Introduction It is shown that a very simple model reproduces the pressure versus beam current characteristics of elongated single-cavity boiling water targets for 18F production surprisingly well. By fitting the model calculations to measured data, values for a single free parameter, namely an overall heat-transfer coefficient, have been extracted for several IBA Nirta H218O targets. IBA recently released details on their new Nirta targets that have a conical shape, which constitutes an improvement over the original Nirta targets that have a cylindrical shape [1,2]. These shapes are shown schematically in FIGURE 1. A study by Alvord et al. [3] pointed out that elevated pressures and temperatures in excess of the saturation conditions may exist in a water target during bombardment. However, as long as the rate of condensation matches the rate of vaporization, the bulk of the system should remain at saturation conditions. Superheated regions are therefore likely to form but also likely to disappear rapidly, typically on the scale of a few milliseconds. Even though the boiling process is generally quite complex, enhanced by radiation-induced nucleation, the presence of fast mixing mechanisms in the water volume justifies some simplifications to be made. Materials and Methods The simplified model assumes that the bulk of the target water has a constant temperature, which is the same as the inner wall temperature of the cavity, Tw. A second simplification is to neglect the temperature difference across the target chamber wall, which is only justified if the wall is thin. The boiling is not explicitly taken into consideration, including the rather complex boiling behaviour at the Havar window, except to acknowledge that it is the main mixing mechanism. Large temperature gradients can briefly exist in the water medium but they also rapidly disappear. A further assumption is that a single, overall convective heat-transfer coefficient can be applied, which is constant over the entire water-cooled surface. As the wall thickness is neglected, the heat-transfer surface is assumed to be the inner surface of the cavity, excluding the surface of the Havar window. One can then write down an energy balance between the beam heating and the convection cooling (Newton’s law of cooling), where Ib is the beam intensity, ΔE is the energy windows of the target (taken as 18 MeV), h is the convective heat-transfer coefficient, A is the inner cavity surface through which the heat has to be transferred from the target-water volume to the cooling water, and T0 is the cooling-water temperature. The saturated vapour pressure of water versus temperature is a characteristic curve, given by the steam tables [4]. Assuming the bulk of the system at saturation conditions, one gets from (1) and (2). The function f is represented by a polynomial. The only unknown in Equation (3) is the overall convective heat-transfer coefficient h. Our approach was to adjust h until a good fit with a set of measured data was obtained. It also has to be mentioned that subtle differences in the physical properties between 18O-water and natural water have been neglected. All in all, quite a few assumptions and simplifications are made in deriving Equation (3) and the system is, admittedly, much more complex. Nevertheless, the results obtained by applying Equation (3) are rather interesting. Results and Conclusion Measured data and corresponding calculations are shown in FIGURE 2 for three different conical targets and one cylindrical target. The extracted convective heat-transfer coefficients are pre-sented in TABLE 1 for the four cavities. As can be seen in FIGURE 2, while there are some differences between the data and calculated curves, especially towards lower beam currents, the overall agreement is remarkably good. It is possible that the better agreement towards higher beam intensities is related to more ebullient boiling and more rapid mixing, i.e. closer to the conditions that the model assumes. The values obtained for the overall convective heat-transfer coefficient are also remarkably similar. This tells us that, by and large, all the cavities perform in a similar way and the performance in terms of maximum operational beam current depends largely on the available surface to effectively remove the heat from. The values of h increase marginally if a smaller value is adopted for the cooling water. Note that the choice of T0 = 30 ᵒC used to obtain the results in TABLE 1 is typical for the room temperature closed-loop cooling system used at iThemba LABS, once it has stabilized under operational conditions. A study by Buckley [5] on a quite different target design reports a value of h = 0.49 W cm−2 ᵒC−1, which is reassuringly similar. That study describes a cylindrical target cavity with a volume of 0.9 cm3, 8 mm deep, cooled with 25 ᵒC water from the back, operated with a 15 MeV proton beam with an intensity of 30 µA. The Nb Nirta targets are typically filled with 18O-water to about 60% of the cavity volume (see refs. [1,2] for the recommended values). The elongated shape, in combination with the ebullient properties of the boiling water, prevents burn-through. All the targets deliver the expected saturation yield. The targets are self-regulating ─ no external gas pressure is required. While the thermosyphon targets seemingly take advantage of a superior concept, we are now questioning whether this is really so in practice? It is not clear to us that the much more complex thermosyphon targets deliver any operational and/or performance advantages compared to the simple elegance of these elongated, single-cavity boiling target designs.

Introduction Batch-style water targets used for F-18 production are known to operate under boiling conditions in the target irradiation chamber, but the distribution of vapor under steady-state conditions was previously unknown. Thermal performance of batch targets has been correlated to average void in the target [1], but the simplified assumptions of such models do not represent the true non-uniform boiling behavior. Visualization targets can be used to observe boiling inside of a target during operation [2–5]. Commercial BTI targets operate at 28–35 bar (400–500 psi) with heat inputs of 0.5 to 3 kW and fill volumes of 1 to 4 mL. Recently, a visualization target featuring two transparent viewing windows was used to observe boiling conditions for realistic operating beam power, target pressure, and fill volume [4]. The same methodology has been applied to three additional visualization targets to examine the effect of target geometry on observed boiling phenomena. Material and Methods The original visualization target featured an aluminum body with a 0.127 mm (0.005 inch) integral aluminum beam window and two viewing windows made of optically clear sapphire (Al2O3). It was operated on an IBA 18/9 cyclotron with 18 MeV protons at beam power up to 1.1 kW, for pressures of 5 to 21 bar (70 to 300 psi), and a fill volume of 2.5 mL. All of the new designs featured a wider chamber to allow for higher beam transmission and an increased chamber height, consistent with cur-rent trends in high power targets. One target featured a reduced chamber depth, and another had a ramp in the back of the chamber to reduce fill volume. Target pressure was limited to a maximum of 14 bar (200 psi) due to the larger diameter beam window. A video camera was used to record the boiling conditions observed for each target under several lighting conditions. During irradiation, the proton beam excites the water molecules, producing visible blue light emissions during de-excitation. These light emissions provide a good indication of beam distribution and penetration depth. A strong backlight can be used to produce clearer images of bubbles generated during boiling. Results and Conclusion Proton range and visible blue light emissions were recorded in dark ambient conditions. The width of the Bragg peak and natural circulation in the bulk fluid were visible with good ambient lighting. Size and distribution of vapor bubbles could be observed by using a strong backlight. The beam current was increased gradually to determine the thermal limit for each target for several fill volumes and pressures. Two thermal limits were observed which resulted in some beam penetration in the top region of the beam. For lower fill volumes, steam ac-cumulates in or around the helium overpressure bubble, causing the helium bubble to move into the upper region of the beam. For higher fill volumes, beam penetration occurs due to excessive voiding, when bubbles produced in the beam region cannot rise quickly enough out of the path of the beam.

Introduction It is shown that a very simple model reproduces the pressure versus beam current characteristics of elongated single-cavity boiling water targets for 18F production surprisingly well. By fitting the model calculations to measured data, values for a single free parameter, namely an overall heat-transfer coefficient, have been extracted for several IBA Nirta H218O targets. IBA recently released details on their new Nirta targets that have a conical shape, which constitutes an improvement over the original Nirta targets that have a cylindrical shape [1,2]. These shapes are shown schematically in FIGURE 1. A study by Alvord et al. [3] pointed out that elevated pressures and temperatures in excess of the saturation conditions may exist in a water target during bombardment. However, as long as the rate of condensation matches the rate of vaporization, the bulk of the system should remain at saturation conditions. Superheated regions are therefore likely to form but also likely to disappear rapidly, typically on the scale of a few milliseconds. Even though the boiling process is generally quite complex, enhanced by radiation-induced nucleation, the presence of fast mixing mechanisms in the water volume justifies some simplifications to be made. Materials and Methods The simplified model assumes that the bulk of the target water has a constant temperature, which is the same as the inner wall temperature of the cavity, Tw. A second simplification is to neglect the temperature difference across the target chamber wall, which is only justified if the wall is thin. The boiling is not explicitly taken into consideration, including the rather complex boiling behaviour at the Havar window, except to acknowledge that it is the main mixing mechanism. Large temperature gradients can briefly exist in the water medium but they also rapidly disappear. A further assumption is that a single, overall convective heat-transfer coefficient can be applied, which is constant over the entire water-cooled surface. As the wall thickness is neglected, the heat-transfer surface is assumed to be the inner surface of the cavity, excluding the surface of the Havar window. One can then write down an energy balance between the beam heating and the convection cooling (Newton’s law of cooling), where Ib is the beam intensity, ΔE is the energy windows of the target (taken as 18 MeV), h is the convective heat-transfer coefficient, A is the inner cavity surface through which the heat has to be transferred from the target-water volume to the cooling water, and T0 is the cooling-water temperature. The saturated vapour pressure of water versus temperature is a characteristic curve, given by the steam tables [4]. Assuming the bulk of the system at saturation conditions, one gets from (1) and (2). The function f is represented by a polynomial. The only unknown in Equation (3) is the overall convective heat-transfer coefficient h. Our approach was to adjust h until a good fit with a set of measured data was obtained. It also has to be mentioned that subtle differences in the physical properties between 18O-water and natural water have been neglected. All in all, quite a few assumptions and simplifications are made in deriving Equation (3) and the system is, admittedly, much more complex. Nevertheless, the results obtained by applying Equation (3) are rather interesting. Results and Conclusion Measured data and corresponding calculations are shown in FIGURE 2 for three different conical targets and one cylindrical target. The extracted convective heat-transfer coefficients are pre-sented in TABLE 1 for the four cavities. As can be seen in FIGURE 2, while there are some differences between the data and calculated curves, especially towards lower beam currents, the overall agreement is remarkably good. It is possible that the better agreement towards higher beam intensities is related to more ebullient boiling and more rapid mixing, i.e. closer to the conditions that the model assumes. The values obtained for the overall convective heat-transfer coefficient are also remarkably similar. This tells us that, by and large, all the cavities perform in a similar way and the performance in terms of maximum operational beam current depends largely on the available surface to effectively remove the heat from. The values of h increase marginally if a smaller value is adopted for the cooling water. Note that the choice of T0 = 30 ᵒC used to obtain the results in TABLE 1 is typical for the room temperature closed-loop cooling system used at iThemba LABS, once it has stabilized under operational conditions. A study by Buckley [5] on a quite different target design reports a value of h = 0.49 W cm−2 ᵒC−1, which is reassuringly similar. That study describes a cylindrical target cavity with a volume of 0.9 cm3, 8 mm deep, cooled with 25 ᵒC water from the back, operated with a 15 MeV proton beam with an intensity of 30 µA. The Nb Nirta targets are typically filled with 18O-water to about 60% of the cavity volume (see refs. [1,2] for the recommended values). The elongated shape, in combination with the ebullient properties of the boiling water, prevents burn-through. All the targets deliver the expected saturation yield. The targets are self-regulating ─ no external gas pressure is required. While the thermosyphon targets seemingly take advantage of a superior concept, we are now questioning whether this is really so in practice? It is not clear to us that the much more complex thermosyphon targets deliver any operational and/or performance advantages compared to the simple elegance of these elongated, single-cavity boiling target designs.

Introduction Batch-style water targets used for F-18 production are known to operate under boiling conditions in the target irradiation chamber, but the distribution of vapor under steady-state conditions was previously unknown. Thermal performance of batch targets has been correlated to average void in the target [1], but the simplified assumptions of such models do not represent the true non-uniform boiling behavior. Visualization targets can be used to observe boiling inside of a target during operation [2–5]. Commercial BTI targets operate at 28–35 bar (400–500 psi) with heat inputs of 0.5 to 3 kW and fill volumes of 1 to 4 mL. Recently, a visualization target featuring two transparent viewing windows was used to observe boiling conditions for realistic operating beam power, target pressure, and fill volume [4]. The same methodology has been applied to three additional visualization targets to examine the effect of target geometry on observed boiling phenomena. Material and Methods The original visualization target featured an aluminum body with a 0.127 mm (0.005 inch) integral aluminum beam window and two viewing windows made of optically clear sapphire (Al2O3). It was operated on an IBA 18/9 cyclotron with 18 MeV protons at beam power up to 1.1 kW, for pressures of 5 to 21 bar (70 to 300 psi), and a fill volume of 2.5 mL. All of the new designs featured a wider chamber to allow for higher beam transmission and an increased chamber height, consistent with cur-rent trends in high power targets. One target featured a reduced chamber depth, and another had a ramp in the back of the chamber to reduce fill volume. Target pressure was limited to a maximum of 14 bar (200 psi) due to the larger diameter beam window. A video camera was used to record the boiling conditions observed for each target under several lighting conditions. During irradiation, the proton beam excites the water molecules, producing visible blue light emissions during de-excitation. These light emissions provide a good indication of beam distribution and penetration depth. A strong backlight can be used to produce clearer images of bubbles generated during boiling. Results and Conclusion Proton range and visible blue light emissions were recorded in dark ambient conditions. The width of the Bragg peak and natural circulation in the bulk fluid were visible with good ambient lighting. Size and distribution of vapor bubbles could be observed by using a strong backlight. The beam current was increased gradually to determine the thermal limit for each target for several fill volumes and pressures. Two thermal limits were observed which resulted in some beam penetration in the top region of the beam. For lower fill volumes, steam ac-cumulates in or around the helium overpressure bubble, causing the helium bubble to move into the upper region of the beam. For higher fill volumes, beam penetration occurs due to excessive voiding, when bubbles produced in the beam region cannot rise quickly enough out of the path of the beam.