Introduction Sr-82 is produced for PET cardiac imaging at the Isotope Production Facility (IPF) with 100-MeV proton beams. During irradiation, the target material (RbCl, Rb) and Inconel capsule are ex-posed for extended periods to intense radiation, thermally and mechanically induced stresses, and chemicals. The structural integrity of the Inconel capsules is of crucial importance to containing the target starting materials and produced Sr-82. Unexpected failure capsules severely affects the reliability of the isotope supply chain and increases in radioactive emission and wastes, maintenance cost, and personnel radia-tion exposure. Knowledge of the structural integrity of a target before irradiation plays an important role in that defects may be identified and rejected prior to irradiation. In the cases of where a breach occurs, the location of the breach can be correlated with the inspected data. Material and Methods RbCl target failure: IPF has a successful irradiation history of RbCl targets at 230 A proton beam current since the facility commissioning in 2004. In 2013 run cycle, three targets irradiated in the medium energy B slot (35–65 MeV) [1] failed unexpectedly. The failure mode was the formation and propagation of cracks at the cor-ner radius along the edge of the target (FIGS. 1a-b). The common failure location was in the rear window relative to the beam direction and at the top of the target. These targets failed relatively early in the course of irradiation and typically after several cycles of beam loss and recovery. Possible failure mechanisms: A calculated von-Mises stress analysis at room temperature of an Inconel capsule under a static pressure load at 4 MPa shows a stress concentration at the corner radius and deformation of the window (FIG. 2). Additionally, a beam loss and recovery process causes the capsule windows to fatique especially at the corner due to a thermal and pressure cyclic loading. Furthermore, there is a thermal stress within the window due a temperature gradient resulting from nonuniform heating by the donut-shaped IPF beam [2]. Finally, Cl vapor in the void region or Rb liquid at the top of the target where the highest temperature of target material (RbCl or Rb) is expected may have contribution to a stress-corrosion cracking. An individual or a combination of these mechanisms aggrevate target failure if defects (voids, cracks, or thinning) exist. When the applied stress exceeds the ultimate tensile strength of Inconel, the target is likely to fail at these locations. Non-destructive evaluation methods: Digital radiographic images were generated using a Philips 450 x-ray source set to 150–190 keV and a Varian panel detector. Ultrasonic (UT) amplitude and time-of-flight (TOF) images were generated with a spherically-focused transducer operated at 50 MHz. Results Inconel capsule halves: Radiographic images of the front and rear parts of 7 RbCl A targets (~65-95 MeV) and 7 RbCl B targets prior to target assembly (FIG. 3). For target A halves (left two columns), there is some variation in thickness between the front and rear parts. Other than thickness variation, no other defects (inclusions, voids, cracks) was detected. For target B halves (right two columns), all rear parts exhibit thinning around their edges, whereas the front parts appear more uniform. UT TOF images were performed on 4 target A halves (155, 156, 157, and 159) and 7 target B halves (154-160). The rear window of 155A appears to thin out (~12.5%) near the rim on the right half. The front of 159A shows a similar thinning (~ 15%) near the rim on the left half. Although there is a thinning along the edges, all parts except 159A front have an average thickness within the stated specification (TABLE 1). Similarly to radiographic data, UT TOF data con-firm a thinning towards the edges of the window on most of target B parts. Only images of 155B are illustrated in FIG. 4. Significant thinning (15%) is observed on 154B (front & rear), and the rear windows of 155B, 157B, 158B, and 159B. Although there is a thinning, all parts have an average thickness within the stated specification (0.0120” ± 0.0005”) except for the rear windows of 154B and 155B. No inclusions or voids are apparent in any of the parts. RbCl filled targets: For comparison purpose, three B (130, 135, 147) and two A (137, 147) filled targets were evaluated. Radiographic data show no defects in the Inconel capsules while the RbCl pucks have numerous features (cracks, voids). The images of targets 130B and 135B illustrate the basic conditions of the RbCl pucks (FIG. 5). UT TOF images of targets 130B and 135B rear and front windows are illustrated in FIG. 6. Average thicknesses of 0.011–0.014” for both rear and front windows of all 5 targets are within the stated specification. However, there is thinning around the edge of the target 135B front window. Rb empty capsule: Radiograph of an unfilled Inconel capsule with and the fill tube is shown in FIG. 7. The predrilled 1-mm OD pinhole on the front window can be easily detected with the instrument’s detection limits of 30-μm pinhole and 5-μm crack. There is no other visible defect or thickness variation. This target was filled with Rb to characterize the reaction released Rb through the pinhole with water and its effects on equipment. Rb metal filled targets: Radiographs of two Rb metal filled targets show the front and side views of Rb distribution and fill tube (FIG. 8). Voids are visible throughout the Rb and small amount of Rb remaining in the fill tube. TOF results indicate the average thicknesses of 0.0201–0.0214” for both rear and front windows of 2 targets. Except the 2B front window, all thicknesses are within stated specification (0.020” ± 0.0005). UT TOF images for the rear and front of each target capsule are shown in FIG. 9. Moiré pat-terns are likely caused by a combination of stress arising in the manufacturing/filling process and some degree of measurement artifact. Target 1B windows exhibit uniform thickness across the bulk of the diameter, with the front window being slightly thinner overall than the rear. There is slight thinning observed near the edges on both windows. Thinning is more pronounced on the left side of the rear window than the right side of the front window. Target 2B shows a more pronounced distortion particularly on the rear window. The rear window appears to have a slightly thinner concentric region approximately one-quarter of diameter in. The front window displays good uniformity, with slight thinning along the inner edge of the left. Both targets 1B and 2B were successfully irradiated up to 230 A for 2 hours. Higher beam current and longer irradiation of Rb targets is underway. Conclusion Radiographic and ultrasonic methods were used in non-destructive evaluation of pre-assembly Inconel parts and fully assembled RbCl and Rb targets. These studies show the potential to identify defective parts and/or targets prior to irradiation, to provide useful information for improving target manufacturing process, and to enable better decision-making in managing risks of target failure. The results also have target quality assurance potential, enable comparison of target features and document data for future interpretation of target failure. The benefits of non-destructive evaluation include improved target reliability, reduced target failure rate, reduced revenue loss and increased productivity of Sr-82.

Strontium-82 (t1/2 = 25.5 d) is one of the medical isotopes produced on a large scale at the Isotope Production Facility (IPF) of the Los Alamos National Laboratory (LANL), employing a high intensity 100 MeV proton beam and RbCl targets. A constant increase in the 82Sr demand over the last decade combined with an established thermal limit of molten RbCl salt targets [1,2] has challenged the IPF’s world leading production capacity in recent years and necessitated the consideration of low-melting point (39.3 °C) Rb metal targets. Metal targets are used at other facilities [3–5] and offer obvious production rate advantages due to a higher relative density of Rb target atoms and a higher expected thermal performance of molten metal. One major disadvantage is the known violent reaction of molten Rb with cooling water and the potential for facility damage following a catastrophic target failure. This represents a significant risk, given the high beam intensities used routinely at IPF. In order to assess this risk, a target failure experiment was conducted at the LANL firing site using a mockup target station. Subsequent fabrication, irradiation and processing of two prototype targets showed a target thermal performance consistent with thermal modeling predictions and yields in agreement with predictions based on IAEA recommended cross sections [6]. Target failure test: The target failure test bed (FIG. 1) was constructed to represent a near replica of the IPF target station, incorporating its most important features. One of the most vulnerable components in the assembly is the Inconel beam window (FIG. 2) which forms the only barrier between the target cooling water and the beam line vacuum. The test bed also mimicked relevant IPF operational parameters seeking to simulate the target environment during irradiation, such as typical cooling water flow velocities around the target surfaces. While the aggressive thermal effects of the beam heating could not be simulated directly, heated cooling water (45 °C) ensured that the rubidium target material remained molten during the failure test. A worst case catastrophic target failure event was initiated by uncovering an oversized predrilled pinhole (1 mm Φ) to abruptly expose the molten target material to fast flowing cooling water. Prototype target irradiations: Two prototype Rb metal target containers were fabricated by machining Inconel 625 parts and by EB welding. The target containers were filled with molten Rb metal under an inert argon atmosphere. Follow-ing appropriate QA inspections, the prototype targets were irradiated in the medium energy slot of a standard IPF target stack using beam currents up to 230 µA. After irradiation the targets were transported to the LANL hot cell facili-ty for processing and for 82Sr yield verification. During the target failure test, cooling water conductivity and pressure excursions in the target chamber were continuously monitored and recorded at a rate of 1 kHz. Video footage taken of the beam window and the pinhole area combined with the recorded data indicated an aggressive reaction between the Rb metal and the cooling water, but did not reveal a violent explosion that could seriously damage the beam window. These observations, together with thermal model predictions, provided the necessary confidence to fabricate and fill prototype targets for irradiation at production-scale beam currents. X-ray imaging of filled targets (FIG. 3) shows a need for tighter control over the target fill level. One prototype target was first subjected to lower intensity (< 150 µA) beams before the second was irradiated at production level (230 µA) beams. During irradiation, monitoring of cooling water conductivity indicated no container breach or leak and, as anticipated given the model predictions, the post irradiation target inspection showed no sign of imminent thermal failure (see FIG. 4). Subsequent chemical processing of the targets followed an established procedure that was slightly modified to accommodate the larger target mass. TABLE 1 shows that post chemistry 82Sr yields agree to within 2 % of the in-target production rates expected on the basis of IAEA recommended cross sections. The table also compares 82Sr yields from the Rb metal targets against yields routinely obtained from RbCl targets, showing an increase in yield of almost 50 %.

Introduction A method based on neutron activation is being developed to assist in resolving discrepancies between the expected yield and actual yield of radionuclides produced with the vertical-beam target station (VBTS) at iThemba LABS. The VBTS is routinely employed for multi-Ci batch productions of the radionuclide pairs 22Na/68Ga and 82Sr/68Ga using standardized natMg/natGa and natRb/natGa tandem targets, respectively [1]. The metal-clad target discs are bombarded with a primary beam of 66 MeV protons at an intensity of nominally 250 µA. The encapsulation materials are either Nb (for Mg and Ga) or stainless steel (for Rb) which serve to contain the molten target materials during bombardment and act as a barrier to the high-velocity cooling water which surrounds the targets in a 4π geometry. The natRb/natGa targets are typically bombarded according to a two-week cycle while natMg/natGa targets are bombarded on an ad-hoc basis, depending on a somewhat unpredictable 22Na demand. A too-large deviation between expected yield and actual yield has at times plagued this programme. These deviations can manifest both as an apparent loss or an apparent gain (relative to the expected yield) by up to about 15% in either direction. The resulting uncertainty of up to 30% (in the worst case) from one production batch to another can be costly and is unacceptable in a large-scale production regimen. This phenomenon is believed to be brought about by two types of problems: (1) Production losses, e.g. during the radio-chemical separation process or incomplete recovery of activated target material during the decapsulation step. (2) Incorrect values obtained for the accumulated proton charge. A problem of type (1) will always result in a loss of yield. A problem of type (2) can manifest as an apparent loss or gain. In an effort to get a handle on this second type of problem, neutron activation of suitable material samples, embedded in a target holder, is being investigated as an independent indicator of the total yield. For this purpose, samples of Co, Mn, Ni and Zn were activated during production runs and Co was found to be the most appropriate. Preliminary results will be presented after first discussing why the determination of the accumulated pro-ton charge is a problem with the VBTS. Materials and Methods The VBTS consists of a central region in which a target holder is located during bombardment as well as two half-cylindrical radiation shields which completely surround the target. The shields can be moved away from the central region on dedicated rails, e.g. when repairs or maintenance is required. FIGURE 1 shows the VBTS with the shields moved to the “open” position. As some components of the station are located below the vault floor, with the target position near floor level, it proved difficult to electrically isolate the VBTS as was done for the two horizontal-beam target stations at iThemba LABS [1]. The VBTS does not act as a Faraday cup like the other target stations. Instead, the beam current and accumulated charge is measured by means of a calibrated capacitive probe [1,2]. There appears to be a variation in the response of the capacitive probe, sensitive to the beam microstructure, in particular a dependence on the beam packet length. This problem is not yet fully resolved. FIGURE 2 (a) shows the beamstop of a VBTS target holder with several Co samples mounted on the outside as well as one each of Ni, Mn and Zn. The samples are small “tablets” with a 10 mm diameter and 1 mm thickness. The reactions of interest are 59Co(n,γ)60Co, 59Co(n,3n)57Co, nat-Ni(n,X)60Co, natNi(n,X)57Co, natZn(n,X)65Zn and 55Mn(n,2n)54Mn. The relevant half-lives are 60Co(5.271 a), 57Co(271.8 d), 65Zn(244.3 d) and 54Mn(312.2 d). The half-life should be long compared to the two-week cycle in order to reduce the dependence on the exact beam history, which is very fragmented over any production period. In this respect, 60Co is considered to be particularly attractive as its long half-life of more than 5 years leads to a negligible effect by the beam history. Note that the tandem targets, shown in FIGURE 2 (b), are mounted just upstream of the beamstop – in fact, the targets and beamstop form a single unit before being fitted into the target holder. At the end of bombardment, all samples were assayed for their characteristic γ-emissions using standard off-line γ-ray spectrometry with an HPGe detector connected to a Genie 2000 MCA. Calculations of the neutron fluence density in the central sample volume on the beamstop were also performed using the Monte Carlo radiation transport code MCNPX. For these calculations, the entire VBTS, a Rb/Ga target and the vault walls were included in the model. Results and Conclusion All samples activated significantly – copious amounts of 60Co were detected in the Co discs after a two-week run. The neutron fluence density for the case of a 250 µA, 66 MeV proton beam on a natRb/natGa tandem target is shown in FIGURE 3. The dominance of low-energy neutrons is evident, which is in part due to the large amount of paraffin-wax shielding material in close proximity to the target. While reactions such as the (n,2n) and (n,3n) would be sensitive to the more energetic part of the neutron spectrum, the (n,γ) capture reaction benefits from the large low-energy component. This explains the copious amounts of 60Co formed. It was therefore decided to only retain the central Co sample for subsequent bombardments, as shown in FIGURE 4. The first results are shown in TABLE 1. The accumulated charge as obtained from the capacitive probe (Q), the specific 60Co activity (A) at the end of bombardment (EOB), and their ratio (A/Q) are presented in the table, together with the deviation of individual ratios relative to their average for the case of the Mg/Ga tandem tar-gets only. Note that all samples were counted until the statistical uncertainties were negligible. Any systematic uncertainties are ignored at this stage as they are considered to remain the same from one batch production to another. For the sake of argument, the average value of the ratio is taken as the expected value. A positive deviation of the A/Q value is then indicative of a too-small value of the accumulated charge obtained from the capacitive probe, leading to a corresponding overproduction. Likewise, a negative value is indicative of a too-large value of the accumulated charge, leading to a corresponding underproduction. It is certainly true that the data in TABLE 1 are currently very limited. It is envisaged, however, that with time the growing database of values will assist in reducing the uncertainty in determining the accumulated charge and reduce the discrepancies between predicted and actual yields significantly. TABLE 1 illuminates the underlying problem satisfactorily. The four Mg/Ga tandem target bombardments, on identical targetry, were performed successively. The neutron activation correlates well the with actual yields, pointing directly to the current integration as the main source of error. The method already proves to be useful. An indication of an over or underprediction can be obtained prior to the target processing by recovering and measuring the Co disc. This in-formation can be used to make a decision concerning the present batch production and/or the subsequent one. One can either add beam to the present production target and/or in-crease/reduce the total beam on the subsequent production target to compensate for an expected overproduction or shortfall. In conclusion, we would like to stress that the capacitive probes show great promise and that better understanding and/or possibly some development of their signal processing algorithm may improve their ability to measure the accumulated charge to the desired accuracy. Segmented capacitive probes used at iThemba LABS and elsewhere for beam position measurement [1,3] are not affected by beam microstructure as only the ratios of the signal strengths on the different sectors are important. In this case, changes in response affect all sec-tors equally and the ratios are unaffected.

Strontium-82 (t1/2 = 25.5 d) is one of the medical isotopes produced on a large scale at the Isotope Production Facility (IPF) of the Los Alamos National Laboratory (LANL), employing a high intensity 100 MeV proton beam and RbCl targets. A constant increase in the 82Sr demand over the last decade combined with an established thermal limit of molten RbCl salt targets [1,2] has challenged the IPF’s world leading production capacity in recent years and necessitated the consideration of low-melting point (39.3 °C) Rb metal targets. Metal targets are used at other facilities [3–5] and offer obvious production rate advantages due to a higher relative density of Rb target atoms and a higher expected thermal performance of molten metal. One major disadvantage is the known violent reaction of molten Rb with cooling water and the potential for facility damage following a catastrophic target failure. This represents a significant risk, given the high beam intensities used routinely at IPF. In order to assess this risk, a target failure experiment was conducted at the LANL firing site using a mockup target station. Subsequent fabrication, irradiation and processing of two prototype targets showed a target thermal performance consistent with thermal modeling predictions and yields in agreement with predictions based on IAEA recommended cross sections [6]. Target failure test: The target failure test bed (FIG. 1) was constructed to represent a near replica of the IPF target station, incorporating its most important features. One of the most vulnerable components in the assembly is the Inconel beam window (FIG. 2) which forms the only barrier between the target cooling water and the beam line vacuum. The test bed also mimicked relevant IPF operational parameters seeking to simulate the target environment during irradiation, such as typical cooling water flow velocities around the target surfaces. While the aggressive thermal effects of the beam heating could not be simulated directly, heated cooling water (45 °C) ensured that the rubidium target material remained molten during the failure test. A worst case catastrophic target failure event was initiated by uncovering an oversized predrilled pinhole (1 mm Φ) to abruptly expose the molten target material to fast flowing cooling water. Prototype target irradiations: Two prototype Rb metal target containers were fabricated by machining Inconel 625 parts and by EB welding. The target containers were filled with molten Rb metal under an inert argon atmosphere. Follow-ing appropriate QA inspections, the prototype targets were irradiated in the medium energy slot of a standard IPF target stack using beam currents up to 230 µA. After irradiation the targets were transported to the LANL hot cell facili-ty for processing and for 82Sr yield verification. During the target failure test, cooling water conductivity and pressure excursions in the target chamber were continuously monitored and recorded at a rate of 1 kHz. Video footage taken of the beam window and the pinhole area combined with the recorded data indicated an aggressive reaction between the Rb metal and the cooling water, but did not reveal a violent explosion that could seriously damage the beam window. These observations, together with thermal model predictions, provided the necessary confidence to fabricate and fill prototype targets for irradiation at production-scale beam currents. X-ray imaging of filled targets (FIG. 3) shows a need for tighter control over the target fill level. One prototype target was first subjected to lower intensity (< 150 µA) beams before the second was irradiated at production level (230 µA) beams. During irradiation, monitoring of cooling water conductivity indicated no container breach or leak and, as anticipated given the model predictions, the post irradiation target inspection showed no sign of imminent thermal failure (see FIG. 4). Subsequent chemical processing of the targets followed an established procedure that was slightly modified to accommodate the larger target mass. TABLE 1 shows that post chemistry 82Sr yields agree to within 2 % of the in-target production rates expected on the basis of IAEA recommended cross sections. The table also compares 82Sr yields from the Rb metal targets against yields routinely obtained from RbCl targets, showing an increase in yield of almost 50 %.

Introduction Sr-82 is produced for PET cardiac imaging at the Isotope Production Facility (IPF) with 100-MeV proton beams. During irradiation, the target material (RbCl, Rb) and Inconel capsule are ex-posed for extended periods to intense radiation, thermally and mechanically induced stresses, and chemicals. The structural integrity of the Inconel capsules is of crucial importance to containing the target starting materials and produced Sr-82. Unexpected failure capsules severely affects the reliability of the isotope supply chain and increases in radioactive emission and wastes, maintenance cost, and personnel radia-tion exposure. Knowledge of the structural integrity of a target before irradiation plays an important role in that defects may be identified and rejected prior to irradiation. In the cases of where a breach occurs, the location of the breach can be correlated with the inspected data. Material and Methods RbCl target failure: IPF has a successful irradiation history of RbCl targets at 230 A proton beam current since the facility commissioning in 2004. In 2013 run cycle, three targets irradiated in the medium energy B slot (35–65 MeV) [1] failed unexpectedly. The failure mode was the formation and propagation of cracks at the cor-ner radius along the edge of the target (FIGS. 1a-b). The common failure location was in the rear window relative to the beam direction and at the top of the target. These targets failed relatively early in the course of irradiation and typically after several cycles of beam loss and recovery. Possible failure mechanisms: A calculated von-Mises stress analysis at room temperature of an Inconel capsule under a static pressure load at 4 MPa shows a stress concentration at the corner radius and deformation of the window (FIG. 2). Additionally, a beam loss and recovery process causes the capsule windows to fatique especially at the corner due to a thermal and pressure cyclic loading. Furthermore, there is a thermal stress within the window due a temperature gradient resulting from nonuniform heating by the donut-shaped IPF beam [2]. Finally, Cl vapor in the void region or Rb liquid at the top of the target where the highest temperature of target material (RbCl or Rb) is expected may have contribution to a stress-corrosion cracking. An individual or a combination of these mechanisms aggrevate target failure if defects (voids, cracks, or thinning) exist. When the applied stress exceeds the ultimate tensile strength of Inconel, the target is likely to fail at these locations. Non-destructive evaluation methods: Digital radiographic images were generated using a Philips 450 x-ray source set to 150–190 keV and a Varian panel detector. Ultrasonic (UT) amplitude and time-of-flight (TOF) images were generated with a spherically-focused transducer operated at 50 MHz. Results Inconel capsule halves: Radiographic images of the front and rear parts of 7 RbCl A targets (~65-95 MeV) and 7 RbCl B targets prior to target assembly (FIG. 3). For target A halves (left two columns), there is some variation in thickness between the front and rear parts. Other than thickness variation, no other defects (inclusions, voids, cracks) was detected. For target B halves (right two columns), all rear parts exhibit thinning around their edges, whereas the front parts appear more uniform. UT TOF images were performed on 4 target A halves (155, 156, 157, and 159) and 7 target B halves (154-160). The rear window of 155A appears to thin out (~12.5%) near the rim on the right half. The front of 159A shows a similar thinning (~ 15%) near the rim on the left half. Although there is a thinning along the edges, all parts except 159A front have an average thickness within the stated specification (TABLE 1). Similarly to radiographic data, UT TOF data con-firm a thinning towards the edges of the window on most of target B parts. Only images of 155B are illustrated in FIG. 4. Significant thinning (15%) is observed on 154B (front & rear), and the rear windows of 155B, 157B, 158B, and 159B. Although there is a thinning, all parts have an average thickness within the stated specification (0.0120” ± 0.0005”) except for the rear windows of 154B and 155B. No inclusions or voids are apparent in any of the parts. RbCl filled targets: For comparison purpose, three B (130, 135, 147) and two A (137, 147) filled targets were evaluated. Radiographic data show no defects in the Inconel capsules while the RbCl pucks have numerous features (cracks, voids). The images of targets 130B and 135B illustrate the basic conditions of the RbCl pucks (FIG. 5). UT TOF images of targets 130B and 135B rear and front windows are illustrated in FIG. 6. Average thicknesses of 0.011–0.014” for both rear and front windows of all 5 targets are within the stated specification. However, there is thinning around the edge of the target 135B front window. Rb empty capsule: Radiograph of an unfilled Inconel capsule with and the fill tube is shown in FIG. 7. The predrilled 1-mm OD pinhole on the front window can be easily detected with the instrument’s detection limits of 30-μm pinhole and 5-μm crack. There is no other visible defect or thickness variation. This target was filled with Rb to characterize the reaction released Rb through the pinhole with water and its effects on equipment. Rb metal filled targets: Radiographs of two Rb metal filled targets show the front and side views of Rb distribution and fill tube (FIG. 8). Voids are visible throughout the Rb and small amount of Rb remaining in the fill tube. TOF results indicate the average thicknesses of 0.0201–0.0214” for both rear and front windows of 2 targets. Except the 2B front window, all thicknesses are within stated specification (0.020” ± 0.0005). UT TOF images for the rear and front of each target capsule are shown in FIG. 9. Moiré pat-terns are likely caused by a combination of stress arising in the manufacturing/filling process and some degree of measurement artifact. Target 1B windows exhibit uniform thickness across the bulk of the diameter, with the front window being slightly thinner overall than the rear. There is slight thinning observed near the edges on both windows. Thinning is more pronounced on the left side of the rear window than the right side of the front window. Target 2B shows a more pronounced distortion particularly on the rear window. The rear window appears to have a slightly thinner concentric region approximately one-quarter of diameter in. The front window displays good uniformity, with slight thinning along the inner edge of the left. Both targets 1B and 2B were successfully irradiated up to 230 A for 2 hours. Higher beam current and longer irradiation of Rb targets is underway. Conclusion Radiographic and ultrasonic methods were used in non-destructive evaluation of pre-assembly Inconel parts and fully assembled RbCl and Rb targets. These studies show the potential to identify defective parts and/or targets prior to irradiation, to provide useful information for improving target manufacturing process, and to enable better decision-making in managing risks of target failure. The results also have target quality assurance potential, enable comparison of target features and document data for future interpretation of target failure. The benefits of non-destructive evaluation include improved target reliability, reduced target failure rate, reduced revenue loss and increased productivity of Sr-82.

Introduction A method based on neutron activation is being developed to assist in resolving discrepancies between the expected yield and actual yield of radionuclides produced with the vertical-beam target station (VBTS) at iThemba LABS. The VBTS is routinely employed for multi-Ci batch productions of the radionuclide pairs 22Na/68Ga and 82Sr/68Ga using standardized natMg/natGa and natRb/natGa tandem targets, respectively [1]. The metal-clad target discs are bombarded with a primary beam of 66 MeV protons at an intensity of nominally 250 µA. The encapsulation materials are either Nb (for Mg and Ga) or stainless steel (for Rb) which serve to contain the molten target materials during bombardment and act as a barrier to the high-velocity cooling water which surrounds the targets in a 4π geometry. The natRb/natGa targets are typically bombarded according to a two-week cycle while natMg/natGa targets are bombarded on an ad-hoc basis, depending on a somewhat unpredictable 22Na demand. A too-large deviation between expected yield and actual yield has at times plagued this programme. These deviations can manifest both as an apparent loss or an apparent gain (relative to the expected yield) by up to about 15% in either direction. The resulting uncertainty of up to 30% (in the worst case) from one production batch to another can be costly and is unacceptable in a large-scale production regimen. This phenomenon is believed to be brought about by two types of problems: (1) Production losses, e.g. during the radio-chemical separation process or incomplete recovery of activated target material during the decapsulation step. (2) Incorrect values obtained for the accumulated proton charge. A problem of type (1) will always result in a loss of yield. A problem of type (2) can manifest as an apparent loss or gain. In an effort to get a handle on this second type of problem, neutron activation of suitable material samples, embedded in a target holder, is being investigated as an independent indicator of the total yield. For this purpose, samples of Co, Mn, Ni and Zn were activated during production runs and Co was found to be the most appropriate. Preliminary results will be presented after first discussing why the determination of the accumulated pro-ton charge is a problem with the VBTS. Materials and Methods The VBTS consists of a central region in which a target holder is located during bombardment as well as two half-cylindrical radiation shields which completely surround the target. The shields can be moved away from the central region on dedicated rails, e.g. when repairs or maintenance is required. FIGURE 1 shows the VBTS with the shields moved to the “open” position. As some components of the station are located below the vault floor, with the target position near floor level, it proved difficult to electrically isolate the VBTS as was done for the two horizontal-beam target stations at iThemba LABS [1]. The VBTS does not act as a Faraday cup like the other target stations. Instead, the beam current and accumulated charge is measured by means of a calibrated capacitive probe [1,2]. There appears to be a variation in the response of the capacitive probe, sensitive to the beam microstructure, in particular a dependence on the beam packet length. This problem is not yet fully resolved. FIGURE 2 (a) shows the beamstop of a VBTS target holder with several Co samples mounted on the outside as well as one each of Ni, Mn and Zn. The samples are small “tablets” with a 10 mm diameter and 1 mm thickness. The reactions of interest are 59Co(n,γ)60Co, 59Co(n,3n)57Co, nat-Ni(n,X)60Co, natNi(n,X)57Co, natZn(n,X)65Zn and 55Mn(n,2n)54Mn. The relevant half-lives are 60Co(5.271 a), 57Co(271.8 d), 65Zn(244.3 d) and 54Mn(312.2 d). The half-life should be long compared to the two-week cycle in order to reduce the dependence on the exact beam history, which is very fragmented over any production period. In this respect, 60Co is considered to be particularly attractive as its long half-life of more than 5 years leads to a negligible effect by the beam history. Note that the tandem targets, shown in FIGURE 2 (b), are mounted just upstream of the beamstop – in fact, the targets and beamstop form a single unit before being fitted into the target holder. At the end of bombardment, all samples were assayed for their characteristic γ-emissions using standard off-line γ-ray spectrometry with an HPGe detector connected to a Genie 2000 MCA. Calculations of the neutron fluence density in the central sample volume on the beamstop were also performed using the Monte Carlo radiation transport code MCNPX. For these calculations, the entire VBTS, a Rb/Ga target and the vault walls were included in the model. Results and Conclusion All samples activated significantly – copious amounts of 60Co were detected in the Co discs after a two-week run. The neutron fluence density for the case of a 250 µA, 66 MeV proton beam on a natRb/natGa tandem target is shown in FIGURE 3. The dominance of low-energy neutrons is evident, which is in part due to the large amount of paraffin-wax shielding material in close proximity to the target. While reactions such as the (n,2n) and (n,3n) would be sensitive to the more energetic part of the neutron spectrum, the (n,γ) capture reaction benefits from the large low-energy component. This explains the copious amounts of 60Co formed. It was therefore decided to only retain the central Co sample for subsequent bombardments, as shown in FIGURE 4. The first results are shown in TABLE 1. The accumulated charge as obtained from the capacitive probe (Q), the specific 60Co activity (A) at the end of bombardment (EOB), and their ratio (A/Q) are presented in the table, together with the deviation of individual ratios relative to their average for the case of the Mg/Ga tandem tar-gets only. Note that all samples were counted until the statistical uncertainties were negligible. Any systematic uncertainties are ignored at this stage as they are considered to remain the same from one batch production to another. For the sake of argument, the average value of the ratio is taken as the expected value. A positive deviation of the A/Q value is then indicative of a too-small value of the accumulated charge obtained from the capacitive probe, leading to a corresponding overproduction. Likewise, a negative value is indicative of a too-large value of the accumulated charge, leading to a corresponding underproduction. It is certainly true that the data in TABLE 1 are currently very limited. It is envisaged, however, that with time the growing database of values will assist in reducing the uncertainty in determining the accumulated charge and reduce the discrepancies between predicted and actual yields significantly. TABLE 1 illuminates the underlying problem satisfactorily. The four Mg/Ga tandem target bombardments, on identical targetry, were performed successively. The neutron activation correlates well the with actual yields, pointing directly to the current integration as the main source of error. The method already proves to be useful. An indication of an over or underprediction can be obtained prior to the target processing by recovering and measuring the Co disc. This in-formation can be used to make a decision concerning the present batch production and/or the subsequent one. One can either add beam to the present production target and/or in-crease/reduce the total beam on the subsequent production target to compensate for an expected overproduction or shortfall. In conclusion, we would like to stress that the capacitive probes show great promise and that better understanding and/or possibly some development of their signal processing algorithm may improve their ability to measure the accumulated charge to the desired accuracy. Segmented capacitive probes used at iThemba LABS and elsewhere for beam position measurement [1,3] are not affected by beam microstructure as only the ratios of the signal strengths on the different sectors are important. In this case, changes in response affect all sec-tors equally and the ratios are unaffected.