Purification of molybdenum oxide, growth and characterization of medium size zinc molybdate crystals for the LUMINEU program

The LUMINEU program aims at performing a pilot experiment on neutrinoless double beta decay of 100Mo using radiopure ZnMoO4 crystals operated as scintillating bolometers. Growth of high quality radiopure crystals is a complex task, since there are no commercially available molybdenum compounds with the required levels of purity and radioactive contamination. This paper discusses approaches to purify molybdenum and synthesize compound for high quality radiopure ZnMoO4 crystal growth. A combination of a double sublimation (with addition of zinc molybdate) with subsequent recrystallization in aqueous solutions (using zinc molybdate as a collector) was used. Zinc molybdate crystals up to 1.5 kg were grown by the low-thermal-gradient Czochralski technique, their optical, luminescent, diamagnetic, thermal and bolometric properties were tested.


Introduction
Neutrinoless double beta (02) decay is a key process in particle physics thanks to its unique ability to test the Majorana nature of neutrino and lepton number conservation, the absolute scale and the hierarchy of neutrino mass [1,2,3,4]. Low temperature scintillating bolometers are considered as extremely promising detectors to search for 02 decay in different nuclei [5,6,7,8,9,10,11]. Recently developed technique to grow large high quality radiopure zinc molybdate (ZnMoO 4 ) crystal scintillators [9,10,12,13,14,15] makes this material advantageous for low temperature bolometric experiments to search for 0ν2β decay of 100 Mo. Here we report further progress in deep purification of molybdenum and growth of ZnMoO 4 crystals for the LUMINEU project. First results of the crystals characterization are presented too.

Production of ZnMoO 4 crystals 2.1 Purification of molybdenum
High purity molybdenum and zinc are required to grow high quality radiopure ZnMoO 4 crystal scintillators. While a high purity zinc oxide is commercially available, molybdenum should be additionally purified. Furthermore, there are no commercially available molybdenum compounds that are tested for the presence of radioactive elements and have the required level of radioactive contamination. Such a test of the raw materials for crystal growth is extremely difficult and requires long measurement procedure. Typically high sensitivity radiopurity tests can be only done after crystal growth using calorimetric method. Moreover, development of efficient purification methods with minimal losses of molybdenum is strongly required to produce ZnMoO 4 crystals from enriched molybdenum, whose contamination is typically on the level of tens  hundreds ppm [16] (keeping also in mind the necessity to recycle the costly enriched material). We have developed a two stages technique of molybdenum purification consisting of sublimation of molybdenum oxide in vacuum (with addition of zinc molybdate) and double recrystallization from aqueous solutions by coprecipitation of impurities on zinc molybdate sediment.

Purification of MoO 3 by sublimation
Sublimation of molybdenum oxide under atmospheric pressure with subsequent leaching in aqueous solutions with ammonia is widely used in the industry of molybdenum. Nevertheless the concentration of impurities, particularly of tungsten (on the level of up to 0.5wt% even in the high purity grade materials) still exceeds the ZnMoO 4 crystal growth requirements. Even additional vacuum sublimation of molybdenum oxide proved to be insufficient. According to [17] separation of tungsten and molybdenum is a well known problem. Besides, the sublimation of MoO 3 is not efficient enough to reduce traces e.g. of Ca, Na and Si to the level below 20  70 ppm.
We have assumed that during sublimation at high temperature the following exchange reaction could occur: Such a reaction should reduce the concentration of tungsten, and therefore can be used for separation of molybdenum from tungsten. To prove this possibility, we have prepared a sample of MoO 3 powder with 10wt% of WO 3 . The concentration of tungsten in the MoO 3 product after sublimation was reduced to 0.1wt%. One more confirmation of the method's efficiency was obtained by chemical and X-ray diffraction analysis of the rests after the sublimations performed with an aim to purify molybdenum for crystal growth (the amount of the rests is typically 1  3 wt% of the initial amount of the purified material). The bottoms after a few sublimation processes were mixed and annealed in the air atmosphere to oxidize residues of metals. Then we have carried out sublimation of the sample in vacuum to reduce presence of MoO 3 . Atomic emission analysis, performed in the analytical laboratory of the Nikolaev Institute of Inorganic Chemistry, gives the following elemental composition of the bottoms: Сa -0.14wt%; Cu -0.011wt%; Fe -0.064wt%; K -1wt%; Mg -0.026wt%; Na -0.13wt%; Si -2.6wt%; Mo -22wt%; W -18wt%; Zn -14wt%. Oxides of molybdenum and silicon, tungstate (in form of tungstate-molybdate) and molybdate of zinc, as well as K 2 Mo 7 O 22 and K 2 MgSi 5 O 12 , have been identified in the bottoms with the help of X-ray diffraction analysis. At the same time, tungsten oxide (present in the initial product) was not detected in the bottoms. The data supported occurring of the exchange reaction (1) and confirmed efficiency of molybdenum oxide sublimation in vacuum with addition of zinc molybdate.
To purify molybdenum for ZnMoO 4 crystal growth, we have added up to 1% of high purity zinc molybdenum (obtained earlier in the course of the R&D) to the MoO 3 prepared for sublimation. The obtained sublimates contained mixture of molybdenum oxides of different composition and color, which hinders their use for ZnMoO 4 synthesis. The sublimates were then annealed in the air atmosphere to obtain yellow color stoichiometric MoO 3 . The sublimates were analyzed by atomic emission spectrometry. The results are presented in Table 1. One can see that the purity level of MoO 3 was improved onetwo orders of magnitude after the double sublimation process. The sublimation also should remove metal oxides, which have a high vapor pressure at temperatures up to a thousand degrees.  6 ]. Role of the central atom, except Si(IV) and P(V), can perform also V(V), Ge(IV), Cr(III), etc., while ligands of the inner sphere can be ions WO 4 2-, VO 3-, CrO 4 2-, TeO 4 2-. Solutions of polymolybdates at pH < 6 are able to dissolve oxides and hydroxides of many metals, e.g. ZnO, Fe(OH) 3 , Ni(OH) 2 , Cu(OH) 2 , etc. As a result, a partial co-crystallization of impurities could occur. Thus, recrystallization of ammo- Besides, a typical concentration of impurities in high purity commercial MoO 3 is relatively low (1  100 ppm). As a result, the impurity sediments appear in a form of fine microcrystals, hardly removable by filtration.
To improve efficiency of the recrystallization process, we have used zinc oxide to initiate precipitation (taking into account that zinc does not affect the crystal quality). ZnO on the level of 1  2 g/L was dissolved in the ammonium para-molybdate solution at рН > 6, then ammonia was added to the solution to reach pH = 7  8. After several hours of exposure precipitation of zinc molybdate occurs. The ZnMoO 4 sediment sorbs impurities from the solution. Further increasing of рН leads to precipitation of contaminants in the form of hydroxides. It should be stressed, the basic solution with pH  8  9 provides the most favorable conditions for thorium and uranium precipitation. After separation of the sediment, the solution was evaporated to 70%. Then ammonium oxalate was added to the solution to bind the residues of iron impurities. Results of the purification are presented in Table 2.
It should be also stressed that using of the additional "wet" chemistry procedure is also encouraged by the fact that large crystal grains of MoO 3 are formed in sublimation process, which provides certain difficulties to produce radiopure ZnMoO 4 powder (an additional procedure of the oxide grains grinding could contaminate the material). Subsequent dissolution of the molybdenum oxide in ammonia allows to obtain high purity MoO 3 perfectly fine for further synthesis of ZnMoO 4 powder.
The molybdenum oxide purified by twice recrystallization procedure from aqueous solutions and high purity grade ZnO produced by Umicore were used to synthesize ZnMoO 4 powder for crystal growth.

ZnMoO 4 crystal growth
Several ZnMoO 4 crystal boules were grown in air atmosphere from the purified input powder by the lowthermal-gradient Czochralski technique [18,19,20] in platinum crucibles 40 and 80 mm (it should be mentioned that, according to the certificates of the platinum crucibles, iron content in the platinum does not exceed 40 ppm). The temperature gradient was kept below 1 K/cm, the rotational speed was in the range of 5  20 rotations per minute with the crystallization rate of 0.8  1.2 mm/hour. A low crystallization rate was kept during growing of upper cone of the crystal boules. Rotational speed was decreased from the start to the end of the growth process in 1.5  2 times. The yield of the produced boules was on the level of 80%, which is an important achievement taking into account the future plans to produce crystals from enriched 100 Mo. Four optical elements (two 20  40 mm and two 35  40 mm with masses 55 g and 160 g, respectively) were cut and polished for low temperature measurements. Several small samples were produced for optical, luminescent, diamagnetic and thermal tests.

Optical absorption
Visible and near infrared absorption spectra of ZnMoO 4 crystal were recorded with a Varian Cary 5000 spectrophotometer. The transmission coefficient, T, was measured on a 2.0 mm-thick single crystal and found to be higher than 0.5 from 327 nm to 4.96 m. The absorption coefficient was calculated as  = logTln10/t, with t the thickness of the crystal. The data are presented in Fig. 1. The cut-off wavelengths are 313 nm ( 3.96 eV) and 5.13 m. The absorption coefficient decreases from 1.47 to 0.89 cm -1 in the wavelength region from 400 nm to 2 m, but there does not seem to have a broad absorption band around 440 nm that could be ascribed to Fe 2+ /Fe 3+ impurities as described in [21] and [22]. This is due to the low Fe concentration in the crystal ( 1.7110 16 cm -3 ) and in fact, it can be stated that a safe detection limit around 440 nm by such transmission experiments is ~ 10 18 atoms of Fe per cm 3 . Such a low absorption coefficient turns out to be lower than that of the orange crystals grown in [23] and [24], which exhibit  abs ( 550 nm)  2.5 cm -1 . The refractive index at 650 nm, obtained from 1+(1-T 2 ) 1/2 /T (no Fresnel losses), is  1.96, close to the value ~ 1.91 given in [25] and to the values 1.87 -2.01 (taking into account biaxiality of the material) obtained in [15] for the wavelengths 406 -655 nm.

Luminescence under X-ray excitation
The luminescence of the ZnMoO 4 crystal sample (10 × 10 × 2 mm) was investigated as a function of temperature between 8 and 290 K under X-ray excitation. The sample was irradiated by X-rays from a BHV7 tube with a rhenium anode (20 kV, 20 mA). Light from the crystal was detected in the visible region by a FEU-106 photomultiplier (sensitive in the wide wavelength region of 300−800 nm) and in the near infrared region by FEU-83 photomultiplier with enhanced sensitivity up to ≈ 1 μm. Spectral measurements were carried out using a high-aperture MDR-2 monochromator. Emission spectra measured at 8, 118 and 290 K are shown in Fig. 2. A broad band in the visible region with a maximum at 550 nm was observed at room temperature. At 8 K luminescence exhibits an emission band with a maximum at  600 nm in agreement with the results of previous studies [15,23,26]. We have also observed a band at approximately 480 nm occurring below the liquid nitrogen temperature.

Magnetic susceptibility
Magnetic susceptibility was measured using a Quantum Design SQUID MPMS XL magnetometer operating in the 4.2 − 350 K temperature range and in the 0 − 5 T magnetic field range. The crystal mass was 210 mg and its volumic mass assumed to be 4.19 g/cm 3 . It was mounted in a capsule placed in a straw and the negligibly small diamagnetic contribution (  10 -7 ) of the capsule was not subtracted from our data. The ZnMoO 4 proved to be weakly diamagnetic with a MKSA  = (8.0  0.2)  10 -6 over the whole temperature range investigated, from 20 to 320 K. Thus, paramagnetic impurities such as Fe 2+ or Fe 3+ could not be evidenced even under higher applied magnetic fields up to 0.2 T.

Specific heat measurements
Specific heat measurements were made on a 3  3  2 mm 3 single crystal to optimize the exchange surface and avoid too much thermal inertia. The crystal was fixed on a sapphire sample holder with vacuum grease. The sample holder was mounted on the measurement shaft of a Quantum Design PPMS equipment interfaced to operate with a 2- pulse-step method corrected for the grease baseline. The results of the measurements are presented in Fig. 3. The phononic contribution could be approximated for temperatures higher than ~ 23 K by means of high-temperature series expansion: (formula (5) from [27], and red curve in Fig. 3), which yielded a high Debye temperature of  625.1 K and the following Bernoulli numbers: B 1 = 1.9091, B 2 = 1.86714, B 3 = -0.96009, B 4 = -0.00907. No long range order (LRO) effect was observed down to 4 K. The C p /3NR ratio at 351 K reaches 0.87 and remains lower than the Dulong and Petit limit, which suggests low anharmonic effects at play at this temperature, consistent with the high Debye temperature obtained by the HTS fit. The inset shows C p /T vs T at low temperature to evidence the absence of any LRO down to 4 K.

Low temperature tests
The operation of the scintillating bolometers with the produced ZnMoO 4 crystals at low temperatures was performed in the cryogenic laboratories of the CSNSM (Orsay). Two ZnMoO 4 samples (55 g and 160 g) were installed in a high-power dilution refrigerator with a large experimental space. A single photodetector, consisting of a 2'' Ge disk and instrumented with a Neutron Transmutation Doped Ge thermistor as a temperature sensor, identical to those attached at the ZnMoO 4 crystals, collected the scintillation light emitted by both samples (see Fig. 4). The copper heat-sink temperature was stabilized at 18 mK, and the detector operation temperature was about 1 mK higher due to sensor biasing. Both samples and light detector performed well, with an excellent signal-to-noise ratio. Unfortunately, aboveground operation is marginally compatible with such large crystals. In the large detector, practically every time window containing a full pulse contains also other pulses. One can see the pile-up effect on Inset of Fig. 5 where 4 seconds streaming data accumulated with a weak 232 Th gamma source are shown. It should be stressed, the pile-up remains substantial also for the background data acquired without calibration source. This affects the energy resolution of the detector, which is expected to be much better in underground operation under heavy shielding with much lower pile-up effect.
In spite of that, a preliminary useful characterization can be performed in terms of signal amplitude, light yield, light quenching factors for alpha particles and crystal radiopurity. The results for the two crystals are summarized in Table 3. The difference in thermal response is due to the intrinsic irreproducibility of the thermal coupling in this type of detectors. The values of the quenching factor and of the light yield are perfectly compatible with those reported in the literature. Fig. 5 shows a 232 Th calibration and Fig. 6 presents a light-heat scatter plot accumulated with the large ZnMoO 4 crystal. In terms of radiopurity, we will give quantitative estimations in a work in preparation. We can anticipate however that, with the exception of the 210 Po line at 5.41 MeV, no internal alpha line emerged in the energy spectrum after about two weeks of data taking.  Table 3. Performance of 55 g and 160 g ZnMoO 4 detectors. QF denote quenching factor of α particle signals with respect to β particle signals for the same deposited energy (~ 5.4 MeV).

Conclusions
The LUMINEU program aims at performing a pilot experiment on neutrinoless double beta decay of 100 Mo using radiopure ZnMoO 4 crystals operated as scintillating bolometers. This problem requires development of methods of molybdenum purification to obtain crystals with desired characteristics.
Different approaches of molybdenum purification for ZnMoO 4 crystals growth were elaborated. A purification using two stages sublimation (with addition of zinc molybdate) and recrystallization from aqueous solutions of ammonium para-molybdate (using zinc molybdate as a collector) is a promising approach to purify molybdenum for high quality radiopure ZnMoO 4 crystals growth.
A first batch of LUMINEU crystals with mass up to 1.5 kg have been successfully grown by by the lowthermal-gradient Czochralski technique, and their optical, luminescent, diamagnetic, thermal and bolometric properties were tested. Characterization of the material is in progress.
In the future, crystals of increasing mass from deep purified precursors will be developed for the LUMINEU experiment, including crystals enriched in the isotope 100 Mo.