VO2-dispersed glass: A new class of phase change material

Energy storage technology is crucial for a sustainable society, and its realisation strongly depends on the development of materials. Oxide glass exhibits high durability. Moreover, the amorphous structure of the glass without periodic ordering demonstrates excellent formability and controllability, thus enabling a large-scale production. These factors provide impetus for the development of new materials for thermal management applications. As vanadium dioxide (VO2) with a strongly correlated electron system exhibits a structural phase transition, leading to a large heat of transition. Therefore, VO2 demonstrates immense potential as a phase change material (PCM). This study reports the fabrication of VO2-dispersed glass and examines its potential as a new latent heat storage material, which can be applied for massive PCM heat storage applications.

and NIPPON DENKO CO., LTD. However, as VO 2 is available as a powder, it is mandatory to use a container if the VO 2 powder is used as the PCM. In addition, vanadium is a rare metal; hence, VO 2 reagents are expensive. Thus, VO 2 exhibits some disadvantages that need to be improved so as to enable large-scale PCM applications.
In this study, a new PCM is proposed, i.e., VO 2 -dispersed glass, where the VO 2 powders embedded in the glass matrix, serves as a durable container-free PCM and prevents the degradation of the VO 2 phase because of oxidation and moisture. Glass exhibits immense advantages from the viewpoint of materials science, e.g., large-scale or mass production, flexibility, and formability. Hence, recently, glass has been extensively examined not only for photonics but also as energy-related materials [14][15][16][17] . Based on this background, VO 2 -dispersed glass exhibits immense potential for PCM on the basis of the solid-solid phase transition. An incorporation method 18 is utilised for fabricating the dispersed glass (Fig. 2), aiming to realise the all-solid PCM.

Results
Selection of the glass matrix composition. First, several glasses were prepared to investigate the composition of the glass matrix suitable for the VO 2 dispersion. In this study, three compositions as candidates for the glass matrix, i.e., 35BaO-65B 2 O 3 , 15B 2 O 3 -10P 2 O 5 -75V 2 O 5 , and 30BaO-10TeO 2 -60V 2 O 5 , were examined. A borate system was selected because of a previously reported study in which vanadate compounds have been embedded in a borate glass matrix 19 . Vanadate systems have been selected because a low-processing temperature is expected for synthesis [20][21][22] . Endothermic or exothermic reaction of the VO 2 phase caused by phase transition. DTA measurement was carried out using reagent-grade VO 2 powder. In the DTA curves, endothermic and exothermic peaks were observed in the heating and cooling processes, respectively, corresponding to the phase transition between the monoclinic and tetragonal (rutile-type) structures at a T c of ~68 °C. Differential thermal analysis (DTA) (Fig. 3(a)) revealed that the BaO-B 2 O 3 and BaO-TeO 2 -V 2 O 5 systems exhibit high thermal stability against crystallisation because the exothermic peak corresponding to the amorphous-crystal phase transformation was absent, whereas a strong, sharp crystallisation peak was observed for the B 2 O 3 -P 2 O 5 -V 2 O 5 system, corresponding to formation of the V 2 O 5 phase (inset). On the other hand, vanadate systems exhibited a considerably lower glass-transition temperature (T g ) compared with that of the borate system. A low T g is preferable for the VO 2 dispersion to proceed at a low temperature with respect to energy savings as well as the prevention of degradation of VO 2 because of high-temperature exposure.
Next, the incorporation method was carried out as a trial to fabricate VO 2 -dispersed glasses, which were subsequently analysed by powder X-ray diffraction (XRD) (Fig. 3(b)). The VO 2 phase in the borate system was transformed into V 3 O 5 (or the so-called Magneli phase) 23 , and the size of the Magneli phase (~10 μm, Supplementary  Fig. S2) is less than that of the VO 2 reagent (~20 μm, Supplementary Fig. S1), suggesting the elution or diffusion of vanadium into the BaO-B 2 O 3 glass matrix. On the other hand, the VO 2 phase was stably retained in case of the vanadate systems. This result led us to expect that vanadate systems are suitable for fabricating VO 2 -dispersed samples. In addition, the water stability test was carried out using the candidate vanadate system glasses. After the immersion of the B 2 O 3 -P 2 O 5 -V 2 O 5 system glass into water, a pale-yellow colour was immediately observed, and then the water became black after 1 week (Fig. 3(c)). Meanwhile, the BaO-TeO 2 -V 2 O 5 system remained transparent, indicative of its high resistively against water or moisture. Thus, the BaO-TeO 2 -V 2 O 5 system demonstrates immense potential as glass matrices, which can be used for the VO 2 -dispersed PCM. Hence, 30BaO-10TeO 2 -60V 2 O 5 is selected as the glass matrix.
Characterisation of the VO 2 -dispersed glass. VO 2 -dispered glasses with different additive amounts (x; mol%) of VO 2 powder, i.e., 30BaO-10TeO 2 -60V 2 O 5 :xVO 2 composites, were fabricated according to the protocol described in the Method section. Scanning electron microscopy (SEM) analyses revealed that the incorporated VO 2 powders are stably embedded in the matrix, with no significant aggregation of the powders (x = 80; Fig. 4(a)). Elemental mapping results revealed a clear boundary between the VO 2 particles and glass phases. In addition, the migration of barium (Ba) and tellurium (Te) surrounding the VO 2 particles and the diffusion of vanadium (V) into the glass matrix were barely observed. To primarily examine the heat storage property, thermal cycling test was also carried out by means of DTA in the VO 2 -dispersed sample. Clear endothermic and exothermic peaks were observed in the temperature range of ~60-80 °C, which are attributed to the phase-transition of VO 2 , during the heating and cooling process, respectively. (Fig. 4(b)). The peaks of 10-cycle could be almost superimposed on that of 1-cycle. In addition, the SEM observation revealed no significant change in the microstructure for the sample after the cycling test ( Supplementary Fig. S3). Since any collapses were hardly observed in the sample subjected to the cycling test, it is suggested that the dispersed glass possesses a high thermal repeatability.
Thermal properties. To examine the latent heat storage function in detail, DSC measurements were carried out. A steep change in the specific heat (peak of the C p -T curve), related to the phase transition of the dispersed VO 2 powder, was observed around the phase-transition temperature of VO 2 (T c~6 8 °C), and then the transition enthalpy (ΔH), which corresponded to the latent heat storage capacity, was estimated (Fig. 5a,b, and c) for x = 60, 80, and 120, respectively). The ΔH values increased with x, and finally the sample with x = 120 exhibited a ΔH value of ~14.3 J/g (Fig. 5(d)), corresponding to ca. 30% of that of the VO 2 reagent (~45 J/g, Supplementary  Fig. S1). Thus, the latent heat storage function is imparted to the glass-based material via the VO 2 dispersion. In addition, the temperature retention property of the VO 2 -dispersed glass was assessed. Small pieces of matrix glass (free from VO 2 , bulk state) and dispersed glass (x = 120, bulk) and VO 2 reagent (powder), with similar weight (~2.2-2.4 g), were heated at around ~100 °C, and then were left at room temperature (RT). Variations in the surface temperature were monitored by thermography. In the matrix glass, the temperature monotonically decreased with time, eventually returning to RT (Fig. 6a)). On the other hand, the VO 2 reagent exhibited a temperature plateau for ~2.5 min, corresponding to the latent heat related to the phase transition, with subsequent cooling to RT. In addition, the dispersed glass exhibited a plateau for ~1.5 min, indicative of the temperature retention properties. Thermographic images also aided in the better understanding of the temperature variation. The dispersed glass and VO 2 reagent exhibited similar temperature distribution after ~1 min (start of the plateau) ( Fig. 6(b)). Notably, the VO 2 reagent exhibited a fluctuation in the image colour, indicative of a non-homogeneous temperature gradient. This fluctuation was related to the difference in the packing density of the VO 2 powder. However, the dispersed glass barely exhibited any fluctuation.  Hence, it is hypothesised that a 3D structure provides a stable network and considerably contributes to the high thermal stability against crystallisation and water resistivity in the BaO-TeO 2 -V 2 O 5 system.

Discussion
The homogeneity in the temperature retention (no fluctuation of temperature distribution, Fig. 5) was observed in the dispersed glass, and the presence of the glassy phase probably contributed to the minimisation of the pores occurring between the VO 2 particles, which necessarily occurred in the powder state. Temperature fluctuations should be enlarged via the expansion of the material in case the PCM powder is used; hence, dispersion or incorporation plays a crucial role in maintaining the homogeneity of the spatial temperature distribution. Moreover, generally, thermal conductivity (Λ) of multicomponent glass possesses Λ ~ 1 Wm −1 K −1 , regardless of its composition/system, and such a low Λ originates in prevention of phonon-propagation due to the random-network (disordered) structure without transition symmetry, e.g., 40P 2 O 5 -60V 2 O 5 glass; Λ ~ 0.8 Wm −1 K −1 (at 350 K or 77 °C) 31 . Although the Λ-value of matrix glass in this study (30BaO-10TeO 2 -60V 2 O 5 ) is considered to be lower than that of VO 2 crystal (Λ ~ 4 Wm −1 K −1 and Λ ~ 5 Wm −1 K −1 in monoclinic and tetragonal systems, respectively) 32 , the low Λ-value of glass is possibly preferable from the viewpoint of heat-retaining property because a gradual cooling is expected in the VO 2 -dispersed glass before/after temperature retention (plateau), and we can also see this trend in Fig. 6(a). The acquisition of the temperature retention properties in glass materials with the dispersion or immobilisation of the VO 2 phase is expected to be valuable to exploit new functions and applications.
There are some reports about heat storage material consisting of polymers, e.g., polymer blend PCM. However, taking that oxide (or ceramic) material basically possess a high mechanical/thermal properties compared to polymer (organic) material into account, glass-based PCM is expected to have a long-term reliability. Furthermore, glass-based PCMs are considered to demonstrate potential for massive thermal storage applications, for example, in space development and terraforming, on the basis of the advantages of glass materials, e.g., large-scale and mass production. In particular, the presence of abundant glass and related minerals (e.g., pyroxene and olivine) on Mars 33 has also vigorously stimulated the study of glass-based PCMs.
In conclusion, a new PCM based on glass materials, i.e., VO 2 -dispered glass in multicomponent systems, is reported, and its latent heat storage and temperature retention properties are demonstrated. On the other hand, because of the dispersed glass still being a prototype, some issues should be overcome, e.g., improvement of the heat storage amount. Nevertheless, as the technology for glass-crystal composites and their industrialisation has been previously reported 34 , the results obtained herein demonstrate significance as the first step in the development of all-solid PCMs.  ) under air. The melts were poured onto a steel plate heated at ~200 °C, followed by pressing using another steel plate to obtain the as-quenched samples (Quenching rate: ~10 1 -10 2 K/sec). Their samples were confirmed to be in the glassy state as evidenced by X-ray diffraction (XRD) analysis.
Dispersion of VO 2 in the glass matrix. To fabricate the VO 2 -dispersed glass, the incorporation method reported in the study by Komatsu et al. 18 was utilised. Figure 2 shows the schematics of the procedure. A powdered matrix glass and VO 2 powder (purity: 99.9%; Kojundo Chemical Laboratory Co., Ltd.) were thoroughly mixed using an alumina mortar. Second, the mixture was added into an alumina crucible and melted under the following conditions: 1200 °C for 10 min (BaO-B 2 O 3 system), 1200 °C for 15 min (B 2 O 3 -P 2 O 5 -V 2 O 5 ), and 900 °C for 10 min (BaO-TeO 2 -V 2 O 5 ) under air. The quenching process was similar to that utilised during the preparation of the matrix glass, finally furnishing VO 2 -dispered glasses with different matrices.
Characterisation of the matrix glass and VO 2 -dipersed glass. In the studied glasses and VO 2 -dispersed samples, the state of the material and crystals were observed by XRD (Cu-Kα radiation). Microscopic observation was carried out by scanning electron microscopy (SEM) equipped with energydispersive X-ray spectroscopy. Water stability was examined by the immersion of the samples (~0.2 g; bulk form) into water.
The thermal properties of the matrix and VO 2 -dispersed glasses were examined by differential thermal analysis (DTA; heating rate of 10 K/min, Rigaku Corporation, Thermoplus TG8120). The transition enthalpy (ΔH), corresponding to the amount of the stored latent heat, of the VO 2 -dispersed samples was evaluated by differential scanning calorimetry (DSC; heating rate of 1 K/min, Seiko Instruments Inc., DSC6220). The specific heat C p [= ∂ ∂ H T ( / ) p ] was measured as a function of temperature (T), and ΔH was estimated on the basis of the thermodynamic relation: The sample state used in the measurement was the bulk form with a weight of ca. 10 mg. The time dependence of temperature in the examined samples was evaluated by a thermography test (Nippon Avionics Co., Ltd.; R300SR-S).