The Effects of Glass Doping, Temperature and Time on the Morphology, Composition, and Iron Redox of Spinel Crystals

Precipitation of large crystals/agglomerates of spinel and their accumulation in the pour spout riser of a Joule-heated ceramic melter during idling can plug the melter and prevent pouring of molten glass into canisters. Thus, there is a need to understand the effects of spinel-forming components, temperature, and time on the growth of crystals in connection with an accumulation rate. In our study, crystals of spinel [Fe, Ni, Mn, Zn, Sn][Fe, Cr]2O4 were precipitated from simulated high-level waste borosilicate glasses containing different concentrations of Ni, Fe, and Cr by heat treating at 850 and 900°C for different times. These crystals were extracted from the glasses and analyzed with scanning electron microscopy and image analysis for size and shape, with inductively coupled plasma–atomic emission spectroscopy and atom probe tomography for concentration of spinel-forming components, and with wet colorimetry and Mössbauer spectroscopy for Fe/Fetotal ratio. High concentrations of Ni, Fe, and Cr in glasses resulted in the precipitation of crystals larger than 100 μm in just two days. Crystals were a solid solution of NiFe2O4, NiCr2O4, and ɤ-Fe2O3 (identified only in the high-Ni-Fe glass) and also contained small concentrations of less than 1 at% of Li, Mg, Mn, and Al. INTRODUCTION The U.S. Department of Energy is building the Waste Treatment and Immobilization Plant at the Hanford Site in Washington State to remediate 55 million gallons of radioactive waste that is being temporarily stored in 177 underground tanks. The plan is to vitrify the waste into a durable borosilicate glass with Joule-heated ceramic melters. To do this efficiently and costeffectively, the waste loading in the glasses must be maximized. The major factor limiting waste loading in Hanford high-level waste (HLW) glasses is the precipitation, growth, and subsequent accumulation of spinel crystals [Fe, Ni, Mn, Zn, Sn][Fe, Cr]2O4 in the glass discharge riser of the melter during idling. These crystals can reach a size of a few hundreds of micrometers and because of their high density (~5.3 × 10 kg/m) settle down fast, forming a thick sludge layer that prevents the discharge of the molten glass into stainless steel canisters. Spinel crystals have a crystal structure of the natural spinel MgAl2O4 and can be described as a cubic close-packed arrangement of oxygen atoms with divalent and trivalent cations at two different crystallographic sites. These sites have tetrahedral and octahedral oxygen coordination. A remarkable feature of spinel structure is that it is able to form a virtually unlimited number of solid solutions. This means that the composition of spinel can be varied significantly without altering the basic crystalline structure. Depending on cation distribution, a spinel can be normal, inverse, or mixed. In a normal spinel (e.g., NiCr2O4 and ZnFe2O4) the tetrahedral and octahedral sites are occupied by divalent and trivalent cations, respectively. If tetrahedral sites are completely occupied by trivalent cations and the octahedral sites are shared by both divalent and trivalent cations, it is an inverse spinel (e.g., NiFe2O4 and ɤ-Fe2O3). However, most spinels are mixed spinels (e.g. MnFe2O4) in which divalent and trivalent cations occupy both tetrahedral and octahedral sites. This paper focuses on investigating the effects of temperature, time, and glass doping with metal oxides on the morphology, composition, and iron redox of spinel crystals in the HLW borosilicate glasses in connection with crystal accumulation in the riser of the melter. Crystals of spinel were precipitated from glasses containing different concentrations of Ni, Fe, and Cr by heat treating at 850 or 900°C for different times. These crystals were then recovered and analyzed with scanning electron microscopy (SEM) and image analysis for size and shape, with inductively coupled plasma atomic emission spectroscopy (ICP-AES) and atom probe tomography (APT) for concentration of spinel-forming components, and with wet colorimetry and Mössbauer spectroscopy for Fe/Fetotal ratio. The following sections describe the experimental approach and the results obtained from analysis of crystals with an array of instrumentation. MATERIALS Three glasses were formulated from previously used baseline glass by varying component fractions of NiO, Fe2O3, and Cr2O3. The remaining components were kept in the same proportions as in the baseline glass. Table 1 shows the compositions of tested glasses (Ni1.5, Ni1.5/Fe17.5, Ni1.5/Cr0.3) in mass fraction of oxides and halogens. Glasses were prepared from AZ-101 simulant and additives (H3BO3, Li2CO3, Na2CO3, SiO2, Cr2O3, NiO, and Fe2O3). These components were hand-mixed in a plastic bag, mixed in an agate mill for 5 min, and melted at 1200°C for an hour in a Pt-10%Rh crucible. The glass melt was quenched, ground in a tungsten carbide mill for 2 min, and remelted at the same temperature for the same time before being poured into double crucible assemblies that were placed inside the furnace at 850 or 900°C. The double crucibles were removed at various times and cross-sectioned. To extract crystals from glasses, accumulated layers were cut out from the bottoms of crucibles and treated with heated (60°C) 20% HNO3 overnight to dissolve the glass, and then with 5% HF to dissolve the residual silica gel. Table 1. Glass compositions in mass fraction of oxides and halogens. Component Ni1.5 Ni1.5/Fe17.5 Ni1.5/Cr0.3 Al2O3 0.0814 0.0784 0.0813 B2O3 0.0792 0.0763 0.0791 CaO 0.0057 0.0054 0.0057 CdO 0.0064 0.0062 0.0064 Cr2O3 0.0017 0.0016 0.0030 Fe2O3 0.1438 0.1750 0.1436 K2O 0.0034 0.0032 0.0034 Li2O 0.0197 0.0190 0.0197 MgO 0.0013 0.0012 0.0013 MnO 0.0035 0.0033 0.0035 Na2O 0.1850 0.1781 0.1848 NiO 0.0150 0.0150 0.0150 P2O5 0.0032 0.0031 0.0032 SiO2 0.3995 0.3847 0.3989 ZrO2 0.0412 0.0397 0.0411 Ce2O3 0.0020 0.0019 0.0020 a Each glass contained 0.0009 BaO, 0.0002 Cl, 0.0001 CoO, 0.0004 CuO, 0.0001 F, 0.0022 La2O3, 0.0018 Nd2O3, 0.0010 SnO2, 0.0008 SO3, 0.0003 TiO2, 0.0002 ZnO. METHODS Whole Rock Analysis Crystals were dissolved using lithium metaborate/tetraborate fusions, and solutions were analyzed with ICP-AES. Two solid standards, National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 278 Obsidian Rock and NIST SRM Basalt Rock, were run with samples to validate the analysis. The recoveries for analytes were in the range from 95 to 105%, except for Ba (78%), K (77%), Zr (80%), and Na (92%). Duplicates, which were used to assess the precision of the preparation process, showed a good agreement for analytes, with relative percent difference of less than 9%. Atom Probe Tomography Crystals collected from high-Ni-Fe glass (Ni1.5/Fe17.5-900°C-7D) were dispersed onto a carbon tape and coated with chromium. A wedge-shaped bar (25 × 3 × 2 μm) was extracted from a (111) facet of the selected crystal surface with a focused ion beam liftout method. Eight specimens with an apex diameter of 80 to120 nm were prepared and analyzed with a Leap 4000X HR 3D Atom Probe Microscope (CAMECA, Gennevilliers, France) in laser pulsing mode. Figure 1 shows SEM images of a crystal with a wedge-shaped bar region extracted and a conical specimen for analysis with APT. The specimen temperature and evaporation rate were fixed at 40 K and 2.5 × 10 ions per pulse, respectively. The laser energy was 32 pJ per pulse. Atom probe tomography provided a three-dimensional (3-D) elemental map and chemical identity of about 60% of the atoms in the analysis volume with a spatial resolution of ~0.1 nm. Since its detection efficiency is independent of atomic number it was also used to detect and quantify Li. Figure 1. SEM images of spinel crystal (Ni1.5/Fe17.5-900°C-7D) showing the wedge-shaped region removed for preparation of a specimen (dashed line rectangle) and a conical specimen for APT analysis. Scanning Electron Microscopy and Image Analysis Crystals were sprinkled onto a carbon tape, sputter-coated with Au/Pd and analyzed for morphology with a JEOL JSM-5900 SEM (SEMTech Solutions Inc., North Billerica, Massachusetts). The microscope was set up at accelerating voltage 15 kV, spot size 42-45, and working distance 12 mm. Clemex Vision PE 6.0 image analysis software (Clemex Technologies Inc., Longueuil, Canada) was used to measure the length of the edge of 50 crystals for each glass composition. Wet Colorimetry For each sample, approximately 15 mg of crystals were gently ground with a sapphire mortar and pestle under acetone. The nitrogen-dried powders were converted into solutions by treating with 6 mL of 48% HF at 94°C for 48 h. The solutions were then neutralized with a N2-sparged cocktail containing 12 mL of 10% H2SO4, 60 mL of 5% H3BO3, 2 mL of a solution of 200 mg of 1,10-phenathroline in 95% ethanol, and 35 mL of deionized water. Phenathroline was added to complex Fe and thereby prevent its oxidation. To determine the quantity of Fe, 1 ml of neutralized digestate was mixed with 10 mL of 1% sodium citrate. To determine the quantity of Fetotal, 1 ml of neutralized digestate was mixed with 10 ml of a solution containing 1% sodium citrate and 1% NH2OH·H2SO4. After the minimum equilibration time of 90 min, the absorptivities of the Fe and the Fetotal were determined at a wavelength of 510 nm with a Shimadzu UV-2401 PC UV-VIS spectrophotometer (Shimadzu, Columbia, Maryland). Several control and standard samples were run with experimental samples. These included an iron-free reagent blank, ferrous diammonium sulfate hexahydrate (FAS, Sigma F-1543, Lot # 58H0465, nominal purity 101.3%), hematite (Fe2O3, Alfa Aesar 14680, Lot # G21U020, nominally 99.945% pure on metals basis), and magnetite (Fe3O4, Sigma-Aldrich 518158, Lot # MKBJ5645V, nominal Fe content 71.8% vs. 72.36% stoichiometric). Fe Mössbauer Spectroscopy Mössbauer analysis was performed on crystals collected from Ni1.5, Ni1.5/Fe17.5, and Ni1.5/Cr03 glasses which were heat-treated at 900°C for 7 days. Approximately 20-50 mg of crystals per sample was mixed with petroleum jelly in a cylindrical holder with inside diameter 1.27 cm and height 0.94 cm. The completely filled holders were sealed at both ends with Kapton tape, which was then snapped into the holders with carbonized polyethyletherketone rings to ensure tightness. Mössbauer spectra were collected at room temperature and 10 K with WissEL Elektronik (Starnberg, Germany) or Web Research Company (St. Paul, MN) instruments, which included an SHI-850 closed-cycle cryostat (Janis Research Company, Inc., Wilmington, MA), a CKW-21 helium compressor unit (Sumitomo Cryogenics of America, Chicago, IL), and an Ar-Kr proportional counter detector (WissEL) or Ritvec (St. Petersburg, Russia) NaI detection system (Web Research Company). A Co/Rh source (Ritvec, Russia) with 50-75 mCi initial strength was used as the gamma energy source. The velocity transducers were operated in a constant acceleration mode (23 Hz, ±12 mm/s). The transmitted counts were stored in a multichannel scalar as a function of energy (transducer velocity) using a 1024-channel analyzer. The data were folded to 512 channels to provide a flat background and a zero-velocity position corresponding to the center shift of 25-μm-thick α-Fe foil (Amersham, Amersham, England) at room temperature. The Mössbauer data were modeled with Recoil software (University of Ottawa, Ottawa, Canada) using a Voight-based structural fitting routine. RESULTS AND DISCUSSION The whole rock analysis revealed that the concentrations of major spinel-forming components Ni, Fe, and Cr varied with glass composition, temperature, and time. The concentration of Fe ranged from 0.7 to 0.8 mol, Ni from 0.3 to 0.4 mol, and Cr from 0.02 to 0.1 mol. Figure 2 illustrates the changes in concentration of Ni, Fe, and Cr over time for different glass compositions and temperatures. Adding more Ni, Fe, and Cr to glasses resulted in increased concentrations of these components in the crystals. Analysis also identified the presence of other spinel-forming components such as Mn, Al, and Mg in concentrations of less than 0.01 mol. Figure 2. Effect of glass composition, temperature, and time on concentration of Fe, Ni, and Cr in spinel crystals. A 3-D reconstructed volume ~60 × 60 × 140 nm of 10 million atoms was obtained for the cone-shaped specimen prepared from spinel crystal that precipitated from high-Ni-Fe glass. Figure 3 shows concentration profiles of Ni, Fe, and Cr along the z-axis from top to bottom. The measured concentrations were about constant over the depth of 140 nm, suggesting nonexistence of short-range composition fluctuations within the reconstructed volume. Observed small-scale fluctuations in concentrations were most likely the result of occasional laser scans during data acquisition, which were performed to maintain the laser alignment on the specimen. Table 2 shows minimum, maximum, and average concentrations of different elements in at% within the reconstructed volume. Good agreement with chemical analysis was obtained for average concentrations of Ni (30.2 vs. 31.6 at%) and Fe (14.5 vs. 14.0 at%). However, the average concentration of Cr from chemical analysis was 1.2 at% compared to 2.1 at% from APT. This suggests formation of more NiCr2O4 at the late stage of crystal growth and possible longrange composition fluctuations. This needs to be confirmed, however, by APT analysis of additional specimens from the interior of the crystal. The analysis also revealed the presence of other spinel-forming components such as Li, Mg, Mn, and Al, which were detected in small concentrations up to 0.9 At%, providing evidence that these components participate in the formation of spinel solid solution in the HLW glasses. 50 100 150 200 250 300 350 0.66 0.68 0.70 0.72 0.74 0.76 0.78 0.80 0.82 Ni1.5 900°C Ni1.5/Cr0.3 900°C Ni1.5/Cr0.3 850°C Ni1.5/Fe17.5 900°C Ni1.5/Fe17.5 850°C