Apparatus and Experiments Towards Fully Automated Medical Isotope Production Using an Ion Beam Accelerator

Abstract

Zirconium-89 (⁸⁹Zr) is a widely used radionuclide in immune-PET imaging due to its physical decay characteristics. Despite its importance, the production of ⁸⁹Zr radiopharmaceuticals remains largely manual, with limited cost-effective automation solutions available. To address this, we developed an automated system for the agile and reliable production of radiopharmaceuticals. The system performs transmutations, dissolution, and separation for a range of radioisotopes. Steps in the production of ⁸⁹Zr-oxalate are used as an exemplar to illustrate its use. Three-dimensional (3D) printing was exploited to design and manufacture a target holder able to include solid targets, in this case an ⁸⁹Y foil. Spot welding was used to attach ⁸⁹Y to a refractory tantalum (Ta) substrate. A commercially available CPU chiller was repurposed to efficiently cool the metal target. Furthermore, a commercial resin (ZR Resin) and compact peristaltic pumps were employed in a compact (10 × 10 × 10 cm³) chemical separation unit that operates automatically via computer-controlled software. Additionally, a standalone 3D-printed unit was designed with three automated functionalities: photolabelling, vortex mixing, and controlled heating. All components of the assembly, except for the target holder, are housed inside a commercially available hot cell, ensuring safe and efficient operation in a controlled environment. This paper details the design, construction, and modelling of the entire assembly, emphasising its innovative integration and operational efficiency for widespread radiopharmaceutical automation.

1. Introduction

Monoclonal antibodies (mAbs) and immunoglobulin fragments are widely used clinically for therapeutic purposes. Imaging using positron emission tomography (PET) is essential in diagnostic and pre-treatment assessment. One of the most useful radionuclides for radiolabelling mAbs [1] for immune-PET for cancer imaging is zirconium-89 (⁸⁹Zr). The half-life of ⁸⁹Zr (t_1/2 = 78.41 h), which matches the circulatory time of mAbs, along with its average (E_ave (β⁺) = 396.9 keV) and maximum (E_max (β⁺) = 897 keV) positron energies, makes it well suited for immune-PET imaging, producing high-resolution images comparable to those obtained with ⁶⁴Cu and ¹⁸F radionuclides [2,3].

Although ⁸⁹Zr can be produced via a commercial solid target system, such as the TRASIS system [4], the ALECO system [5], or the NITRA system [6], the target is usually transferred to the hot cell manually post-irradiation. Gelbart and Johnson designed a solid target system with in situ target dissolution. Although dissolution can be accomplished without manual transfer, the system lacks a chemical separation unit [7]. Imura et al. reported that the manual purification of ⁸⁹Zr hindered them from producing adequate ⁸⁹Zr for their work on ⁸⁹Zr-PSMA-617 [8], highlighting a key challenge in manual radiopharmaceutical processing. Automation in radiopharmaceutical production has become essential for improving efficiency, reducing radiation exposure, and ensuring reproducibility. By replacing manual steps with automated systems, production workflows can be streamlined, enabling safer and more reliable large-scale isotope production for both diagnostic and therapeutic applications.

Developing robust target and dissolution systems involves the following considerations:

• Target holder fabrication

A procedure that enables the production of complex geometries to support effective irradiation and thermal management.

• Target physical form

Designed to enhance cooling, minimise neutron activation, and ensure resistance to acid.

• Efficient chemical separation

Essential for isolating the desired radionuclide with high purity while minimising impurities.

• Photolabelling system integration

Combines vortex mixing, UV activation, and controlled heating to enhance labelling efficiency.

• Application of 3D printing technologies

Offers practical, cost-effective, and sustainable solutions that are highly adaptable for the design of target holders, separation systems, and photolabelling components [9].

The design of a target and dissolution system that enables production of radionuclides in an agile, reliable, and fully automated way is presented. The typical manufacturing process and the modular integration of irradiation, dissolution, separation, and labelling stages are illustrated schematically in Figure 1. The system consists of a polypropylene target holder produced using a 3D printer, a target design that maximises sample cooling and minimises neutron activation—enabling post-irradiation handling and compact peristaltic pumps for securely transferring liquids, as well as a heating/vortex-mixing unit to support downstream radiopharmaceutical processing and applications. This approach can be applied to produce a variety of isotopes necessary for preclinical and clinical work in the future.

2. Materials and Methods

2.1. Target Holder and Translator

The target holder was designed using a computer-aided design package (CAD; Fusion 360) and printed in-house from polypropylene (PP) from RS Components (RS part no. 183-0211) using a 3D printer (S3 Ultimaker). It comprises a bottom plate and a top plate. Each plate has dimensions of 108 × 108 mm², 20 mm thickness and eight holes (M6 clearance), two in each corner to allow M6 bolts and nuts to hold the top plate and bottom plate together.

The bottom plate is designed to hold a CPU chiller in a groove and other components, as illustrated in Figure 2. The tantalum sheet acts to protect the aluminium of the CPU chiller from acid during the dissolution phase (see below), as well as holding the metal target. The top plate has a conical structure that faces the ion beam. A conical sheet of tantalum was fabricated from 0.5 mm thick tantalum sheet by spot welding. It was inserted into the conical recess to catch any accidentally misdirected ions from the transmutation beam. The current can be monitored on this conical sheet to signal to the accelerator operators that the beam is being mis-steered (Figure 3). To seal the liquid inside the target holder, an O-ring groove was designed on the back surface of the top plate (Figure 4). The top plate also contains narrow tunnels, one from each side. As is illustrated in Figure 3, one tunnel is to move chemical reagents to/from the metal target post-irradiation for the target dissolution. The other tunnel is for the helium gas for cooling and sheathing the top surface of the ⁸⁹Y disc. These two pathways start from the top plate sides and end inside the O-ring groove passing through the top plate. The helium jet pathway was designed to cool and sheath the top surface of the target. This sheathing avoids the ¹⁴N (p, n)¹⁴O reaction that could occur if the proton beam were to pass through the air. The target holder is connected to the chemical separation unit via a fitting from RS Components (RS stock no. 419-7259) to allow the liquid to be delivered to the target holder when in dissolution configuration. The liquid can be delivered into the 3D target holder via a peristaltic pump (Thomas, part no. 2030104), connected to an Arduino (RS stock no. 715-4084), so the number of steps and the direction (clockwise/anticlockwise) can be chosen to determine the amount of liquid added or removed. The amount of acid or water can be determined by coding, whereby a specific amount of liquid can be delivered automatically for the dissolution step or for the other steps in the chemical separation process.

To prepare the target for irradiation, a 12-mm 89Y-disc (Goodfellow special metal with 99% purity; pro e: Y-00-FL-000110) is punched from a 0.125-mm-thick Y-foil. Before irradiation, the Y-disc is spot welded to a tantalum sheet from Goodfellow (99.9% purity, product code: TA00-FL-000183) measuring 50 × 50 mm², 0.5 mm thick, utilising a resistance spot welder fitted with copper electrodes. The welded Y disk on the Ta is glued using silver paste (Agar Scientific Ltd., Rotherham, UK; catalogue no. AGG3304) to a cool plate (Mouser, part no. 984 ATS TCP1000). The pair of plates of the 3D target holder are held together with pairs of bolts and nuts, one pair for each of the four corners. Once assembled, the target holder is transferred and mounted in front of a vacuum-air exit window on the accelerator beamline, using a repurposed 3D printer, described below. The target can be cooled to remove the heat from the proton beam by passing chilled water through the cool plate connected to a KI chiller (Applied Thermal Control Ltd., Leicestershire, UK; part no. K117000) and on the front side by helium gas, which passed through the top plate of the target holder. This chiller has a rated cooling capacity of 1750 W and maintained the coolant at 5 °C during irradiation to ensure thermal stability.

A low-cost Clarity Ender-3 3D printer from Box Co. (part no. 5056143614876) was repurposed as a translator to move the target holder in the X and Y directions, i.e., horizontally and vertically perpendicular to the beam. This facilitates automated alignment of the target before irradiation begins. The extruder part of the 3D printer was replaced with a 3D-printed U-shaped holder to support the target holder assembly as is illustrated in Figure 5. This holder was constructed from a 6.35 mm aluminium rod, which passes through the bottom plate of the target holder and is connected to the 3D U-shaped holder and a servo horn from Farnell (part no. 3769950). The servo horn is connected to the Al rod via a small 3D-printed piece used as a connector between the aluminium rod and the servo motor. This motor can efficiently rotate the entire target holder through 90 degrees, rotating the target holder so the target foil is vertical during irradiation and horizontal during its dissolution. The radiation dose rate at the electronics location was measured to be well below 1 mGy/h, primarily due to secondary positron and gamma emissions, ensuring minimal exposure to sensitive components during operation. The control of that translator and the servo motor is performed automatically by connecting each stepper motor of the translator to a controller from UI ROBOT (part no. UIM240L02) and then to an Arduino board (RS Stock No. 715-4084) so the target’s orientation can be automated and controlled via a software interface.

2.2. Chemical Separation Unit

An automated synthesis unit was designed so that it all fits into a 10 × 10 × 10 cm³ cube for the automated separation and purification of ⁸⁹Zr. By incorporating different resins, the unit could also be used to separate various target material/medical isotope pairs. Its design is shown in Figure 6. The unit comprises six peristaltic pumps from Thomas (part no. 2030104), six controllers from URROBOT (part no. UIM24002/04/08), a reaction vial, and two vials for the final product (⁸⁹Zr-oxalate) and any waste liquids. All vials were manufactured from borosilicate glass. A ready-to-use commercial hydroxamate resin was obtained from Triskem©, and a pinch valve was procured from OEM AUTOMATIC (part no. S204 07-ZE30A).

All liquids are transferred though PHARMED^® tubing (ID:2.54 mm, OD:4.24 mm, from VWR, part no. 228-1708) and controlled via peristaltic pumps. The hydroxamate resin and all liquids used in the ⁸⁹Zr production are preloaded before transmutation, in a non-active environment. The unit can be located within a lead hot cell with the reagents (HCl, H₂O, and oxalic acid) and their peristaltic pumps being located outside the hot cell. It operates automatically through an Arduino-controlled system (RS Stock No.: 715-4084). A schematic diagram of the separation unit is shown in Figure 7. The chemical separation unit is installed in a Hot Cell H600 from Trasis (Product No: S12569) and operates under fully automated control. The separation and dissolution procedures described here were performed using the automated chemical separation unit illustrated in Figure 7.

2.3. Photolabelling, Vortex Mixing, and Controlled Heating Unit

This unit was fabricated using rapid prototyping techniques, specifically 3D printing and laser cutting. The materials utilised in its construction include PETG (polyethylene terephthalate glycol) for the 3D printed components, produced using a Pursa Mk4s printer, and acrylic (Perspex) for the Laser cut parts, processed with an HPC C02 laser cutter. It incorporated three main functions: vortex mixing, photolabelling, and controlled heating. The core of the unit is a custom-designed glass beaker (reaction vessel), crafted to support each function with carefully integrated components tailored for specific applications. The outer casing design is shown in Figure 8.

In the sections below, each function is illustrated with its respective components and their specific uses:

2.4. Vortexing and Controlled Heating

The vortex-mixing and controlled heating functions were designed to operate together or independently, providing efficient mixing and heating capabilities tailored to the specific chemical reaction, incubation time, or synthesis requirements. The design features a cylindrical heating block with an outer diameter of 62 mm, a height of 35 mm, and a 14 mm cutout section. This cutout enables UV lights (described in the subsequent section) to reach the reaction vessel effectively, ensuring uniform exposure during photochemical reactions. The heating block houses a thermistor from E3D (model no: E-SEMITEC-50-MOLEX-INC-CABLE), embedded within the structure for precise temperature regulation (Figure 9A).

The vortex mixer has been successfully operated at temperatures as high as 95 °C with water. The heater can raise the temperature to approximately 140 °C for any liquid that does not boil before this point and is compatible with quartz glass. The reaction vessel is custom-made with dimensions of 124 mm in height, a top circular opening of 30 mm in diameter, and a narrow bottom opening of 3 mm in diameter for connection to a peristaltic pump (Figure 9A). This narrow bottom opening offers benefits compared to V-shaped flasks typically used as no liquid is left behind upon its removal. Because the bottom of this flask is connected to a peristaltic pump via narrow flexible tubing, liquid collects, with an airlock below it, at the bottom of this vessel to allow heating, mixing, or photolabelling to take place.

The stirring assembly is integrated beneath the reaction vessel. The 3D-printed stirring assembly houses two magnets (10 × 3 mm²) for efficient liquid mixing. The stirring process is powered by a 3500 RPM motor from Maxon (part no: 642221), ensuring consistent agitation of the reaction mixture (Figure 9B).

Liquid addition and removal are facilitated by a peristaltic pump from Thomas (part no. 2030104) positioned adjacent to the stirring assembly. The pump connects to the reaction vessel via the 3 mm bottom opening, enabling precise and efficient liquid transfer during operation. The assembly of the components is shown in Figure 9C.

2.5. Photolabelling

The photolabelling function enables the labelling of antibodies to radioactive substances, depending on the labelling condition requirements, with a design that incorporates several key features. Three Luminus UV-C High Power Starboards from (MOUSER ELECTRONICS, Mfr No: LST1-01G08-UV01-01) are mounted vertically on a machined copper plate, which facilitates heat dissipation. The plate is mounted using 3D-printer mounts measuring 20 × 25 mm². These UV lights are secured with six hexagonal socket head cap screws (M2.5 × 0.45 × 5 mm², steel grade 4.6, plain) from RS (Part No: 293-303), along with nuts to ensure stability. For effective cooling, a CPU cooler from The Pie Hut (Sudbury, UK; SKU: EP-0107) is mounted at the back of the 3D-printed holder. This cooler is mounted to the rear side of the LED heatsink plate using four hexagonal socket head cap screws (M2.5 × 0.45 × 5 mm², steel grade 4.6, plain) from RS (Part No: 293-303), also with nuts for secure attachment. The CPU cooler is secured using 3D-printed top and bottom mounts, measuring 20 × 25 mm², and 40 × 16 mm², respectively. These mounts ensure stable integration within the unit alongside other components. Each mount is attached using two hexagonal socket button screws (M4 × 20 mm) from RS (Part No: 822-9108) (Figure 10).

The UV lights directly face the reaction vessels and heating block, where the photolabelling is conducted. The vessel is made of quartz glass to facilitate good penetration of the UV light. To enhance cooling efficiency, the middle section of the unit integrates a heat core cover equipped with bilateral fans (40 × 40 mm²) from RS (Part No: 668-8801) on both sides. These fans are positioned to provide additional airflow and thermal regulation for the entire setup, ensuring consistent performance during operation (Figure 11).

All operations, including vortex mixing, controlled heating, and photolabelling, are automated and managed through an electronic board mounted vertically next to the system (Figure 11). The automation is controlled by Python-based control code. The photolabelling hardware was validated in this work; a full photolabelling experiment is planned for future inclusion.

3. Results

3.1. Irradiation

The nuclear reaction ⁸⁹Y (p, n)⁸⁹Zr was employed to produce ⁸⁹Zr. This nuclear reaction is commonly used because of the high ⁸⁹Zr yield that can be produced from naturally isotopically pure ⁸⁹Y [10,11,12]. To prevent the synthesis of considerable amounts of long-lived ⁸⁸Zr as a side product, the proton energy must be kept below 13 MeV [13]. During the irradiations two thermal cameras were used to measure the temperature.

A 5-MV tandem ion accelerator (NEC Pelletron 15SDH-4) was used for irradiation. In an initial test, the 12-mm ⁸⁹Y-disc was irradiated for a total of 30 min (in three 10 min fractions) at an initial proton energy of 8.89 MeV. The total integrated current delivered during the irradiation was estimated to be between 180 µC and 1800 µC, producing ⁸⁹Zr via the ⁸⁹Y (p, n)⁸⁹Zr nuclear reaction. During irradiation, the target was cooled on the back side by chilled water through the CPU chiller plate connected to a chiller (Applied Thermal Control Ltd., part no. K117000) and on the front side by helium gas, which passed through the top plate of the target holder. Figures S1–S3 in the Supplementary Information illustrate the setup of the irradiation apparatus.

The irradiated Y disc is shown in Figure 12. A high-resolution thermal camera was used to image the local thermal environment of the Y-foil. It can be seen being heated by the ion beam, as the black/purple region in Figure 12A. A lower-resolution thermal camera (FLIR model no: KIT-15948) was used to take a thermal image with a wider field of view. Three thermocouples were additionally attached to the edge of the Y-disc during some test irradiations as shown in Figure 12B.

A new vacuum window has since been commissioned, enabling safe irradiation at a beam current of 50 µA. Based on the current yield estimates, this would allow the production of approximately 50 GBq/h of ⁸⁹Zr, subject to internal permit limits. This output would be sufficient to support clinical-level radiotracer production, demonstrating the system’s scalability and its potential for routine implementation in radiopharmacy applications.

The thermal camera and the thermocouples showed the temperature to be 5 °C or less. Analysis of the data from these cameras tells us that the target head was always at a temperature of below 40 °C across the entire Y-foil. The largest thermal increase observed at the back of the Ta sheet was less than 1 °C for a 1 µA beam current. Since this temperature will rise sub-linearly with beam current and the whole system can withstand temperatures well in excess of 200 °C, this demonstrates that the target holder design is fit for much higher beam currents. These higher beam currents will be realised once we fit a new vacuum window which is currently under construction.

After irradiation, the target was placed against a high-purity germanium (HPGe) detector (Ortec, Oak Ridge, TN, USA) for gamma-ray quantification. Spectra measured using the detector are shown in Figure 13 and Figure 14.

The spectrum presented in Figure 13 was collected 50 min after partial irradiation of the target to verify the expected transmutations. It shows the presence of 511 keV and 909 keV (⁸⁹Zr), and 588 keV (^89mZr) gammas. Figure 13 shows a spectrum collected 41 h after the end of irradiation, and as expected the peak related to the ^89mZr contaminants has completely disappeared, while the longer-lived ⁸⁹Zr (511 keV and 909 keV) remains. Due to proton-induced reactions on yttrium targets at energies relevant to the production of ⁸⁹Zr, two Zr contaminants, ^89mZr (t_1/2 = 4.2 m) and ⁸⁸Zr (t_1/2 = 83.4 d), could be formed from the target irradiation. These are chemically indistinguishable from ⁸⁹Zr so cannot be separated. Due to its short 4.2 min half-life ^89mZr, it will entirely decay within 1 h and hence is not regarded as a problematic contaminant [2,14]. To allow decay of short-lived isotopes, the target is given an hour to cool after the end of the irradiation (EOI) [2]. The amount of ⁸⁸Zr is small (0.0005%) at the energies ranging from 11 MeV to 15 MeV, generally used for ⁸⁹Zr synthesis [13], and is expected to be even less at the energy used here. We detected a weak signal from ^89mZr, 50 min after irradiation, using an HPGe detector, as shown in Figure 13.

3.2. Zr/Y Separation

The Zr separation was examined using non-active materials rather than radioactive ⁸⁹Zr, due to ion accelerator upgrades precluding further transmutation and to establish the procedure safely without any radioactive species being present. These non-radioactive experiments were conducted as proof-of-principle studies, allowing optimisation and validation of the automated process in preparation for future radioactive runs once irradiation access is restored.

A simulant solution of Zr:Y (9:500) atom ratio was employed to assess Zr separation. The quantity of the ⁸⁹Zr radioisotope produced in the irradiated yttrium target was estimated using this ratio. A total of 10 g of yttrium chloride (YCl₃; 99.5%, trace metal basis, Sigma-Aldrich, St. Louis, MA, USA) was dissolved in 25 mL of 32 wt % hydrochloric acid (HCl; FCC grade, Sigma-Aldrich) and 25 mL of water to create an yttrium stock solution. A volumetric flask was filled to capacity with 25 mL of yttrium chloride and 0.0881 g of solid Zirconium tetrachloride (ZrCl₄; ≥99.5%, trace metal basis, Sigma-Aldrich UK) was added, with a final concentration of 8 mM ZrCl₄/0.5 M YCl₃ in the mixture. Throughout the ZR resin (Triskem©) separations, a peristaltic pump with a continuous flow of 1 mL/min (4 rpm) and PHARMED^® peristaltic pumps tubing were used.

The approach reported by Larenkov et al. [15] for Zr-Y separation was conducted using the ZR resin, as is summarised in Figure 15. Briefly, 1 mL of ZR resin was primed in a 1 mL cartridge with a 2 M HCl initial load using commercial hydroxamate resin purchased from Triskem© (Step 1 in Figure 15). In a second step, 1 mL of material was loaded and then washed with 3 mL of 2 M HCl and 3 mL of water to eliminate Y (Step 3). Zr-oxalate was eluted using 0.1 M or 1 M oxalic acid (COOH)₂ (89%, Sigma-Aldrich) in Step 4. Next, 12 fractions of 1 mL each were recovered after washing with deionised water and a second separation with ZR resin using 1.0 M oxalic acid as the Zr eluent. Oxalic acid separations at 0.1M and 1M were acquired for optimisation of Zr extraction.

The oxalate and chloride salt of zirconium were precipitated from the pooled fractions after application of drops of 20% w/w NaOH until pH > 10. White sol-gel precipitate formed in fractions of 2–5 (Y wash step, 2–5 mL total eluent quantity from the commencement of separation) and 10 (Zr elution step, 10 mL total eluent amount from the start of separation). In addition, 1.0 M oxalic acid separation fractions (10–12) contained soluble Na–oxalate salt that was removed by rinsing with Milli-Q^® water as the pH approached 7, leaving purified sol-gel of the Zr oxalate. Fraction 2 (Y wash) and fraction 10 (Zr elution) drops from each experiment were placed on aluminium studs with a carbon tape coating and allowed to dry in the fume hood. To assess the Y-to-Zr ratio contained in these samples, they were analysed on an SEM (Scanning Electron Microscope) (model no: Quanta 250 FEG) using an EDX detector. In addition, these samples were dissolved in 2% HCl or 2% HNO₃ at a concentration of 0.5 mg per mL for ICP-MS analysis.

No yttrium was found in the eluted zirconium fractions, even though yttrium was present in a significant excess ratio to zirconium. Notably, some zirconium elution with yttrium occurred in the Y wash fractions. Repeating the separation process on these fractions is necessary to prevent radionuclide loss in this step. The ICP-MS results (Figure 16) confirm the SEM findings of the ZR:Y ratio. There were low amounts of yttrium relative to zirconium in the Zr-elution fractions of ZR resin separations (fraction 10).

These results demonstrate the optimisation and adaptation of ZR resin in ⁸⁹Zr separation from ⁸⁹Y in our automated chemical separation unit. According to the comparison of the various oxalic acid concentrations used for elution, a larger concentration does not necessarily result in zirconium separation with higher purity. A lower initial concentration of oxalic acid is more acceptable for radiopharmaceutical usage since oxalate must be further separated from ⁸⁹Zr due to its toxicity [16].

3.3. Photolabelling, Vortex Mixing, and Controlled Heating

This automated system successfully integrated vortex mixing, controlled heating, and photolabelling, ensuring precise control over reaction conditions. The heating block, with its cylindrical design, regulated temperature effectively through an embedded thermistor, allowing for consistent and accurate heating.

The reaction vessel’s custom design (Figure 9B) facilitated efficient liquid transfer, with its narrow bottom opening preventing residual liquid retention, which is commonly observed in traditional V-shaped flasks.

The stirring assembly, equipped with a high-speed Maxon motor and dual magnets, achieved uniform mixing, enhancing reaction homogeneity. The peristaltic pump system demonstrated efficient liquid handling, enabling precise addition and removal of reagents without significant loss.

The system was tested using aqueous solutions and dilute mineral acids. Peristaltic pumps were calibrated to handle up to 50 mL per cycle, ensuring accurate and repeatable liquid transfer.

The UV photolabelling setup functioned optimally, with three Luminus UV-C High Power Starboards mounted to provide direct UV irradiation to the reaction vessel. The cooling system, including a CPU cooler and bilateral fans, successfully maintained stable thermal condition during UV exposure, preventing overheating and ensuring reproducible labelling efficiency.

Automation of the entire system using Python-controlled electronics enabled fully automated, seamless operation, reducing manual intervention and improving process reliability. The integration of multiple functions within a compact system demonstrated the feasibility of a fully automated chemical synthesis and photolabelling unit.

4. Discussion

4.1. Target Holder

Target holders frequently come in the form of stainless steel or aluminium. Aluminium is a popular material choice because of its low neutron absorption cross-section and the short half-lives of the radionuclides produced from aluminium. Additionally, because aluminium has strong thermal conductivity, heat generated inside the target can quickly be transported to the cooling system [17]. However, these metals are not suitable choices for our approach as they are susceptible to attack by the acid used in the dissolution step.

Previously, aluminium or niobium holders have been employed to hold metal targets, such as yttrium foil [2]. Alternatively, yttrium can be deposited onto copper or niobium substrates [14,18]. These target holders have been used by several research groups [10,17,19,20]. Although these substrates can be used as target holders, they must be moved manually post-irradiation to the hot cell for the chemical separation process, which does not lend itself to automation. A solid target system has been reported that can be used for remote irradiation and dissolution. However, no chemical separation unit is included, and it is expensive [7]. Therefore, we used a low-cost 3D-printed target holder printed from polypropylene. To the best of our knowledge, no one has previously used a 3D-printed combined target holder and dissolution unit. The filament we used was made from PP, which can withstand heat and acid [21].

Our target has several advantages, which include holding the ⁸⁹Y, with efficient thermal coupling to the cool plate connected to a chiller where it cools the ⁸⁹Y disc spot welded to a Ta sheet by thermal conduction. It also contains a pathway for helium gas for cooling the target from the top surface of the ⁸⁹Y disc and it has a pathway for liquid flow, permitting dissolution of the target post-irradiation and then pumping the dissolved sample directly into the hot cell for the chemical separation process.

This approach was adapted from Ellison et al. with some modifications. These modifications include automated target transfer after irradiation to the hot cell, as it is connected to the chemical separation unit in the hot cell and the target dissolution step occurs in the target holder. Moreover, they used direct chilled water on the Y spot welded to the Ta sheet where chemical attack risks might occur. However, we used a commercial, low-cost cold plate where we glued the spot-welded Y to the Ta. Tantalum as a substrate seems to be a good choice for our design because it has a high melting point (3017 °C) [22] and resistance to high concentrations of HCl [16]. The cooling system we use cools efficiently. The temperature rise when the disk was irradiated was small. We did not observe damage to the ⁸⁹Y target disc, as the ⁸⁹Y was cooled by chilled water connected to the chiller and by helium gas from the top surface of the target holder.

4.2. Chemical Separation Unit

⁴⁸V, ¹⁵⁶Tb, ⁶⁵Zn, and ⁵⁶Co are additional long-lived radioactive pollutants that could be created from contaminating metals in the ⁸⁹Y target and target holders [23]. However, the hydroxamate resin used in ⁸⁹Zr purification has low affinity for these metals which consequently wash out of the column [13]. Accordingly, they can be simply removed from the desired product during the target’s processing in the separation and purification of the target. The naturally abundant ⁸⁹Y target we use is available from a variety of commercial providers as foil of various thicknesses. It is also possible to use the sputtering technique to deposit yttrium onto a substrate [14,18]. Moreover, when the cyclotron cannot irradiate solid targets, liquid targets that use yttrium solutions, such as Y(NO₃)₃ and YCL₃, have been used [13]. Interestingly, the liquid target has the advantage of not requiring target dissolution in the chemical separation process post-irradiation but due to the lower target atom density the production of ⁸⁹Zr is considerably less efficient.

The short column of the ZR resin helps keep the design compact. ZR resin is also more suited for future optimisation because the shorter column allows quicker separation. Separation was accomplished in less than 15 min. Additionally, the manufacturer’s approach did not require a lengthy equilibration process for the resin, making its handling safer and suitable for automation. The timing is determined by the capacity of the ZR resin cartridges, which hold only 1 mL of material, and the flow rate of the peristaltic pumps, which transfer fluid at 1 mL per minute. This slow transfer rate is due to the compact design of the peristaltic pumps. However, these pumps appear to be ideal for this application. During calibration, we found that they could consistently deliver liquids with a precision of 1 microlitre.

4.3. Photolabelling, Vortex Mixing, and Controlled Heating

Our compact, fully automated system eliminates the need for intermediate purification and manual transfers, ensuring a seamless and reproducible workflow while reducing radiation exposure for operators, aligning with previous findings [1,24,25].

While Klingler et al. [25] developed a photolabelling module tailored specifically for ⁸⁹Zr-labelled antibodies, our system differs in being part of a modular platform that integrates irradiation and dissolution (within the target holder), followed by chemical separation and photolabelling in dedicated units. The photolabelling module itself is compact, versatile, and adaptable to multiple isotopes and reaction conditions. This modularity enhances flexibility and allows streamlined configuration for various radiopharmaceutical applications.

Ultimately, this automated unit provides a robust, efficient, and reproducible platform for chemical synthesis and photolabelling, offering a solution for high throughput labelling in both clinical and research applications.

5. Conclusions

We have reported on the design of a compact, fully automated system capable of starting with a foil target and carrying out the transmutation, dissolution, and chemical separation. We have demonstrated the application of the system using an yttrium foil target to produce, in liquid form, purified ⁸⁹Zr all without the need for human presence in the same room. The system exploits readily available, low-cost components such as a CPU chiller, peristaltic pumps, and a repurposed 3D printer as an X–Y translator, with bespoke component parts predominantly being manufactured using 3D printing.

The vortex-mixing function allows for efficient reagent homogenisation, which is a crucial step in many labelling and synthesis processes. The photolabelling setup features high-power UV LEDs positioned to maximise light exposure to the reaction vessel, ensuring efficient conjugation. By integrating multiple critical functions into a single platform, this system represents a step forward in automating radiopharmaceutical synthesis.

In a follow-on project, the system has been further upgraded and successfully used to produce medical isotopes of copper and scandium, with complete test runs of the full process demonstrated for both isotopes. These upgrades and additional innovations will be reported in a future publication.