On-line monitoring for process control and safeguarding of radiochemical streams at spent fuel reprocessing plants

Samuel A. Bryan, T.G. Levitskaia, A.J. Casella

Pacific Northwest National , 902 Battelle Boulevard, PO Box 999, Richland, WA, 99352, USA

Abstract. The International Atomic Energy Agency has established international safeguards standards for fissionable material at spent nuclear fuel reprocessing plants to ensure that significant quantities of weapons-grade nuclear material are not diverted from these facilities. Currently, methods to verify material control and accountancy (MC&A) at these facilities require time-consuming and resource- intensive destructive assay. Leveraging new on-line nondestructive assay (NDA) techniques in conjunction with the traditional and highly precise destructive assay methods may provide a more timely, cost-effective and resource-efficient means for MC&A verification at such facilities. Pacific Northwest National Laboratory (PNNL) is developing on-line NDA process monitoring , including a -based monitoring system, to potentially reduce the time and resource burden associated with current techniques. The spectroscopic monitor continuously measures chemical compositions of the process streams including actinide metal (U, Pu, Np), selected fission products, and major non-radioactive flowsheet chemicals using ultraviolet, visible, near infrared (UV- vis-NIR) and . This paper provides an overview of the methods and reports PNNL’s ongoing efforts to develop and demonstrate the technologies. PNNL’s ability to identify material intentionally diverted from a liquid-liquid contactor system was successfully tested using on-line process monitoring as a means to detect the amount of material diverted. A chemical diversion and detection of that diversion, from a solvent extraction scheme was demonstrated using a centrifugal contactor system operating a tributyl phosphate-based extraction. A portion of the feed from a counter-current extraction system was diverted while a continuous extraction experiment was under way; the spectroscopic on-line process monitoring system was simultaneously measuring the feed, raffinate, and organic products streams. The amount observed to be diverted by on-line spectroscopic process monitoring was in excellent agreement with values based from the known mass of sample directly diverted from system feed solution.

1. Introduction

Raman spectroscopy [1-6] and ultraviolet-visible-near infrared (UV-vis-NIR) [5-11] are analytical techniques that have been extensively used for measuring the concentrations of various organic and inorganic compounds including actinides. The corresponding spectrometers used under laboratory conditions are easily convertible to process-friendly configurations allowing remote measurements under flow conditions. In the proposed monitoring system, fibre optic Raman and UV-vis-NIR probes allow monitoring of the chemical components encountered in various aqueous and organic streams within tri-butyl phosphate- (TBP) based extraction flowsheets. A Raman probe is applied to monitor high concentration species, including metal oxide ions such as uranyl , components of the organic solvent such as n-dodecane and TBP, and inorganic oxo-anions. Other actinides and lanthanides are monitored by UV-vis-NIR spectroscopy. Toward this goal, Pacific Northwest National Laboratory (PNNL) performed spectroscopic measurements of actual commercial used fuel, and simulated fuel reprocessing solutions.

The scheme depicted in Fig. 1 shows a typical separation process using co-decontamination (CoDECON), a modified plutonium uranium extraction (PUREX) for U, Pu, and Np separation followed by transuranium extraction (TRUEX) to remove fission products, then by trivalent actinide- lanthanide separation by phosphorus reagent extraction from aqueous komplexes (TALSPEAK) to

1 separate the lanthanides from actinides. The general concept is to place an on-line monitoring system with probes at feed inlet and product outlet streams, which then can yield real-time concentration and inventory of specific metals and reagents of interest. Every process step contains Raman and/or UV- vis-NIR active species amenable to spectroscopic on-line process monitoring and therefore is non- flowsheet specific.

dissolved fuel

U CoDECON U, Pu, Np Tc



rare earths

FIG. 1. Typical separation process showing the flow of material starting with dissolved fuel with various product streams. Process monitoring can be deployed at various points in the flowsheet to provide mass balance for specific material.

2. Experimental

All reagents including Nd(NO3)3·6H2O, NaNO3, HNO3, TBP, and n-dodecane were purchased from Sigma Aldrich and used as received, except for the TBP/n-dodecane solvent, which was water-washed prior to use. UO2(NO3)2 salts were from in-house stocks. The Approved Testing Material (ATM-109) fuel used in this analysis was part of a General Electric zirconium barrier lead test assembly [12]. The segment of fuel used in this work had an initial enrichment of 3 percent and a burnup of approximately 70 MWd/kgM, a relatively high burnup value for a commercial used fuel.

Raman spectra were collected on an InPhotonics Inc. RS2000 echelle spectrograph. The system was equipped with a stabilized 670-nm, 150-mW, visible diode laser as the excitation source. Samples were measured with an InPhotonics focused fibre-optic probe (RamanProbe™) with a thermoelectrically cooled, charge-coupled device detector. Vis-NIR measurements were performed using an Ocean Optics USB2000 Miniature Fiber Optic Spectrometer (500 to 1200 nm wavelength detection range). NIR measurements in the 900 to 1700 nm range were performed using Ocean Optics NIR-512 temperature-regulated NIR spectrometer.

Chemometric modelling of the data utilized a commercial software package partial least squares (PLS) toolbox for MATLAB (version 7.14, The MathWorks Inc., Natick, Massachusetts) to perform the multivariate analysis using an approach similar to that described in Casella et al. [13].

3. Results and discussion

The goal of this study is to demonstrate the utility of optical process monitoring methods for the direct measurement of dissolved fuel feed, organic product, and raffinate phases, employing Raman and vis- NIR spectroscopic methods. The following sections 1) contain results demonstrating the ability to use Raman and vis-NIR spectroscopy for the quantitative determination of metals within actual used commercial fuel solutions, 2) give an overview of the on-line spectroscopic measurement

2 methodology, and 3) provide a summary of a process monitoring demonstration showing diversion detection within a centrifugal contactor based fuel simulant reprocessing process.

3.1. Spectroscopic measurement of spent commercial fuel samples

Solutions designated for spectroscopic analysis using ATM-109 commercial fuel, were prepared in a shielded hot cell facility at PNNL. The fuel segments underwent decladding, dissolution, solids removal, and acid adjustment to yield five solutions with varying nitric acid concentration (0.3, 1.3, 2.5, 3.8 and 5.1 M HNO3), each with approximately 0.7 M uranium metal content. Each aqueous feed was contacted with an organic extractant phase, containing 30% vol TBP in n-dodecane. Fig. 2 is a photograph of the organic phase after contact with the five aqueous acid phases, showing the resulting solution sufficiently clear for purposes of optical spectroscopic analysis.

FIG. 2. Photograph of organic product (TBP/n-dodecane phase) after contact with ATM-109 fuel feed solution.

The original feed, organic product and raffinate phases were analyzed by vis-NIR and Raman spectroscopy. Fig. 3 depicts the vis-NIR spectra for the five aqueous fuel feed solutions, showing that plutonium, in both Pu(IV) and Pu(VI) oxidation states is observed in the commercial fuel feed, with varying concentrations of each depending on the HNO3 concentration of the feed. Neptunium as Np(V) and Np(VI) is also evident in the fuel feed solutions as depicted in Fig. 3.



Nd(III) Pu(IV) Pu(IV) + H2O Np(V) Np(VI) + H O absorbance Np(V)-UO 2 Nd(III) 2 Pu(VI) Pu(VI)


wavelength, nm FIG. 3. Visible-NIR absorption spectra of dissolved ATM-109 Fuel showing Pu(IV, VI), Np(V, VI), and Nd(III) species in solution. Solutions were adjusted to variable nitric acid as indicated in the figure.

3 Table I contains the concentrations for U, Pu, Np, and Nd for the ATM-109 feed as determined by spectroscopic analysis. This table also contains the comparison with the standard inductively coupled plasma (ICP-MS) analysis for these species. The ratio of the spectroscopic/ICP method is shown (ratio = 1.0 is an ideal match) indicating good agreement between spectroscopic and ICP method. UV-vis-NIR was used for Pu, Np, and Nd analysis, and Raman spectroscopy was used to determine U concentrations. Details of the fuel dissolution and spectroscopic analysis are given in Bryan et al. 2011 [5].

Table I. Concentrations of U, Pu, Np, and Nd for ATM-109 fuel feed as determined by spectroscopic analysis and comparison with ICP-MS analysis. Data from Bryan et al. 2011 [5]. Concentrations in Molar units. ATM-109 U Pu Np Nd ICP-MS 0.721 8.99E-03 4.7E-04 0.84E-02 Spectroscopy 0.719 8.90E-03 4.7E-04 1.10E-02 Spectroscopic / ICP ratio 1.0 0.99 1.0 1.3

3.2 On-line spectroscopic measurement methodology The methodology for developing and using models for on-line monitoring and analysis is shown in Fig. 4 using Pu(IV) as a specific example. Initially, training set spectra of known composition of species of interest are acquired for model building, as shown in Fig.. 4A. This figure contains variable Pu(IV) spectra (0.1 – 10 mM) in a 0.8 M HNO3 solution containing 1.33 M UO2(NO3)2. Additional spectra are measured containing variable Pu(IV) under a variety of acid and other solutions species. These training set data are then used to develop chemometric models for the prediction of analytes from their respective spectra, as shown in Fig. 4B. The models are then applied to new spectra, collected individually, or collected in consecutive order, yielding time-dependent concentration information.

In the example shown in Fig. 4C, vis-NIR absorption measurements were taken for a period of approximately 1 hour using a solution containing approximately 0.9 mM Pu(NO3)4 in 1.33 M UO2(NO3)2 and 0.8 M HNO3, representing 230 individual spectra taken over that time span. The vis- NIR absorption data were subjected to chemometric analysis, with the results of the analysis shown in Fig. 4D, displaying Pu(IV) concentration (mM) versus time. The chemometric analysis yielded a mean concentration of 0.929 mM plutonium (IV) with a standard deviation of ± 0.016 mM, which represents a 1.7 percent relative standard deviation for the plutonium (IV) measurement over the 1 hour time of data collection. In this example, the for plutonium (IV) was determined to be 0.08 mM, sufficient for analyzing Pu at relevant fuel reprocessing concentrations.

A similar analytical approach was used for other oxidation states of plutonium (Pu(III) and Pu(VI)) as well as the common oxidation states of neptunium (Np(III), Np(IV), Np(V) and Np(VI)). Uranium, as 2+ the uranyl ion, [UO2] , was determined using Raman spectroscopy.

3.3 Process monitoring demonstrating diversion detection using centrifugal contactor system To test the on-line and near real-time aspects of process monitoring performance during a mock- diversion experiment, a counter-current liquid-liquid extraction testing apparatus for use with spent fuel separations processing was assembled. The counter-current design is based on a single bank of four 2-cm centrifugal contactors. Fig. 5A shows a schematic of a centrifugal contactor detailing the inlets and outlets for the countercurrent aqueous and organic phases. The bank was instrumented with vis-NIR and Raman spectroscopy probes, and with density/flow meters connected on the influent and effluent for both the organic and aqueous lines. Fig. 5B is a photograph of the centrifugal contactors. Fibre-optic cables are used to connect the spectroscopic instrumentation to the solution probes attached to the centrifugal contactors. Fig. 5C contains a schematic representation of the bank of contactors used with the locations of feed, raffinate, organic inlet, and loaded organic product streams shown, along with the location of a diversion valve situated between the spectroscopic probes for feed

4 and the inlet line to the bank of contactors. The vis-NIR and Raman monitoring probes are positioned on the feed, raffinate, and organic product streams. Pu chemometric model PLS model for Pu(IV) using UV-vis data 12 0.7 0.7 542 Bnm y = 0.9997x + 0.0006 A 2 0.6 0.6 10 R2 = 0.9999 R = 0.998 659 nm

0.5 0.5 797 nm 8 R2 = 0.999 0.4 0.4 854 nm

R2 = 0.998 , predicted , 6 R2 = 0.997

0.3 0.3


Absorbance Absorbance absorbance 0.2

0.2 4 Pu(IV), mM predicted mM Pu(IV),


0.1 [Pu(IV)], 2 0 0 0 2 4 6 8 10 12 450 550 650 750 850 950 Pu(IV)0 added, mM wavelength,wavelength, nm nm 0 2 4Pu(IV),6 mM8 10 12 [Pu(IV)],Pu(IV), mM mM



Pu(IV), absorbance

time, min wavelength, nm time, min FIG. 4. (A) Absorption Spectra of 0.1 – 10 mM Pu(IV) in 0.8 M HNO3 and 1.33 M UO2(NO3)2. (B) PLS prediction of Pu(IV) in simulant feed solution. (C) Visible-NIR spectra for a solution of approximately 0.9 mM Pu(NO3)4 in a matrix of 1.33 M UO2(NO3)2 and 0.8 HNO3. (D) The results of chemometric analysis of spectra containing Pu(NO3)4 as a function of time. Analysis yielded a mean concentration of 0.929 mM plutonium (IV) with a standard deviation of ± 0.016 mM, which represents a 1.7 percent relative standard deviation.

The centrifugal contactor system test was initially started with an aqueous feed composed of water and an organic phase of 30 vol% TBP in n-dodecane at flow rates of approximately 11 and 10.5 mL/min, respectively. When the aqueous feed/raffinate and organic solvent/product flows reached steady state (i.e., the same volumes flowing in and out of the system), the aqueous feed was switched from water to the feed solution containing 20 mM Nd(NO3)3 in 4 M NaNO3/0.1M HNO3. Switching from water to the neodymium feed solution signalled the start of the test. After the steady state flow for both aqueous and organic streams was sustained for 87 minutes, a diversion valve (shown in schematic in Figure 5C) was opened at the entrance into the contactor feed inlet, and a fraction of the feed solution was diverted at an average flow rate of 3.3 mL/min over the time of diversion. After approximately 47 minutes, the diversion of material was stopped (i.e., 134 minutes after the start of experiment), and the normal feed flow (i.e., with no diversion) into the contactor system was re-established.

Spectroscopic monitoring of the feed, raffinate, and organic product steams were recorded during the entire flow test. Fig. 6A contains the vis-NIR measurements of the aqueous feed during the continuous contactor experiment. The characteristic absorbance bands for the Nd3+ ion are shown in this series of spectra taken over the entire time (0 to 150 minutes). The location of the spectroscopic probes on the feed inlet is upstream from the point of sample diversion during this experiment, and therefore the series of absorbance spectra for the feed are not changed during the time of diversion. This is not the case for the spectra associated with the organic and raffinate streams. Fig. 6B and 6C contain the series of vis-NIR spectra measured at the organic product and raffinate locations during the extraction experiment, respectively. The vis-NIR spectra show the typical absorbance spectra associated with

5 Nd3+ ion in solution, and show the increase in Nd3+ in the organic product and raffinate phases after the initiation of the experiment and a plateau of absorbance after about 20 minutes after the start of the experiment. However, at the approximate point of diversion (87 minutes, for both organic product and raffinate) the absorbance value for the Nd3+ band significantly decreases, and stays at a suppressed level until the diversion was stopped (at time = 134 minutes), after which the measured absorbance increased to its pre-diversion value.

A A B B C raffinate organic product organic aqueous solvent feed

FIG. 5. (A) Schematic diagram of a centrifugal contactor showing the inlet and outlets for the aqueous and organic phases. (B) Photograph of the centrifugal contactor system; the bank of centrifugal contactors is visible in the photo. (C) Schematic representation of the bank of contactors used in our study; the feed, raffinate, and loaded organic product streams are instrumented with vis-NIR and Raman probes as well as flow meters, locations shown as blue rectangles in figure. The location of the feed diversion valve is shown in between the feed spectroscopic probe, and the inlet to the bank of contactors.

Concentrations of the Nd3+ ion in the feed, organic product, and raffinate phases from the spectra in Figure 6 were determined by employing PLS model analysis using Nd3+ standards measured in both aqueous and organic phases. By combining the flow rate information with the concentration of Nd3+ in each phase, the cumulative total (integrated amount) of Nd3+ detected in each stream was determined and is shown in Fig. 7A. This figure shows the integrated total of Nd measured in the feed, raffinate, and organic product over the entire experiment. The sum of the Nd3+ amount in the organic product and raffinate steams is also shown (labelled “raffinate + organic”, in Fig. 7A) should equal the total amount of Nd3+ measured for the feed. The curve for the “raffinate + organic” is parallel with that for the feed measurement prior to the diversion point at 87 minutes into the experiment. There is a constant difference between the “raffinate + organic” and feed because of in-process amount of Nd3+, which is still within the contactor system and not yet measured by the spectroscopic probes on the outlet of the system. Fig. 7B shows the difference (delta) between the inlet (feed + solvent) minus the outlet (raffinate + organic product) streams during the solvent extraction experiment.

diversion of 3 ml/min started at 87 min diversion of 3 ml/min diversion stopped diversion stopped started at 87 min A at 134 min B at 134 min C

diversion of 3 ml/min started at 87 min diversion stopped

at 134 min


absorbance absorbance

wavelength (nm) wavelength (nm) time (min) wavelength (nm) time (min) time (min) FIG. 6. Visible spectra showing Nd3+ in (A) feed, (B) organic product, and (C) raffinate as a function of time during centrifugal contactor experiment.

6 This difference in measurement is labelled “delta from in-process” within Figdelta. 7 B,from and diversion is the difference in moles of Nd3+ between the two curves (delta-y on xy-plot). After the start of diversion, beginning at 87 minutes, the “raffinate + organic” curve in Fig. 7A further deviates from the “feed” curve, and during the time in which diversion is occurring (between 87 and 134 minutes) the two curves are no longer parallel. After diversion is stopped (at 134delta min fromutes in-process) the “raffinate + organic” curve then returns to being parallel with the “feed” curve. By extrapolating a line from the “organic + raffinate” curve prior to diversion, one can then measure the amount of Nd3+ material diverted by subtracting the extrapolated value (prior to diversion) from the measured value after diversion. The “delta from diversion” is also shown within Figure 7B. This difference was measured to be 3×10-3 moles of Nd3+ diverted based on the analysis of the spectroscopic data in Figures 6 and 7. This value is in excellent agreement with the value measured by inductively coupled plasma optical emission spectrometry (ICP-OES) (2.9×10-3 moles of Nd3+).

A ddeltaelta from from diversion B

delta from diversion deltadelta from from in -inprocess-process

delta from in-process

FIG. 7. Nd3+ extraction and measured in feed, raffinate, and organic product streams showing (A) detection of Nd from contactor feed and product stream concentration profile as a function of time; and (B) differencedelta from of diversion integrated total amount of Nd between feed and product streams as a function of time.

4. Conclusions delta from in-process

PNNL is developing a spectroscopy-based NDA monitoring system, capable of measuring metal concentrations within dissolved commercial used nuclear fuel and fuel simulant reprocessing solutions. The system has been demonstrated for quantitative analysis for U, Pu, Np, and Nd metals within actual used commercial fuel, with results comparable to ICP-MS analysis. Satisfactory prediction of the analyte concentrations in these preliminary experiments warrants further development of the spectroscopy-based methods for radiochemical process control and safeguarding.

Additionally, using a centrifugal contactor system instrumented with on-line process monitoring probes, the ability to identify and quantify material intentionally diverted from a liquid-liquid extraction contactor system was successfully demonstrated. During a counter-current extraction experiment designed to mimic a PUREX-type extraction, a portion of the feed solution was diverted; during this diversion experiment, spectroscopic monitoring probes simultaneously measured feed, raffinate, and organic products streams. The amount diverted was determined by spectroscopic on-line process monitoring (3 x 10-3 mol Nd3+) which is in excellent agreement with the value based on the known sample diverted from the system feed solution measured by ICP-OES (2.9 x 10-3 mol Nd3+). PNNL concludes that near real-time spectroscopic process monitoring provides a tool for the immediate detection of material being diverted from nuclear material process streams.

5. Acknowledgments

This research was supported in part by the Material Recovery & Waste Form Development Campaign within the U.S. Department of Energy’s Fuel Cycle Research and Development Program (NE-5); the U.S. National Nuclear Security Administration (NNSA) Nonproliferation and International Security (NA-24) office with the U.S. Department of Energy (DOE); and by the Laboratory Directed Research

7 and Development Program at the Pacific Northwest National Laboratory operated by Battelle for the U.S. Department of Energy under Contract DE-AC05-76RL01830. Pacific Northwest National Laboratory is a multi-program national laboratory operated by Battelle for the U.S. Department of Energy.


[1] MADIC, C., G.M. BEGUN, D.E. HOBART, AND R.L. HAHN, Raman-Spectroscopy of Neptunyl and Plutonyl Ions in Aqueous-Solution - Hydrolysis of Np(VI) and Pu(VI) and Disproportionation of Pu(V). Inorganic , 1984. 23(13):1914-1921. [2] MADIC, C., D.E. HOBART, AND G.M. BEGUN, Raman Spectrometric Studies of Actinide(V) and Actinide(VI) Complexes in Aqueous Sodium-Carbonate Solution and of Solid Sodium Actinide(V) Carbonate Compounds. , 1983. 22(10):1494-1503. [3] NGUYENTRUNG, C., G.M. BEGUN, AND D.A. PALMER, Aqueous Uranium Complexes .2. Raman-Spectroscopic Study of the Complex-Formation of the Dioxouranium(VI) Ion with a Variety of Inorganic and Organic-Ligands. Inorganic Chemistry, 1992. 31(25):5280-5287. [4] BRYAN, S.A., T.G. LEVITSKAIA, A.J. CASELLA, J.M. PETERSON, A.M. JOHNSEN, A.M. LINES, E.M. THOMAS, AND C. ORTON, Spectroscopic On-Line Monitoring for Process Control and Safeguarding of Radiochemical Streams, in Advanced separation techniques for nuclear fuel reprocessing and radioactive waste treatment. K.L. Nash and G.J. Lumetta, Editors. 2011, Woodhead Publishing Ltd., CRC Press LLC., Cornwall, UK. [5] BRYAN, S.A., T.G. LEVITSKAIA, A.M. JOHNSEN, C.R. ORTON, AND J.M. PETERSON, Spectroscopic monitoring of spent nuclear fuel reprocessing streams: an evaluation of spent fuel solutions via Raman, visible, and near-. Radiochimica Acta, 2011. 99(9):563-571. [6] BRYAN, S.A., T.G. LEVITSKAIA, J.M. SCHWANTES, C.R. ORTON, J.M. PETERSON, AND A.J. CASELLA, Monitoring, Controlling and Safeguarding Radiochemical Streams at Spent Fuel Reprocessing Facilities, Part 1: Optical Spectroscopic Methods. International Journal on Nuclear Energy Management and Safety, 2012. [7] SCHMIEDE, H. AND E. KUHN, Automatic Measurement and Control of Nuclear Fuel Reprocessing by Spectrophotometry and Conductivity Measurements. Chemie Ingenieur Technik, 1972. 44(3)104. [8] ERTEL, D. AND G. HORN, Analytical Methods in the Purex Process. Atomkernenergie- Kerntechnik, 1985. 46(2):89-94. [9] ERTEL, D., P. GROLL, G. KNITTEL, AND W. THESSIS, Process Analysis in Purex Method. Journal of Radioanalytical Chemistry, 1976. 32(2):297-314. [10] BAUMGARTNER, F. AND D. ERTEL, The Modern Purex Process and Its Analytical Requirements. Journal of Radioanalytical Chemistry, 1980. 58(1-2):11-28. [11] COLSTON, B.J. AND G.R. CHOPPIN, Evaluating the performance of a stopped-flow near- infrared spectrophotometer for studying fast kinetics of actinide reactions. Journal of Radioanalytical and , 2001. 250(1):21-26. [12] WOLF, S.F., D.L. BOWERS, AND J.C. CUNNANE, Analysis of high burnup spent nuclear fuel by ICP-MS. Journal of Radioanalytical and Nuclear Chemistry, 2005. 263(3):581-586. [13] CASELLA, A.J., L.R. HYLDEN, E.L. VALERIO, J.M. PETERSON, G.J. LUMETTA, T.G. LEVITSKAIA, AND S.A. BRYAN. Advances in On-Line Monitoring for Weak Acid Based Nuclear Fuel Reprocessing Schemes in American Nuclear Society, Winter Meeting. 2013. Washington, D.C.