**1. Introduction**

Zirconium-89 (89Zr), a positron-emitting isotope, has emerged as an attractive radionuclide in the development of pre-clinical and clinical radiopharmaceuticals for positron-emission tomography (PET) imaging. This is due to the favourable physical characteristics of the isotope which decays via positron emission (*T*1/2: 78.4 h, β <sup>+</sup> 22.7%, Eβ+max = 901 keV; average Eβ+max = 396 keV) [1,2] and electron capture (EC 77%, Eγ = 909 keV) to the stable yttrium-89 (89Y). The longer half-life matches the need for large biomolecules such as antibodies, antibody fragments and nanoparticles which require prolonged circulation time in order to reach optimal target accumulation [3,4]. Furthermore, the β + branching ratio, average β <sup>+</sup> energy and short positron range (Rave = 1.23 mm) provide high spatial resolution and image quality of the <sup>89</sup>Zr PET probes [5].

The combination of antibody-based targeting vectors and PET-based imaging, known as immuno-PET, is rapidly becoming a powerful tool for highly selective imaging agents [6,7]. Radiolabeling of the targeting vector requires a stable radiometal chelate that can be readily conjugated

to an antibody or other proteins. At present, the bifunctional derivative of desferrioxamine (DFO), *p*-isothiocyanatobenzyl-desferoxamine (*p-SCN-Bz*-DFO) is considered the gold-standard and with facile reaction chemistry has had widespread application for preclinical research and clinical trials [8]. However, recently there have been some concerns surrounding the in vivo instability of the <sup>89</sup>Zr-DFO complex, presumably due to the unsaturated coordination sphere, which has led to the generation of new chelators with enhanced stability of the resulting <sup>89</sup>Zr complex [9–11].

The most common <sup>89</sup>Zr production method is via the <sup>89</sup>Y(p,n)89Zr transmutation reaction. This route, employing <sup>89</sup>Y targets that is in 100% natural abundance and commercially available, is readily accessible by most medical cyclotrons. In general, the target is irradiated with incident proton beam energies of 13–16 MeV. At these low energies, there is a trade-off between radionuclidic purity and efficiency of the process owing to the competing reactions to produce <sup>88</sup>Zr via the <sup>89</sup>Y(p, 2n)88Zr reaction and <sup>88</sup>Y via the <sup>89</sup>Y(p, pn)88Y reaction at threshold energy of 13.3 MeV. The average maximum cross section value for <sup>88</sup>Zr is 786.9 <sup>±</sup> 1.5 at 22.98 MeV and for <sup>88</sup>Y is 298.0 <sup>±</sup> 5.7 mb at 28.30 MeV [1]. The production of <sup>88</sup>Zr and <sup>88</sup>Y is minimized at 12.8 MeV due to small cross sections and threshold energy. Following dissolution of the irradiated target in concentrated HCl the isolation and purification of <sup>89</sup>Zr proceeds via cation exchange chromatography using hydroxamate-modified resin [1,12]. Zirconium-89 is selectively eluted from the column using 1 M oxalic acid. With optimized conditions, this standardized method of separation can achieve high recovery, radionuclidic purity and effective specific activity (ESA) [1].

The increased interest in <sup>89</sup>Zr-immuno-PET imaging probes for use in preclinical and clinical studies has led to a rising demand for <sup>89</sup>Zr. Although <sup>89</sup>Zr is commercially available, the lengthy transportation time and high costs are a barrier for preclinical and routine clinical applications of <sup>89</sup>Zr immuno-PET. Furthermore, the highly penetrating 511 and 909 keV photons emitted by <sup>89</sup>Zr deliver an undesirably high radiation dose, which makes it difficult to produce large amounts manually. So far, the processing of <sup>89</sup>Zr from the target has mainly been achieved manually or has been based on semi-automated systems. These have not been widely adopted because they are based on in-house developed systems instead of commercially available equipment [13–15]. Moreover, regulatory demands are increasing and consequently the application of <sup>89</sup>Zr in clinical studies is becoming challenging. To maintain the high quality of the <sup>89</sup>Zr, a reliable process with quality control and good manufacturing practice-compliant automated module in the production is required. Additionally, there are several advantages of using automation in radiochemistry, such as lower radiation exposure to the production personnel, high reproducibility, more precise control of the parameters and ease of handling. Mostly, automated modules are placed inside dedicated lead shielded hot cells to protect the operator and environment from radiation exposure. Commercially available automated modules can be classified into two types, namely, those that consist of fixed tubing in which fluid flows are regulated by inert gas, and those that consist of sterile disposable cassettes where fluid flow is regulated by a syringe pump. The main disadvantage of using fixed tubing is bacterial contamination in the product if a validated cleaning method is not implemented. Usually this tubing is used for multiple productions and needs to be cleaned after each synthesis. Also, a single system cannot be used for multiple radioisotopes or for the production of other radiopharmaceuticals. These issues can be solved by using sterile, single-use cassettes manufactured in current Good Manufacturing Practices (cGMP)-compliant clean rooms.

The focus of this present work is the implementation of the semi-automated purification of <sup>89</sup>Zr using a commercially available automated synthesis unit that could fulfil regulatory requirements.
