Liquid-liquid solvent extraction processes have been widely used for aqueous radioactive material separations. These processes employ two immiscible liquid phases, aqueous and organic, to preferentially extract from one phase to the other the desired components of the feed. A solvent extraction cycle consists of up to four operations: extraction, scrubbing, stripping and washing, which are illustrated in Fig. 3.11 and exemplified by the reprocessing of UNF.

Reprocessing entails the separation of plutonium and unused uranium from the fission products and other transuranics in UNF. In this reprocess­ing application, the UNF is dissolved in nitric acid and the aqueous solution is mixed with an immiscible organic liquid or solvent to form a dispersion of aqueous or organic phase drops in the continuous other-phase. The solvent comprises the extractant, tri-butyl phosphate that will preferentially extract uranium and plutonium, dissolved in a diluent, which is typically a kerosene. The aqueous and organic phases are separated after adequate time mixing to ensure sufficient extraction of the desired species. A small quantity of fission products and other transuranics are inevitably extracted but these are separated from the organic phase in the scrub step by contact­ing it with a relatively small volume of dilute nitric acid. Uranium and plutonium are then stripped into dilute nitric acid by further contacts with the solvent. The solvent is finally washed of acidic degradation products by contacting it with sodium carbonate and dilute nitric acid to make it ready for further extraction.

In industrial practice, each step of the solvent extraction cycle step is repeated a number of times to achieve the desired separation of uranium and plutonium. The liquid phases are mixed and separated in process

3.11 Typical solvent extraction cycle.

equipment collectively known as contactors. The cycle is additionally con­figured so the aqueous phase flows from contactor to contactor in the opposite direction to the organic phase, or counter-currently, to increase efficiency.

A wide range of contactors for solvent extraction have been used in general industry and these fall into two overall types: stage-wise and dif­ferential. Stage-wise contactors, typified by mixer-settlers, are composed of a number of discrete stages in which the two phases are mixed together and brought to chemical equilibrium before being physically separated and passed counter-currently to the adjacent stages. In differential contactors, the two phases do not reach equilibrium at any point in the contactor. Instead, mass transfer occurs throughout the unit driven by a continuous concentration gradient as the phases flow counter-currently. The phases are only separated as they exit the contactor.

For successful nuclear facility application, contactors must:

• develop sufficient interfacial area between the two phases to promote the desired transfer of the extractable components;

• facilitate counter-current flow of the two phases, while avoiding exces­sive entrainment of one phase in the other as they leave the contactor;

• have flexibility to operate over a range of phase flow rates, ratios and feed concentrations;

• be reasonably compact with low hold up of the process solutions;

• be mechanically reliable and/or easy to maintain.

Reliability is particularly important and a maintenance-free design with no moving parts within the hot cell is preferable. A low holdup of process solutions is important for nuclear applications, since this reduces the aque­ous-solvent contact time and hence the degradation of the solvent by the radioactivity of the aqueous phase. It also facilitates designs that are safe by geometry from nuclear criticality at higher than very dilute concentra­tions of fissile material.

A summary of solvent extraction contactors that are currently in use, or that have been proposed, is shown in Table 3.1. It can be seen that, in the nuclear industry, only three types of contactor are in current use: mixer — settlers, pulsed perforated plate columns and centrifugal contactors.

With the above brief explanation of the underlying chemistry, the explana­tion of the codecontamination cycle is more fruitful. The reader’s attention is re-directed to the decontamination cycle in Fig. 6.2 for the ensuing discussion.

The dissolved fuel solution enters the main extraction contactor at mod­erate nitric acid concentration of ~3 M HNO3. As the aqueous flows coun­ter-currently to the organic stream, U and Pu are extracted virtually quantitatively into the organic phase. Under these conditions, Zr, Ru, and Tc are also extracted to varying degrees in accord with the above descrip­tions of their chemistries. Also noteworthy is that the conditions of the dissolved solution favor the formation of neptunium in the higher, i. e., Np(V) and Np(VI), oxidation states; the former is inextractable while the

latter is very extractable by TBP into the organic phase. The proportion — ation between these Np oxidation states is a complex function of the exact NO3-/NO2- concentrations and redox kinetics, which can even be influenced by the type and design of the process equipment. It has been reported that ~25% of the Np is rejected, as Np(V), to the raffinate (waste) in the La Hague plant (Dinh 2008); however, the point is that some fraction of the Np in the dissolved feed is extracted into the organic phase and subse­quently carries through into the downstream processes.

Cyanex 301 was investigated for separating trivalent actinides from the TRUEX product. (Fig. 7.10) (Vandegrift, 2004). Cyanex 301 is a commercial product supplied in an impure form by Cytec Industries, Canada. The pre­dominant ingredient is bis(2,4,4-trimethylpentyl)dithiophosphinic acid, which must be extensively purified before it can be used for actinide/lan — thanide separations. The purified Cyanex 301 was dissolved in a mixture of TBP and n-dodecane. The scrub feed is the acid form of the weak complex — ant in the feed and the strip feed is an ammonium salt of a powerful com — plexant. Because of the tendency of the Cyanex 301 to decompose both as a solid and in solution, it was not deemed a suitable material for industrial — scale actinide/lanthanide separations. Subsequent investigations focused on using the TALSPEAK process to separate the trivalent lanthanides from the actinides.

Since the TRUEX strip is a buffered lactic acid feed similar to the TALSPEAK feed, only a minor pH adjustment is required to prepare the TRUEX strip solution for feed to TALSPEAK. The TALSPEAK process segment has three sections: extraction, scrub, and strip, as shown in Fig. 7.11 (Pereira, 2007b). The actinide:lanthanide separation achieved by TALSPEAK is based on the preferential complexation of actinides by aminopolyacetic acids and effective extraction of trivalent lanthanides by bis(2-ethylhexyl)phosphoric acid (HDEHP). In the extraction section, the more weakly complexed lanthanides are extracted, while the stronger actinide/DTPA complexes remain in the aqueous phase. The scrub removes the small fraction of Am and Cm that are co-extracted by HDEHP. The lanthanides are stripped from the solvent with moderately concentrated nitric acid.

The potential for using the UNEX-extractant for the treatment of Idaho acidic HLW required substantiating all process characteristics, primarily all its hydrodynamic properties, as applied to extraction equipment on an industrial scale.

For this purpose, the main operations of the UNEX process work flow were checked by using Idaho simulated HLW on the EZR125 commercial

centrifugal contactor at the Research and Construction Institute of Assembling Technology (NIKIMT, Moscow). The operating conditions for the experiments corresponded to those of the previously performed tests for the improved work flow using real HLW. A basic diagram of the test rig for the UNEX process using a commercial centrifugal contactor is pre­sented in Fig. 9.8.

The tests of the UNEX process using industrial equipment provided the following evidence:

• the centrifugal contactor EZR125, working under UNEX process condi­tions, has a flowrate of 800 L/hour in total for all phases of the process (this corresponds to with the PUREX process system, with 30% TBP in kerosene and 2 M HNO3);

• entrainment of the aqueous phase into the organic was no more than 0.1%, while the aqueous phase did not practically contain any of the organic phase;

• adjusting the best interphase position in the contactor rotor for each of the specific process operations allows entrainment of the aqueous phase into the organic phase to be eliminated.

Probes for detecting the height of the salt/gas interface or salt/cadmium interface by resistivity measurement have been used in electrorefiner

10.28 Percentage deviation of predicted salt volume relative to measured salt volume.

experiments. The mass of the salt inventory in the electrorefiner was obtained from the product of the fluid volume measured by level probes and the density, assuming additive volumes. Figure 10.28 shows a time plot comparing the predicted salt volume with the measured salt volume of an electrorefiner in INL. For the Mark IV electrorefiner, the average percent­age deviation of the predicted salt volume relative to the measured salt volume was reported to be 0.18% with a standard deviation of 1.26%, while it was 0.04% with a standard deviation of 1.25% for the Mark V electrore­finer (Vaden, 2007). This indicates that calculating the salt density via addi­tive volumes is a satisfactory method.

Efforts to produce a solid-phase extraction media applicable to separations in the Th-U fuel cycle have led to several interesting materials. An amide grafted polymer, p-amino-N, N-dihexylacetamide (ADHA), anchored on high surface area Amberlite XAD-16 resin has been synthesized and char­acterized by Maheswari and Subramanian (2005). The ADHA ligand was reacted with aminated XAD-16 resin (AXAD-16), in the presence of anhy­drous K2CO3 dissolved in dimethyl formamide, to form the AXAD-16- ADHA solid phase extraction resin. The novel material was tested with a synthetic nuclear acidic waste solution to evaluate its efficacy for separating U(VI) and Th(IV). Multiple bench-scale column tests showed the grafted resin matrix to be very effective at partitioning U(VI) from Th(IV) and Ln(III) elements in nitric acid. The Th(IV) and Ln(III) elements were completely washed from the column with 0.1M HNO3 and U(VI) was quantitatively stripped with 0.1 M (NH4)CO3. It is also interesting to note

that all metals could be completely removed from the column with de­ionized water. Raju and Subramanian (2005, 2007a, 2007b) and Maheswari and Subramanian (2005) have used similar grafting techniques to attach other ligands, such as CMPO, dimethyl amino-phosphono-methyl phos — phinic acid (DAPPA), N, N,-di-hexyl succinamic acid (DHSA) and 4-ethoxy — N, N-dihexylbutanamide (EDHBA) to Merrifield chloromethylated (MCM) or AXAD-16 polymer resins. The authors report the physicochemical prop­erties of these grafted ligand-polymer resins and performed bench-scale tests to evaluate their utility for effecting various U, Th and Ln(III) element separations. Chemical structures of the MCM-CMPO and AXAD-16- EDHBA resins are shown if Figs 13.12 and 13.13, respectively.

In each case, the solid-phase extraction resins derived from grafting the desired ligand to a polystyrene-di-vinyl benzene support showed rapid dif­fusion kinetics and high capacity for the target metals. Columns of the grafted resins were tested for up to 20 loading-elution cycles without showing significant degradation in performance.

A method for immobilizing CMPO within a polyacrylonitrile (PAN) polymer has been reported (Kamenik et al., 2006, Tranter et al., 2003, Mann et al., 2002). The CMPO and PAN polymer are dissolved in concentrated HNO3 or dimethyl sulfoxide and subsequently sprayed through a nozzle into a water bath to precipitate solid-phase extraction beads of a desired size range according to methods published by Sebesta (1997). It is posited

that the CMPO exists as a finely dispersed solid trapped within the PAN matrix. Loadings of approximately 30% (w/w) CMPO in the polymer were obtained and the highly porous structure of the polymer composite facili­tates mass transfer. A scanning electron microscopy (SEM) image of the CMPO-PAN composite is shown in Fig. 13.14.

Batch experiments with simulated acidic waste solutions showed good kinetic properties and high distributions for the An and Ln elements tested. Nitric acid dependencies for the An(III) and Ln(III) elements, however, differed from those seen with CMPO in the TRUEX process or CMPO on solid supports, with or without TBP. Distribution coefficients for these ele­ments on the CMPO-PAN material remained higher than 103 mL g-1 at 10-3 M HNO3 as shown in Fig. 13.15. Work to explain this behavior is
ongoing, but it suggests some interesting separation schemes may be possible.

The coordination chemistry of actinides in aqueous solutions can be split into two groups: lower oxidation state (di, tri and tetravalent) and higher oxidation state (penta, hexa — and heptavalent) ions.

The coordination number and geometry of their aqueous complexes is determined by the electronic configuration and steric size and shape of the ligands. While ionicity is the predominant characteristic of both lanthanide and actinide bonding (Choppin, 2002), an appreciable covalency, stronger in the actinide bonds, has been confirmed by many spectroscopic studies and attributed to the 6d orbital interactions with the ligands, which are significantly stronger than the 5f interactions (Clark, 2006).

The actinide ions are relatively large cations. Their ionic radii range from 0.112 to 0.095 nm for the trivalent and from 0.094 to 0.082 nm for tetrava — lent cations. The actinides are found to have high coordination numbers, from 6 to 14. For a majority of the actinides, the exact numbers of water molecules that are bound to the metal centers in the hydrated metal ions are still controversial; the uncertainty in the structures can be explained by the limited number of crystal structures that exist for their aquo complexes. This lack of data is related to the difficulty in crystallizing materials from aqueous solutions (Keogh, 2005).

Typical examples of coordination geometries for An3+ and An4+ are the Structures 2.1-2.6 for their octa — and nona-aquo ions An(H2O)8/93+/4+. Structures (2.1-2.3) are octa-coordinate with a cubic, square antiprism, and a bicapped trigonal prism arrangement of the ligands (water molecules), respectively. The nine-coordinate structure (2.4) is a tricapped trigonal prism. Higher coordination numbers are observed with multidentate ligands, such as carbonate and nitrate. Ten-coordinate species include, for example, anionic pentacarbonate species of tetravalent actinides (Structure 2.5) with an irregular geometry with two trans carbonate ligands at the axial sites and three nearly planar carbonate ligands in a pseudo-equatorial plane that is reminiscent of the structure of dioxocation complexes of hexavalent actinides. The hexanitrate anion An(NO3)62- is extremely important in sepa­ration of actinides, for example plutonium, under conditions when other metals are in cationic form. Cation exchange resins have a strong affinity for the hexanitrato species Pu(NO3)62-. Structure 2.6 represents the coordi­nation geometry for this anion, which has six bidentate nitrate ligands, giving the central Pu4+ ion a coordination number of 12. A single crystal XRD study (Spirlet et al. ,1992) performed on (NH4)2Pu(NO3)6 confirmed

the structure (6) of the icosohedral Pu(NO3)62- unit characterized by three mutually perpendicular planes formed by the trans NO3- groups giving virtual symmetry Th. The twelve Pu-O bond distances average 2.487(6) angstroms (Clark, 200

For the aqueous species of penta — and hexavalent actinides, the typical species includes the linear dioxo unit, AnO2+/2+, with two oxygen atoms positioned at 180° with an average M-O distance of 0.175-0.180 nm for hexavalent cations and 0.181-0.193 nm for pentavalent cations. All “second­ary” ligands are coordinated in the perpendicular equatorial plane with typical M-X bond distances of 0.24-0.26 nm. The bonding for these ions has significant covalency with the axial An-O ligands, while the bonding for the majority of the ligands residing in the equatorial plane is primarily ionic (Keogh, 2005). As a result of this dual behavior (covalency and ionicity) of the trans dioxo ions, the linear dioxo unit is unperturbed (with the exception of bond distance changes) in all of the aqueous-based complexes. The coordination numbers of the central actinide cation are defined by the equatorial size of ligands and their electronic properties. The structures of a variety of aqueous-based coordination complexes have been observed (Structures 2.9-2.12). Compounds with tetragonal syStructure 2.9

Structure 2.12

The penta-aqua ion (Structure 2.11) and pentafluoro complex (Structure 2.12) for the hexavalent actinides are seven-coordinate structures, prevalent in actinide chemistry, and are the highest coordination numbers achievable with all monodentate ligands; however, coordination complexes with eight atoms bound to the actinide are achievable. The most well-studied aquo ion of the actinides is UO2(H2O)52+ (Structure 2.11). From EXAFS, structural data on the aquo ions have been obtained for the hexavalent ions, AnO2(H2O)52+ (An = U-Am). For calibration purposes, the bond dis­tance for the oxo ligands of the UO22+ species obtained from XAFS and single-crystal analyses show a nearly identical length. The bond An = O distance was found to be 0.176, 0.175; 0.174 and 0.18 nm for An(VI) = U, Np, Pu, and Am, respectively. The An-OH2 distance for the same complexes was found to be 0.242, 0.242, 0.241 and 0.24 nm, respectively (Keogh, 2005). For neptunyl and plutonyl aqua ions of pentavalent Np and Pu, 4, 5 and 6 water ligands were identified by XAFS. The bond distances for both An = O (0.183 nm for both Np and Pu) and An-OH2 (0.251 and 0.250, respec­tively) expand in the pentavalent ions in line with an increase in the ionic radii with the change in oxidation state.

Structures with 9-12 molecules of water have been proposed for tetrava — lent actinides in aqueous solutions. In general, the most accepted values for the number of H2O molecules bound to the metal center are 10 for Th and 9 for U to Pu. The An-OH2 distances in these ions range from 0.25 to 0.24 nm (Keogh, 2005). Trivalent plutonium with nine molecules of water (Matonic, 2001) was crystallized in a tricapped trigonal prismatic geometry.

The UREX Simple Feed simulant was subjected to Raman spectroscopic measurements. The Raman spectrum in Fig. 4.1 (blue spectrum) shows two strong bands due to UO22+ (870 cm-1) and NO3- (1047 cm-1). The UREX simple feed stimulant was contacted with 30% TBP/n-dodecane, resulting in an organic phase loaded with UO2(NO3)2. The Raman spectrum of the loaded solvent shown in Fig. 4.1 (red spectrum) contains bands due to dodecane, TBP, uranyl, and nitrate were observed. The UO22+ (859 cm-1) and NO3- (1029 cm-1) bands were both shifted to lower energy in the

organic phase spectrum compared to the aqueous phase spectrum due to the UO2(TBP)2(NO3)2 complex formation. The spectrum of the loaded solvent was compared to the spectrum of water-washed TBP/n-dodecane, and it was found that the bands due to the solvent do not interfere with the uranyl or nitrate Raman bands. It was demonstrated that the depletion of uranyl ions from the aqueous phase upon extraction can be easily followed using Raman spectroscopy. The intensity of the uranyl band (870 cm-1) decreased from the initial spectrum to the final spectrum, indicating a reduced concentration of UO22+ in the aqueous phase after extraction.

In order to evaluate the detection limit for U(VI) and nitrate in the UREX Simple Feed system, a series of feed simulant solutions containing 0.0003-1.31 M UO2(NO3)2 in 0.8 M HNO3 were prepared and subjected to Raman spectroscopic measurements. The spectral overlay obtained is shown in Fig. 4.2 (left). A linear relationship was established between the Raman response of the respective UO22+ (870 cm-1) and NO3- (1047 cm-1) bands and the concentration of UO2(NO3)2 and nitrate in the 0.8 M HNO3 solution. A treatment recommended by the International Union of Pure and Applied Chemistry (IUPAC) was used for the evaluation of the detec­tion limit (Long 1983). In this treatment, the detection limit is calculated using equation 4.1,

DL = kSb 4.1

m + tSm

where DL is the detection limit, k is a numerical coefficient, m is the slope, Sb and Sm are the standard errors for the intercept and slope of a calibration plot, respectively, and t is Student’s value for (n — 2) degrees of freedom at the chosen confidence level. In accord with IUPAC recommendations, a k value of 3 was applied, which in turn calls for a 99.87% confidence level. This confidence level was used in the linear regression analysis, and the denominator in equation 4.1 was taken as the upper 99.87% value of the
slope. This treatment yielded the detection limit of 3.1 mM for UO22+ and 2.6 mM for NO3- under the applied measurement conditions. The analogous Raman calibration measurements using extraction solvent containing 30 vol% TBP/n-dodecane loaded with UO2(NO3)2 afforded detection limits of

1.9 and 21 mM for UO22+ and nitrate, respectively. The Raman spectra of variable UO2(NO3)2 extracted into TBP/n-dodecane solvent is shown in Fig.

4.2 (right). To account for variable baseline shifts in the organic solvent system, the intensity of the UO22+ and nitrate bands (858.9 cm-1 and 1029 cm-1, respectively) were normalized to the intensity of the dodecane solvent band at 1300.9 cm-1. The 1300.9 cm-1 Raman band in dodecane is a strong vibrational band ascribed to the -(CH2)n — in-phase twist character­istic of n-alkanes (Lin-Vien 1991). Due to its constant concentration as the diluent, this dodecane band is used as an internal standard.

Initial chemometric analysis of the Raman spectral data was undertaken using PLS analysis of the di-component uranyl nitrate — nitric acid solutions. Predictive models based on PLS analysis of Raman spectral data (contain­ing variable UO22+/total nitrate/proton concentrations) showed linear response over the 0-1.3 M and 0-3.5 M range for uranyl and nitrate species, respectively. Results of the PLS modeling based on Raman UO22+, NO3-, and H+ measurements are depicted in Fig. 4.3.

Hydroxamic acids are organic ligands which have the ability to form stable five membered chelate rings with metal ions as indicated Fig. 6.4. An example of the complexation reaction between Pu(IV) and acetohy- droxamic acid, denoted HL, is shown below with the coordinated nitrates omitted for clarity:

Pu4+ + nHL W PuLn4-n + nH+ 6.12

where n = 1, 2 or 3 AHA molecules coordinated to each tetravalent metal ion (Carrott 2007). Both AHA and FHA are hydrophilic molecules that are not extracted by themselves into the TBP solvent (Taylor 1998). Larger or longer chain hydroxamic acids may be susceptible to distribution between the two phases; consequently, only the smaller molecules (FHA and AHA) have been considered for advanced PUREX applications.

Since the introduction of hydroxamic acid as a potential reduction/com — plexation agent in the PUREX process, several research groups have made complementary investigations on different aspects of AHA chemistry. Both formo — and acetohydroxamic acid are prone to hydrolysis in nitric acid media, FHA more so than AHA. Both hydroxamic acids break down to hydroxylamine and carboxylic acid according to the reaction below (Taylor 1999, Carrott 2008).

CH3CONHOH + H2O + H+ ^ CH3COOH + NH3OH+ 6.13

This degradation potentially creates a problem in the process due to a decrease in the hydroxamic acid available for Np and Pu complexation/ reduction. However, the hydroxylamine produced will reduce Pu(IV) to Pu(III) in accord with the earlier discussion (see Eq. 6.10), and would have a beneficial effect on the process. The rate of degradation of AHA in acidic media translates to a destruction half-life of around 300, 100 and 60 minutes
at 1, 3 and 5 M HNO3, respectively (Nunez 2001). This degradation rate is only of concern for relatively long process times. If future processes use centrifugal contactors with short residence times, this will be a minor or negligible problem. Furthermore, the aqueous strip/scrub solution contain­ing the hydroxamic acid will likely not be recycled after leaving the process; instead the remaining free hydroxamic acid and AHA-Np/Pu complexes will be deliberately destroyed downstream to facilitate the extraction of neptunium and plutonium away from the other fission and activation prod­ucts, e. g., NPEX (vide infra).

The complexation and reduction of pertechnetate by AHA was investi­gated to see if AHA addition interferes with the extraction of technetium (Gong 2008). The study showed that it is possible for AHA to reduce and complex technetium; however, the rate of reduction at moderate AHA concentrations is low and, although an important observation, it is not an issue for short contact times in centrifugal contactors. Studies have also shown that pertechnetate does not react with hydroxylamine to any large extent (Koltunov 2000).

The reduction of Np(VI) to Np(V) by AHA is a rapid reaction and has been suggested to proceed by the following route (Chung 2005):

2NpO22+ + 2CH3CONHOH ^ 2NpO2+ + 2CH3COOH + N2 + 2H+ 6.14

This reaction was shown to be kinetically rapid enough for successful use of centrifugal contactors (Taylor 1998, 1999).

Even though the addition of AHA will not affect the extraction of U(VI) (Taylor 1998) the presence of a uranyl-hydroxamate complex in the TBP solvent has been identified (Tkac 2008). This will most likely not cause any problems downstream in the uranium purification stage, see UREX below.

The incorporation of AHA into the PUREX flow sheet for Np/Pu separa­tion has potential benefits, and the results thus far from several research groups indicate no major disturbances to process performance. The small — scale process tests carried out to date have been mostly successful (Vandegrift 2004, Birkett 2004), see below, and this chemistry may be con­sidered to be very close to ready for full-scale testing. Of course, a major byproduct from AHA is acetic acid (refer to equations 6.13 and 6.14 above), the behavior and impact of which (e. g., effluent treatment and acid recycle) must yet be evaluated. Finally, the political climate and possible use of plutonium and neptunium as reactor fuel will still determine how the final process will be shaped.

The reduction in particle size greatly accelerates the rate of dissolution. The higher oxidation state of the uranium reduces the nitric acid requirement

a Zr is apparently from residual cladding fines.

b Prior to dissolution, 50% of Ru and Rh apparently was removed by volatilization during the voloxidation or dissolution process. Of the non-volatilized portion, 80-90% was in the UDS.

and reduces the amount of NOx evolved (see Table 8.2).919 When produced, UO3 powder dissolves readily in >0.3 M HNO3 acid with only trace levels of NOx generated.