硅酸盐学报

碱金属钼酸盐黄相在高放废液玻璃固化过程中难以处理，严重影响玻璃的化学稳定性。为了对高钼废液固化玻璃的化学稳定性进行较为准确的预判，从而加速配方研发进程，利用玻璃结构基因模拟法(GSg M)研究了高Mo (2.6%～3.3%，质量分数)模拟高放废液硼硅酸盐玻璃的化稳性和钼分相累积情况，建立了B、Li、Na、Mo浸出率、玻璃转变温度和钼分相累积程度的结构预测模型，并进行了模型验证。研究显示，在本工作的成分研究范围内，钼分相成分主要为CsLiMoO₄(水溶性黄相)及CaMoO₄(耐水性晶相)。成分–结构–性质关系表明，提高Li_2O和CaO会显著促进钼分相的形成。518～528 cm^-1 Si—O—Si振动增强，则钼分相加剧，但[BO₃]基团中B—O—B振动增强，可有效减少钼分相的析出。模型验证结果表明，各性质的结构预测模型可对设计玻璃的化稳性及钼分相累积程度进行快速预测，以筛选设计玻璃配方是否有进一步研究的必要。

Introduction The solubility of Mo in borosilicate glasses is relatively low.Therefore,molybdate is easy to separate from glass,forming Mo-yellow phase (alkali molybdates,such as Na_2Mo O₄,CsLiMoO₄,and Li_2Mo O₄) or CaMoO₄ crystals.The crystal CaMoO_4performs a superior water resistance ability,which can effectively reduce the Mo leaching rate of a high Mo content and high level waste (HLW) solidification glass.On the contrary,alkali-Mo-yellow phase is highly water-soluble,leading to a serious Mo-leaching,and thus decreases the chemical durability of the products.The duration of the chemical durability test of solidification glass is 7 or 28days,thus retarding the whole cycle for the investigation of HLW solidification glass.Effective predicting Mo-yellow phase and the Mo leaching rate can accelerate the research via saving time on glass preparation and chemical durability test.Based on the glass structural gene modeling (GSgM),this paper proposed the Structure-Property (S-P) prediction models of chemical durability and the possibility of Mo-phase separation (PS,its value is x _PS) for a simulated high Mo content (MoO₃ 2.6-3.3%) HLW borosilicate solidification glass.The key elements (i.e.,Na,Li,B,and Mo) leaching rate r _Na,r _Li,r _B,r _Mo as well as the Mo-phase separation x _PS were predicted by the models.The reliability and practicability of the models were proved.In addition,combined with the Structure-Composition (S-C) modeling,the C-S-P analysis was also carried out for x _PS and r _Mo.Methods Fifteen borosilicate glasses were designed by an one-component-at-a-time (OCAT) method.Glasses with different weights (130 g and 800 g) were prepared for each composition.Small samples with the mass of 130 g were melted and quenched.Large glass samples with the mass of 800 g were melted,and annealed.The properties of the samples were determined by X-ray diffraction (XRD),Fourier transform infrared spectroscopy (FTIR) with KBr,differential scanning calorimetry (DSC) and PCT-7 day chemical durability test,respectively.Mo-phase (x _PS) was evaluated and scored for each sample by naked eyes (from 0-100%,high score means high probability of Mo-phase separation).The structural data used for modeling were derived from FTIR peak fitting area A_i.Each FTIR spectrum was decomposed into 10 Gaussian bands by a software named GRAMS AI 32 according to the same fitting rules.The S-P and S-C models were constructed by a software named JMP based on the Cornell first-order linear mixture formula.The sample (glass MV) for model validation was also prepared and measured by the same procedure.Results and discussion No Mo-phase appeats in the small samples (130 g),while a serious Mo-phase separation (i.e.,yellow phase and white phase) occurs in some large samples (800 g),which can be attributed to the Mo-phase accumulation and the slow cooling rate of large size glasses.The XRD patterns indicate that the main components of Mo-phase are CsLiMoO₄ (i.e.,water-soluble yellow phase,yellow) and CaMoO₄ (i.e.,water-resistant crystal,white).The S-P models with satisfied statistical significance (i.e.,P<0.0001)and model accuracy (i.e.,R_sq≥0.87) are proposed based on the A_i and glass properties.The corresponding prediction formula of the target properties (i.e.,Na,Li,B,Mo leaching rate r_i,T_g,x _PS) arer_B=0.085-0.036A_3+0.024A_4+0.002A₅(1)r_{L i}=0.13-0.052A_3+0.005A_5+0.005A₆(2)r_{N a}=0.38-0.086A_3+0.008A₆-0.003(A_6+A₇)(3)r_{M o}=3.16-1.28A₃(4)T_g=456.1+0.57A_7+1.51A_8+1.31A_{1 0}(5)x _PS=57.1+68.07 A₂-42.37 A₃(6)In order to test the reliability of the models,a validation glass (i.e.,MV) is designed.The A_i values of MV (130 g,without Mo phase separation) are used to calculate the target properties by formulas (1) to (6),and the predicted values of each property arer _B=0.05 g·m^-2·d^-1,r _Li=0.06 g·m^-2·d^-1,r _Na=0.04 g·m^-2·d^-1,r _Mo=1.49 g·m^-2·d^-1.T_g=524.1℃,x _PS=83%.The model prediction results show that MV glass can perform high Mo leaching rate,exceeding the EJ-1186 standard (r_i<1 g·m^-2·d^-1).The Mo-phase possibility is 83%,indicating that MV glass has heavy Mo-phase separation.The r _B,r _Li,r _Na meets the standard requirement.The large sample MV (800g) is prepared.Large area of MV glass is covered by Mo phase separation,which is scored by70%by naked eyes,and the r _B,r _Li,r _Na is qualified to EJ-1186.However,the r _Mo is as high as 0.91 g·m^-2·d^-1,which is close to the standard limit.Considering the test error,MV glass is not valuable for further investigation and measurement (such as viscosity,electrical resistance,liquidus temperature,etc.).The measured properties arer _B=0.06 g·m^-2·d^-1,r _Li=0.14 g·m^-2·d^-1,r _Na=0.15g·m^-2·d^-1,r _Mo=0.91 g·m^-2·d^-1.T_g=524.4℃,x _PS=70%.The S-P models indicate that within the designed glass composition,the Mo phase separation can prevent effectively with the decrease of Si—O—Si vibrations at 518-528 cm^-1,and the increase of B—O—B bending vibrations in[BO₃]units.In order to figure out the components that affect the target structure as well as x _PS and r _Mo,a detailed exploration of the components associated with the structure units that screen for PS and r _Mo from S-P models is simulated via S-C modeling.The results indicate that Li_2O and CaO enhancement can increase the formation of Mo-phase,in which CaO can lead to the formation of the water resistance white crystal CaMoO₄,while Li_2O intensifies the precipitation of water-soluble Mo-yellow phase CsLiMoO₄.SiO₂ can affect the formation of CaMoO₄ and CsLiMoO₄.Certainly,Li_2O can promote the Mo leaching rate.Conclusions A series of high Mo (MoO₃ 2.6%-3.3%),HLW borosilicate solidification glass was investigated by Glass Structural gene Modeling (GSgM) to establish the prediction models for the key elements (i.e.,Na,Li,B,Mo) leaching rate r _Na,r _Li,r _B,r _Mo and the Mo-phase separation.The r _Na,r _Li,r _B,r _Mo and x _PS with a satisfied accuracy were predicted.The results were summarized as bellow:1) The main components of Mo-phase were water-soluble yellow phase CsLiMoO₄ (yellow) and water-resistant white crystal CaMoO₄.2) The S-P models for all the target properties exhibited high accuracy (R_sq≥0.87) and satisfied statistical significance(P<0.000 1).The model validation proved the reliability of the models.3) Composition-structure-property analysis showed that SiO₂ variation could affect the concentration of Si—O—Si (at 518-528 cm^-1)and B—O—B bending vibrations in[BO₃]units,inducing two opposite effects on the Mo-phase separation,which could be evaluated via the simulated parameter calculation.Li_2O enhancement increased the formation of water-soluble Mo yellow phase Cs Li Mo O₄,and promoted the Mo leaching rate consequently.CaO could lead to the formation of the water resistance white crystal CaMoO₄.