Journal of Non-Crystalline Solids 358 (2012) 674–679

Contents lists available at SciVerse ScienceDirect

Journal of Non-Crystalline Solids

j ourna l homepage: www.e lsev ie r .com/ locate / jnoncryso l

Structural analysis of some sodium and alumina rich high-level nuclear waste glasses

Ashutosh Goel a,⁎, John S. McCloy a, Kevin M. Fox b, Clifford J. Leslie c, Brian J. Riley a,Carmen P. Rodriguez a, Michael J. Schweiger a

a Pacific Northwest National Laboratory, Richland, WA, 99354, United Statesb Savannah River National Laboratory, Aiken, South Carolina, 29808, United Statesc Department of Materials Science and Engineering, University of Washington, Seattle, WA, 98195, United States

⁎ Corresponding author. Tel.: +1 509 371 7143; fax:E-mail address: ashutosh.goel@pnnl.gov (A. Goel).

0022-3093/$ – see front matter © 2011 Elsevier B.V. Alldoi:10.1016/j.jnoncrysol.2011.11.015

a b s t r a c t

a r t i c l e i n f o

Article history:Received 30 September 2011Received in revised form 6 November 2011Available online 3 December 2011

Keywords:Nepheline;Nuclear waste glass;Molecular structure;Infrared spectroscopy;Borosilicate glass

Sodium- and aluminum-rich high-level nuclear waste glasses are prone to nepheline (NaAlSiO4) crystalliza-tion. Since nepheline removes three moles of glass-forming oxides (Al2O3 and SiO2) per mole of Na2O,the formation of this phase can result in severe deterioration of the chemical durability in a given glass.The present study aims to investigate the relationships between the molecular-level structure and the crys-tallization behavior of sodium alumino-borosilicate-based simulated high-level nuclear waste glasses withinfrared spectroscopy (FTIR) and X-ray diffraction, respectively. The molecular structure of most of the inves-tigated glasses comprise a mixture of Q2 and Q3 (Si) units while aluminum and boron are predominantly pre-sent in tetrahedral and trigonal coordination, respectively. The increasing boron content has been shown tosuppress the nepheline formation in the glasses. The structural influence of various glass components onnepheline crystallization is discussed.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

The Hanford site is home to one of the greatest concentrations ofradioactive wastes in the world mostly generated from plutoniumproduction reactors during the timeframe, 1944–1987. The most sig-nificant challenge at Hanford is stabilizing the 2.0×105 m3 (5.3×107

U.S. gallons) of high-level radioactive waste (HLW) stored in 177 un-derground tanks. Approximately a third of these tanks have leakedwaste into the soil and groundwater. As of 2008, most of the liquidwaste had been transferred to more secure double-shelled tanks;however, 1.1×104 m3 (2.8×106 U.S. gallons) of liquid waste, togetherwith 1×105 m3 (2.7×107 U.S. gallons) of salt cake and sludge, re-mains in the single-shelled tanks. The waste was originally scheduledto be removed by 2018 but the revised deadline is 2040 [1].

Hanford's current strategy is to vitrify the HLW in borosilicateglasses because of the high chemical durability and moderate proces-sing temperatures (1000–1200 °C) of this glass system. It is due tothis reason that Bechtel National, Inc. is designing, constructing, andcommissioning the world's largest radioactive Tank Waste Treatmentand Immobilization Plant (WTP) for the U.S. Department of Energy(DOE) at the Hanford site. The WTP will use vitrification technologyto immobilize the radioactive waste and is expected to be operationalby 2019. The vitrification technology at WTP will include blendingthe waste with glass-forming materials, heating it to 1100–1200 °C,

+1 509 372 5997.

rights reserved.

and then pouring the mixture into stainless steel canisters to cooland solidify.

Neutralized HLW from processing of spent nuclear fuel containshigh concentrations of sodium, nitrate, and aluminum (listed inorder of abundance) [1,2]. The fuel cladding is the source for mostof the aluminum in the tank waste since the aluminum claddingwas dissolved in nitric acid and then neutralized with NaOH, creatingwastes rich in both Na and Al. This is a major cause of concern since inthe presence of glass-forming SiO2, the melt rich in both Na2O andAl2O3 is prone to the crystallization of nepheline (NaAlSiO4), especial-ly during the pouring of glass into stainless steel canisters. As the meltslowly cools in the center of the canister (compared to the melt nearto the walls of canister) it may be reheated by hot glass additions,thus promoting conditions favorable for crystallization. Nephelineformation can result in severe deterioration of the chemical durabilityconsidering that it removes three moles of glass network-forming ox-ides (Al2O3 and 2 SiO2 per each mole of Na2O) [3].

Therefore, the need to design waste glass compositions outsidethe primary crystallization field of nepheline led to an intensiveresearch effort at Pacific Northwest National Laboratory (PNNL) andSavannah River National Laboratory (SRNL) that resulted in the for-mation of the “nepheline discriminator” (ND) [3]. According to ND,in the nepheline primary phase field in the SiO2–Al2O3–Na2O ternarysystem, if the SiO2 content in the SiO2, Al2O3 and Na2O submixtureis≥62 wt.% in the final glass, nepheline is unlikely to form:

NSi ¼WSi

WSi þWAl þWNa≥ 62 ð1Þ

675A. Goel et al. / Journal of Non-Crystalline Solids 358 (2012) 674–679

Here, NSi is the normalized silica concentration, and Wi representsthe wt.% of the corresponding i-th oxide. The ND criterion is highlyuseful in designing nepheline-free glass compositions and has beenincorporated into predictive models for Hanford [4] and DefenseWaste Processing Facility (DWPF) [5,6]. However, from a composi-tional standpoint, the ND constraint is a significant impedimenttowards achieving higher waste loading, particularly for wastes highin Na and Al. Since the cost of managing these legacy high-levelwastes is directly related to the amount of HLW glass generated,it is desirable to maximize the waste loading to decrease overallglass volume, and thereby the cost of vitrified waste for storage anddisposal. Also, it has been recently shown that some glass composi-tions with NSib0.62 do not precipitate nepheline [7] while the pres-ence of B2O3 and/or some alkali/alkaline-earth cations in waste glasscompositions suppresses nepheline formation [5,8]. Unfortunately,the ND criterion cannot take into account the influence of theseoxides while designing the waste glass compositions. In order toovercome the limitations faced by the ND model, McCloy et al. [9]suggested a different metric for influence of oxides on the propensityfor nepheline formation in glasses through consideration of ‘opticalbasicity’ (OB) concept. According to the OB concept, more basiccations (e.g., CaO, K2O) are more likely to cause aluminosilicatecrystallization as they readily donate valence electrons and thus getremoved from the covalent glass network as opposed to more acidiccations (e.g., B2O3).

Since the development of technologically useful glass composi-tions is based on an understanding of the relationships between themolecular-level glass structure, crystallization kinetics, and importantphysical properties, the present study aims at understanding theimpact of glass structure on nepheline formation in a series ofHLW glasses. The results presented here are expected to help improvethe HLW glass formulation with a positive impact on both wastevitrification economy and the waste product quality.

2. Design of glass compositions

The glasses in the current study were designed and developed byFox et al. [10] in the ternary system Na2O–Al2O3–SiO2 based on thecomposition of sludge batch 5 (SB5) processed at DWPF. The glasseswere designed such that all of them had NDb0.62 but containedvarying amounts of CaO and B2O3. Table 1 lists the compositions ofall the investigated glasses along with their computed ND and OBvalues. Further, in accordance with the composition of SB5, a set ofoxides (labeled as ‘Others’ in Table 1) including Ce2O3, Cr2O3, CuO,Fe2O3, K2O, La2O3, MgO, MnO, NiO, PbO, SO4

2−, ZnO, TiO2 and ZrO2

were added to the glass compositions while U3O8 and ThO2 were ex-cluded from the glass batch, since the glasses were meant to be non-radioactive simulants. It should be noted that these oxides (‘Others’)

Table 1Composition of glasses (mol.%) along with their nepheline discriminator (ND) andoptical basicity (OB) values.

Glass SiO2 Al2O3 B2O3 Na2O CaO Others ND OB

2 44.16 4.34 17.87 21.41 2.46 9.76 0.6 0.57510 46.78 13.78 18.93 7.56 2.61 10.34 0.6 0.54311 48.45 14.28 4.1 7.83 15.27 10.07 0.6 0.60412 38.54 11.36 18.48 6.23 15.30 10.09 0.6 0.57017 45.90 10.82 4.06 26.70 2.52 10.00 0.5 0.63119 39.27 9.26 3.99 22.84 14.85 9.79 0.5 0.65825 49.35 23.26 4.37 9.57 2.71 10.74 0.5 0.58530 29.58 4.36 17.95 35.84 2.48 9.79 0.4 0.63133 38.16 16.87 4.22 27.75 2.62 10.38 0.4 0.64434 31.34 13.85 19.02 22.78 2.62 10.39 0.4 0.59335 32.46 14.35 4.12 23.6 15.34 10.13 0.4 0.67036 25.80 11.4 18.55 18.76 15.36 10.13 0.4 0.61639 34.60 25.48 4.39 8.38 16.35 10.80 0.4 0.626

were added to the glass compositions in minor (0.03–1 wt.%) andfixed quantities except Fe2O3 that was added in an amount of10.77 wt.%.

3. Experimental

The glasses were synthesized using a standard melt-quench tech-nique. In accordance with the compositions presented in Table 1,homogeneous mixtures of batches (150 g) were obtained by millingoxides (SiO2, Al2O3, Ce2O3, Cr2O3, CuO, Fe2O3, La2O3, MnO, NiO, PbO,TiO2 and ZrO2), carbonates (Na2CO3, MgCO3, CaCO3), sulfates(Na2SO4) and boric acid (H3BO3) in an agate mill with an agatepuck in a vibratory fixture. The batches were preheated in Pt–Rhcrucibles at 900 °C for decarbonization and then melted in the rangeof 1150–1450 °C for a dwell time period of 1 h. Table 2 lists the melt-ing temperature of each of the glass batches. Finally, glasses werequenched by pouring onto a stainless steel plate.

In order to understand the crystalline phase evolution in theglasses, monolithic glass pieces were heated isothermally at 950 °Cfor 24 h. This heat treatment regime was chosen considering theprocessing constraint that will be faced by the WTP with respect tospinel formation in glasses i.e. the temperature at which the spinelcontent in glass is~1 vol.%. should be≤950 °C. Qualitative and quan-titative assessments of the amorphous/crystalline nature of theas-quenched glasses as well as heat-treated glasses/glass-ceramicswere made with X-ray diffraction analysis (XRD) with a conventionalBragg–Brentano diffractometer (Bruker D8 Advance; Bruker AXS Inc.,Madison, WI, USA) coupled with a LynxEye detector. The sampleswere crushed to powder (particle sizeb63 μm) and doped with5 wt.% CaF2 as an internal standard. The mixtures were ground in anagate mortar and loaded into plastic holder in order to minimizethe preferred orientation. Data were recorded in 2θ range=of 5–70° with a 0.001486° step size and a 0.05 s dwell (divergence andanti-scattering slits were 0.300 and 0.0500, respectively). The qualita-tive crystalline phase analysis of the as obtained XRD data was madewith Jade 6.0 Software (MDI, Materials Data Incorporated, Livermore,CA, USA). Quantitative crystalline phase fractions were extracted withRietveld refinements using RIQAS® 4 (MDI) and were rescaled on thebasis of the absolute mass of CaF2 originally added to their mixturesas an internal standard, thereby becoming internally renormalized.The background was successfully fitted with a Chebyshev functionwith a variable number of coefficients depending on its complexity.The peak profiles were modeled using a pseudo-Voigt function withone Gaussian and one Lorentzian coefficient. Lattice constants,phase fractions, and coefficients corresponding to sample displace-ment and asymmetry were also refined.

Infrared (IR) spectra of the as-quenched glasses were obtainedwith a Fourier transform IR spectrometer (FTIR, Thermo Nicolet

Table 2Batch melting temperature (TM) and quantitative crystalline phase assemblage of melt-quenched glasses.

Glass TM (°C) Crystalline phase Vol.%

2 1150 Amorphous –

10 1450 Magnetite (Fe3O4) 8.3 (5)11 1300 Amorphous –

12 1300 Amorphous –

17 1300 Amorphous –

19 1300 Amorphous25 1450 Trevorite (NiFe2O4) 11.8 (4)30 1150 Amorphous –

33 1300 Carnegieite low (NaAlSiO4) 10.7 (1)Trevorite (NiFe2O4) 1.3 (7)

34 1300 Amorphous –

35 1300 Trevorite (NiFe2O4) 14.7 (1)36 1300 Amorphous –

39 1450 Lithium iron oxide (Li0.5Fe0.95O2) 9.2 (3)

2

30

10

34

466

460

730

1000

983

700

715

1080

1020

1270

1190 1435

12751430

400 800 1200 1600

Tra

nsm

itta

nce

(a.u

.)

a

11

35

12

36

465

705

695

720

1030

940

1000 1230

1260

1420

1280

1230

1430

1410

400 800 1200 1600

Tra

nsm

itta

nce

(a.u

.)

b

Wave number (cm-1)

Wave number (cm-1)

Fig. 1. FTIR spectra of glasses with varying Na2O/SiO2 ratio.

34

35

10

11

465

12

19

715

695

1010

950

1270

1235

1430

1410

700

1080

10301275 1430

700

1000

980

1235

1260 1420

1390

400 800 1200 1600

Tra

nsm

ittan

ce (

a.u.

)

Wave number (cm-1)

Fig. 2. FTIR spectra of glasses with varying CaO/B2O3 molar ratio (10 and 11; 34 and 35)and Na2O/B2O3 molar ratio (12 and 19).

676 A. Goel et al. / Journal of Non-Crystalline Solids 358 (2012) 674–679

6700, Thermo Fisher Scientific, Waltham, MA, USA). For this purposeglass powders were mixed with KBr in the proportion of 1:150(by weight) and pressed into a 10-mm diameter pellet with a uniaxialpress at 8.9×108 Pa (1.3×105lbs/in2). Sixteen scans in the 400–4000 cm−1 range were collected with a resolution of 2 cm−1 forboth the background and the sample.

3. Results

3.1. Glass-forming ability

Among the thirteen glass compositions investigated in the presentstudy, eight resulted in glasses after quenching while iron-containingspinel phases could be observed in almost all the remaining five com-positions. Table 2 presents the quantitative crystalline phase analysisof the investigated glass compositions.

In an aluminosilicate glass system, Al2O3 acts as a network formerat low concentrations and occurs as AlO4 tetrahedra in glass structure.This structural unit improves the glass stability and hence the chem-ical durability. Alkali and alkaline-earth cations are located near AlO4

tetrahedra and balance their negative charge so that they are nolonger modifiers in a silicate glass network [11]. However, increasingAl content at the expense of alkali/alkaline-earth (at constant silicacontent) in glasses will lead to an increase in the abundance of non-bridging oxygens (NBOs) as a result of Al3+ changing from tetrahe-dral to octahedral coordination in the melt [12].

On the other hand, in iron-containing aluminosilicate glasses,Fe3+ exists predominantly in tetrahedral coordination [13]. There-fore, depending on the Al2O3/(Na2O+CaO) ratio in the presentglasses the following two scenarios are possible: (i) some of thealkali or alkaline-earth cations that charge-balance Al3+ in theseglassesmay be transferred to Fe3+ or (ii) Fe3+ in four-fold coordinationforms some form of a complex with Fe2+ [14]. In the first scenario, wecan expect an amorphous glass while the second scenario is most likelyto result in the crystallization of an iron-containing spinel phase.

3.2. Glass structure

The room temperature FTIR transmittance spectra of all the inves-tigated glasses are shown in Figs. 1–3. In general, the IR spectra ofall the investigated glasses exhibit four broad transmittance bandsin the region of 400–1600 cm−1. This lack of sharp features is indica-tive of the general disorder in the silicate network mainly due toa wide distribution of Qn (polymerization in the glass structure,where n denotes the number of bridging oxygens) units occurringin these glasses. The most intense bands lie in the 800–1300 cm−1

region, the next between 300 and 600 cm−1 and 1350–1500 cm−1,while the least intensive lies between 650 and 800 cm−1. The broadbands in the 800–1300 cm−1 are assigned to the stretching vibrationsof the SiO4 tetrahedron with a different number of bridging oxygenatoms, while the bands in the 300–600 cm−1 region are due to bend-ing vibrations of Si–O–Si and Si–O–Al linkages [15]. The transmittanceband in the region 1350–1500 cm−1 corresponds to B–O vibrations inthe [BO3] triangle [16]. The bands in the 300–600 cm−1 region maybe related to the bending vibrations of Si–O–Si linkages as well aswith B–O bonds. The transmittance band in the 650–800 cm−1 regionmay be attributed to the bending vibrations of bridging oxygenbetween trigonal boron atoms and it may also be related to thestretching vibrations of the Al–O bonds with Al3+ ions in four-fold co-ordination. It is noteworthy that although seven out of thirteen glasscompositions exhibited crystallization immediately after quenchingof glass melts, we could not observe any bands with sharp intensityin the IR spectra of these glasses which are typical for the crystallinematerials. In particular, no bands were observed around~580 cm−1

and 400 cm−1 which are characteristic for magnetite (Fe3O4) [17].This may be attributed to the surface crystallization of magnetite

11

39

17

33

470 715

1030

1000

1410

1285

400 800 1200 1600

Tra

nsm

ittan

ce (

a.u.

)

Wave number (cm-1)

Fig. 3. FTIR spectra of glasses with varying SiO2/Al2O3 ratio.

677A. Goel et al. / Journal of Non-Crystalline Solids 358 (2012) 674–679

which allowed the crystals to grow superficially on the glass surface,thus, leaving the core of the bulk glass amorphous.

3.2.1. Na2O/SiO2 ratioA total of eight glass compositions were designed with varying

Na2O/SiO2 ratio. Further, these eight compositions were divided infour groups with each group comprising two glasses based on thesimilarity in their chemical compositions. The glasses that will bediscussed in this section are as follows: (i) 2 and 30; (ii) 10 and 34;(iii) 11 and 35; (iv) 12 and 36 (Table 1). The FTIR spectra of allthese glasses are presented in Fig. 1.

While glasses 2 and 30 are XRD amorphous (Table 2), the broadFTIR transmittance band in the spectra for these glasses in the 800–1300 cm−1 region is split in two parts (Fig. 1a). In the case of glass2, the most dominant band in the region of 800–1200 cm−1 was cen-tered at~1000 cm−1 depicting the dominance of a mixture of Q2 andQ3 (Si) units [18] while the same band was centered at a slight lowerwave number of ~983 cm−1 in glass 30 indicating depolymerizationin the silicate glass network with an increasing Na2O/SiO2 ratio. Fur-thermore, the second part 1180–1300 cm−1 band exhibits a broadshoulder at ~1270 cm−1 and 1190 cm−1 for glasses 2 and 30, respec-tively. This feature has been attributed to the presence of sulfur in the6+ oxidation state (i.e., sulfate or SO4

2−) as 0.51 mass% SO42− has

been added to all the glasses (Table 1). Sulfate anions in mineralsmay exhibit bands in the IR at ~1050–1250 (v3), ~1000 (v1), ~500–700 (v4), and ~400–500 (v2) cm−1 due to asymmetric and symmetricstretching and bending of the SO4

2− anion [19]. Although, this featurecould also have been attributed to the presence of Q4 (Si) units inglasses since we cannot neglect the contribution of these units inthe glass structure. However, we did not find this assignment logicaldue to the following reasons: (i) the glass compositions 2 and 30contain ~44 and 30 mol.% SiO2, respectively, which is very low incomparison to glasses with Q3 or Q4 as predominant structural units[20], (ii) Al creates Q3 units at the expense of Q4 and Q2 units insilicate glasses [11], and (iii) owing to the multicomponent natureof these glasses, it will be difficult to predict the formation of Q4

units in the glass structure. Also, the possibility for the existence of

SiOnS4−n (n=1 to 3) groups is negligible due to low sulfur contentin glasses but cannot be completely neglected with certainty [21].

In the case of glasses 10 and 34, the glass 10 exhibited the crystal-lization of~8 vol.% magnetite while glass 34 was XRD amorphous innature after quenching of glass melts (Table 2). However, the magne-tite crystallization in glass 10 did not seem to affect the glass structuresignificantly owing to the fact that this iron content did not contributeto the glass network formation due to scarcity of alkali cations in thisglass as has been explained in Section 3.1. The FTIR spectra of glasses10 and 34 as presented in Fig. 1a depict higher degree of polymeriza-tion in silicate glass network for glass 10 in comparison to glass 34 aswell as all other glasses investigated in the present study with Q3 (Si)units being the predominant species. An increase in Na2O/SiO2 ratioin glass 34 led to the formation of NBOs, thus, depolymerizing theglass structure. A similar scenario was also observed in glasses 11and 35 where glass 35 exhibited a depolymerized glass structure incomparison to glass 11 due to the high Na/Si ratio in latter while nosignificant changes in glass structure could be observed in glasses12 and 36 with varying Na/Si ratio.

3.2.2. CaO/B2O3 and Na2O/B2O3ratioA total of four glass compositions were designed with varying

CaO/B2O3 ratios (10, 11, 34, 35) and two compositions with varyingNa2O/B2O3 ratios (12, 19). Further, these six compositions were divid-ed in three groups with each group comprising two glasses based onthe similarity in their chemical compositions. The compositions thatwill be discussed in this section are: CaO/B2O3: (i) 10 and 11, (ii) 34and 35; Na2O/B2O3: (i) 12 and 19. The FTIR spectra of all these glassesare presented in Fig. 2.

With respect to the variation in alkali/alkaline-earth content inthese glasses at the expense of boron, the investigated glasses exhib-ited depolymerization in glass network with an increasing (Na,Ca)/Bratio. Borosilicate glasses are prone to phase separation and tend toseparate into silica-rich and borate-rich phases [22]. Also, the pres-ence of glass network-modifying cations (Na and Ca in the presentcase) in the glass promotes phase separation in accordance with theionic field strength of the cation where the higher the ionic strength,the more pronounced the phase separation [23]. Also, it has been ex-perimentally observed that alkali/alkaline-earth network modifyingcations have a higher affinity towards the borate-rich phase [24],thus, creating a deficit of Na and Ca in the silica-rich phase. Therefore,decreasing Na/B or Ca/B ratio in glasses will lead to higher tendencytowards phase separation in glasses with majority of modifier cationsassociating with borate-component in the glass structure. This createsa deficit of Na and Ca in the silica-rich phase that, consequentially,results in re-polymerization of this phase. Hence, the FTIR spectrafor 19, 11 and 35 show that these glasses are more depolymerizedthan the corresponding pair glasses 12, 10 and 34, respectively.

3.2.3. SiO2/Al2O3

Fig. 3 presents the FTIR spectra of glasses with varying SiO2/Al2O3

ratios in glasses (11 and 39; 17 and 33). As is evident from Fig. 3,the FTIR spectra of glasses 11 and 39 are almost identical to eachother. Likewise, glasses 17 and 33 are also similar. This illustratesthat varied SiO2/Al2O3 ratios did not affect the glass structure signifi-cantly and the types of coordination and atomic configurations offramework cations remained rather constant. As is expected fromthe bond-valence concept, more charge-balancing cations are associ-ated with more negatively-charged oxygen sites. Charge balancingcations are not randomly distributed but are more selectively concen-trated near Al–O–Al sites in lower silica glasses and near Al–O–Si sitesin higher silica glasses. On the other hand, oxygens interact withcharge-balancing cations, to some degree with varying SiO2/Al2O3

ratio, implying that clustering of framework cations may beimprobable [25]. Similar results on sodium aluminosilicate glasses

Table 3Quantitative crystalline phase analysis of glass ceramics (950 °C/24 h).

Glass Crystalline phase Vol.%

2 Amorphous –

10 Magnetite (Fe3O4) 21.1 (2)11 Sodium calcium aluminosilicate (Na0.34Ca0.66Al1.66Si2.34O8) 6.6 (4)

ZnFe2O4 16.8 (3)12 Magnetite (Fe3O4) 11.8 (1)17 Nepheline 26.9 (6)

MgFe2O4 4.4 (1)Lazurite [Na8.56(Al6Si6O24)(SO4)1.56] 5.1 (1)

19 Nepheline 5.6 (3)Fe3O4 4.7 (1)Lazurite [Na8.56(Al6Si6O24)(SO4)1.56] 4.1 (4)

25 Lithium iron chromium oxide (Li2Fe5Cr5O16) 22.2 (5)Magnesium iron aluminum oxide (MgFe0.2Al1.08O4) 11.6 (3)Aluminum oxide (Al2O3) 3.7 (1)

30 Amorphous –

33 Sodium aluminosilicate (Na7.11Al7.2Si8.8O32) 62.0 (3)Hauyne [Na6Ca2Al6SiO24(SO4)2] 2.9 (1)Magnetite (Fe3O4) 17.9 (4)

34 Magnetite (Fe3O4) 8.8 (5)35 Sodium aluminosilicate (Na1.45Al1.45Si0.55O4) 32.0 (3)

Magnetite (Fe3O4) 6.8 (1)Na7.11(Al7.2Si8.8O32) 10.1 (2)Jadeite (NaAlSi2O6) 15.8 (3)Sodalite [Na8(AlSi6O24)(MnO4)1.46(OH)0.54] 1.4 (1)

36 Magnetite (Fe3O4) 7.7 (9)39 Nepheline (K0.24Na6.00Al6.24Si9.76O32) 29.3 (6)

Magnetite (Fe3O4) 8.8 (2)

678 A. Goel et al. / Journal of Non-Crystalline Solids 358 (2012) 674–679

and calcium aluminosilicate glasses have been reported by Lee andStebbins [26].

3.2.4. Na2O/Al2O3 ratioAlthough the variation in the SiO2/Al2O3 ratio did not affect

the glass structure, the increase in the Na2O/Al2O3 ratio in glasses:(i) 17 and 25; (ii) 30 and 34 did depolymerize the silicate glassnetwork (Fig. 4). This was due to increasing number of NBOs withdecreasing Al concentration in glasses that allowed the alkali cationsto act as network modifiers instead of acting as charge-balancing cat-ions. Also, it is noteworthy that the field strength of a modifier cationplays a key role in deciding the aluminosilicate glass structure as theextent of deviation from Al-avoidance, and thus the configurationalentropy, increases with increasing cation field strength. The Al–O–Alsite population in these glasses, the disorder in the framework cationdistribution, and therefore possible topological arrangements are alsoproportional to the field strength of modifier cation [26]. Therefore,substituting equimolar concentration of CaO for Na2O in these glasseswould have resulted in higher number of NBOs in the glass structure.Thus, glasses 30 and 17 are more depolymerized than their pairglasses 34 and 35, respectively.

3.3. Crystallization behavior of glasses

The quantitative crystalline phase analysis of all the glasses afterheat treatment at 950 °C for 24 h is presented in Table 3. Glass com-positions 2 and 30 were still amorphous due to their low aluminacontent, preventing nepheline formation since Na+ acts as a charge-balancing cation for Fe3+ species and associates with the boron-richphase. All other glass compositions exhibited the crystallization ofspinels or/and sodium aluminosilicates. No boron-containing crystal-line phases were observed in these glasses. The crystallization ofsodium aluminosilicates, nepheline, in particular, has been observedin these glasses to depend on their boron content. The presence of17–19 mol.% B2O3 in these glasses (i.e., 2, 10, 12, 30, 34, and 36)seems to inhibit the formation of any Na–Al–Si–O based crystallinephase. This role of B2O3 in retarding the nepheline formation inglasses can be attributed to its tendency to induce phase separation

17

25

30

34

465

715

990

1020

1275

1290

1420

600

980

1010

730

720

1190

1330

1450

400 800 1200 1600Wave number (cm-1)

Tra

nsm

ittan

ce (

a.u.

)

Fig. 4. FTIR spectra of glasses with varying Na2O/Al2O3 ratio.

in these glasses as has been explained in Section 3.2.2. Since, alkaliand alkaline-earth cations have higher affinity for borate-rich compo-nent in borosilicate glasses, increasing boron concentration in theseglasses strips off the modifier cations from the silicate glass network,inducing re-polymerization and stabilizing the amorphous phase.

4. Discussion

The knowledge of molecular-level glass structure is pertinent toour understanding of their physical and thermochemical propertiesas it is known to control their nucleation and crystallization behavior[27,28]. However, the development of a structural model elucidatingsuch relationships is an arduous task considering the compositionalcomplexity of nuclear waste glasses. It is probably due to this reasonthat no structural model has been put forward that describes therelationship between glass structure and nepheline formation eventhough thermochemical models based on the freezing-point depres-sion of quasicrystalline precursors have been proposed to explainthe nepheline-spinel liquidus in nuclear waste glasses [29,30].

In the present study, an attempt has been made to correlate themolecular structure of glasses with their nucleation and crystalliza-tion behavior. The structural complexities of these glasses still requireextensive effort to reveal the key atomic-level processes that controlthe performance of nuclear waste glasses for a wide range of compo-sitions and structural features. In particular, additional emphasisshould be given on understanding the sulfur speciation and coordina-tion in these glasses. Apart from that, further information about theSi-, Al-, and B-coordination in these glasses will be highly beneficialin order to validate the perspective presented in this article.

With respect to the influence of glass structure on the propensityof nepheline (or sodium aluminosilicate based crystalline phases)formation, the structural coordination and phase separation due tothe presence of boron proved to be the most effective parameters.Although high-boron glasses did exhibit the formation of spinels(Fe3O4, MgAl2O4, etc.), spinel crystallization is known to have littleor no effect on durability of nuclear waste glasses [5] because thecrystal structure does not require glass network-forming compo-nents. Also, CaO content did not play significant role in suppressing

679A. Goel et al. / Journal of Non-Crystalline Solids 358 (2012) 674–679

the nepheline formation, as glasses 11, 19 and 35 with higher CaO/B2O3 ratio exhibited the crystallization of nepheline-related crystal-line phases. Similar results with respect to the influence of CaO andB2O3 on tendency towards nepheline formation in nuclear wasteglasses have also been explained by McCloy et al. [9] on the basis ofcorrelation between ND and OB concept. We could not obtain clearevidence with respect to the impact of Al2O3 on nepheline crystalliza-tion in glasses as a function of high-boron content because the glasseswith high Al2O3 were poor in boron. Although, literature suggeststhat in alumino-borosilicate glasses, the width of the miscibility gapbetween silica-rich and borate-rich amorphous phases decreasesmore rapidly with the B–Al substitution than with the Ca–Na substi-tution, implying that interactions between network-forming ele-ments have a greater effect on borate-silicate unmixing than thenature of network-modifying cations [24]. However, the impact ofsuch a structural change in glasses on their crystallization behavioris yet to be investigated.

Furthermore, although the concept of correlation between ND andOB, recently proposed by McCloy et al. [9] to explain the tendency ofnepheline formation in HLW glasses, is not violated by any of theglasses that precipitated nepheline (17, 19, 33, 35, 39). However, itfails to explain the absence of nepheline crystallization in the otherglasses. According to McCloy et al. [9], nepheline precipitation isexpected to be suppressed at high SiO2 levels (ND>0.62) and atlow optical basicity (OBb0.55–0.58), though the authors point outthat there are some glasses which have NDb0.62 and OB>0.58 andstill do not show nepheline crystallization. It should be noted thatND and OB concepts were designed to screen compositions whichwere likely to have problems with nepheline crystallization, butthey were not intended to be absolute criteria requiring nephelinecrystallization in conditions of low SiO2 or high basicity. In fact,McCloy et al. [9] suggest that a better set of criteria would includeposition in the Na2O–Al2O3–SiO2 ternary system (defined by ND anda similar metric defining the normalized Na2O concentration) alongwith the OB criterion to account for effects of basic cations (CaO,K2O, etc.) and acidic glass formers (B2O3). Therefore, detailed and rig-orous structural analysis of the low SiO2, high-OB glasses free fromnepheline is needed to understand the structural impact of these fac-tors on sodium aluminosilicate crystallization. This understandingwill further enable the design of high waste-loading glasses withhigh Al2O3–Na2O compositions for nuclear waste immobilization.

5. Conclusions

The structure and crystallization behavior of sodium- andaluminum-rich HLW glasses have been studied with FTIR and XRD,respectively. An attempt has been made to study the glass structuralinfluence on the susceptibility of nepheline formation. All the inves-tigated glasses exhibit the dominance of Q2 and Q3 (Si) structuralunits while Al3+ primarily exists in tetrahedral coordination. It hasbeen observed that the presence of high amount (≥17 mol.%) ofB2O3 in glasses inhibits the crystallization of sodium aluminosilicatebased phases owing to amorphous phase separation while CaO

content did not play a significant role in suppressing the nephelineformation.

The addition of alkali/alkaline-earth cations in the glasses led tothe depolymerization of the silicate glass network while variation inSiO2/Al2O3 content in glasses did not affect the molecular structure.Most of the glasses did not violate the ND and OB criterion whilesome glasses with low ND value and high OB did violate it. This leavesa scope for detailed structural analysis of similar glasses that are freefrom nepheline so the glass structure can be better correlated to theND and OB criteria. With those results, it would be possible to comeup with a rigorous structural model which will enable the design ofhigh waste-loading glasses with high Al2O3–Na2O compositions fornuclear waste immobilization.

References

[1] National Research Council, Committee on long-term research needs for radioac-tive waste at Department of Energy Sites, Washington, D.C, 2001.

[2] M.F. Kupfer, Boldt Al, K.M. Hodgson, Lockheed Martin Hanford Corporation, Richland,WA, 1999.

[3] H. Li, J.D. Vienna, P. Hrma, D.E. Smith, M.J. Schweiger, Mater. Res. Soc. Symp. Proc.465 (1997) 261–268.

[4] J.D. Vienna, A. Fluegel, D.S. Kim, P. Hrma, Glass Property Data and Models for Es-timating High-Level Waste Glass Volume, PNNL-18501, Pacific Northwest Nation-al Laboratory, Richland, WA, 2009.

[5] K.M. Fox, T.B. Edwards, D.K. Peeler, Int. J. Appl. Ceram. Tech. 5 (2008) 666–673.[6] K.M. Fox, D.K. Peeler, T.B. Edwards, Adv. Mater. Sci. Environ. Nucl. Technol. 222

(2010) 77–90.[7] D.S. Kim, J.D. Vienna, D.K. Peeler, K.M. Fox, A. Aloy, A.V. Trofimenko, K.D. Gerdes,

WM 2008 Waste Management Conference, Phoenix, AZ, 2008.[8] H. Li, P. Hrma, J.D. Vienna, M.X. Qian, Y.L. Su, D.E. Smith, J. Non-Cryst. Solids 331

(2003) 202–216.[9] J.S. McCloy, M.J. Schweiger, C. Rodriguez, J.D. Vienna, Int. J. Appl. Glass Sci. 2

(2011) 201–214.[10] K.M. Fox, J.D. Newell, T.B. Edwards, D.R. Best, I.A. Reamer, R.J. Workman, Refinement

of the Nepheline Discriminator: Results of a Phase I Study, ” U.S. Department ofEnergy Report WSRC-STI-2007-00659, Revision 0, Washington Savannah RiverCompany, Aiken, SC, 2008.

[11] M.I. Ojovan,W.E. Lee, Immobilisation of RadioactiveWastes inGlass, An Introductionto Nuclear Waste Immobilisation, Elsevier, Oxford, 2005, pp. 213–249.

[12] B.O. Mysen, D. Virgo, C.M. Scarfe, Am. Mineral. 65 (1980) 690–710.[13] B.O. Mysen, Geochim. Cosmochim. Acta 70 (2006) 2337–2353.[14] D. Virgo, B. Mysen, Phys. Chem. Minerals 12 (1985) 65–76.[15] S.-L. Lin, C.-S. Hwang, J. Non-Cryst. Solids 202 (1996) 61–67.[16] L. Stoch, M. Sroda, J. Mol. Struct. 511–512 (1999) 77–84.[17] B. Grzeta, M. Ristic, I. Nowik, S. Music, J. Alloys Compd. 334 (2002) 304–312.[18] A. Goel, D.U. Tulyaganov, V.V. Kharton, A.A. Yaremchenko, J.M.F. Ferreira, Acta

Mater. 56 (2008) 3065–3076.[19] M.D. Lane, Am. Mineral. 92 (2007) 1–18.[20] H.R. Fernandes, D.U. Tulyaganov, A. Goel, M.J. Ribeiro, M.J. Pascual, J.M.F. Ferreira,

J. Eur. Ceram. Soc. 30 (2010) 2017–2030.[21] T. Tsujimura, X. Xue, M. Kanzaki, M.J. Walter, Geochim. Cosmochim. Acta 68

(2004) 5081–5101.[22] M. Fábián, E. Sváb, G. Mészáros, Z. Révay, T. Proffen, E. Veress, J. Non-Cryst. Solids

353 (2007) 2084–2089.[23] D. Ehrt, R. Keding, Phys. Chem. Glasses 50 (2009) 165–171.[24] I.V. Veksler, A.M. Dorfman, D.B. Dingwell, N. Zotov, Geochim. Cosmochim. Acta 66

(2002) 2603–2614.[25] S.K. Lee, J.F. Stebbins, J. Non-Cryst. Solids 270 (2000) 260–264.[26] S.K. Lee, J.F. Stebbins, J. Phys. Chem. B 104 (2000) 4091–4100.[27] E. Muller, K. Heide, E.D. Zanotto, J. Non-Cryst. Solids 155 (1993) 56–66.[28] V.R. Mastelaro, E.D. Zanotto, N.C. Lequeux, R. Cortes, J. Non-Cryst. Solids 262

(2000) 191–199.[29] C.M. Jantzen, K.G. Brown, J. Am. Ceram. Soc. 90 (2007) 1866–1879.[30] C.M. Jantzen, K.G. Brown, J. Am. Ceram. Soc. 90 (2007) 1880–1891.