# Fileset

[s41526-025-00520-w.pdf](https://mdr.nims.go.jp/filesets/19c977ca-d60b-4ecf-beaa-9a0db012f279/download)

## Creator

[Atsunobu Masuno](https://orcid.org/0000-0003-0667-9782), Chihiro Koyama, [Shinji Kohara](https://orcid.org/0000-0001-9596-2680), Shunta Sasaki, Satoshi Izumi, Tomoharu Matsuya, Yuki Mikami, Kenta Yoshida, Hirotaka Kobayashi, Yuki Watanabe, Akitoshi Mizuno, Hirohisa Oda, Yuta Shuseki, Manabu Watanabe, Junpei T. Okada, Takehiko Ishikawa

## Rights

[Creative Commons BY Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/)

## Other metadata

[Glass-forming ability of La2O3–Nb2O5 evaluated via thermophysical properties under microgravity](https://mdr.nims.go.jp/datasets/eb683e09-3336-4d1b-b3df-60988a716278)

## Fulltext

Glass-forming ability of La2O3–Nb2O5 evaluated via thermophysical properties under microgravitynpj | microgravity ArticlePublished in cooperation with the Biodesign Institute at Arizona State University, with the support of NASAhttps://doi.org/10.1038/s41526-025-00520-wGlass-forming ability of La2O3–Nb2O5evaluated via thermophysical propertiesunder microgravityCheck for updatesAtsunobu Masuno1,2,3 , Chihiro Koyama4, Shinji Kohara3, Shunta Sasaki1,2, Satoshi Izumi2,Tomoharu Matsuya2, Yuki Mikami2, Kenta Yoshida2, Hirotaka Kobayashi2, Yuki Watanabe5,Akitoshi Mizuno6, Hirohisa Oda4, Yuta Shuseki1,3, Manabu Watanabe1, Junpei T. Okada7 &Takehiko Ishikawa8The La2O3–Nb2O5 binary system is a unique glass-forming system without conventional networkformer oxides, exhibiting two distinct glass-forming regions: La2O3-rich and Nb2O5-richcompositions. To evaluate its glass-forming ability, the temperature dependence of density, viscosity,and surface tension was measured using the electrostatic levitation furnace aboard the InternationalSpace Station (ISS–ELF). Melt density showed linear temperature dependence, and thermalexpansion coefficients at 2000 Kvaried from2.5 × 10−5 to 4.0 × 10−5 K−1. Substantial undercoolingwasobserved for glass-forming compositions. Viscosity measurements above the melting point revealedthat both La2O3-rich and Nb2O5-rich melts behave as fragile liquids. Activation energy derived fromviscosity data was higher for glass-forming compositions. These results suggest that glass-formingability can be assessed based on undercooling and activation energy across a wide compositionalrange, including non-glass-forming melts. The ISS–ELF experiments provide a valuable platform forunderstanding glass formation in systems inaccessible by terrestrial techniques.Oxide glass science is based on the concept of three-dimensional randomnetwork formation through corner-sharing tetrahedral units of networkformer oxides (NWFs), such as SiO2 and P2O51,2. Typically, a high con-centration of NWFs is required for glass formation, limiting the range ofchemical compositions explored in glass science. However, recentadvancements in levitation techniques have started to overcome these tra-ditional constraints3–6. By preventing heterogeneous nucleation at theinterface between the melt and its container, these techniques allow fordeeper undercooling of melts7. As a result, even materials with very lowglass-forming ability can solidify into glasses without crystallizing duringcooling. In the past two decades, levitation techniques have successfullyproduced systems such as AO–SiO2 and R2O3–B2O3 glasses containingminimal amounts of NWFs, where A denotes alkaline earth metals and Rdenotes rare-earth elements or Y4,8–12. Additionally, several glasscompositionswithoutNWFshavebeendeveloped, includingAl2O3-, TiO2-,Nb2O5-,WO3-,MoO3-, Ga2O3-, and Ta2O5-based binary glasses13–22. Theseunconventional glasses often exhibit exceptional physical properties, whichare attributed to their densely packed glass structures that lack conventionaltetrahedral networks23,24. This emerging class of glasses is expanding thehorizons of glass science, paving the way for new possibilities in both fun-damental research and practical applications25,26.Despite these advancements, the mechanisms behind unconventionalglass formation without three-dimensional networks are still not wellunderstood. Because glass forms by cooling from a melt, it is essential togather temperature-dependent thermophysical data—such as density,viscosity, and surface tension—across a broad temperature range, fromabove the melting point to the supercooled state. Levitation techniquesenable access to this wide temperature range and allow for accurate1Graduate School of Engineering, Kyoto University, Kyotodaigaku-Katsura, Nishikyo-ku, Kyoto, 615-8520, Japan. 2Graduate School of Science and Technology,Hirosaki University, 3 Bunkyo-cho, Hirosaki, Aomori, 036-8505, Japan. 3Center for Basic Research onMaterials, National Institute forMaterials Science (NIMS), 1-2-1 Sengen, Tsukuba, Ibaraki, 305-0047, Japan. 4Human Spaceflight Technology Directorate, Japan Exploration Agency, 2-1-1 Sengen, Tsukuba, Ibaraki, 305-8505, Japan. 5AdvancedEngineering ServicesCo., Ltd., 1-6-1 Takezono, Tsukuba, Ibaraki, 305-0032, Japan. 6National Institute of Technology, HakodateCollege,14-1 Tokura-cho, Hakodate, Hokkaido, 042-8501, Japan. 7Institute forMaterials Research, TohokuUniversity, 2-1-1Katahira, Aoba-ku, Sendai,Miyagi, 980-8577,Japan. 8Institute of Space and Astronautical Science, Japan Aerospace Exploration Agency, 2-1-1 Sengen, Tsukuba, Ibaraki, 305-8505, Japan.e-mail: masuno.atsunobu.3k@kyoto-u.ac.jpnpj Microgravity |           (2025) 11:58 11234567890():,;1234567890():,;http://crossmark.crossref.org/dialog/?doi=10.1038/s41526-025-00520-w&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41526-025-00520-w&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41526-025-00520-w&domain=pdfmailto:masuno.atsunobu.3k@kyoto-u.ac.jpwww.nature.com/npjmgravmeasurements of thermophysical properties27,28. Among the various levi-tation methods, the electrostatic levitation furnace (ELF) is particularlyadvantageous for accurate measurements because it maintains the levitatedmelt in an almost perfectly spherical shape29. The ELF has been widelyapplied to metals and provided valuable thermophysical data30. However,levitating oxidematerials on Earth presents considerable challenges becauseachieving the high electric field necessary to counteract gravity requires ahigh vacuum environment, which in turn leads to considerable evaporationfrom oxide melts31,32. Oxide samples can be levitated under a pressurizedenvironmentwithout evaporation, but it is very hard tomaintain the samplecharge at high temperature for stable levitation. Only a few oxide sampleshave been successfully levitated, melted to obtain their thermophysicalproperties in the pressurized ELFs33–35. To overcome these challenges,advancements in ELF technology have been made, including performingexperiments in space. Notably, the ELF installed on the Japanese Experi-mentModule “Kibo” aboard the International Space Station (ISS) allows forthe measurement of thermophysical properties of oxide melts under con-ditions close to atmospheric pressure36,37.The thermophysical properties of oxidemelts are combined with theiratomic structural data measured by X-ray and neutron diffraction experi-ments on the ground, as well as theoretical approaches for a better under-standing of their glass-forming abilities6,38,39. These investigations haveoffered valuable insights into the structural characteristics of non-glass-forming melts at high temperatures, such as ZrO2, Al2O3, UO2, andR2O340–44.In this study, we investigate the La2O3–Nb2O5 binary system.Using anaerodynamic levitation furnace, we found that certain compositions pro-duced colorless, transparent glasses with exceptionally high refractiveindices and low wavelength dispersion, even in the absence of NWFs18,45,46.Notably, the glass-forming region is divided into twodistinct areas: a La2O3-rich region (39–42mol% Nb2O5) and a Nb2O5-rich region (60–75mol%Nb2O5), with a non-vitrifying composition range separating them45,47,48.This indicates that the primary components driving glass formation changedepending on the composition. Figure 1 shows the phase diagram of thebinary system, highlighting two distinct glass-forming regions. The thermalproperties, including glass transition temperatures, show considerablevariation between these regions. Detailed structural analyses revealedmarked changes in connectivity across these regions47. Consequently, wepropose that the melt properties contain key indicators for understandingthe distinct glass-forming abilities and non-vitrifying behaviors in theLa2O3–Nb2O5 system. This study thoroughly investigates the chemicalcomposition range of La2O3–Nb2O5 melts to better understand thesephenomena.ResultsDensity and undercooling temperatureFigure 2 shows the temperature dependence of density for (99−x)La2O3–xNb2O5–1Fe2O3 (x = 29, 39, 49, 59, 69, 79, 89, and 99) melts. Datawere typically collected in the temperature range of 1500–2000 K, whilecompositions with higher melting points provided data up to 2500K. Forx = 39, a bending behavior was observed at approximately 2000 K, whereasthe other compositions showed a linear relationship between density andtemperature. It is worth noting that the bending temperature for x = 39 isclose to the liquidus temperature of 1868K. Fromthe linearfit to thedata, anapproximate formula for the temperature dependence of melt density wasLa 3NbO7LaNbO4LaNb3O920 40 60 80 1000La2O3 Nb2O5x22002000180016002400Temperature (K)La 2Nb12O33x = 39 42 60 75Fig. 1 | Phase diagram of the La2O3–Nb2O5 binary system45,48.Red arrows indicateglass-forming regions observed using the aerodynamic levitation technique.Density(g/cm3) x = 2939495969798999ΔT u,nor(K)xFig. 2 | Temperature dependence of the density of (99−x)La2O3–xNb2O5–1Fe2O3melts (x= 29, 39, 49, 59, 69, 79, 89, and 99). The downward arrows and trianglesrepresent the glass transition temperature and the density of 30La2O3–70Nb2O5 and60La2O3–40Nb2O5 glasses, corresponding to x = 69 and 39, respectively. The relativeerror is estimated to be 2.5% (not shown)37. The inset depicts the compositiondependence ofΔTu,nor, with yellow regions indicating the glass-forming regions. Thedashed line serves as a guide to the eye.Table 1 | Parameters for the density equationρ(T) = ρ(TL)+ (dρ(T)/dT)·(T− TL) for (99−x)La2O3–xNb2O5–1Fe2O3 meltsx ρ(TL) (g cm−3) dρ(T)/dT (10−4 g cm−3 K−1) TL (K)29 5.488 −5.9 200239 5.552 −5.0 186849 5.424 −5.4 189359 5.289 −5.2 181569 5.137 −6.0 164479 4.868 −5.7 161389 4.584 −5.7 164299 4.273 −4.6 1803https://doi.org/10.1038/s41526-025-00520-w Articlenpj Microgravity |           (2025) 11:58 2www.nature.com/npjmgravobtained as ρ(T) = ρ(TL)+ (dρ(T)/dT)·(T− TL) where ρ(TL) is the densityat the liquidus temperature TL. Table 1 summarizes the parameters of theequation. There was no considerable composition dependence observed forthe temperature coefficient dρ(T)/dT. As x increased, density decreasedmonotonically at a given temperature, reflecting the substitution of theheavier molecular weight La2O3 with Nb2O5.The lowest temperature measured for each composition correspondsto the crystallization temperature from the melt, except for those in theglass-forming regions. Some of the recovered samples exhibited a whiteappearance, indicating the evaporation or reduction of Fe during levitation.No significant differences were observed in the physical properties betweensamples that retained Fe and those from which Fe was depleted. Therefore,the impact of Fe2O3 on the physical properties in the present system isconsidered negligible. Accordingly, the liquidus temperatures (TL) of themelts were referenced from the La2O3–Nb2O5 binary phase diagram, asshown in Fig. 1. The maximum undercooling temperature (ΔTu) wasdetermined as the difference between the liquidus temperature (TL) and thelowest temperature (Tlw) reached before crystallization. A smaller ΔTuindicates a higher tendency for crystallization. The reduced undercoolingtemperatureΔTu,nor is defined asΔTu,nor =ΔTu/TL. The inset of Fig. 2 showsthe compositiondependenceofΔTu,nor, highlighting twopeaks atx = 39and69, which correspond to compositions capable of glass formation. Addi-tionally, ΔTu,nor values for x = 49 and 59 were higher than those for x = 29,79, 89, and 99 compositions. This suggests that the x = 49 and 59melts havea greater potential for glass formation compared to the x = 29, 79, 89, and 99melts. Notably, a sample with x = 49 formed glass in space, even though thiscomposition lies outside the glass-forming regions on the ground. Unlikelevitation using an ADL, where the melt often experiences violent rotationowing to gas flow, levitation in the ISS–ELF is highly stabilized, enablingcompositions at the boundary of non-vitrification to potentially form glass.In glass science, the temperaturedifference (ΔT) between the glass transitiontemperature and the crystallization temperature, typically measured usingthermal analysis methods such as DSC or DTA, serves as an indicator ofglass stability against crystallization. The ΔTu,nor determined through alevitation technique provides a more intrinsic and reliable measure forassessing glass-forming ability. However, this remains an experimentalobservation and does not yet reach the level of a comprehensive theoreticalframework that integrates historical metrics of glass-forming ability.Downward arrows in Fig. 2 indicate the glass transition temperature(Tg) of 60La2O3–40Nb2O5 and 30La2O3–70Nb2O5 glasses, correspondingto x = 39 and 69 melts, respectively. A distinct bend in the temperaturedependence of density, indicative of the glass transition, was frequentlyobserved for the glass-forming liquids in the ISS–ELF experiments49. Atx = 39, however, the data approached Tg, but no distinct bending of thedensity curve was observed. For both x = 39 and x = 69, density data for thecorresponding glasses at room temperature are available and are plotted astriangles in the figure. The dashed lines connecting the points of the glasstransition temperature and room temperature exhibit a gentler slopecompared to the high-temperature melt data. This behavior is consistentwith expectations, as the thermal expansion coefficient of solids is typicallysmaller than that of liquids.Thermal expansion coefficientThe thermal expansion coefficient (β) was determined from the volumedata. First, the temperature dependence of the volume was fitted to a linearequation: V(T) = AT+ B, where A and B are constants. The thermalexpansion coefficient β2000 at 2000 K was then calculated using Eq. (1),whereV2000 represents the volume at 2000 K. The linear thermal expansioncoefficient (α) was subsequently derived as α2000 = (1/3) β2000. These para-meters are summarized in Table 2. Previous studies50–52 reported β values atits melting point of 2.23 (Al2O3), 5.35 (Y2O3), 4.5 ± 0.5 (Gd2O3), 3.7 ± 0.4(Tb2O3), 4.6 ± 0.4 (Ho2O3), 10.0 ± 0.8 (Er2O3), 2.2 ± 0.6 (Tm2O3), 6.5 ± 0.9(Yb2O3) and5.0 ± 0.9 (Lu2O3) inunits of 10−5K−1. Theβ2000 valuesobtainedin this study were consistent with those previously reported data. Figure 3shows the composition dependence of α2000 for the melts, with valuesranging from2.5×10−5 to 4.0×10−5K−1.While some scatterwasobserved, alinear fit was applied, resulting in the equation α(x) = (1.75 ± 0.45) × 10−7 ×x+ (2.28 ± 0.28) × 10−5. From the extrapolation of the fitted line, the linearthermal expansion coefficients at 2000 K of La2O3 and Nb2O5 were esti-mated asαLa2O3 = (2.28 ± 0.28) × 10−5 K−1 andαNb2O5 = (4.03 ± 0.73) × 10−5K−1, respectively.β2000 ¼1V2000dVdTð1ÞViscosityFigure 4 shows the temperature dependence of viscosity for (99−x)La2O3–xNb2O5–1Fe2O3 (x = 29, 39, 49, 59, 69, 79, 89, and 99) melts.Viscosity measurements were performed using the drop oscillationmethodwith the ISS–ELF, which has ameasurable viscosity range of 10−1–10−3 Pa s.As a result, viscosity datawere collected at temperatures considerably higherthan both the glass transition and crystallization temperatures. AlthoughTable 2 | Thermal expansion coefficient parameters for (99−x)La2O3–xNb2O5–1Fe2O3 meltsx A(10−4 mm3 K−1)B (mm3) V2000(mm3)β2000(10−5 K−1)α2000(10−5 K−1)29 3.73 2.79 3.54 10.5 3.5139 2.44 2.39 2.87 8.48 2.8349 3.84 3.19 3.96 9.71 3.2459 3.29 2.68 3.33 9.87 3.2969 4.82 3.48 4.45 10.8 3.6179 3.34 2.33 3.00 11.2 3.7289 4.38 2.81 3.69 11.9 3.9699 3.53 2.68 3.39 10.4 3.47α(10−5K−1)xFig. 3 | Linear thermal expansion coefficient α at 2000 K for (99−x)La2O3–xNb2O5–1Fe2O3 melts. The dashed line represents the fitted trend.https://doi.org/10.1038/s41526-025-00520-w Articlenpj Microgravity |           (2025) 11:58 3www.nature.com/npjmgravsome compositions had a limited number of data points, the results con-sistently indicated that viscosity decreased as temperature increased.Surface tensionFigure 5 summarizes the temperature dependence of the surface tension for(99−x)La2O3–xNb2O5–1Fe2O3 melts. The surface tension of these meltsshows a mild linear temperature dependence, as depicted in Fig. 5a. Astemperature increases, the surface tension decreases. The values range from250 to 550 mN/m, which are consistent with previously reported data forglass-forming oxide liquids53. Figure 5b shows the composition dependenceof surface tension at 2000 K, clearly showing that the surface tensiondecreases monotonically as the Nb2O5 content increases. The dashedline represents a linear fit to the equation γ2000(x) =(−3.66 ± 0.23)·x+ (627 ± 21). Based on this equation, γ2000_La2O3 andγ2000_Nb2O5 are estimated to be 627 ± 21 and 262 ± 46 mN/m, respectively.The surface tension of La2O3 at themelting point under 1 atmwas obtainedas 572.5 ± 27 mN/m by using a pendant droplet method54. The surfacetension of Nb2O5 was determined to be 269 mN/m at 1934 K by using themaximum bubble pressuremethod55. Both values obtained in this study areconsistent with previously reported data54,55. The plotted data in Fig. 5(b)showa slight deviation from the linearfit, potentially indicating thepresenceof excess enthalpy. However, owing to the limited data, confirming non-linearity remains challenging.DiscussionTo analyze the viscosity data, the Andrade equation (Eq. (2)) was used,where η is the viscosity, D is the pre-exponential factor, E is the activationenergy,R is the gas constant, andT is the temperature. A linear fit of the ln ηdata based on the Andrade equation allowed for the determination of E.Figure 6 shows the composition dependence of E, with yellow regionsindicating the glass-forming regions. The results show that the activationenergy for x = 39 was considerably higher than for other compositions.Fig. 4 | Temperature dependence of viscosity for (99−x)La2O3–xNb2O5–1Fe2O3 melts. a x = 99. b x = 89. c x = 79. d x = 69. e x = 59. f x = 49. g x = 39. h x = 29.Fig. 5 | Surface tension for (99−x)La2O3–xNb2O5–1Fe2O3 melts. a Temperaturedependence of the surface tension. b Compositiondependence of the surface tension at 2000 K. Thedashed line represents a linear fit. The surface ten-sion values for La2O3 and Nb2O5 are shown as atriangle and a diamond, respectively, for refs. 54,55.γ 2000(mN/m) x(b)(a) x = 2939495969798999https://doi.org/10.1038/s41526-025-00520-w Articlenpj Microgravity |           (2025) 11:58 4www.nature.com/npjmgravSimilarly, E for x = 49, 59, and 69 was also higher compared to x = 29,79, 89, and 99. This trend aligns with the glass-forming region, wherehigher E values indicate an enhanced glass-forming ability. A lower Esuggests that ions (or flow units) can move more freely, promotingviscous flow, which correlates with a reduced glass-forming abilityand an increased tendency for crystallization. The higher E indicatesrestricted ion mobility, preventing the atomic structure from easilytransitioning to a crystalline state. Instead, the structure remains“frozen” in a liquid-like state, which supports glass formation duringcooling. The glass-forming ability inferred from the viscosity data isconsistent with the trends observed in the ΔTu,nor values obtainedfrom the cooling curves. This consistency confirms the relationshipbetween high activation energy, glass-forming ability, and the crys-tallization resistance of the melts.η ¼ D expERT� �ð2ÞThe temperature dependence of viscosity is a key parameter forevaluating glass-forming ability, as proposed by Angell56,57. Figure 7shows the Angell plot for x = 39, 59, and 69 melts, which lie in theglass-forming regions. Data for other compositions are not showndue to the absence of glass transition temperature. The vertical axisrepresents the logarithmic viscosity (η), while the horizontal axisshows the inverse of temperature, normalized by the glass transitiontemperature (Tg). The Tg values were obtained from the literature45.The plotted data strongly indicate that all the melts are fragile liquidsbecause they deviate from the linear relationship. An estimation of mis made using the Mauro–Yue–Ellison–Gupta–Allan (MYEGA)model (Eq. (3))58:log10η Tð Þ ¼ log10η1 þ 12� log10η1� �TgTexpm12� log10η1� 1� �TgT� 1� �� �ð3Þwhere log10η∞ represents the high-temperature limit of viscosity. However,accurately determining the fragility index (m) from viscosity data only athigher temperatures is challenging. For rough estimation, the simplifiedversion of the MYEGA equation (Eq. (4))59, where log10η∞ is fixed at−3, isoften employed:log10η ¼ �3þ 15TgTexpm15� 1�   TgT� 1� �� �ð4ÞThe dashed curve in Fig. 7 was calculated using the equationwith m = 80. Most of the experimental data align closely with thiscurve. One notable feature is that the temperature dependence ofviscosity in the experimental data appears slightly weaker than thatpredicted by the simplified MYEGA model. Although the modelassumes a sharp increase in viscosity as temperature decreases, itremains unclear whether adjusting only log10η∞ in the originalMYEGA formulation is sufficient to capture this behavior. Futurework should therefore focus on accurate viscosity measurements atlower temperatures, near the glass transition temperature, usingalternative methods.The density, viscosity, and surface tension of La2O3–Nb2O5 binarymelts were measured based on their temperature dependence using theISS–ELF. The compositions studied ranged from x = 29 to 99. Densitymeasurements were taken from temperatures above 2300 K down to deeplyundercooled states, with the degree of undercooling varying by compositionand being most pronounced for glass-forming compositions. The linearthermal expansion coefficientα, estimated at 2000K,was approximately 2.5× 10−5 to 4.0 × 10−5 K−1. Viscosity measurements were performed at tem-peratures above the melting point, in the range of approximately 10−1–10−3Pa·s. The activation energywas determinedusing theAndrade equation andwas found to be dependent on the composition. Higher activationenergy values were found in glass-forming regions, indicating thelimited ion mobility that facilitates glass formation. The fragilityindex (m) was estimated using the simplified MYEGA equation,revealing that the melts are highly fragile liquids, with m valuesexceeding 70. Surface tension demonstrated a linear relationship withcomposition, further supporting the connection between melt prop-erties and glass-forming ability. These findings show that the glass-forming ability can be effectively assessed using thermophysicalparameters obtained through the ISS–ELF.E(kJ/mol)xFig. 6 | Composition dependence of E for (99−x)La2O3–xNb2O5–1Fe2O3 melts.Yellow regions indicate glass-forming regions.log10η(Pa·s)Tg/Tm = 80Fig. 7 | Angell plot for (99−x)La2O3–xNb2O5–1Fe2O3 melts (violet: x= 39,magenta: x= 59, cyan: x= 69). The dotted curve represents the calculation usingthe simplified MYEGA equation with m = 80.https://doi.org/10.1038/s41526-025-00520-w Articlenpj Microgravity |           (2025) 11:58 5www.nature.com/npjmgravMethodsSample preparationHigh-purity powders of La2O3, Nb2O5, and Fe2O3 were stoichiometricallymixed to achieve compositions of (99−x)La2O3–xNb2O5–1Fe2O3 (x = 0, 9,19, 29, 39, 49, 59, 69, 79, 89, and99). Fe2O3was added to improve absorptionefficiency for the 980-nm semiconductor laser used in the ISS–ELFexperiments. Themixtures were pressed into pellets and sintered at 1000 °Cfor 12 h in air. The sintered pellet was then crushed to prepare target piecesfor the aerodynamic levitation (ADL) furnace, which is used to producespherical ceramics for the ISS–ELF. In theADL furnace, a piece of the targetmaterial was levitatedusing anO2 gasflow.ACO2 laserwas used tomelt thelevitated sample for several seconds, and the melt was rapidly cooled toroom temperature by turning off the laser. This process solidified thematerial into spherical ceramics. Depending on their composition in thepreviously reported glass-forming regions45, some compositions formedglasses, while others crystallized. The diameter of the resulting sphericalceramics ranged from 1.6 to 2.1mm, which is optimized for the ISS–ELFsample holder36. Preliminary tests were performed to ensure that thespherical ceramics maintained their shape during rocket transport andstorage in space. Samples with a high La2O3 content (x = 0, 9, and 19) wereexcluded owing to their high deliquescence and fragility. Three samplesfrom each of the remaining compositions (x = 29, 39, 49, 59, 69, 79, 89, and99) were weighed, placed in the sample holders designed for the ISS–ELF,and then transported to the ISS aboard the H-II Transfer Vehicle 7(HTV7)60. Upon arrival, astronauts installed the sample holders in theISS–ELF. The experiments were remotely controlled by our team from theTsukuba Space Center of JAXA in Japan37.Measurement conditions at ISSThe ISS–ELF chamber was filled with dry air, N2, or Ar at a pressure of2 atm. Each sample was pushed from the holder into the center of thechamber using a rod and levitated by an electric field. The sample, carryingpositive charges in the range of 10−11–10−12 C, was stabilized at the centerusing three pairs of orthogonally arranged electrodes. Its position wasmonitored by two He–Ne laser position-sensing systems. Voltage adjust-ments between the electrodes ensured the stable positioning of the sample.The levitated samples were heated by four 40W, 980 nm diode lasersarranged tetrahedrally for uniform heating. As the temperature increased,the charge on the sample occasionally reversed, leading to instability inlevitation. High-speed feedback control was used to maintain stable posi-tioning. Once melted, the oxide samples formed nearly perfect sphericalshapes owing to surface tension in the microgravity environment. Magni-fied images of the samples were captured using ultraviolet backlighting ateach temperature37,61.Temperaturemeasurements were performed using a pyrometer with awavelength range of 1.45–1.8 μm. After the laser was turned off, the melttemperature decreased quickly below the melting point and then increasedto themelting point owing to the release of latent heat during crystallization.The measured temperatures were corrected by adjusting the emissivityvalues to align the peak temperature after the recalescense with the meltingtemperature reported in the literature. Figure 8 shows a representativecooling curve obtained in the ISS-ELF experiments for the50La2O3–49Nb2O5–1Fe2O3melt, corresponding to a composition that doesnot vitrify under ground-based conditions. The temperature rapidlyincreased after crystallization, occurring 2.15 s after the laser was turned off.The dashed line indicates the melting point, approximately 1893 K. Theundercooling temperature ΔTu reached approximately 400 K, a value thatconventional crucible-basedmelt-quenchingmethods cannot achieve. Thisis because the levitatedmelt is highly stabilized in the ISS-ELF,whereas in anADL on the ground, the melt is constantly rotated and distorted by the gasflow, which promotes crystallization. While it may be of interest to inves-tigate the effect of gravity on crystallization or glass formation, such com-parisons are difficult due to numerous differences in experimentalconditionsbeyondgravity between ISS-ELF andground-based systems.TheISS–ELF effectively extended the accessible temperature range both aboveand below the melting point Tm. Thermophysical property data were col-lected from temperatures below Tm just before crystallization occurred.Density analysisThe density ρ of themelts was calculated using the formula ρ =ms/V, wherems is the mass of the sample and V is its volume. The volume V wasdetermined by analyzing magnified images taken during the experiments.Before the melt experiments, pixel measurements in the images were cali-brated to actual sample dimensions (in mm) using reference images of alevitated stainless-steel ball with a known diameter of 2.0mm, levitated inthe ISS–ELF. While the melts were nearly spherical, slight distortions wereobserved. To accurately calculate the volume, 400 edge points were identi-fied from the nearly spherical sample image and converted into polarcoordinates (R, θ). These coordinates were fitted to sixth-order sphericalharmonic functions using Eq. (5), where Pn(cosθ) represents the n-th orderLegendre polynomials and cn are the coefficients (n = 0–6) determined byminimizing the F value from Eq. (6). The total volume V of thesample was then calculated using Eq. (7). During the cooling process,magnified images of the specimen were captured at approximately10-K intervals within a few seconds. The time-dependent volumechange obtained in the ISS-ELF experiments is shown in Fig. 2. Themelt volume decreased monotonically in accordance with thermalexpansion. The temperature dependence of density was derived fromthe data series. The lowest measurable temperature varied dependingon the sample composition. The mass (ms) of each specimen wasmeasured before transport to the ISS. While this value was used fordensity calculations during the space experiments, it was later cor-rected upon return to Earth to account for any minor evaporationthat may have occurred during high-temperature melting. Theuncertainty in the density measurements has been thoroughly dis-cussed in several previous studies and is estimated to be approxi-mately 2–2.5%37,50,51,62. In the present work, we adopted anuncertainty of 2.5% based on the prior report37.R θð Þ ¼X6n¼0cnPn cos θð Þ ð5ÞF ¼X400j¼1Rj � Rj θð Þn o2ð6ÞV ¼ 2π3Z π0R3 θð Þ sin θ dθ ð7ÞTemperature (K) Volume (mm3)Time (sec)Fig. 8 | Cooling curve of the 50La2O3–49Nb2O5–1Fe2O3 melt in the ISS-ELFexperiments. The time axis begins at the moment the laser is turned off. The dashedline represents the melting point Tm of 50La2O3–50Nb2O5, as determined from thephase diagram45.https://doi.org/10.1038/s41526-025-00520-w Articlenpj Microgravity |           (2025) 11:58 6www.nature.com/npjmgravSurface tension and viscosity analysisThe surface tension γ and viscosity η of the melts were measured using thedrop oscillation method37,63. A collimated laser beam, used to sense thesample’s position, was split by a beam splitter, projecting the sample’sshadow onto a powermeter. Sinusoidal voltages were applied to the verticalelectrodes to induce oscillatory deformation in the sample.The deformationwas detected as fluctuations in the power of the He–Ne laser beam receivedby the power meter. Sample oscillations began when the excitation voltagefrequency neared the mode-2 oscillation frequency f2 of the sample. Afterthe excitationvoltagewas removed, the oscillations gradually decayedowingto the viscosity of the melt. Signals from the power meter were recordedstarting 1 s before the excitation stopped, with a time resolution of 5000Hz.Figure 9a shows a typical powermeter signal for the 99Nb2O5–1Fe2O3melt.The oscillationdecay time τwas extracted from the data as 0.064 s. Figure 9bshows the corresponding Fast Fourier Transform (FFT) result, with theresonance frequency f2 determined to be 127.83Hz. Surface tension γ andviscosity η were calculated using the following equations:γ ¼ ρr3 2πf 2� �28ð8Þη ¼ ρr25τð9Þwhere r is the radius of the melt. For the case shown in Fig. 3, the surfacetension γ and viscosity ηwere calculated to be 272.6mN/m and 11.3mPa s,respectively. During the drop oscillation measurements, sample tempera-ture is kept at a constant temperature. The drop excitation and signalmeasurements are conducted several times by sweeping the correct reso-nance frequency. Then, the heating laser powers are adjusted to the nexttemperature where drop oscillations are conducted. This sequence isrepeated with the temperature interval around 20–30 K. These measure-ments provided accurate evaluations of the thermophysical properties,contributing to a better understanding ofmelt behavior undermicrogravityconditions49,50,64. The error evaluation for surface tension and viscositymeasurements in drop oscillation experiments conducted using theISS–ELF has been thoroughly discussed in previous studies37. In the presentwork, the same error estimation methodology was applied.Data availabilityThe authors declare that the data supporting the findings of this study areavailable within the paper. Should any raw data files be needed in anotherformat they are available from the corresponding author upon reasonablerequest.Received: 28 March 2025; Accepted: 10 August 2025;References1. Zachariasen, W. H. The atomic arrangement in glass. J. Am. Chem.Soc. 54, 3841–3851 (1932).2. Sun, K.-H. Fundamental condition of glass formation. J. Am. Ceram.Soc. 30, 277–281 (1947).3. Weber, J. K. R., Felten, J. J., Cho, B. & Nordine, P. C. Glass fibres ofpure and erbium- or neodymium-doped yttria–alumina compositions.Nature 393, 769–771 (1998).4. Kohara, S. et al. Glass formation at the limit of insufficient networkformers. Science 303, 1649–1652 (2004).5. Weber, J. K. R. The containerless synthesis of glass: the containerlesssynthesis of glass. Int. J. Appl. Glass Sci. 1, 248–256 (2010).6. Benmore, C. J. &Weber, J. K. R. Aerodynamic levitation, supercooledliquids and glass formation. Adv. Phys.: X 2, 717–736 (2017).7. Kato, K., Masuno, A. & Inoue, H. Containerless solidification ofundercooled SrO–Al2O3 binary melts. Phys. Chem. Chem. Phys. 17,6495–6500 (2015).8. Kalampounias, A. G., Nasikas, N. K. & Papatheodorou, G. N. Glassformation and structure in the MgSiO3–Mg2SiO4 pseudobinarysystem: from degraded networks to ioniclike glasses. J. Chem. Phys.131, 114513 (2009).9. Nasikas, N. K., Chrissanthopoulos, A., Bouropoulos, N., Sen, S. &Papatheodorou, G. N. Silicate glasses at the ionic limit: alkaline-earthsub-orthosilicates. Chem. Mater. 23, 3692–3697 (2011).10. Nasikas, N. K., Edwards, T. G., Sen, S. & Papatheodorou, G. N.Structural characteristics of novel Ca–Mg orthosilicate andsuborthosilicate glasses: results from 29Si and 17O NMRspectroscopy. J. Phys. Chem. B 116, 2696–2702 (2012).11. Masuno, A. et al. High refractive index La-rich lanthanum borateglasses composed of isolated BO3 units. Dalton Trans. 48,10804–10811 (2019).12. Sasaki, S. et al. Structural origin of additional infraredtransparency and enhanced glass-forming ability in rare-earth-richborate glasses without B–O networks. Inorg. Chem. 59,13942–13951 (2020).13. Weber, J. K. R., Abadie, J. G., Hixson, A. D., Nordine, P. C. & Jerman,G. A. Glass formation and polyamorphism in rare-earth oxide-aluminum oxide compositions. J. Am. Ceram. Soc. 83, 1868–1872(2004).14. Weber, J. K. R. et al. Synthesis and optical properties of rare-earth-aluminum oxide glasses. J. Am. Ceram. Soc. 85, 1309–1311 (2004).15. Watanabe, Y., Masuno, A. & Inoue, H. Glass formation of rare earthaluminates by containerless processing. J. Non-Cryst. Solids 358,3563–3566 (2012).16. Yu, J. et al. Fabrication of BaTi2O5 glass−ceramics with unusualdielectric properties during crystallization. Chem. Mater. 18,2169–2173 (2006).Fig. 9 | Data obtained by the droplet oscillationmethod for the 99Nb2O5–1Fe2O3 melt. a Theoscillation signal. bThe FFT of the oscillation signal.https://doi.org/10.1038/s41526-025-00520-w Articlenpj Microgravity |           (2025) 11:58 7www.nature.com/npjmgrav17. Arai, Y., Itoh, K., Kohara, S. & Yu, J. Refractive index calculation usingthe structural properties of La4Ti9O24 glass. J. Appl. Phys. 103,094905 (2008).18. Masuno,A. & Inoue,H.High refractive indexof 0.30La2O3–0.70Nb2O5glass prepared by containerless processing. Appl. Phys. Express 3,102601 (2010).19. Yoshimoto, K.,Masuno, A., Inoue,H. &Watanabe, Y. Transparent andhigh refractive index La2O3-WO3 glass prepared using containerlessprocessing. J. Am. Ceram. Soc. 95, 3501–3504 (2012).20. Masuno, A. et al. Crystal-like atomic arrangement and opticalproperties of 25La2O3–75MoO3 binary glasses composed of isolatedMoO42–. Inorg. Chem. 63, 5701–5708 (2024).21. Yoshimoto, K. et al. Low phonon energies and wideband opticalwindows of La2O3-Ga2O3 glasses prepared using an aerodynamiclevitation technique. Sci. Rep. 7, 45600 (2017).22. Rosales-Sosa, G. A. et al. High elastic moduli of a 54Al2O3-46Ta2O5glass fabricated via containerless processing. Sci. Rep. 5, 15233(2015).23. Masuno, A. Functionalities in unconventional oxide glasses preparedusing a levitation technique. J. Ceram. Soc. Jpn. 130, 563–574 (2022).24. Masuno,A.Structureof denselypackedoxideglassespreparedusinga levitation technique. J. Phys. Soc. Jpn. 91, 091003 (2022).25. Suzuki, F. et al. Large Faraday effect of borate glasses with high Tb3+content prepared by containerless processing. Opt. Mater. 76,174–177 (2018).26. Suzuki, F., Fujita, N. &Sato, F. Small short-wavelength optical isolatorusing Tb3+-rich magneto-optical glass. Proc. SPIE 10914, 1091412(2019).27. Haumesser, P.-H., Bancillon, J., Daniel, M., Perez, M. & Garandet, J.-P. High-temperature contactless viscosity measurements by thegas–film levitation technique: Application to oxide and metallicglasses. Rev. Sci. Instrum. 73, 3275–3285 (2002).28. Langstaff, D., Gunn, M., Greaves, G. N., Marsing, A. & Kargl, F.Aerodynamic levitator furnace for measuring thermophysicalproperties of refractory liquids. Rev. Sci. Instrum. 84, 124901 (2013).29. Rhim, W. et al. An electrostatic levitator for high-temperaturecontainerless materials processing in 1-g. Rev. Sci. Instrum. 64,2961–2970 (1993).30. Ishikawa, T., Paradis, P.-F., Okada, J. T. & Watanabe, Y. Viscositymeasurements of molten refractory metals using an electrostaticlevitator.Meas. Sci. Technol. 23, 025305 (2012).31. Kelton, K. et al. Containerless solidification studies in TiFeSiO andTiZrNi polytetrahedral phase-forming liquids. in AIAA 2002-0765https://doi.org/10.2514/6.2002-765.32. Rogers, J. R. Thermophysical property measurement and materialsresearch in theNASA/MSFC electrostatic levitator. InAIPConferenceProceedings Vol. 552, (ed. El-Genk, M. S) 332–336 (AIP, 2001).33. Paradis, P.-F. et al. Contactless densitymeasurement of superheatedand undercooled liquid Y3Al5O12. J. Cryst. Growth 249, 523–530(2003).34. Paradis, P., Yu, J., Ishikawa, T. & Yoda, S. Contactless densitymeasurement of liquid Nd-doped 50%CaO–50%Al2O3. J. Am.Ceram. Soc. 86, 2234–2236 (2003).35. Paradis, P.-F., Yu, J., Ishikawa, T., Aoyama, T. & Yoda, S. Contactlessdensity measurement of high-temperature BiFeO3 and BaTiO3. Appl.Phys. A 79, 1965–1969 (2004).36. Tamaru, H. et al. Status of the electrostatic levitation furnace (ELF) inthe ISS-KIBO.Microgravity Sci. Technol. 30, 643–651 (2018).37. Ishikawa, T., Koyama, C., Oda, H., Saruwatari, H. & Paradis, P.-F.Status of the electrostatic levitation furnace in the ISS—surfacetensionand viscositymeasurements. Int. J.MicrogravitySci. Appl.39,12 (2022).38. Kohara, S., Ohara, K., Ishikawa, T., Tamaru, H. & Weber, R.Investigation of structure and dynamics in disordered materials usingcontainerless techniques with in-situ quantum beam andthermophysical property measurements. Quantum Beam Sci. 2, 5(2018).39. Kohara, S. Probing order within disorder in oxide glasses and liquidsby quantum beam diffraction. J. Ceram. Soc. Jpn. 130, 531–544(2022).40. Kohara, S. et al. Atomic and electronic structures of an extremelyfragile liquid. Nat. Commun. 5, 5892 (2014).41. Shi, C. et al. The structure of amorphous and deeply supercooledliquid alumina. Front. Mater. 6, 38 (2019).42. Skinner, L. B. et al. Molten uranium dioxide structure and dynamics.Science 346, 984–987 (2014).43. Skinner, L. B. et al. Low cation coordination in oxidemelts.Phys. Rev.Lett. 112, 157801 (2014).44. Koyama, C. et al. Very sharp diffraction peak in nonglass-formingliquid with the formation of distorted tetraclusters. NPG Asia Mater.12, 43 (2020).45. Masuno, A., Inoue, H., Yoshimoto, K. & Watanabe, Y. Thermal andoptical properties of La2O3-Nb2O5 high refractive index glasses.Opt.Mater. Express 4, 710 (2014).46. Yoshimoto, K., Masuno, A., Inoue, H. & Watanabe, Y. Thermalstability, optical transmittance, and refractive index dispersion ofLa2O3-Nb2O5-Al2O3 glasses. J. Am. Ceram. Soc. 98, 402–407 (2015).47. Masuno,A., Kohara, S., Hannon,A.C., Bychkov, E. & Inoue,H.Drasticconnectivity change in high refractive index lanthanum niobateglasses. Chem. Mater. 25, 3056–3061 (2013).48. The American Ceramic Society. ACerS-NIST Phase EquilibriaDiagrams Database version 3.2. https://ceramics.org/news-publications/phase-equilibria/49. Wilke, S. K. et al. Microgravity effects on nonequilibrium meltprocessingof neodymium titanate: thermophysical properties, atomicstructure, glass formation and crystallization. npj Microgravity 10, 26(2024).50. Koyama, C. et al. Densities of liquid lanthanoid sesquioxidesmeasured with the electrostatic levitation furnace in the ISS. J. Am.Ceram. Soc. 104, 2913–2918 (2021).51. Ishikawa, T. et al. Densities of liquid Tm2O3, Yb2O3, and Lu2O3measured by an electrostatic levitation furnace onboard theInternational Space Station.Metals 12, 1126 (2022).52. Oda,H., Shimonishi, R., Koyama,C., Ito, T. & Ishikawa, T.Determiningthe density of molten Y2O3 using an electrostatic levitation furnace inthe International Space Station. High Temperatures High Pressures52, 341–350 (2023).53. New Glass Forum. International Glass Database System INTERGLAD(New Glass Forum, 2019).54. Paras, J., Takeda, O., Wu, M. & Allanore, A. The surface tension anddensityofmoltenSc2O3, La2O3,Y2O3, Al2O3, andMgOmeasuredviaapendant droplet method.Met. Mater. Trans. B 53, 2077–2087 (2022).55. Ikemiya, N., Umemoto, J., Hara, S. & Ogino, K. Surface tensions anddensities of molten Al2O3, Ti2O3, V2O5 and Nb2O5. ISIJ Int. 33,156–165 (1993).56. Angell, C. A. Formation of glasses from liquids and biopolymers.Science 267, 1924–1935 (1995).57. Angell, C. A. Glass formation and glass transition in supercooledliquids, with insights from study of related phenomena in crystals. J.Non-Cryst. Solids 354, 4703–4712 (2008).58. Mauro, J. C., Yue, Y., Ellison, A. J., Gupta, P. K. &Allan, D. C. Viscosityof glass-forming liquids. Proc. Natl Acad. Sci. 106, 19780–19784(2009).59. Zheng, Q., Mauro, J. C., Ellison, A. J., Potuzak, M. & Yue, Y.Universality of the high-temperature viscosity limit of silicate liquids.Phys. Rev. B 83, 212202 (2011).60. Miki, Y. et al. Development of theH-II transfer vehicle (HTV).MitsubishiHeavy Industries Tech. Rev. 47, 58–64 (2010).61. Ishikawa, T., Paradis, P.-F. & Koyama, C. Thermophysical propertymeasurements of refractory oxide melts with an electrostatichttps://doi.org/10.1038/s41526-025-00520-w Articlenpj Microgravity |           (2025) 11:58 8https://doi.org/10.2514/6.2002-765https://doi.org/10.2514/6.2002-765https://ceramics.org/news-publications/phase-equilibria/https://ceramics.org/news-publications/phase-equilibria/https://ceramics.org/news-publications/phase-equilibria/www.nature.com/npjmgravlevitation furnace in the International Space Station. Front. Mater. 9,954126 (2022).62. Koyama, C., Ishikawa, T. & Watanabe, Y. Density measurements ofmolten Y3Al5O12 and Y2O3-Al2O3 compounds using an electrostaticlevitation technique. J. Am. Ceram. Soc. 106, 5648–5653 (2023).63. Fraser, M. E., Lu,W. K., Hamielec, A. E. &Murarka, R. Surface tensionmeasurements on pure liquid iron and nickel by an oscillating droptechnique. Met. Trans. 2, 817–823 (1971).64. Wilke, S. K. et al.Measuring thedensity, viscosity, andsurface tensionof molten titanates using electrostatic levitation in microgravity. Appl.Phys. Lett. 124, 264102 (2024).AcknowledgementsA part of the research was conducted under the FY2016 Kibo feasibilitystudy theme"Theoriginof fragility inhigh-temperatureoxide liquids - towardfabrication of novel non-equilibrium oxide materials: Fragility" (PI: ShinjiKohara). The authors are grateful to the ISS crew members and the groundoperation staff for their support during the onboard experiments. This studywas supported in part by JSPS KAKENHI (Grant numbers JP18K18928,JP19H05163,JP20H02429,JP20H05880, JP20H05882,andJP21K18800).Author contributionsA.Ma. designed the research, analyzed data, andwrote themanuscript withthe support of C.K. and T.I. S.K. planned the space experiment project.A.Ma., C.K., S.K., A.Mi., J.T.O. and T.I. conductedmicrogravity experimentsusing the ISS-ELFwith thehelpofY.W.andH.O.A.Ma.,S.S., S.I., T.M., Y.M.,K.Y., and H.K. prepared samples. Y.S. and M.W. contributed to the dis-cussion on melt properties. All authors reviewed the manuscript.Competing interestsThe authors declare no competing interests.Additional informationCorrespondence and requests for materials should be addressed toAtsunobu Masuno.Reprints and permissions information is available athttp://www.nature.com/reprintsPublisher’snoteSpringerNature remainsneutralwith regard to jurisdictionalclaims in published maps and institutional affiliations.Open Access This article is licensed under a Creative CommonsAttribution 4.0 International License, which permits use, sharing,adaptation, distribution and reproduction in anymedium or format, as longas you give appropriate credit to the original author(s) and the source,provide a link to the Creative Commons licence, and indicate if changeswere made. The images or other third party material in this article areincluded in the article’s Creative Commons licence, unless indicatedotherwise in a credit line to the material. If material is not included in thearticle’sCreativeCommons licence and your intended use is not permittedby statutory regulation or exceeds the permitted use, you will need toobtain permission directly from the copyright holder. To view a copy of thislicence, visit http://creativecommons.org/licenses/by/4.0/.© The Author(s) 2025https://doi.org/10.1038/s41526-025-00520-w Articlenpj Microgravity |           (2025) 11:58 9http://www.nature.com/reprintshttp://creativecommons.org/licenses/by/4.0/www.nature.com/npjmgrav Glass-forming ability of La2O3–Nb2O5 evaluated via thermophysical properties under microgravity Results Density and undercooling temperature Thermal expansion coefficient Viscosity Surface tension Discussion Methods Sample preparation Measurement conditions at ISS Density analysis Surface tension and viscosity analysis Data availability References Acknowledgements Author contributions Competing interests Additional information