Production and Purification of ¹⁶⁵Er from Pressed Ho₂O₃:Al Targets on a 16.5 MeV Cyclotron

Abstract

Erbium-165 (¹⁶⁵Er) is an Auger electron emitter with 7.2 electrons per decay and very few other emissions, making it an interesting candidate for Auger electron therapy. We present here a procedure for producing ¹⁶⁵Er by the ^natHo(p,n)¹⁶⁵Er nuclear reaction on a 16.5 MeV medical cyclotron. The target was prepared by pressing a Ho₂O₃:Al 1:1 (w/w) powder mixture on a Ag disc with a cylindrical depression in the center. With a 0.1 mm Nb foil in front, degrading the energy to 15 MeV, and water cooling at the back of the Ag disc, the target could withstand irradiation at currents up to 45 µA without showing any signs of damage. The beam tolerance of the target was also estimated by calculating the temperature and heat dissipation in the target via the numerical solution of the heat transport equations. For a 180 mg target, the production yield was 12.3 ± 1.9 MBq/µAh. The separation of two neighboring lanthanides is challenging, which led us to study the distribution coefficients for Er and Ho on commercially available LN2 resin for both HNO₃ and HCl eluents. Based on these values, we propose a purification procedure involving two successive LN2 columns for separating the ¹⁶⁵Er from Ho and Al, followed by a small TK221 column to concentrate the final eluate. No radionuclidic impurities were detected, and the chemical impurities found in the final formulation were traces of Ho, Er, Ca, Pb, and Fe. For three different chelators (DOTA, DTPA, and CHX-A″-DTPA), the effective molar activity of the final formulation was measured. The stability of the three complexes formed was also assessed upon incubation in mouse serum for 28 h.

1. Introduction

Auger electron (AE) emitters are increasingly attracting attention due to their potential for targeted radionuclide therapy. The high energy deposition within a very short range associated with an AE cascade holds good promise for efficiently killing targeted cancer cells while sparing the surrounding healthy tissues [1]. The therapeutic properties of AEs are believed to be especially suitable for the elimination of single cells and small metastatic tumors. The efficient elimination of such cancer cells could greatly increase the outcome of targeted radiation therapy by potentially lowering the risk of tumor regrowth and relapse of the disease after treatment [2]. This is supported by the fact that Auger and conversion electrons were shown to have a positive effect on the therapeutic efficacy of terbium-161 (¹⁶¹Tb), compared to the more commonly used β^- emitter luthetium-177 (¹⁷⁷Lu) [3,4]. Both radionuclides have similar half-lives and emit β^- radiation of a similar energy range, but ¹⁶¹Tb emits a considerable amount of Auger and conversion electrons compared to ¹⁷⁷Lu. ¹⁶⁵Er (T_½: 10.36 h) decays by electron capture and is an AE emitter with 7.2 electrons per decay [5], with very few other emissions. It does not emit any γ-rays nor conversion electrons upon decay and only low-energy X-rays [6,7]. This is an advantage in terms of a decreased dose to surrounding healthy tissue and less exposure to the personnel handling the radiopharmaceutical. The emission of γ-rays with an energy of 60–300 keV can be advantageous for dosimetry calculations and single-photon emission computed tomography (SPECT) [5], and a recent study has demonstrated that the X-ray emissions of ¹⁶⁵Er were sufficient to perform SPECT imaging in small animals, making it a potential theranostic radionuclide for translational AE therapy studies [8]. ¹⁶⁵Er has also been considered for bimodal imaging as an analogue to gadolinium [9]. An additional advantage of ¹⁶⁵Er is that the chelator, DOTA, can be labeled, which is commonly used for radiopharmaceuticals [8].

¹⁶⁵Er can be produced by the (p,n) nuclear reaction from natural Ho, which is monoisotopic. The maximal cross section for the ^natHo(p,n)¹⁶⁵Er reaction is around 160–180 mb in the 10–11 MeV range [10,11,12,13], enabling the ¹⁶⁵Er production on a 16.5 MeV medical cyclotron. Another production route that has been reported is via the ^natEr(p,xn)¹⁶⁵Tm reaction. With its 30.06 h half-life, the thulium-165 (¹⁶⁵Tm) can act as a ¹⁶⁵Er generator afterwards. The advantage of the ^natEr(p,xn)¹⁶⁵Tm production route is the higher cross-section of the reaction with a maximum around 450 mb at 23 MeV [14]. It, however, leads to the production of a mixture of Tm radionuclides. The use of an enriched ¹⁶⁶Er target and subsequent ¹⁶⁶Er(p,2n)¹⁶⁵Tm nuclear reaction can reduce the amount of produced Tm isotopes [15,16]. The use of a deuteron beam to produce ¹⁶⁵Er by the ¹⁶⁵Ho(d,2n)¹⁶⁵Er nuclear reaction has also been reported. The maximal cross section is close to 600 mb, but this route requires cyclotrons with higher energy than for the (p,n) reaction route, which are less available [17].

The most used target material for producing ¹⁶⁵Er by the ^natHo(p,n)¹⁶⁵Er reaction is metallic Ho, mostly as foil [9,12,18] but also as pressed powder [12] or pressed and melted chips [19]. However, Er is a common impurity in metallic Ho, which cannot be chemically separated from ¹⁶⁵Er, leading to reduced molar activity. A pre-purification of the target material to remove Er and a regeneration of Ho in metallic form are difficult. Ho₂O₃ is an alternative and has the advantage of being available in high purity (99.999%). It is also possible to purify the target material and recover the Ho₂O₃ [12]. The properties of Ho₂O₃ as a cyclotron target material are not as suitable as metallic Ho, due to a lower thermal conductivity and the presence of oxygen that will stop some of the proton beam and decrease the ¹⁶⁵Er production yield. To increase the mechanical properties and thermal conductivity of a pressed powder target, we have previously added Al powder to a [¹³⁵Ba]BaCO₃ target for lanthanum-135 (¹³⁵La) production with success, making it possible to increase the beam current on the target [20]. Al was chosen due to its relatively high thermal conductivity, low stopping power, and the fact that it is a soft metal, making it possible to press the target material in the Ag discs used as target backing. Additionally, Al can be chemically separated from lanthanides, and no long-lived radionuclides are produced by the (p,n) nuclear reaction.

In this work, we describe the cyclotron production of ¹⁶⁵Er from pressed targets made of a combination of Ho₂O₃ and Al powder and study the stability and beam tolerance of the target experimentally and theoretically by partial differential equations. The separation of two neighboring lanthanides comes with some challenges due to their similar chemical behaviors. The distribution of Er and Ho on commercially available LN2 resin was measured, and the ¹⁶⁵Er purification was performed on three successive columns that led to a purity suitable for radiolabeling. The apparent molar activity was determined from inductively coupled plasma optical emission spectroscopy (ICP-OES), and the effective molar activities were measured by titration with the DOTA, DTPA, and CHX-A′′-DTPA chelators due to their known ability to form stable complexes with the lanthanides [21,22,23]. The stability of the three ¹⁶⁵Er complexes was then assessed after incubation at 37 °C in mouse serum over 28 h.

2. Materials and Methods

The chemicals were purchased from Merck (Darmstadt, Germany) unless otherwise specified and used without further purification. Ho₂O₃ of high purity (99.999%, trace rare earth analysis ≤15.0 ppm, Sigma-Aldrich (St. Louis, MO, USA)) and HCl and HNO₃ of “Suprapur” quality were used for the experiments. The Al powder (99.9%, 60 µm max. particle size) and Nb foil were purchased from Goodfellow (Huntingdon, England). The water was of 18 MΩ MilliQ-grade (Roskilde, Denmark), except for the labeling experiments, in which case ultra-trace water was used. 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA), diethylene triaminepentaacetic acid (DTPA), and p-NO₂-Bn-CHX-A″-DTPA ([(R)-2-Amino-3-(4-nitrophenyl)propyl]-trans-(S,S)-cyclohexane-1,2-diamine-pentaacetic acid (CHX-A′′-DTPA)) were purchased from Macrocyclics (Plano, TX, USA). Mouse serum was purchased from Biowest (Nuaillé, France). The targets were irradiated on a GE PETtrace cyclotron with a 16.5 MeV proton beam. For the purification, the LN2 (50–100 µm), branched DGA (50–100 µm), and TK221 (50–100 µm) resins were purchased from TRISKEM (Bruz, France). The columns were run using a BT100-2J LongerPump peristaltic pump (supplied by Drifton, Hvidovre, Denmark). Non-radioactive metal impurities in the final sample were quantified by ICP-OES with a Thermo Scientific iCAP 6000 and 7000 Plus Series ICP-OES apparatus (supplied by Thermo Fisher Scientific, Roskilde, Denmark), and Er, Ho, and multielement 1 ICP standard solutions were used for the quantification. Sample activities were measured using a Canberra GC 2020 high-purity germanium (HPGe) detector gamma spectrometer (Olen, Belgium), a Canberra GL 0055P X-ray detector (Olen, Belgium), and a CRC-55tR, CII Capintec, Inc., (Florham Park, NJ, USA) dose calibrator. The HPGe detector for gamma spectrometry was calibrated with certified ¹³³Ba and ¹⁵²Eu sources and the X-ray detector with certified ²⁴¹Am and ²¹⁰Pb sources. Spectra were analyzed using Genie 2000 software. For the effective molar activity measurements, the thin-layer chromatography (TLC) plates (TLC Silica gel 60 F₂₅₄ plates (aluminum sheets), Supelco, Darmstadt, Germany) were read using a Perkin Elmer Cyclone^® Plus Storage Phosphor System apparatus (Perkin Elmer, Singapore).

2.1. Target Preparation and Irradiation

For the production of ¹⁶⁵Er, different mixtures of Ho₂O₃:Al (without Al, 7:3, 1:1, and 1:2 w/w, 80–270 mg total weight) were pressed in a Ag disc (29 mm diameter, 5 mm thickness) with a cylindrical well in the center (9 mm diameter, 3 mm depth). A hydraulic press (1 ton) was used with a matching stainless steel pressing tool. The Ag disc was placed in a target holder with water cooling on the back of the disc. A 0.1 mm-thick Nb foil (45–50 × 45–50 mm dimensions) was placed in front of the target, degrading the incoming proton beam to 15.0 MeV. For target testing and work on the purification procedure, low activity batches were produced by irradiating the target for 10–15 min at 10–50 µA. For high activity batches, targets were irradiated up to 3 h at 25 µA. To be able to track both Ho and Er during optimization of the purification procedure, we also used radioactive Ho made on-site by neutron activation. For this, 100 mg Ho₂O₃ in an Eppendorf tube was immersed in a bucket filled with water (4–5 L) placed just behind a standard ¹⁸F target. The water bucket acted as a moderator to thermalize the neutrons generated by the ¹⁸O(p,n)¹⁸F reaction. A routine ¹⁸F production led to the generation of small amounts of ¹⁶⁶Ho (T_½: 26.8 h). The bucket was retrieved after the cyclotron run, and the Ho sample contained 17–42 kBq of ¹⁶⁶Ho, which was dissolved in 5 mL 4 M HCl.

2.2. Simulation of the Thermal Performance of the Target

To estimate the maximum beam current that the Ho₂O₃:Al cyclotron target can tolerate, the heat dissipation and thermal performance of the target were simulated in MATLAB R2025b using similar calculations, as previously described [24,25]. The temperature in the target was calculated from partial differential equations defined for the geometry of the target, including the Nb degrader foil, the Ho₂O₃:Al target mixture, the air between the foil and the target material, the Ag backing, and the water cooling on the back of the Ag disc. A steady-state model for the conductive heat transfer was used, where the temperature does not change with time. The target was defined as a heat source Q and calculated from Equation (1), where k is the thermal conductivity and T the temperature [24].

$$\nabla \cdot (-k\nabla T)=Q$$

(1)

The thermal conductivity of the target material, Ag backing, Nb foil, and the air between the target and the foil were used in the model. The thermal conductivity of the Ho₂O₃:Al 1:1 (w/w) mixture was not known and was initially set to 30 W/m·K, since the thermal conductivity of Al is 273 W/m·K [26] and the thermal conductivity of Ho₂O₃ is expected to be below the value for metallic Ho (16 W/m·K).

The boundary conditions were set to 20 °C. The water-cooling flow rate was set to 1 L/min, and the tubing diameter was 7 mm. The convective heat transfer coefficient for the water cooling was calculated from Equation (2), where D is the hydraulic diameter, ρ the density, ν the velocity of flow, μ the dynamic viscosity, and C_p the specific heat capacity of water [24].

$$h=0.023\left({\displaystyle \frac{k}{D}}\right){\left({\displaystyle \frac{\rho \nu D}{\mu }}\right)}^{0.8}{\left({\displaystyle \frac{\mu C_p}{k}}\right)}^{0.4}$$

(2)

The incoming power from the proton beam was calculated from the beam energy and current. The beam tolerance of the target was assessed by calculating the temperature of the Nb degrader and of the Ho₂O₃:Al 1:1 (w/w) target material at different beam currents. The limit of the target was assumed to be at the beam current where the temperature of the Nb foil or the target material exceeded the melting point of Nb or Al, respectively. The experimental data were subsequently used to calibrate the code and calculate the thermal conductivity of the Ho₂O₃:Al 1:1 (w/w) target material.

2.3. Distribution Coefficient Determination

The distribution coefficients (K_D) at different HCl and HNO₃ concentrations were measured to design the separation procedure for Er/Ho with LN2 resin. The K_D values were measured by the batch method. Ten mL of solutions of different concentrations of HCl and HNO₃ (0.05, 0.08, 0.1, 0.3, 0.5, 1.0, and 2 M) containing Er or Ho (1–2 mg) were mixed with 0.1 g LN2 resin. The mixtures were shaken overnight, and the solution was separated from the resin. The resin was washed with 5 mL of the corresponding HCl or HNO₃ concentration and combined with the solution. The amount of Er or Ho in the solution was measured with ICP-OES. The resin was then washed with 15 mL 2 M HNO_3, and the amount of Er or Ho on the resin was measured by analyzing these fractions. The K_D values were calculated using Equation (3).

$$K_D={\displaystyle \frac{m_{wash of resin}}{m_{mobile phase}}}\cdot {\displaystyle \frac{V}{w}}$$

(3)

m_{wash of resin} and m_{mobile phase} are the mass of Er or Ho in the 2 M HNO₃ wash fraction of the resin and in the mobile phase at equilibrium, which is the initial 10 mL of solution and the 5 mL rinse solution with the corresponding HCl or HNO₃ concentration. w is the weight of dry resin (g), and V is the volume of the initial 10 mL solution (mL).

The separation factor (α) was calculated from the K_D values with Equation (4).

$${\alpha }_{batch,Er,Ho}={\displaystyle \frac{K_{D,Er}}{K_{D,Ho}}}$$

(4)

2.4. Full-Scale Separation Procedure

A 180 mg Ho₂O₃/Al (1/1 w/w) target had a suitable thickness for the ¹⁶⁵Er production. The target was dissolved in 6 mL of 4 M HCl at 60 °C. The pH of the solution was adjusted to 2 with 2 M NaOH (around 6 mL). Alternatively, the dissolved target material was evaporated to dryness at 130 °C under argon flow and redissolved in 3–5 mL 0.01 M HNO₃. The solution was loaded on a 4 g LN2 column that was previously washed with 60 mL 4 M HNO₃ and conditioned with 120 mL 0.01 M HNO₃. The column was washed with 10 mL 0.01 M HNO₃, 40 mL 0.3 M HNO₃, 60 mL 0.4 M HNO_3, and 30 mL 0.5 M HNO₃. ¹⁶⁵Er was eluted in 20–30 mL 4 M HNO₃. The column with LN2 resin could be regenerated with 60 mL 30% HCl, conditioned with 120 mL 0.01 M HNO_3, and reused. Before storing the column for a longer time, it was washed with 120 mL of water. The eluate containing ¹⁶⁵Er was evaporated to dryness and redissolved in 3–5 mL 0.01 M HNO₃. Alternatively, the eluate was loaded on a column containing 1.2 g branched DGA (b-DGA) resin previously washed with 0.01 M HNO₃ and conditioned with 4 M HNO₃. The column was washed with 30 mL 0.5 M HNO₃, and ¹⁶⁵Er and traces of Ho were eluted in 20 mL 0.01 M HNO₃. This solution was then loaded on a second 4 g LN2 column, which was washed with 4 M HNO₃ and conditioned with 0.01 M HNO₃. The column was then washed with 10 mL 0.01 M HNO₃, 20 mL 0.6 M HNO₃, 50–60 mL 0.7 M HNO_3, and 5–10 mL 0.8 M HNO₃. ¹⁶⁵Er was eluted in 10–20 mL 4 M HNO₃. A column with 0.1 g of TK221 resin has previously been used to concentrate and change the acid concentration of ¹⁶⁵Er solutions [19]. The eluate from the second LN2 column was loaded directly on the TK221 column that had previously been washed with 10 mL 0.05 M HCl and conditioned with 10 mL 4 M HNO₃. The column was washed with 10 mL 0.1 M HNO₃. The ¹⁶⁵Er was eluted in six 0.2 mL fractions with 0.05 M HCl.

The amount of Ho and Al in the different fractions was analyzed by ICP-OES. The separation factor for the dynamic separations using columns was calculated by Equation (5), where m_Ho,before and m_Ho,after are the mass of Ho before and after the separation, and A_165Er,before and A_165Er,after are the activity of ¹⁶⁵Er before and after the separation.

$${\alpha }_{column,Er,Ho}={\displaystyle \frac{m_{Ho,before}/m_{Ho,after}}{A_{{165}_{Er,before}}/A_{{165}_{Er,after}}}}$$

(5)

2.5. Radiochemical Purity, Chemical Purity, and Effective Molar Activity Determination

The samples were analyzed by X-ray spectrometry by spotting 10–50 µL of the samples on a piece of paper in a zip-lock plastic bag or by gamma spectrometry. The resolution of the two spectrometers was not sufficient to distinguish between all the emitted X-rays. For the X-ray spectrometry, ¹⁶⁵Er was quantified from the 46.7, 47.5, and 55.3 keV lines, and the 53.6 and 53.9 keV lines were combined into one peak at 53.9 keV. For gamma spectrometry, the lines 46.7 and 47.5 keV were combined into one peak at 47.6 keV, and the 53.6 and 53.9 keV lines were also combined and found at 53.9 keV. The dose calibrator was mainly used to give a preliminary estimation of the activity. A calibration factor from a ¹⁶⁵Er sample measured with the X-ray detector and the dose calibrator was applied. The calibration factor was 216 for samples in 10 mL glass vials. To determine the amount of stable metal impurities and the apparent molar activity, the samples were analyzed by ICP-OES. The samples were diluted in 1% HNO₃ and analyzed against ICP-OES standards containing Er, Ho, Ag, Al, Ba, Ca, Cd, Co, Cr, Cu, Fe, Mg, Mn, Na, Ni, Pb, and Zn. The effective molar activity was measured for three different chelators: DOTA, DTPA, and CHX-A″-DTPA. Samples from the final formulation (10 µL, 14.9 MBq decay corrected to end of bombardment (EOB)) were combined with 230 µL 0.1 M ammonium acetate (pH 5–6) and increasing amounts of chelators (0.1 mM solutions) in 0.1 M ammonium acetate buffer (0.1, 0.2, 0.3, 0.4, 0.5, 0.7, and 1 nmol chelator). The samples containing DOTA were heated to 90 °C for 30 min, while the samples containing DTPA and CHX-A″-DTPA were left at room temperature for 30 min, and the reaction mixtures were spotted on normal phase TLC plates. A 50:50 methanol–water mixture of solvents containing 5% w/v ammonium acetate was used as the mobile phase. The R_f value was 0.4 for the ¹⁶⁵Er-DOTA complex, 0.6 for the ¹⁶⁵Er-DTPA complex, 0.7–0.8 for the ¹⁶⁵Er-CHX-A″-DTPA complex, and 0 for non-chelated ¹⁶⁵Er. The amount of metals in the samples was calculated by dividing the amount of added chelator by the ratio of the formed complex. The activity of the sample was divided by the amount of metals in the sample to determine the effective molar activity. Only samples showing 15–85% of complex formed were used for the calculation, as linearity was assumed in this region, and the average of these results was used for the effective molar activity calculation.

2.6. Serum Stability of ¹⁶⁵Er Complexes

The stability of complexes prepared by mixing 150 µL of the final formulation (215.6 MBq ¹⁶⁵Er decay corrected to EOB) and 15–50 µL of each ligand (DOTA, DTPA, and CHX-A′′-DTPA) solution (1.0 mM in acetate buffer 0.1 M, 15–50 nmol) and 100–150 µL 0.1 M ammonium acetate was evaluated. Full complexation was not seen with one of the samples containing CHX-A″-DTPA, so an additional 15 µL of CHX-A″-DTPA solution was added (30 nmol CHX-A″-DTPA in total), and the sample was heated to 90 °C for 30 min. After confirmation that full complexation was achieved by TLC analysis, 60–80 µL samples were mixed with mouse serum 420–540 µL and incubated for 27–28 h at 37 °C. The mixtures were analyzed by TLC after 0, 1, 3–4, 20–22, and 27–28 h using the same conditions as described in the previous section.

3. Results

3.1. Production of ¹⁶⁵Er from Ho₂O₃:Al Target

Different amounts of Al powder were added to Ho₂O₃ to find the optimal target composition and irradiation conditions (

Table 1. The target composition, irradiation conditions, and production yield for the tested targets. The targets were irradiated for 10–180 min with a 0.1 mm Nb foil, degrading the proton beam energy.

Ho2O3 (mg)	Al (mg)	Current (µA)	165Er Yield (MBq/µAh)	Target in Figure 1
91	0	10	4.4	a
91	38	10	6.4	b
90	90	10–50	12.3 ± 1.9	c
67	133	15	8.4	-
90	180	10–15	7.8 ± 0.6	d

). The Al powder was needed to prevent thermal damage to the target material otherwise observed even at a 10 µA beam current. Targets without Al and with a Ho₂O₃:Al ratio of 7:3 (w/w) became powdery (Figure 1), and a Ho₂O₃:Al ratio of 1:1 (w/w) was found to be the optimal ratio since larger amounts of Al decreased the production yield. Targets with a 1:1 ratio were stable up to 45 µA. When a beam current of 50 µA was applied for 15 min, the target material started to be discolored and damaged (Figure 2). The Nb foil was also more affected by the irradiation at 50 µA than at 45 µA, but was still intact (Appendix A). For irradiations of more than 15 min, a beam current of 25 µA was used, but the test of the maximal beam tolerance suggested that currents up to 40 µA would be acceptable for longer irradiations.

The ¹⁶⁵Er production resulted in the coproduction of several short-lived radionuclidic impurities. ^164g,mHo (T_½: 28.8 min (ground state) and T_½: 36.6 min (metastable state)) were produced from the ¹⁶⁵Ho(p,pn)¹⁶⁴Ho nuclear reaction and ¹⁶⁶Ho from neutron activation. These Ho isotopes were produced in small amounts and followed the ^natHo in the separation procedure and were not found in the final product. ¹⁰⁷Cd (T_½: 6.5 h) was detected after the irradiation and was expected to be produced due to the use of Ag target backings. ⁵⁶Co (T_½: 77.2 d), ⁶⁹Ge (T_½: 38.9 h), and ^66,67Ga (T_½: 9.6 h and T_½: 3.3 d) were expected to be produced from impurities in the target material by the (p,n) nuclear reaction from Fe, Ga, and Zn, respectively [6]. The radionuclidic impurities were also separated from ¹⁶⁵Er with the purification procedure and did not affect the final radionuclidic purity.

3.2. Simulated Thermal Performance of the Target

The application of a cyclotron target consisting of a mixture of pressed powder made it challenging to predict the target stability and beam tolerance. For the calculation, the thermal conductivity was needed to calculate the heat dissipation and, therefore, had to be estimated for the Ho₂O₃:Al mixture. From the known thermal conductivity of Al and the estimated value for Ho₂O₃, we chose 30 W/m·K as our initial estimation for the 1:1 (w/w) mixture. From this calculation, the 180 mg target would reach a temperature of 645 °C by irradiation with a beam current of 52 µA, which could result in damage to the target, since the temperature would be at the limit of the melting point of Al (660 °C). The temperature of the Nb degrader at this beam current was calculated to be 1127 °C, which is below the melting point of Nb (2477 °C), indicating that the Ho₂O₃:Al 1:1 (w/w) target material was the limiting factor according to the calculations. Damage to the target was observed at a 50 µA beam current (Figure 2b), while the Nb foil was still intact (Appendix A). This indicated that the calculated maximal beam current tolerance was close to the experimental value despite the thermal conductivity of the target material being an estimation and experimental variations, e.g., for the target mass, mixture ratio, and beam profile shape that can affect the target performance. The data from the experiment were additionally used to adjust the value for the thermal conductivity of the Ho₂O₃:Al 1:1 (w/w) mixture, and by using a value of 24 W/m·K, the calculation matched the observed beam current tolerance.

3.3. Theoretical Yield Calculations

The maximum cross section for the ^natHo(p,n)¹⁶⁵Er reaction is at 10 MeV according to cross section studies by Tárkányi et al., Gracheva et al., and Červenák et al. [11,12,13]. It was chosen to degrade the incoming proton beam to 15 MeV using a 0.1 mm Nb foil. According to SRIM-2013 software calculations [27], the target thickness of a Ho₂O₃:Al 1:1 (w/w) target needs to be 276 g/cm², corresponding to 175 mg with a 9 mm diameter target to cover the energy range from 15 MeV to 8 MeV. The calculation is associated with some uncertainty since the density of the Ho₂O₃:Al mixture is not known. The yield was calculated to be 19.5 MBq/µAh using the cross-section measurements by Tárkányi et al. [11] compared to 12.3 MBq/µAh for the experimental yield for the 180 mg target, which corresponds to 63% of the calculated value. The added Al decreases the production yield since it also dilutes the amount of Ho present in the target material, and the Al contributes to the stopping of the proton beam. It was calculated how much higher beam current the target should be able to tolerate to compensate for the added Al powder. A Ho₂O₃ target that covers the same energy range as the 175 mg Ho₂O₃:Al 1:1 w/w target would have a thickness of 348 mg/cm² and a theoretical production yield of 49.2 MBq/µAh, which suggests that the target containing Al should withstand a proton beam current 2.5 times higher than the target without Al. Our initial experiments showed that the Ho₂O₃ target already showed signs of damage at 10 µA, while the target containing Al in a 1:1 ratio could withstand a beam current of 45 µA. This shows that the amount of produced ¹⁶⁵Er is improved by adding Al powder to the Ho₂O₃ target material.

3.4. ¹⁶⁵Er Purification Procedure

3.4.1. K_D Values and Separation Factor for Batch Separations

From the K_D studies, it was found that Er has a higher affinity for the LN2 resin compared to Ho at HCl and HNO₃ concentrations below 1 M (Figure 3). The binding affinity of Er was relatively low at HCl and HNO₃ concentrations above 1 M, which suggested that it would be possible to elute the Er effectively after eluting Ho at a lower acid concentration. In general, a higher separation factor was seen with HNO₃ compared to HCl (Figure 4). For the dynamic separations, both HCl and HNO₃ were tested.

3.4.2. Full-Scale Procedure

The separation of Er/Ho was tested in small-scale on LN2 columns to estimate the resin capacity, eluent concentrations, and flow rate, and to compare HCl and HNO₃ as an eluent. The details for these experiments can be found in Appendix B. From the initial experiments, a purification procedure for a full-scale ¹⁶⁵Er production was developed using three columns: two with LN2 resin and one column containing TK221 resin. An overview of the purification method can be seen in Figure 5.

The target could be dissolved in 4 M HCl at 60 °C within an hour and was loaded on the column after pH adjustment. HNO₃ was chosen as an eluent due to the higher separation factor, and it was found that 4 g of LN2 resin was suitable for the purification of ¹⁶⁵Er from the 180 mg Ho₂O₃:Al 1:1 (w/w) target. From the capacity study (Appendix B), it was expected that 2 g of LN2 resin would be sufficient, but due to the relatively large amount of Al, 4 g of resin was chosen and found to give an acceptable ¹⁶⁵Er/Ho separation. The first column was mainly separating the bulk Ho and Al from ¹⁶⁵Er. It was attempted to maintain a high ¹⁶⁵Er recovery from the first column since a higher separation factor could be obtained after removing a large amount of Ho and the Al. The highest separation factor from the K_D study was at 0.08–0.1 M HNO₃, but for the dynamic separation procedures using columns, the elution of Ho at this concentration would require large volumes of eluents due to the high K_D value. Increasing the concentration of HNO₃ from 0.3 to 0.5 M made it possible to elute the majority of Ho with limited co-elution of ¹⁶⁵Er. The ¹⁶⁵Er could subsequently be eluted in 20–30 mL 4 M HNO₃ (Figure 6).

The separation of Er from Ho was not sufficient, with one LN2 column containing 4 g of resin. And it was decided not to increase the amount of LN2 resin for the first column, since it was expected to be more efficient to use multiple LN2 columns. To load the eluate from the first LN2 column to a second one, the acid concentration needed to be reduced from 4 M to around 0.01 M. The eluate from the first LN2 column was evaporated to dryness and redissolved in 0.01 M HNO₃ and loaded on a second LN2 column. To avoid an evaporation step in the purification procedure, a column with DGA resin can be used to change the acid concentration. The DGA column could also be used to remove remaining Al, since Al had a very low adsorption to DGA resin over a broad range of HNO₃ concentrations. Er and Ho bind to DGA resin over a broad range of HNO₃ concentration, making it possible to load the eluate from the LN2 column and wash the DGA column with 0.5 M HNO₃ without eluting Er and Ho. The elution with 0.01 M HNO₃ was quantitative, and the eluate could be loaded directly on the second LN2 column. It was possible to reuse the same 4 g LN2 column, which reduced the amount of resin needed. Ho was eluted in 0.6–0.8 M HNO₃, and ¹⁶⁵Er started eluting with 0.8 M HNO₃, which meant that a balance between a high recovery and a high purity needed to be found. ¹⁶⁵Er was eluted in 10–20 mL of 4 M HNO₃ (Figure 7).

The large volume and high HNO₃ concentration of the eluate fractions from the second LN2 column were not suitable for ¹⁶⁵Er radiolabeling studies, so the 4 M HNO₃ eluate fractions were passed through a small column containing TK221 resin. ¹⁶⁵Er had a high affinity for the resin in 0.5–4 M HNO₃ and could be eluted in a small volume of 0.05 M HCl (400–600 µL, 89.9 ± 9.0% ¹⁶⁵Er recovery). This step also reduced the amount of contaminants, such as Fe and Ca. The main chemical impurities according to ICP-OES were Ho, Ca, and Fe, and in one batch, Pb was observed as an additional contaminant. The purification procedure from EOB to the end of purification (EOP) took around 8 h, which is close to one half-life of ¹⁶⁵Er. An overview of the metal impurities in the final ¹⁶⁵Er product, together with the radionuclidic purity, separation factor, and the recovery, can be found in

Table 2. Overview of the radionuclidic purity, ¹⁶⁵Er recovery, chemical impurities in the final ¹⁶⁵Er product, and the separation factors. The separation factors were calculated for each individual purification run and then averaged. The values for Al and Zn were below the limit of detection (LOD).

Radionuclidic purity at EOP (%)	>99.9%
Overall 165Er separation recovery (%)	77.9 ± 21.2
Er (nmol)	1.8 ± 0.1
Ho (nmol)	10.5 ± 8.9
Al (nmol)	<17 (LOD)
Zn (nmol)	<0.1 (LOD)
Fe (nmol)	1.3 ± 1.2
Ca (nmol)	16.1 ± 8.8
Pb (nmol)	21.2 ± 28.8
αcolumn,Er,Ho first LN2 column	37.4 ± 25.5
αcolumn,Er,Ho second LN2 column	1.9·103 ± 4.7·102
Overall αcolumn,Er,Ho	4.9·104 ± 3.2·104

. The X-ray spectrum of ¹⁶⁵Er after purification can be seen in Figure 8, and the gamma spectrum in Appendix C. No radionuclidic impurities were detected in the final product. In the X-ray spectrum, ghost peaks between 30–40 keV were detected, arising from escape of the Ge K-shell X-ray emitted following photoelectric interactions of the 47 and 48 keV photons.

3.5. Effective Molar Activity Determination

The effective molar activity of the final ¹⁶⁵Er was calculated from titration with the three chelators, DTPA, CHX-A″-DTPA, and DOTA. The percentage of formed complex can be seen in Figure 9. The effective molar activity was decay corrected to the time of EOB and was 17.8 ± 4.6 MBq/nmol for DTPA, 15.9 ± 5.3 MBq/nmol for CHX-A″-DTPA, and 19.4 ± 2.7 MBq/nmol for DOTA. The apparent molar activity of the same ¹⁶⁵Er batch calculated from ICP-OES was 24.9 MBq/nmol, and the detected metal impurities were Ho, Er, Fe, Ca, and Pb. Ca was not expected to compete with ¹⁶⁵Er for the binding to the three chosen chelators, and the apparent molar activity, excluding Ca, was 36.3 MBq/nmol. The effective molar activity was lower than the apparent molar activity determined from ICP-OES, which could suggest that metal impurities might be present in the sample that were not included in the ICP-OES analysis.

3.6. Serum Stability of ¹⁶⁵Er-Complexes

The [¹⁶⁵Er]Er-DTPA, [¹⁶⁵Er]Er-CHX-A″-DTPA, and [¹⁶⁵Er]Er-DOTA complex formation was confirmed by radio-TLC, and the complexes were mixed with mouse serum and incubated at 37 °C for up to 28 h. All three complexes were stable under the test conditions, and no free ¹⁶⁵Er or other decomplexation was observed.

4. Discussion

The use of pressed Ho₂O₃:Al cyclotron targets was suitable for ¹⁶⁵Er production. The addition of Al powder to the Ho₂O₃ target material ensured high robustness and reliability of the target, even though the production yield was compromised. The higher beam current tolerance compensated for this effect, since the current could be increased 4–4.5 times. Due to the limited data available for pressed mixed powder targets, we performed calculations using the numerical solution of the heat transport equations for the target geometry and an estimated thermal conductivity to predict the maximal beam current tolerance of the Ho₂O₃:Al 1:1 (w/w) target. The calculations gave a beneficial indication and could also be helpful for the further development of pressed powder targets. The quantification of ¹⁶⁵Er came with some difficulties due to the absence of γ-ray emission. Small samples spotted on pieces of paper in plastic bags were used for the quantification to simulate point sources and limit absorption of the low-energy X-rays. The resolution on the detectors was not sufficient to resolve all the emitted X-rays, but an acceptable agreement between the different X-ray lines was achieved by combining some of the lines.

The rather low cross-section of the reaction ^natHo(p,n)¹⁶⁵Er poses challenges for the scalability of ¹⁶⁵Er production on a medical cyclotron. However, the production of radionuclides by irradiation on medical cyclotrons has two advantages: First, competing nuclear reactions leading to the production of multiple radionuclides of the same element can usually be avoided. Second, in comparison to facilities running high-energy beams, there is a larger number of medical cyclotrons available. They offer a reliable and regular supply of the desired radionuclide, as well as shorter transportation times to the receiving medical facilities, which is especially important for radionuclides with half-lives shorter than a day. The half-life of ¹⁶⁵Er is relatively short for application in therapy, making local processing and application preferable to obtain the highest possible molar activity. It is also important to apply radiopharmaceuticals with pharmacokinetics that match the half-life of ¹⁶⁵Er to obtain high uptake and efficient clearance [5]. The relatively short half-life of ¹⁶⁵Er and the similar chemical properties of Er and Ho are challenges for the separation process, but the ¹⁶⁵Er/Ho separation is possible with two LN2 resin columns, one b-DGA resin column, or an evaporation step and one small column containing TK221 resin. In cases where the final chemical purity of the ¹⁶⁵Er can be compromised, a shorter purification procedure using, e.g., only one LN2 column could be considered. Higher purity can potentially be obtained by using additional LN2 resin columns, but large volumes (300–400 mL) of eluents are required due to the large amount of resin needed, and the acid concentration needs to be adjusted between the columns, which is time-consuming. We estimate to be limited to around 1 GBq of ¹⁶⁵Er at EOP with the current procedure, which is expected to be sufficient for preclinical studies. Despite the challenging scalability and separation procedure, ¹⁶⁵Er is highly relevant for studying the AE effect, since only X-rays are emitted together with the AEs. Furthermore, non-enriched target material can be used for production, and conventional chelators can easily be labeled with ¹⁶⁵Er. The labeling of the three tested chelators, DOTA, DTPA, and CHX-A″-DTPA, with ¹⁶⁵Er was efficient, with a high radiochemical purity and similar effective molar activity. The three ¹⁶⁵Er complexes showed no degradation in mouse serum for up to 28 h, which enables the possibility of applying a large variety of radiotracers with one of the three chelators facilitating labeling both at room temperature and at elevated temperatures.

5. Conclusions

A procedure for ¹⁶⁵Er production using pressed Ho₂O₃:Al cyclotron targets on a 16.5 MeV cyclotron was developed together with a ¹⁶⁵Er/Ho separation method using LN2 and TK221 extraction resins. The cyclotron target shows higher stability towards the proton beam when Al powder is added, and a 180 mg Ho₂O₃:Al 1:1 (w/w) can withstand beam currents up to 45 µA, which could also be simulated by temperature and heat dissipation calculations. A ¹⁶⁵Er production yield of 12.3 ± 1.9 MBq/µAh was reached. To obtain ¹⁶⁵Er in a purity suitable for further radiolabeling and application, two relatively large LN2 resin columns were used for the separation, followed by a small TK221 resin column to concentrate the final ¹⁶⁵Er product. The final ¹⁶⁵Er reached an apparent molar activity decay corrected to EOB of 24.9 MBq/nmol according to ICP-OES and was suitable for the labeling of DOTA, DTPA, and CHX-A″-DTPA chelators, which formed ¹⁶⁵Er complexes with high stability in mouse serum. The target development and separation procedure are part of increasing the availability of ¹⁶⁵Er for potential further studies of AE effect and therapeutic efficacy.