, F., Beyer, G.J., 1990. Tempera-
ture course in small volume [18O]water targets for [18F]F-
production. Int J Rad Appl Instrum [A] 41, 753–756.
Tewson, T.J., Berridge, M.S., Bolomey, L.A., Gould, K.L.,
1988. Routine production of reactive fluorine-18 salts from
an oxygen-18 water target. Nucl. Med. Biol. 15, 499–504.
van der Jagt, P., van Leuffen, P., van Langevelde, V., Kruijer,
P., Ghyoot, M., Schmitz, F., Verbruggen, R., Vamecq, F.,
Destexhe, C., Jongen, Y., (2000). Multi Curie fluorine- in
IBA titanium target. Proceedings of the Eighth Workshop
on Targetry and Target Chemistry, St. Louis, MO. 44–45.
Zeisler, S.K., Becker, D.W., Pavan, R.A., Moschel, R., Ruhle,
H., 2000. A water-cooled spherical niobium target for the
production of [18F]fluoride. Appl Radiat Isot 53, 449–453.