A chlorination reaction was proposed as a cladding hull waste treatment technique owing to its ability to selectively recover zirconium through a simple reaction between metallic zirconium and chlorine gas [1-3]. In addition to the simple reaction, the sublimation of the reaction product (zirconium tetrachloride, ZrCl4) at a relatively low temperature of 331℃ is a great merit of the chlorination process, because it enables an easy separation of reaction product from residual used nuclear fuel (UNF). As is well known, cladding hull waste is not free of UNF after acid washing or dry decladding processes [4-6], and zirconiumbased cladding hulls (normally Zircaloy-4 and ZIRLO) are composed of more than 97wt% of zirconium, which produces beta-decay isotopes after being irradiated in nuclear power plants (NPPs). Thus, achieving UNF-free zirconium from the cladding hull waste is a key to successful commercialization of the chlorination process. Although the cladding hull waste might be classified as an intermediate level waste according to the Korean regulation, it would provide economical merits if zirconium can be selectively recovered to be disposed of as a low level waste or to be recycled because a large amount of cladding hull waste (25 wt% of UNF) is generated from the pyroprocessing or wet reprocessing.
A recent work performed in a hot cell using an actual UNF cladding hull by the Oak Ridge National Laboratory (USA) exhibited promising results with high decontamination factors (higher than 1200 for total radiation) . The Korea Atomic Energy Research Institute (KAERI) is also intensively working on the chlorination process for several years as a pyroprocessing metal waste treatment technique . After the theoretical calculations and preliminary experiments [1, 3, 7-9], KAERI is now focusing on achieving fundamental data using a specially designed thermogravimetric analysis (TGA) system for hull chlorination (TGA-HC), which can handle cladding hulls in the g-scale (up to 5 cm length) while monitoring weight change under the chlorination reaction condition [10, 11]. The chlorination reaction kinetics of bare Zircaloy-4  and ZIRLO  cladding hulls were investigated using the TGA-HC system, and it was identified that the reaction rate is under control of gas phase diffusion and chlorine gas supply in the case of ZIRLO. However, the effects of chlorine partial pressure and reaction temperature were analyzed for the Zircaloy-4 case. These results are expected to act as a fundamental background for a scale-up of the chlorination system.
In the present study, an interesting approach was made using the TGA-HC system. It is widely known that micrometer- thick oxide layers are formed on the surface of cladding hulls during NPP operation. In addition, our previous results clearly showed that the oxide layers reduce the chlorination reaction rate or completely protect the hulls from chlorine gas [9, 12]. In other words, side walls of actual cladding hulls might not be available for the chlorination reaction. However, we do have a fresh surface in the cladding hulls: during pre-treatment of the pyroprocessing (or wet processing), fuel rods are cut into pieces (about 1 inch length) so that nuclear materials can be separated from the hardware and cladding hulls. Here, metallic surfaces are exposed on both ends of the cladding hull cuttings, and surely they will react with chlorine gas. In the present study, the effect of the cross-section opening was quantitatively investigated using the TGA-HC system.
The TGA-HC system developed in our group [10, 11] was employed in this study. Briefly, the TGA-HC system is composed of four compartments: 1) the weighing system that employs a balance put on the top of the system. The balance is connected with a computer system to control the balance and record the measured values. The balance is covered by an acryl box connected to an argon feed system to protect the balance from an up-stream of chlorine gas. 2) Cladding hull samples are put in the middle of the reaction part by hanging stainless steel wire and a quartz basket to a hook of the balance. The reaction part was made of quartz and the temperature was controlled by a heating furnace. 3) In the bottom of the system, a cooling part was connected to collect reaction products (chlorides), and gaseous components (argon and un-reacted chlorine) were introduced to a 4) scrubbing system.
The ZIRLO cladding hull samples were prepared to have different cross-section openings. First, a 3 cm-long ZIRLO cladding hull was oxidized at 500℃ for 24 h under an air atmosphere (ZIRLO-I). Second, a 5 cm-long ZIRLO cladding hull was oxidized at an identical condition and then cut into two pieces to have lengths of 3 and 2 cm. The 3 cm-long sample which has one fresh cross-section was employed for the experiments (ZIRLO-II). Third, a 5 cmlong ZIRLO cladding hull was oxidized at 500℃ for 24 h as ZIRLO-I and -II. Both ends of the oxidized ZIRLO cladding hull were cut by 1 cm to produce a 3 cm-long ZIRLO cladding which has fresh cross-sections on both ends (ZIRLO-III). The chlorination reaction experiments were performed at 400℃ by flowing 100 mL/min of argon and 20 mL/min of chlorine gas (16.9 kPa of Cl2 partial pressure (pCl2)).
3.Results and discussion
The protective role of oxide layers was investigated using the ZIRLO-I sample which does not have a metallic surface. A comparison between ZIRLO-I and the bare ZIRLO cladding hull  was made in Fig. 1, where it is clear that ZIRLO-I was not reactive with chlorine gas under the condition of the present study. Based on the previous report , the thickness of oxide layer on the ZIRLO-I sample was calculated as 3.22 μm, and Fig. 1 clearly shows that this thickness is enough to prevent the chlorination reaction. Here, it needs to be discussed that ZIRLO cladding hulls oxidized at an identical condition of this work was reactive with chlorine gas in our previous work . The only remarkable difference between the two experiments is gas flow: 50 mol% of Cl2-Ar gas was fed at a total flow rate of 40 mL/min in Ref. 14, while 17 mol% of Cl2-Ar was supplied at a rate of 120 mL/min in the present study. Thus, it can be suggested that increasing chlorine partial pressure from 16.9 to 50.7 kPa can overcome the 3.22 μm thick oxide layer. But, this oxide layer worked as an excellent protect layer in the reaction condition of this study where the chlorine concentration is limited because of system capacity.
The chlorination reaction results of ZIRLO-II and -III are shown in Fig. 2. Here, progress of the reaction was expressed as α which is defined as:
where m0 is the initial weight, mt is the weight at time t, and m∞ is the weight at the end of the reaction. Interestingly, the effect of oxidation was not so serious considering that the surface areas of the bare ZIRLO, ZIRLO-II, and ZIRLO-III were 18.3, 0.209, and 0.418 cm2, respectively. These results suggest that the opening of the cross-section is a key factor in determining the reaction rate, and the reaction through the side walls plays a minor role. The minor role of the side walls can be explained by heat treatments and mechanical processes applied to the ZIRLO tubes for manufacturing. Before deriving the reaction rate equations for ZIRLO-II and –III, it needs to be noted that the chlorination reaction of the bare ZIRLO under the condition of this study (400℃, 120 mL/min total flow rate, and 16.9 kPa of pCl2) lies within the gas phase transport limited region . Normally the gas phase transport phenomenon in a gas-solid reaction is divided into two categories of gas phase diffusion through the boundary layer and the supply of the reactant gas . In addition, the effect of the gas phase diffusion can be calculated using the Ranz-Marshall equation [16-18] as follows:
where dnCl2/dt is the chlorine molar flow rate, DAr-Cl2 is the binary diffusion coefficient for the Ar-Cl2 system, NRe is the Reynolds number (NRe = U·L/ν, U is the linear fluid velocity, and ν is the kinematic viscosity), NSc is the Schmidt number (NSc = ν/DAr-Cl2), L is the sample characteristic dimension (diameter of cladding hulls (9.5 mm) was employed), pCl2 is the chlorine partial pressure at the bulk and sample surface (subscript 0 and S, respectively), Rg is the gas constant, T is the absolute temperature, and A is the sample external surface. The calculation and experimental results for ZIRLO-II and –III are listed in Table 1 with those of bare ZIRLO . It should be discussed here that the measured chlorine molar flow rates are higher than the calculated values for the ZIRLO-II and –III cases. Considering that the Ranz-Marshall equation assumes a freely flowing gas condition, the actual molar flow rate of chlorine for a sample in a crucible might be in the range of 1/10 – 1/100 of the calculated value [19, 20]. In other words, the measured values should be about 1/10 – 1/100 of the calculated values. This result provides an important hint that the surface areas employed for the calculations (fresh cross-sections) should be revised. The slight decrease in the reaction rate (Fig. 2) and the calculation results (Table 1) suggest that the crosssection openings work as a starting point for the chlorination reaction and the side walls become available as the reaction proceeds, although it cannot be verified at this time.
The experimental results for ZIRLO-II and –III were further analyzed using Sharp-Hancock plots  to identify the suitable geometry function that describes the morphological changes of the cladding hull during the reaction [15, 22]. The Sharp-Hancock plot is expressed using the equation of
where B is a constant and m is a slope that describes the best geometry function of the system. In addition, the Sharp- Hancock plot reveals a range where the geometry function is available. The Sharp-Hancock plots of ZIRLO-II and –III are shown in Fig. 3. The m values were identified as 1.04 and 1.05 for ZIRLO-II and ZIRLO-III, respectively, which means that the contracting volume model (slope = 1.07) is a suitable one for these cases as our previous works for the bare Zircaloy-4  and ZIRLO  cladding hulls. Here, it needs to be noted that the linear fitting was available only for a limited region of 0 ≤ α ≤ 0.55 for both ZIRLO-II and –III. This range is somewhat narrower than the case of the bare ZIRLO (0 ≤ α ≤ 0.73), and it indicates that the reaction mechanism is changed to a different one earlier in the oxidized samples than the bare one. The reaction mechanism in the range of 0.55 ≤ α is not decisive because the slope increased with time. Using the contracting volume model, the reaction rate equation can be written as:
where kapp is the apparent reaction constant and t is reaction time at the hourly scale. Assuming that ZIRLO II and –III follow an identical dependence on the chlorine partial pressure with the bare ZIRLO, and that the effect of the reaction temperature is constant at 400℃, the following reaction rate equation can be derived:
where 0.873 came from our previous work on the bare ZIRLO cladding hulls  and k represents a constant that includes the Arrhenius equation. The k value was achieved by fitting the experimental results using Eq. (6), which is a re-arrangement of Eq. (5).
The k values identified were 0.00579 and 0.00623 for ZIRLO-II and –III, respectively, and the fitting results are shown in Fig. 4. According to Eq. (6), a larger k value implies a faster reaction rate, and it needs to be mentioned that the k value of the bare ZIRLO was 0.00738 . Thus, it can be concluded that the oxide layers on the side walls of ZIRLO-II and –III played negative role on the reaction rate point of view. Using the fitting results, the reaction equations for ZIRLO-II and –III can be finalized as follows:
for ZIRLOO-II(one side cross-section open)
for ZIRLOO-III(two side cross-section open)
Here, it should be noted that these reaction equations are available only in a limited condition of the reaction temperature (400℃), chlorine partial pressure (16.9 kPa) and α (0 ≤ α ≤ 0.55). The effects of the reaction temperature and chlorine partial pressure can be a future work for complete reaction rate equations.
As mentioned above, the reaction rate equations derived using the geometry function are not applicable for the 0.55 ≤ α region. Thus, a reaction rate equation study for a high α region was not conducted in the present study. However, it is clear from Fig. 2 that both ZIRLO-II and –III took 8 hours to complete the reaction while it took 7 hours for the bare ZIRLO, which is a 14% delay in completing the reaction. These results lead to the conclusion that the crosssection opening can provide reaction pathways for chlorine gas to completely convert metallic zirconium into ZrCl4, but the reaction completion time is increased by 14%. In addition, it should be emphasized that no significant changes were observed for the cladding hulls which have cross-sections on one or both sides.
The effect of cross-section openings on the chlorination reaction rate of the ZIRLO cladding hulls was quantitatively identified, resulting in the reaction equations listed below:
for ZIRLOO-II(one end cross-section open)
for ZIRLOO-III(both end cross-section open)
Although the reaction equations above are available only under limited conditions, it was verified that severely oxidized cladding hulls can be chlorinated if a metallic crosssection is provided. In addition, the reaction time to complete the chlorination reaction was extended by 14% for the oxidized hulls, which have cross-sections at one or both ends. These results clearly proved that the oxide layers formed on the surface of the cladding hulls during the NPP operation are not troublesome for the commercialization of the chlorination process.