Fresh and oxidized parts of rhyolitic pumice tuff, as a generic host rock for low and intermediate level waste, were differentiated on the basis of their physical, chemical and mechanical properties. Mineralogical analysis was performed by powder X-ray diffractometer using RIGAKU Rint Ultima-III system at operating conditions

of continuous scanning between 3˚ - 75˚ at 4˚/min, X-ray target of CuKα, tube voltage of 40 kV, tube current of 40 mA, and slit arrangement of 2/3˚ - 0.3 mm - 2/3˚ - 10 mm. Scanning Electron Microscope (SEM) analysis was conducted on small fresh and oxidized pumice tuff fragments using JEOL JSM-5600LV at 20 kV to take Secondary Electron Image (SEI). As for mechanical properties, porosity and Pore Size Distribution (PSD) were analyzed by Mercury Intrusion Porosimetry (MIP) system of Micromeritics Shimadzu AutoPore IV through low and high pressure analysis. Four samples each from fresh and oxidized pumice tuffs were used to obtain pore properties.

Distribution coefficient, K_d, of a certain nuclide is the amount of that nuclide remaining in the solution after the interaction with rock or mineral surface. It is usually expressed in the logarithmic values of m³/kg yielding from batch sorption experiments carried out with the crushed materials using homogenous grain sizes due to the reported significant effect of particle size and surface area of rock body [30] . As such, powder of 150 - 300 μm size pumice tuffs was used for batch preparation in the present research. The detail experimental method has been described in our previous studies [22] [27] . This study presents wider variation of geochemical parameters like initial Cs concentration of 10⁻⁴, 10⁻⁵, 10⁻⁶, 10⁻⁷ mol/dm³ (M), pH of 4 to 12 and ionic strength of 0, 0.1, 1.0 and 3.0 M NaClO₄. For aging, all the samples were kept in argon filled desiccator to avoid carbonate contamination. During the aging process, the pH shifted until 8 - 10 weeks after the start due to the chemical oxidation of pyrite. Therefore, the contact time was taken up to 14 weeks to get equilibrium condition. Due to the pH adjustment, minor changes of ionic strength occurred which was found to be as low as 0.003 times and so, the minimum ionic strength was considered as 0.003 M instead of 0. The K_d was calculated using the following equation considering the probable dissolved concentration of Cs during aging period [27] ;

where, Conc.[ini], Conc.[diss] and Conc.[fin] are initial, dissolved and final concentration of the nuclide in solution, respectively; V and W are the volume of the solution in ml and weight in g, respectively. Cs concentration was determined by Inductively Coupled Plasma Mass Spectrometry (ICP-MS, Elan DRC II, PerkinElmer) considering that the detection range of ICP-MS for Cs is (10⁻⁸ - 10⁻⁴ M). Since the ionization efficiency of MS for Cs slightly decreases with increasing ionic strength, the matrix effect was considered in the determination of the raw concentration. Moreover, having initial high ionic strength solution, it was necessary to dilute the samples before ICP-MS measurement. The experimental K_d values were then used in simple sorption models to recalculate the K_d values to compare with literature. To determine the dissolved concentration of Cs in the solution from pumice tuff, following separate experiment was carried out.

The dissolution experiment was conducted by complete immersion technique with the same grain sizes and solid-solution ratios as those used for the adsorption experiments. The pumice tuff powder of a weight of 8 g was added in 80 ml di-ionized water in a plastic container and kept in an incubator at constant temperature of 25˚C. Initially, the solid was mixed with the solution by mild hand shaking for about 3 minutes. The pH and electrical conductivity (EC) were measured every 24 h using a Horiba pH meter (Model D-51) and a Yokogawa EC Meter (Model SC82), respectively, until reaching equilibrium. After every measurement, samples were mixed by hand shaking. The measurement continued for one month since there were subtle changes in EC even after reaching near equilibrium (Figure 1). After equilibrium, Cs concentration was measured by ICP-MS.

The physical properties provide the differences in oxidation condition of the pumice tuff (Figure 2). Tuff is a type of rock consisting of consolidated volcanic ash ejected from vents during a volcanic eruption, and pumice is a common product of explosive eruptions and has a high silica content and porosity. Visual observation is noticeably different as fresh part of studied pumice tuff is light grey to grey whereas oxidized part is yellowish

brown probably due to being iron (Fe³⁺) rich (Figure 2(a)). Large pumice grains and dark color minerals are distinctive in fresh part while they are not visible in oxidized one. The mineralogical composition of fresh and oxidized parts are similar with the presence of quartz (SiO₂), plagioclase ((Na, Ca) (Si, Al)₄O₈) (dominantly albite), and pyrite (FeS₂). The broad halo of the X-ray spectra in the 2θ range of 15˚ - 35˚ indicates amorphous glass (Figure 2(b)). According to [22] , iron-titanium oxide such as ilmenite, clinopyroxene, and apatite can also be observed as the minor minerals in pumice tuff.

Regarding the pore characteristics of solid samples, both types of pumice tuffs have similar PSD beyond 10 and 0.3 μm although the pore sizes of 0.3 - 0.03 μm sizes are higher in fresh tuff (Figure 2(c)). Numbers of pores of similar diameter within 10 - 0.3 μm size are higher in fresh tuff. Higher incremental volume in oxidized pumice tuff indicates that those similar size pores are distributed within smaller range of 7.9 - 10 μm. Fresh tuff has higher pore area (average 4.6 m²/g in fresh tuff to 2.6 m²/g in oxidized tuff) though the porosity does not vary significantly (around 49% - 50%).

While observing through SEM, general surface area characteristics of both tuffs are similar (Figure 3(a), Figure 3(b)). However, the well-developed pore areas in fresh tuffs (Figure 3(c)) are sometimes filled up by either matrix materials or products from dissolution of weak materials in oxidized part (Figure 3(d)). These secondary materials may have destroyed smaller size pores in oxidized tuffs. Moreover, the large pumice grains are clearly seen in fresh tuffs with connected pore like features (Figure 3(e)). Such elongate shaped features give the larger surface area which can be evident in the difference in pore area. Since highly porous materials are reported to be damaged by oxidation [31] and by reduction of micro pores during weathering [32] , the pores in oxidized tuff are considered to be destroyed. This higher volume of similar size pores in fresh tuff creates higher surface area responsible for water-rock interaction. Twice the total pore area in fresh tuff also indicates

the presence of higher surface area available for adsorption.

Generally, pH values shift to neutral unless affected by external factors. In this study, fresh pumice tuff solution showed that tendency when pH initially (in the 2nd week) shifted towards neutral, around 7 - 8. On the contrary, pH shifted downward to 3.5 - 4 for oxidized pumice tuff due to the oxidation of pyrites [26] . In field condition, pyrites, an important sulphur mineral constituent in pumice tuff, can be oxidized in suitable geochemical environment according to following overall reactions [34] -[36] ;

After the 2nd week, the pH shifting decreased because of adjustment in the previous week and then every other week, the decrease of pH reduced more due to re-adjustments. We observed that pH shifting was not significant after 12 weeks except for several samples (with pH 6 and 8) and hence considered it had reached equili-

brium condition. Samples with initial pH of 6 and 8 were still found to be shifting even in the 14th week because of being at near neutral pH region where only small amount of pH adjusting solution has a large effect.

Three parameters―pH, ionic strength, and the initial nuclide concentrations were expected to be responsible for the variation in the K_d values. The dissolved concentration of Cs at equilibrium pH and EC was found to be insignificant comparing to the initial concentration. For both types of pumice tuffs, they were determined in the order of 5.5 - 5.9 × 10⁻⁹ M. This concentration is essentially near to the background value of analyzing solution. Hence, this negligible dissolved concentration of Cs in Equation (1) was considered as 0.

Ionic strength or salt concentration of the solutions was found to be the most influential parameter whereas the pH and the initial Cs concentration were found to have little effect on the K_d values (Figure 4). Experimental results showed that with the increase of ionic strength, K_d of Cs apparently decreases, indicating the competition of the electrolyte Na⁺ with the sorption of Cs⁺. Such ionic strength dependency was observed for both fresh and oxidized pumice tuff. For low ionic strength (less than or equal to 1.0 M), both fresh and oxidized pumice tuff yielded similar order of K_d for all initial Cs concentrations which is slightly higher than higher ionic strength (3.0 M) solution in Cs concentration of 10⁻⁴ M. This variation becomes more significant at lower Cs⁺ concentrations like 10⁻⁵ M. For 10⁻⁶ and 10⁻⁷ M Cs⁺ concentration at higher ionic strength (3.0 M), K_d values could not be determined, probably due to out of detection limit of Cs measurement. When initial Cs concentration becomes as low as 10⁻⁶ to 10⁻⁷ M, more clear influence of ionic strength was observed. For example, the K_d values were inversely proportional to ionic strength at initial Cs concentration of 10⁻⁶ to 10⁻⁷ M. Independency on initial

concentration suggests that saturation of the surface sites didn’t occur until the investigated highest Cs concentration (10⁻⁴ M).

Overall, fresh and oxidized pumice tuff did not show any major difference in K_d until initial Cs concentrations of 10⁻⁶ and 10⁻⁷ M where slightly decreased values were observed in oxidized part. This difference is better observed at ionic strength less than 1.0 M which is probably due to the difference in pore size distribution in oxidized pumice tuff. The MIP study showed that smaller pores have reduced considerably in oxidized part which may have reduced the sorption surface sites. Though weathering of mineral surface may create new surface area for adsorption [10] , lack of appropriate mineral surface in pumice tuff might be the reason for not having enough surface area. Weathering generated materials (e.g. clay particles) were also found at negligible quantity. Moreover, oxidation phenomena have destroyed the regular surfaces of fresh tuff as observed in SEM.

Furthermore, ionic strength dependency is reduced in Cs concentration more than 10⁻⁵ M. Especially at 10⁻⁴ M, similar K_d values in different ionic strength is assumed to be due to saturation of surface sites by Cs⁺. This also can be observed with increasing Na⁺ in Cs concentrations 10⁻⁶ and 10⁻⁷ M. All these characteristics are found similar for both fresh and oxidized pumice tuff and in good agreement with Cs adsorption on fresh tuff in previous study [22] . Several very high values above neutral pH at low Cs concentration may indicate small pH dependency if low pH K_d values are considered (Figure 4(c), Figure 4(d)). Although, as fresh pumice tuff solution didn’t have any pH less than neutral, such dependency cannot be confirmed.

Pumice tuff is generally composed of some microcrystalline mineral components of mostly plagioclase feldspar, quartz and others. The mixture of fine mineral particles in this volcanic glass is very difficult to isolate from each other. As such, simple one site surface complexation model for predicting K_d values for tuff has been successfully applied in some previous researches [22] [33] . Considering pumice tuff as the same solid phase, the analogous mechanism from [22] was assumed for the present case and the K_d values were recalculated (Figure 4). The results show good agreement for few samples, whereas not good for few. One reason for poor matching could be due to the number of surface site difference (as evident in the large difference of the surface area between fresh and oxidized tuff). The determination of precise surface sites is needed for further study. The recalculated K_d values also indicate that pH dependency may appear for higher K_d values at low ionic strength. It also confirms that the surface complexation model is applicable for high ionic strength too.