Pilot-Scale Treatment of a Spent Uranium Catalyst Formally Used in the SOHIO Process: Pilot Plant Verification of the SENSEI Process

Abstract

Approximately 7000 drums of waste uranium catalyst are currently present in Korea and require an appropriate treatment and waste management strategy. Recently, one such process has been developed and has proven successful at both laboratory and bench scales. The success of the process has culminated in its verification at a pilot plant scale. The purpose of this paper is to describe the catalyst treatment process and present results obtained from the pilot plant study that may be applicable to other such wastes. The individual unit technologies have been tested and verified, enabling process scale-up to be successfully proven. The final volume reduction of up to 80% has been confirmed with the successful separation, encapsulation, and immobilization of residue wastes, representing a potential cost saving of US$70 million compared to the direct disposal. The inactive silica component of the waste catalyst was purified and confirmed to be free of uranium. All effluents generated during the process were treated and satisfy the appropriate Korean release criteria. The process employs the concept of Selective Extraction of Nonradioactive Species, Encapsulation, and Immobilization, and is therefore introduced as the SENSEI process.

1. Introduction

1.1. SOHIO Process Catalysts

Acrylonitrile is a key chemical in the acrylic fiber used to manufacture a wide variety of materials.¹ The economics of acrylonitrile production was greatly improved by the development of catalysts capable of single-step selective catalytic oxidation of the raw materials, propylene, ammonia, and air over a fluidized bed catalyst, thus eliminating the need for expensive multistep processes.¹⁻⁵ The development and implementation of such catalysts in the 1950s by the Standard Oil of Ohio company, or SOHIO, paved the way for a rapid and inexpensive acrylonitrile production process that came to be known as the SOHIO process.^1,6

A variety of different catalysts have been used as part of the SOHIO process.² A catalyst containing a uranium-antimonite active phase (U = 3–7 wt %, Sb = 15–25 wt %) on a silica support (Si = 22.9 wt %, approximately 70% silica by volume) was historically used worldwide, including Japan,⁷⁻¹⁰ Korea,^11,12 and Taiwan.¹² The catalyst is no longer used in the process, and the used catalyst is regarded as radioactive waste. The presence of uranium poses a number of disposal issues, largely due to its radiotoxicity and international regulation on radioactive waste disposal. Uranium-containing wastes require appropriate management strategies to enable safe treatment and disposal, thus reducing human and environmental risks; while taking economic feasibility into consideration.

In addition to uranium, the presence of toxic heavy metals within the catalyst, most notably antimony (Sb), is also a cause for concern regarding the treatment of the used catalyst and final release. Antimony is found in the catalyst formulation due to its role as an α-H abstraction component (Sb(III)) and as an olefin chemisorption and oxygen or nitrogen insertion component (Sb(V)).² The presence of antimony in the catalyst, which is subsequently found in process effluents,¹³ is therefore also the cause for concern because it is considered toxic to the environment¹⁴ and is associated with cancer development.¹⁵ Korean release regulations mean that a suitable method to remove both uranium and antimony from the process effluent is required to gain regulatory approval.¹³

1.2. Spent Uranium Catalyst—A Korean Case Study

Approximately 7000 drums (200 L per drum) of spent uranium catalyst were generated by a private company in Korea over a 10 year period until its use was suspended in 2004.^11,12 At the time of writing, it still remained in temporary onsite storage awaiting an appropriate waste management strategy.

The catalyst is regarded as a radioactive waste (activity of catalyst: 590–1340 Bq g^–1) consisting of depleted uranium (U-234: 0.001%; U-235: 0.194%; U-238: 99.804%; U_total activity: 14 600 Bq g^–1; U_total: 8.5 wt %) on a silica support (70 vol %). In addition, the catalyst is composed of several metal oxides, including antimony (Sb: 24.6 wt %), iron (Fe: 4.4 wt %), molybdenum (Mo: 0.55 wt %), and aluminum (Al: 0.13 wt %). The physical conditions of the waste vary from dry unused particles to moist sludges that contain a mix of used catalysts in various states of degradation and a tar-like organic substance, a remnant of acrylonitrile synthesis. The tar consists of volatile organics (up to 33.9 wt %) and fixed carbon (up to 3.4 wt %). Moisture contents vary up to 40.6 wt % depending on waste conditions. A complete description of the spent uranium catalyst has been previously reported.¹²

Low-level radioactive waste (LLW) disposal is possible in Korea at the Gyeongju disposal site. The site can receive an array of waste types, such as plastics, metals, clothing, etc., with a total capacity for 100 000 drums (200 L per drum). For disposal approval to be granted, wastes must meet a number of waste acceptance criteria (WAC). The criteria applicable to the spent uranium catalyst are listed in Table 1. Disposal costs for radioactive waste in Korea are prohibitively expensive, currently around US$12 500 per 200 L drum as of 2020. In light of this, it is highly desirable that a suitable volume reduction treatment process is found that also generates a suitable wasteform that meets the WAC for disposal. In addition, the process should be simple to operate, have minimal secondary waste generation, be low cost, reliable, and reproducible.

Table 1. Waste Acceptance Criteria (WAC) Set by the Gyeongju Low-Level Waste (LLW) Disposal Site for α Emitting Waste^a.

waste acceptance criteria (WAC)	Gyeongju disposal site criteria
free-standing water	<0.5 wt %
total radioactivity	<3700 Bq g–1
particulate fines (10 μm)	<1 wt %
particulate fines (200 μm)	<15 wt %
compressive strength	>3.5 MPa
antimony oxidation state	+5

^aCriteria applicable to the spent uranium catalyst.

Recently, we have developed a process, which has been successfully demonstrated at the laboratory scale, to treat the catalyst waste and enable disposal of the uranium-bearing wastes at the geological disposal site, while facilitating the clearance of the silica support to specialist landfill.¹⁶⁻¹⁸ Options to treat secondary liquid effluents generated throughout the process have also been developed to ensure that neither uranium^17,19 nor other heavy metals, namely, antimony,¹³ are released to the environment above allowable release limits (applicable release limits for Korea as of January 1, 2020, U < 1 mg L^–1, Sb < 0.2 mg L^–1, PO₄^3– < 8 mg L^–1).²⁰ The process uses the concept of Selective Extraction of Nonradioactive Species, and Encapsulation and Immobilization of the final uranium-containing solid waste, referred to as the SENSEI process. For regulatory acceptance, a successful demonstration of process scale-up to pilot plant was required.

1.3. Conceptual Process

As silica constitutes the largest volume fraction of the spent uranium catalyst, this was targeted for removal via dissolution and purification, thus, if successful, providing a volume reduction of 70%. The approach to treat the catalyst is simple: the selective dissolution of the silica support; its purification and release as clearance (<1 Bq g^–1 α); and immobilization of the uranium-containing mixed oxide residues. To accomplish this, the process is split into four major stages (Figure 1):

• Catalyst pretreatment and dissolution (stage 1).

• Silica precipitation, handling, and purification (stage 2).

• Effluent handling and treatment (stage 3).

• Uranium residue immobilization (stage 4).

Figure 1.

Schematic process for the treatment of the spent uranium catalyst waste.

The key to the treatment process is the solubility of silica under alkaline conditions while remaining insoluble at circumneutral pH and the solubility of uranyl carbonate species. This allows for dissolution of silica under alkaline conditions while most oxides remain insoluble, which can then be separated by filtration (stage 1). Formation of soluble uranyl peroxocarbonate species then permits retention of uranium in the supernatant at circumneutral pH while the silica precipitates (stage 2). These steps are key to selectively remove the silica from the catalyst and ensure that the silica is free of uranium before release as clearance to landfill. Uranium-containing effluents are then treated via phosphate dosing to promote precipitation (stage 3), before final immobilization of the uranium-containing mixed oxide residues (stage 4) for disposal at the Gyeongju LLW disposal site. Implementation of the process makes use of precipitation, solid–liquid separation, and glass–ceramic formation techniques, all of which are common methodologies.

2. Experimental Procedures

2.1. Pilot Plant Design and Procedure

Each unit technology was built to be fully accessible throughout the pilot plant trials, which also enabled automatic or manual addition of reagents. The operation of the pilot plant was performed in a semicontinuous manner over the course of six months. At that time, the methods and procedures were largely kept unchanged from previously reported laboratory and bench-scale trials. All plant personnel were adequately trained prior to trials, and all participated in plant health and safety protocols. Plant visits were also subject to adequate safety measures.

Due to public concern, regulator restrictions were in place, prohibiting the transportation of the uranium-containing catalyst in large volumes from the storage site to the pilot plant; subsequently, it was not possible to use the real catalyst. To simulate the spent uranium catalyst behavior and to verify each stage of the SENSEI process, an alternative uranium-free catalyst (U-free catalyst) was used. The alternative catalyst is physicochemically identical to the real one with the exception of uranium and is in fact currently used in the SOHIO process as a replacement for the original uranium-containing catalyst. Uranyl nitrate, used to study uranium behavior and routing throughout the process, was then added to the process after the catalyst dissolution stage to mimic the levels of codissolved uranium found in the aqueous process stream post silica dissolution. This approach (U-free catalyst + uranyl nitrate ≈ uranium-containing catalyst) was found to be a suitable approximation for the real spent uranium catalyst. Results from the pilot plant are comparable to both laboratory and bench-scale tests. The earlier laboratory and bench-scale tests had been performed using the real spent uranium catalyst.¹⁶ Hereinafter, the “spent catalyst” refers to the U-free catalyst.

2.2. Quality Control, Sampling, and Analysis

Sampling was conducted throughout every stage of the process and during repeated operations to ensure both process efficiency and consistency. Analysis was conducted off-site. Solution concentrations of contaminants were determined by inductively coupled plasma optical emission spectrometry (ICP-OES) (Analytikjena PQ9000 Elite, [U]_LOD = 5 μg L^–1) in a 5% nitric acid media. Bulk mineral phases and solid contents were determined by X-ray powder diffraction (XRD) (Bruker, D2 Phaser) and X-ray fluorescence (XRF) (Olympus, Delta professional), respectively. The moisture content was determined by thermogravimetric analysis (TGA) (TA Instruments Korea, SDT Q600) up to 150 °C at a ramp rate of 10 °C min^–1. Radioactivity of the final silica cake was determined by an α-spectrometer (Alpha Analyst, CANBERRA). Scanning electron microscopy coupled with energy-dispersive X-ray spectroscopy (SEM/EDS, Bruker Nano, Xflash Detector 410-M) was used for morphological and elemental analysis of particulate samples. Solution nephelometric turbidity units (NTU) was measured using a Hanna HI 98703 turbidity meter as an indication of solution clarity.

3. Results and Discussion

3.1. Catalyst Pretreatment and Dissolution

Catalyst pretreatment to remove the tar-like volatile organic matter is imperative to prevent complications during subsequent stages. A 100 L agitator furnace ensured adequate mixing throughout heat treatment (Figure S1A). The spent catalyst was loaded and heated between 550 and 600 °C for 4 h while mixing at 100 rpm. Off-gases were treated using a cyclone burner and cooled in a heat exchanger followed by filtration using a bag filter and a high-efficiency particulate air (HEPA) filter system before being discharged (Figure S2). The pretreated catalyst was then discharged from the agitator furnace and transferred to the dissolver tank.

A 100 L Teflon-coated tank with an external heating jacket was used to dissolve the catalyst in 4 M sodium hydroxide added at a volume of 5 mL g^–1 of catalyst (Figure S1B). Dissolution was performed over a period of 1 h with constant mixing (300 rpm) at a temperature of 105 °C. At this point, uranyl nitrate was introduced to the process. The resultant water glass solution (Na₂SiO₃·H₂O), including uranium ([U] ≈ 200 mg L^–1) and mixed oxide slurry, was passed through a six chamber filter press (Figure S1C) preloaded with a diatomite filter aid.¹⁶ The solid–liquid filtration stage separates the dissolved silica from undissolved residual catalyst solids. Completed filter cakes were discharged from the filter press and considered for waste immobilization (Figure 2A; Section 3.4), while the water glass solution was transferred to the first of two precipitation tanks (Figure S1D; Section 3.2).

Figure 2.

Generated solid wastes. (A) Filter cake of the undissolved mixed oxide slurry (stage 1), (B) precipitated silica filter cake (stage 2), (C) precipitated uranyl phosphate (stage 3), and (D) final wasteform (stage 4).

Thermal pretreatment and catalyst dissolution proceeded as expected. A volume reduction of 47% was obtained based on mass and density differences before and after dissolution. This represented a 5% increase over the laboratory scale (42% volume reduction), yet a minor decrease in the bench-scale tests (48% volume reduction).¹⁶ Solid–liquid separation using the filter press was successful and generated a compacted filter cake composed of the mixed metal oxide residue and a small portion of undissolved silica (Figure 2A). It is possible to add a second dissolution step to recover additional silica, but this resulted in elevated uranium concentrations in the supernatant during preliminary testing.¹⁶ Furthermore, the presence of small amounts of silica in the filter cake is actually found to be beneficial for the formation of the glass–ceramic wasteform and reduces the need to add fresh SiO₂ solids as a glass former material.¹⁸

3.2. Silica Precipitation, Handling, and Purification

Silica was precipitated from the water glass solution by adjusting the pH to below 10 via the addition of sulfuric acid. Precipitation was performed in a 450 L tank at a constant mixing rate of 200 rpm. A portion of uranium was carried over into the solution as part of the prior dissolution stage,¹⁶ simulated by the introduction of uranyl nitrate (Section 3.1). Soluble uranyl peroxocarbonate (UO₂(O₂)(CO₃)₂^4–)²¹ was formed by the addition of hydrogen peroxide and sodium carbonate prior to pH adjustment. Therefore, as the pH decreased by the addition of sulfuric acid, silica precipitated from the solution over the course of 1 h while the uranyl peroxocarbonate remained in the solution. This was to prevent the coprecipitation of uranium with the silica, which would lead to contamination of the silica cake, rendering it unfit for release as clearance. Solid–liquid filtration was performed to separate the precipitated silica from the uranium laden effluent (Figure S1E). The off-white solid silica cake was washed in situ in the filter press twice with water, once with sulfuric acid and further two times with water to further purify the silica for clearance (<1 Bq g^–1 α) (Figures 2B and S3). The remaining color in the silica was because of molybdenum and antimony (Table 2; Figures 3 and S3A). A small amount of uranium (approximately 0.03 wt %, 5 Bq g^–1) within the pores of the filter cake was also present and required removal before free release could be permitted.

Table 2. Typical Contaminant Solution Concentrations Prior to (A) Silica Precipitation and (B) Uranium-phosphate Precipitation^a.

contaminant	U	Si	Sb	Mo	Fe	Al	Mg
(A) silica precipitation-stage 2 (mg L–1)	200	20 000	1800	300	30	30	20
(B) uranium precipitation-stage 3 (mg L–1)	28*	107	38	44	6	4	10

^aConcentrations determined by ICP-OES (* represents that uranium concentration is lower due to dilution).

Figure 3.

SEM/EDS analysis of the precipitated silica prior to purification.

Prior to discharge from the filter press, the silica cake was washed in situ in the filter press several times. First, the cake was flushed with the remaining supernatant solution from the silica precipitation step (referred to as W1 and W2). A water wash followed (W3) before a sulfuric acid (2 M) wash was performed (W4). A second water wash was completed (W5), after which the silica cake was squeezed to remove as much free-standing fluid from within the cake as possible (W6), producing a purified silica cake acceptable for clearance (Figure 2B).

At first, a high NTU value was recorded, indicating that a significant amount of particulate material was released from the filter cake (Figure 4A and Table S1). Washing released further particulate material but the amount decreased as indicated by a reduction in the measured NTU value during repeated washing. Similarly, the uranium concentration post-filter press initially remained high owing to the presence of soluble uranyl peroxocarbonate in the supernatant from the silica precipitation stage that was used to flush the cake (W1 and W2) (Figure 4B). Following the first water wash (W3), the amount of uranium released from the silica cake was low. This indicated that little supernatant, containing uranyl peroxocarbonate, remained within the pores of the silica. Furthermore, it also indicated that any residual uranium remaining within the silica cake was not leached owing to the solution pH being too high to wash the remaining uranium species, insoluble at these pH values, from within the cake; due to low uranium solubility in neutral alkali conditions (Table S1). Residual uranium was successfully leached from the cake during the sulfuric acid washing stage (sulfuric acid = 2 M) (W4) (Figure 4B). Analysis of the subsequent water wash and supernatant after pressing (W5 and W6) (Figure 4B), both with a pH ≤ 1.0, showed that only trace levels of uranium remained, and the concentration of which is technically low enough for free release to the environment (≤1 ppm for uranium-bearing liquid wastes).²⁰ However, the presence of ultrafine particles results in an NTU value greater than the release limit of 1; therefore, additional effluent screening would be required, regardless of the uranium concentration being below the release limit, before its release. The collected washes were transferred to the uranium precipitation tank to await treatment (Section 3.3).

Figure 4.

(A) Supernatant NTU and (B) uranium concentrations post-filter press generated during silica purification. Washing stages are as follows: remaining supernatant solution (W1 + W2), water wash (W3), sulfuric acid wash (W4), water wash (W5), and the supernatant after the filter press squeeze (W6).

The silica was precipitated as amorphous silicon dioxide (Figure S4). Phase identification of the remaining contaminants was not possible owing to either contaminant phases being amorphous or the phase concentrations being lower than the detection limit of the XRD. The freshly prepared cake possesses a high moisture content (76.7%) but this was significantly reduced after air drying (Figure S5), where thermogravimetric analysis confirmed a residual moisture content of 10.6%. A tap density of 0.3454 g mL^–1 was recorded for the air-dried powder. To confirm that the silica was free from uranium and acceptable for clearance, the uranium content was analyzed by an α-spectrometer. The radioactivity of the silica was recorded at 0.335 Bq g^–1 U, which meets the clearance criteria of 1 Bq g^–1 as guided by IAEA.²² The results obtained at the pilot plant were an improvement on bench-scale tests in which the first washed silica cake radioactivity was recorded at 3.45 Bq g^–1 U that required further purification before yielding an activity of 0.56 Bq g^–1 that satisfied the release criteria.¹⁶

The radioactivity of the silica was sufficiently low for clearance; therefore, it does not require disposal at the Gyeongju disposal site. However, due to the Korean law regarding the disposal of such an industrial waste, the silica should be sent to a controlled landfill site. Trace impurities such as Mo (3.32 wt %) and Sb (1.66 wt %) were found to remain within the silica cake despite washing (Figure 3). However, this is of limited concern as the silica cake is destined for a controlled landfill where such impurities do no present an issue.

3.3. Effluent Handling and Treatment

The uranium laden effluent from stage 2 along with the water and acid washes generated during purification of the precipitated silica were mixed in a 1000 L precipitation tank (Figure 1, stage 3; Figure S1F). The stainless steel precipitation tank was equipped with a two-tier quad-bladed overhead stirrer. The combined effluent from the silica precipitation stage was treated with the addition of phosphate. Potassium dihydrogen phosphate was added to reach a total phosphate concentration of 2 mM. The pH was then adjusted to 6.2 by the addition of potassium hydroxide. Fast mixing was performed throughout the phosphate and potassium hydroxide additions before being reduced to a slow stirring for 1 h to aid precipitate formation. Filtration was carried out with a 3 chamber bench-top filter press. Uranyl phosphate solids were collected, while the effluent was screened and collected. Final uranium-free effluents were stored in 1000 L plastic tanks awaiting free release upon project completion.

3.3.1. Uranium-Phosphate Precipitation

The incoming effluent typically totaled approximately 220 L per run, containing 28 mg L^–1 U due to dilution (Table 2). The pH, which was adjusted to 6.2, drifted to 6.4 over 1 h. The formation of lemon-yellow precipitates, also indicated by a change in turbidity, was almost instantaneous. Settling of the precipitates occurred rapidly (<10 min), owing to the particles being relatively dense. Analysis of the unfiltered supernatant after a 1 h settling period showed the uranium concentration to be 1.2 mg L^–1. Sampling the supernatant through filtration removed residual ultrafine particles, thus reducing the uranium concentration to 0.06 mg L^–1, well below the allowable release limit. The recorded uranium concentration of 0.06 mg L^–1 also corresponded to the theoretical thermodynamic solubility limit for meta-ankoleite under the conditions used based on a previous thermodynamic study published by the authors.¹⁷

X-ray fluorescence spectroscopy of the formed solids showed the ratio of uranium to phosphate to be slightly lower than ideal (approximately 1:0.9). It is speculated that the formation of uranium hydroxide accounts for this slight difference, a phenomenon seen during preliminary testing under such precipitation conditions.²³ In contrast, potassium was found in excess, a result of using potassium hydroxide for pH control (Table 3). X-ray fluorescence spectroscopy of the formed solids also showed that coprecipitation of Si, Sb, and Fe had occurred. The Si is likely to be ultrafine particles present from the earlier silica precipitation step. Antimony was also found to be present in the solids owing to its low solubility at circumneutral pH.

Table 3. XRF Analysis of the Solids Formed during Uranium Precipitation (Stage 3).

		U	K	P	Si	Fe	S	Sb	LE
theoretical	wt %	58.8	9.6	7.7					23.7
	ratio U:X	1	1	1
pilot plant solids	wt %	7.02	4.39	0.82	8.35	1.70	1.33	0.34	75.6
	ratio U:X	1	3.82	0.90

X-ray powder diffraction analysis revealed the formation of meta-ankoleite (KUO₂PO₄·3H₂O) (Figure 5), as seen previously at both laboratory and bench scales, suggesting that the same precipitation mechanism occurs regardless of scale.^17,23 Iron was confirmed to have been precipitated as iron phosphate (Figure 5). No XRD peaks could be found for antimony compounds; therefore, it is concluded that those compounds must be amorphous. Minor peaks corresponding to sodium potassium sulfate, present due to the use of sulfuric acid, sodium hydroxide, and potassium hydroxide throughout the process, were also identified but had significant overlap with the larger peaks of meta-ankoleite and iron phosphate.

Figure 5.

XRD pattern for the solids formed during the uranium-phosphate precipitation stage and the corresponding database reference pattern for meta-ankoleite.

3.3.2. Uranium-Phosphate Filtration

Working at the pilot scale gave rise to the opportunity to test a number of filtering scenarios to determine the effective separation of the uranium-phosphate precipitates. Filtering was performed with a filter press, with and without a diatomite filter aid, and also via a single column pleated filter cartridge.^16,17 An approximately 1–2 mm layer of diatomite filter aid was coated on the filter cloth in the filter press prior to the filtration of uranium-phosphate. For filtering, four scenarios were trialed:

• (1)Bulk: The solution, while stirred, was passed through the filter press (Figure S6A).
• (2)Supernatant: The precipitates were allowed to settle for 1 h before the supernatant was passed through the filter press (Figure S6B).
• (3)Settled: The settled precipitates were passed through the filter press (Figure S6C).
• (4)Pleated Filter: The supernatant was passed directly through a single column pleated filter cartridge, completely bypassing the filter press. In this scenario, the effluent was cycled through the pleated filter a total of four times with uranium concentrations in the solution being analyzed after each run (Figure S7).

Under the bulk filtration scenario, the use of no filter aid afforded the fastest initial filtration rates but these quickly retarded due to filter membrane blockages (Figure S6A-1). The average flow rates of 20 and 13 L h^–1 were obtained with and without the use of a filter aid, respectively (Figure S6A-1,A-2). The uranium concentration of the filtrate was initially above allowable release limits both with and without the use of a filter aid but this soon dropped below 1 mg L^–1 within 5 min (Table 4). The use of a filter aid affords greater filtration rates consistently and the generation of good, layered filter cakes (Figure 2C).

Table 4. Uranium Concentration (ppm) of the Filtrate under Different Filtration Scenarios^a.

		time (min)
filter press		1	2	4	8	16	30
without filter aid	bulk	5.8	2.7	0.9	0.2	0.1	0.09
	supernatant	1.2	1.2	0.8	0.8
	settled	5.0	0.9	0.2	0.2	0.1
filter aid	bulk	2.8	1.8	0.7	0.4	0.4	0.07
	supernatant	0.4	0.3	0.3	0.3
	settled	3.7	1.1	0.6	0.4	0.1
background [U] approximately 0.06 mg L–1

^aItalic font indicates uranium concentrations unacceptable for release (>1 mg L^–1).

Opting only to filter the supernatant yielded the best filtration rates due to a lower concentration of particles and, therefore, a slower rate at which particles build up at the membrane surface, impeding filtration (Figure S6B-1,B-2,). The application of the filter aid also removed uranium below the release limit by the time of the first sample point at 1 min (Table 4). For the fourth scenario, results indicate that a single pass directly through a pleated filter cartridge is sufficient, thus bypassing the need to filter the supernatant through the filter press first (Figure S7 and Table 5).

Table 5. Uranium Concentration in the Supernatant after Subsequent Passes through a Pleated Filter System^a.

	pass through
pleated filter	first	second	third	fourth
supernatant	0.12	0.10	0.07	0.06

^aUranium concentrations in mg L^–1.

Filtering the particles after they had been allowed to settle was by far the slowest filtration method with an average rate of 2.3 and 2.1 L h^–1 with and without the use of filter aid, respectively (Figure S6C-1,C-2). Additionally, uranium concentrations post-filtration were comparable with the bulk filtration scenario (Table 4).

Therefore, the recommended filtering method is to allow the particles to settle (1 h) before the supernatant is directly passed through a column pleated filter, bypassing the filter press. The settled particles are then filtered through the filter press with the filter aid present for the formation of a layered filter cake. The resulting supernatant from the filter press is then also passed through the pleated filter before release. Under normal operating conditions the formed filter cake would be sent for immobilization (Section 3.4) as confirmed previously.¹⁸

3.4. Residue Immobilization

The formation of a B₂O₃–SiO₂ glass–ceramic wasteform was previously shown to be the best matrix for immobilization of the uranium residue waste at both laboratory and bench scales.¹⁸ A glass former loading of 3.4 wt % (boron trioxide, B₂O₃) sintered at 1100 °C for a period of 2 h with an initial green-body formation pressure of 30 MPa was found to be optimal with regard to final wasteform volume reduction (45.69%), strength (89.45 MPa), and uranium leachability (1.1 × 10^–3 g m^–2 d^–1), which are all important parameters for determining wasteform validity.¹⁸

Undissolved residual catalyst solids from stage 1, including the diatomite filter aid, were oven-dried at 100 °C for 6 h (Figure 6A), removing excess moisture before being ground to a fine powder in a super-mixer rotary mill together with the glass former (B₂O₃) and water at 5 wt % (Figures 6B and S8A). Previously, the diatomite filter aid has been shown to be a sufficient source of silica required for the formation of the sintered B₂O₃–SiO₂ glass–ceramic.¹⁸ Discharged powders were transferred to a 200 ton custom-built hydraulic press fitted with a quarter circle shaped mold (radius = 200 mm) (Figure S8B,C). The mold was pre-coated with the zinc stearate lubricant for easy ejection of the green body. The powders (3 kg) were pressed into a green body (radius = 200 mm, height = 35–40 mm) under a pressure of approximately 60 MPa for 10 min. Ejected green bodies were then heated at 200 °C for 2 h before being sintered at a temperature of 1100 °C in air for 4 h in a custom-built muffle furnace (Figure S8D). Figure 6C,D show an example green body and final sintered body, respectively. The radius and height of the sintered bodies measured approximately 170–175 mm and 30–35 mm, respectively. This represented average radial and axial shrinkage of 13.55% (± 1.25%) and 12.00% (± 1.57%), respectively, across five samples compared to the green bodies. The volume of the sintered bodies decreased by approximately 35–40% compared to the corresponding green bodies while showing isotropic shrinkage; this was slightly lower than the laboratory tests in which a 45% volume decrease was obtained.^18,24 A compressive strength of 195 MPa was measured, far exceeding the required WAC of 3.44 MPa, while antimony was confirmed to be present as Sb(V) in the form of antimony(V) oxide (Sb₂O₅) and the mixed oxide tripuhyite (FeSbO₄) as seen previously (Figure S9).¹⁸

Figure 6.

Waste immobilization steps. (A) Oven-dried residual waste filter cake from stage 1. (B) Blended waste and glass former ejected from the rotary mill. (C) Molded green body from the hydraulic press. (D) Final sintered wasteform prepared from the green body.

The main goal at this stage of the process was to evaluate the scale-up potential of the earlier laboratory and bench-scale trials in terms of the physical parameters of the wasteform. Uranium was omitted from the final wasteform, primarily for two reasons. First, the pilot-scale wasteform manufacturing equipment was housed outside of the radiation supervised area and, therefore, handling of uranium was not permitted. Second, as the supplied catalyst was a U-free catalyst, no uranium was present within the residual solids of the dissolved catalyst. Although omitting uranium is not ideal, its absence was deemed to have a small impact on the final wasteform. Uranium-containing wastes have been successfully immobilized during our earlier laboratory and bench-scale tests in which the measured leaching rate of uranium was of the order of 10^–3 to 10^–4 g m^–2 d^–1,¹⁸ comparable to that of SYROC developed for immobilization of high-level radioactive waste.²⁵

4. Conclusions

The spent catalyst treatment process was successfully scaled up and verified at the pilot plant scale following previous success at both laboratory and bench scales. The catalyst support material (SiO₂) was successfully removed from the catalyst by selective dissolution, separated, and was sufficiently treated to meet relevant clearance criteria (≤ 1 Bq g^–1 α) (Figure 2B). The codissolved uranium that remained in the solution was successfully removed via meta-ankoleite formation, precipitation, and filtering, enabling the free release of the treated effluents ([U] ≤ 1 mg L^–1). Waste residues remaining after dissolution of the silica were sintered to form a stable glass–ceramic composite matrix wasteform suitable for its disposal according to Korean regulatory guidelines (Figure 2D). The volume of the sintered body decreased by approximately 35–40% compared to the green body while showing isotropic shrinkage; this was slightly lower than the laboratory tests in which a 45% volume decrease was obtained. A compressive strength of 195 MPa was measured, while antimony was confirmed to be present as Sb(V), as seen previously. For final disposal, it is proposed that the wasteforms will be stacked in a 200 L drum before being sent to the Gyeongju disposal site. The final volume reduction yield of the immobilized waste was up to 80% with respect to the volume of the initial catalyst waste. This would potentially lead to fewer than 1500 of the original 7000 drums being sent for disposal at the Gyeongju disposal site representing a potential cost saving of around US$70 million, not factoring for plant construction or operating costs. Process acceptance has been granted by the private company that has the uranium catalyst waste, with process commercialization expected to commence in 2021; final regulatory approval pending.

Glossary

Abbreviations Used

EDS

energy-dispersive X-ray spectroscopy

HEPA

high-efficiency particulate air

IAEA

International Atomic Energy Agency

ICP-OES

inductively coupled plasma optical emission spectrometry

KAERI

Korea Atomic Energy Research Institute

LLW

low-level waste

LOD

limit of detection

NTU

nephelometric turbidity units

SEM

scanning electron microscope

SENSEI

Selective Extraction of Nonradioactive Species, Encapsulation, and Immobilization

SOHIO

Standard Oil of Ohio Company

TGA

thermogravimetric analysis

WAC

waste acceptance criteria

XRD

X-ray powder diffraction

XRF

X-ray fluorescence

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.0c00723.

• Summary of solution parameters after silica cake washing (Table S1); main pilot plant equipment used for the treatment of the spent uranium catalyst (Figure S1); schematic of the off-gas treatment system (Figure S2); silica cake before and after purification, and supernatant samples collected during silica washing (Figure S3); XRD pattern of the precipitated silica (Figure S4); TGA analysis of the precipitated silica cake (Figure S5); and diagram of the three filtering scenarios employed for the removal of uranium-phosphate precipitates from solution (Figure S6); diagram of the fourth filtering scenario employed for the removal of uranium-phosphate precipitates from solution; supernatant passed directly through a pleated filter cartridge (Figure S7); main pilot plant equipment used to produce the final wasteform (Figure S8); XRD pattern of the final sintered wasteform produced for final waste immobilization (Figure S9) (PDF)

Author Contributions

This manuscript was written through contributions of all authors. R.I.F.: Lead manuscript author, pilot plant experiments, characterization and data analysis; M.O.: Pilot plant experiments and sample analysis; K.L.: Team leader, project coordination; K.-W.K.: Project lead, manuscript preparation and revision. All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.