Journal of Earth Science  2018, Vol. 29 Issue (2): 245-254   PDF    
Time Scale of Partial Melting of KLB-1 Peridotite: Constrained from Experimental Observation and Thermodynamic Models
Wei Du1,2, Li Li2, Donald J. Weidner2    
1. State Key Laboratory of Ore Deposit Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences, Guiyang 550081, China;
2. Mineral Physics Institute, Department of Geosciences, University of New York at Stony Brook, Stony Brook NY 11790, USA
Abstract: Partial melting experiments were carried on KLB-1 peridotite, a xenolith sample from the Earth's upper mantle, at 1.5 GPa and temperatures from 1 300 to 1 600 ℃, with heating time varies from 1 to 30 min. We quantify the axial temperature gradient in the deformation-DIA apparatus (D-DIA) and constrain the time scale of partial melting by comparing experimental observations with calculated result from pMELTS program. The compositions of the liquid phase and the coexisting solid phases (clinopyroxene, orthopyroxene, and olivine) agree well with those calculated from pMELTS program, suggesting that local chemical equilibrium achieves during partial melting, although longer heating time is required to homogenize the bulk sample. The Mg# (=Mg/(Mg+Fe) mol.%) of olivines from the 1-minute heating experiment changed continuously along the axial of the graphite capsule. A thermal gradient of 50 ℃/mm was calculated by comparing the Mg# of olivine grains with the output of pMELTS program. Olivine grains at the hot end of the graphite capsule from the three experiments heated at 1 400 ℃ but with different annealing time show consistence on Mg#, indicating that partitioning of Fe2+ between the olivine grains and the silicate melt happened fast, and partial melting occurs in seconds.
Keywords: peridotite    partial melting    temperature gradient    pMELTS program    

The low velocity zone (LVZ) in the Earth's upper mantle has been correlated with partial melting of mantle peridotite (Li and Weidner, 2013; Harmon et al., 2009; Anderson and Sammis, 1970). However, to directly relate partial melting to the low seismic velocity zone, the time scale of partial melting need to be confirmed, because the seismic P-wave and S-wave velocity will be softened by the partial melting process only if the time scale to accumulate certain amount of melts is close to or lower than the seismic period.

The solidus of peridotite was determined by experimental observations of the presence and absence of glass or quenched crystals (e.g., Zhang and Herzberg, 1994; Hirose and Kushiro, 1993; Takahashi, 1986). However, because volatiles components affect the melting temperature of the bulk sample, different groups proposed different methods to get volatile-free solidus of mantle peridotite (Lesher et al., 2003; Herzberg et al., 2000; Walter, 1998). On the other hand, the theoretical model MELTS (Ghiorso and Sack, 1995) shows consistent result with experimental studies on KLB-1 peridotite in the spinel faces (Herzberg et al., 2000); and its revised version pMELTS has a number of improvements on the calculation of phase relations and major element partitioning between residual solid phases and liquid phase, although large discrepancy occurs at higher pressure domains (> 3 GPa) (Lesher et al., 2003; Ghiorso et al., 2002; Hirschmann, 2000; Asimow and Ghiorso, 1998; Hirschmann et al., 1998; Ghiorso and Sack, 1995). Specifically, the output of pMELTS program is applicable to predict compositional trends of near-solidus mantle melts, the effect of temperature changes on the extend of partial melting, and the mode change of the residual phases for fertile peridotite for all oxides (Ghiorso et al., 2002).

Forsterite content (Mg#) in olivine is an indicator of the degree of partial melting because the highly incompatible of Fe (Agee and Walker, 1990), therefore the temperature difference (thermal gradient) inside heater during partial melting experiments can be estimated from the Mg# of olivine grains (Herzberg and Zhang, 1996); and the thermal gradient developed during partial melting experiments was widely used to explore the solidus and liquidus of peridotite at different pressures and the origin of basalts from different degree of partial melting (e.g., Zhang and Herzberg, 1994; Herzberg et al., 1990; Takahashi, 1986; Ohtani, 1979). A potential problem with the use of temperature gradient to constrain phase diagram is the Soret diffusion, which was reported by Walker and Delong (1982), and if occur, will change the composition of the liquid phase continuously through diffusion and make the static phase equilibrium between the solid phases and the coexist liquid phase impossible (Lesher and Walker, 1988). However, the Soret diffusion only happens in melting experiments with long duration or experiments with melt content more than 10%, and for experiments with running time as short as 3 min, no effect of Soret diffusion on major elements was observed along temperature gradient (Herzberg et al., 1990; Kato et al., 1988). Therefore, we designed an experiment to constrain the thermal gradient in D-DIA cell during partial melting by cutting the heating time to 1 min to avoid possible Soret diffusion inside the heater.

The chemical analysis on KLB-1 peridotite (xenolith from Kibourne Hole, New Mexico) carried by different groups show consistence on its bulk composition (Davis et al., 2009; Zhang and Herzberg, 1994; Takahashi, 1986). The natural peridotite sample served as a good analog of the Earth's upper mantle for studying the origin of all kinds of basaltic magma (Davis et al., 2011; Hirschmann, 2010; Dasgupta et al., 2007; Yoshino et al., 2005; Hirose and Fei, 2002; Hirose, 1997; Zhang and Herzberg, 1994; Takahashi et al., 1993; Takahashi, 1986; Ito and Kennedy, 1967). And as mentioned above, these experimental studies were used as references to evaluate the MELTS program, which makes KLB-1 peridotite a good sample for both partial melting experiments and theoretical calculations with pMELTS program.

In this study, we present laboratory studies on partial melting of KLB-1 peridotite at 1.5 GPa but different temperatures from 1 300 to 1 600 ℃ and partial melting experiments at the same temperature 1 400 ℃ but with different heating time (1, 10, and 30 min) using D-DIA multi-anvil device. The chemical composition of olivine grains from the cold end to the hot center of the partial melted sample was determined with microprobe analysis through liner travers method. The temperature gradient inside the graphite capsule was constrained by comparing the Mg# of olivine grains with thermodynamic calculation result from pMELTS program (Smith and Asimow, 2005). The chemical composition of the residual solid phases (Cpx, Opx, and olivine) and the coexisting liquid phase were compared with the calculated result from pMELTS program to evaluate the local chemical equilibrium and the time scale of partial melting.

1 EXPERIMENTAL AND ANALYTICAL METHODS 1.1 Starting Material and Cell Assembly

The aim of this study is to examine the time scale of partial melting of mantle peridotite. For this purpose, we used the same KLB-1 peridotite powder as starting material, which we did not put special effort to eliminate H2O. As suggested by Lesher et al. (2003), KLB-1 peridotite could have ~150 ppm H2O in the bulk composition, and this amount of H2O could influence the near-solidus phase relations of partial melting KLB-1 peridotite, showing difference compared with results from anhydrous partial melting experiments. The structurally bounded OH in the anhydrous silicate and the possible CO2 from the sealed graphite capsules do not contribute to ambiguities in this study because: (1) we have no attempt to constrain the solidus and the composition of the liquid phase as long as we got the KLB-1 sample partially melt within the temperature uncertainty of the solidus; and (2) to reduce the sources of experimental uncertainty for our object of this study, to keep the identical of the starting material is as important as to use the same cell assembly within the same pressured cell.

KLB-1 peridotite powder was packed into a graphite inner capsule, which was isolated from the graphite heater by a boron nitride capsule. The cell assembly is the same as Du et al. (2014), which has been widely used in DIA high-pressure device for deformation experiments (Fig. 1). The crushable Al2O3 isolator, which is relatively high thermal conductivity at high temperature range 800–1 400 ℃ (e.g., Munro, 1997), will enhance heat loss during heating and cause a marked axial thermal gradient inside the cell assembly. By using X-ray transparent anvils, the lattice spacing of Al2O3 can be determined by in situ X-ray spectra, which was adopted to characterize the axial thermal gradient inside the cell assembly (Raterron et al., 2013), serving as a great reference to confine the possible thermal gradient developed during our partial melting experiments.

Figure 1. Cell assembly used in D-DIA deformation experiments. The outside graphite heater used here has an OD 2.36 mm, ID 2.10 mm, and length 6.15 mm. The inside graphite sleeve works as oxygen buffer with an OD 1.55 mm, ID 1.08 mm, and length 3.15 mm, which was described in Li (2009).
1.2 Multi-Anvil Experiments

All experiments were performed in a D-DIA device at the X17B2 beamline of the National Synchrotron Light Source (NSLS), Brookhaven National Laboratory, USA. In order to minimize the possible thermal effect caused by thermocouple and to better compare with the reported experimental result on the same cell assembly during deformation experiments, we use the consumed power correlated with the thermocouple reading, T=f(W), to evaluate the heating temperature as was used by Raterron et al. (2013). The T=f(W) relation was determined by two calibration experiments (KLB-1_22 and KLB-1_27) with W3%Re-W25%Re thermocouple, which show identical result with previous calibrations. Each cell assembly was pressured cold to the same pressure 1.5 GPa, followed by heating up to 1 000 ℃ (200 W) and annealed at this temperature for half an hour. After annealing, the KLB-1 samples were heated up to the target temperatures (Table 1). All experiments were quenched to room temperature by shutting down power, and the temperature inside the cell dropped to ~100 ℃ in seconds.

Table 1 P-T-t conditions of partial melting experiments on KLB-1 peridotite in D-DIA device
1.3 Scanning Electron Microscope and Electron Microprobe

Run products were impregnated with epoxy and cut to cross-sections. After polishing with loose alumina powder, the cross-sections were coated with a thin conductive gold layer and checked by scanning electronic microscope (SEM) at Stony Brook University. The quenched samples' structure and phase relation were described by Du et al. (2014). Quantitative analysis of the chemical composition of the quenched glasses and the solid phases, for example, olivine and their neighboring clinopyroxe (Cpx) and orthopyroxe (Opx) around a triple junction were performed using wavelength dispersive analyses on Jeol JXA-8200 electron microprobe at the Department of Earth and Planetary Sciences of Rutgers University with an accelerating voltage of 20 keV, current of 20 nA, and a focused spot (~1 mm diameter). Peak intensity data were collected from Si, Na, Fe and Mg for 20 s, Ca for 40 s and Ti for 50 s. Intensity data from Na and K was corrected for Time Dependent Intensity (TDI) using a self-calibrated correction on the Probe for Windows program (Donovan, 2012). The composition of the olivine grains from the cold end to the center of the graphite heater was carefully measured through linear traverse method. Considering the effect of the neighboring silicate liquid (glass) on compositional measurement and the possible reaction between the solid residual and the liquid phase during quenching as discussed by previous researches (e.g., Herzberg et al., 2000; Takahashi, 1986), analyses with totals > 101% or < 99% were discarded for measurement on individual olivine grain giving poor stoichiometry calculated on a calculated on a 4-oxygen basis. A shorter traverse was applied on the hot spot (center) of the graphite capsule, where there was larger amount of melt, to check the chemical equilibrating status between the liquid phase and its coexisting solid phases.

2 RESULTS 2.1 Partial Melting of KLB-1 Spinel Peridotite

(1) Modeling result: Bulk composition of KLB-1 peridotite reported by Davis et al. (2009) was adopted as starting material for pMELTS program to predict the phase relation of partial melting spinel peridotite at 1.5 GPa and temperature up to 1 700 ℃. By adding 200 ppm of H2O to the bulk composition of KLB-1, pMELTS program shows the effect of H2O on partial melting (Fig. 2): the existence of the 200 ppm H2O lowers the melting temperature of KLB-1 peridotite by 100 ℃; at 1 300 ℃, anhydrous KLB-1peridotite melts 0.98 vol.%, while the addition of 200 ppm H2O increases the extent of melting to 1.74 vol.%, thus doubles the degree of partial melting; by increasing the temperature to 1 400 ℃, these two values increase to 3.48 vol.% and 5.24 vol.%, respectively. However, this effect diminishes rapidly due to dilution as the extent of melting increases, for example, pMELTS program predicts that dry KLB-1 peridotite melts 14.74 vol.% at 1 450 ℃ and the addition of 200 ppm H2O increases the extent of melting only by 1%. Considering that the aim of this study is the time scale of partial melting rather than the solidus of spinel peridotite and 200 ppm H2O content is beyond the upper limit of the H2O content in KLB-1 peridotite (Hirschmann, 2010; Lesher et al., 2003), we take the pMELTS modeling result from dry KLB-1 peridotite to compare with our laboratory experimental data.

Figure 2. Partial melting of KLB-1 peridotite (bulk composition as reported by Davis et al., 2009) at 1.5 GPa by using pMELTS (Ghiorso et al., 2002). As indicated by the dash line, the existence of the 200 ppm H2O lowers the melting temperature of KLB-1 peridotite by 100 ℃; by increase temperature from 1 300 to1 400 ℃, anhydrous KLB-1 peridotite will be melted from 0.98 vol.% to 3.48 vol.%; while the addition of 200 ppm H2O would increase the extent of melting from 1.74 vol.% to 5.24 vol.%. But after 1 450 ℃, there is almost no difference between the dry system and the hydrous system.

(2) Experimental observation: The back scattered electron images (BSE) of run products from the 30 min heating experiments KLB-1_22, 23, 24 showed the temperature effect on the phase relation of partial melting (Fig. 3). At lower temperature 1 300 ℃ (KLB-1_22), the solid phases olivine, clinopyroxene (Cpx), orthopyroxene (Opx), and spinel evenly distributed inside the graphite capsule and the chemical composition of these minerals are comparable with the starting composition (Table 2), suggesting undetectable or incipient partial melt (Figs. 3A1 and 3A2). When heating temperature increased to 1 400 ℃ (KLB-1_24), a heterogeneous melting texture developed due to the thermal gradient inside the graphite heater: the solidus phases assembly Cpx+Opx+olivine+spinel only distributed at the cold end, and small liquid pools were observed at the hottest center of the graphite capsule (Figs. 3B1 and 3B2). The segregation of liquid phase was observed when the heating temperature is about 1 600 ℃ (KLB-1_23), in which case there is no spinel or clinopyroxene left in the sample and the silicate liquid migrated to the corner or the wall of the graphite sleeve and there are only a few orthopyroxene grains left at the cold end (Figs. 3C1 and 3C2). The composition of olivine grains from KLB-1_23 showed very little variation, indicating the homogenization of the partial melted sample.

Figure 3. SEM photographs of quenched samples from experiments KLB-1_22, KLB-1_24, and KLB-1_23. These samples were heated for 30 min at 1.5 GPa and different temperatures. By increasing temperature from 1 300 to 1 600 ℃, KLB-1 peridotite sample started to melt at the triple junction, and melt content increased with heating temperature, showing as dissolving of spinel, Cpx, and Opx to the liquid phase. At 1 600 ℃ (310 W), liquid phase migrated to the up left corner of the cell and there are only olivine grains were detected in that area, and orthopyroxene grains were found at the cold end of the capsule.
Table 2 Chemical composition of KLB-1 peridotite used in this study and sample quenched from partial melting experiment at 1.5 GPa and 1 300 ℃
2.2 Thermal Gradients inside D-DIA Cell Assembly

For experiment KLB-1_43 (1.5 GPa/~1 500 ℃/1 min), the phase assembly clinopyroxene+orthopyroxene+olivine+liquid showed in the back scattered electron (BSE) images (Fig. 4). And the temperature gradient can be eyed from the images: more clinopyroxene grain in the cold end and more liquid phase was found in the center of the cell and the hot spot close to the wall of the graphite sleeve. Although there is no solidus at the cold end as observed by the many other melting experiments (Herzberg and Zhang, 1996; Zhang and Herzberg, 1994; Takahashi et al., 1993; Takahashi, 1986), because the temperature (1 500 ℃) was much higher than the solidus temperature (1 300 ℃) of KLB-1 at 1.5 GPa (Takahashi, 1986), there is a trend of distribution of clinopyroxene grains along the temperature gradient, for example the dash line in Fig. 4.

Figure 4. SEM photograph of sample quenched from experiment KLB-1_43. This experiment was done at 1.5 GPa and 1 500 ℃ for only 1-minute heating. There is more clinopyroxene grains at the cold end and more silicate liquid merged at the grain boundaries at the hot end, indicating that temperature gradient developed during short time heating. The red line showed the position of the linear traverse measurements with microprobe and Mg# (=Mg/(Mg+Fe) mol.%) of olivine grains along the line. The magnified square showed the distribution of liquid phase among olivine and orthopyroxene grains; chemical composition changes along this short traverse was present in Fig. 7.
Figure 7. Chemical composition changes along short traverse from experiment KLB-1_43. The measurement on small area shows the distribution of major elements between silicate liquid phase and olivine, except Mn, which shows almost no changes along the travers line.

Linear traverse method was used to get the chemical composition changes along thermal gradient (Fig. 4). The Mg# (=Mg/(Mg+Fe) mol.%) of olivine grains along the linear traverse were calculated from their chemical compositions (Table 3). Because the existence of cracks due to decompression after heating, we have to make some stops in front of each crack and separate the long traverse to four parts. The average values of Mg# of olivine grains along each small traverse were plot on Fig. 4. And the Mg# of all olivine grains along these traverses increase from the cold end to the hot end, indicating the degree of partial melting, thus temperature increased along the traverse (Fig. 5).

Figure 5. (a) The Mg# of olivine grains along long traverse (Y stage coordinate); and (b) along short traverse (X stage coordinate). The Mg# of olivine increase from the cold end to the hot end inside the graphite heater, indicating the increase of degree of partial melting along thermal gradient developed during 1-minute heating.

The composition of the solid and liquid phases at 1.5 GPa and temperature from 1 300 to 1 600 ℃ were calculated with pMELTS program. The measured composition of the solid phases (olivine, clinopyroxene, orthopyroxene) and the coexisting liquid phase accumulated in the triple junctions agree well with those calculated from the pMELTS program, even for the 1-minute heating experiment KLB-1_43 (Table 3), although there is variation among the measured compositions of each solid phase due to different degree of partial melting. Considering the fact that the thermal dynamic parameters used in pMELTS came from equilibrated melting experiments, this agreement indicates that local chemical equilibrium was achieved during the 1-minute heating. pMELTS program was then accepted to determine the temperature along the thermal gradient inside the cell. A temperature difference about 180 ℃ between the cold end and the center of the graphite sleeve were found for the 1-minute heating experiment (KLB-1_25), and 70 ℃ for the longer heating experiment (KLB-1_24, 30 min) (Fig. 6), which is consistent with the recently experimental result (Raterron et al., 2013).

Figure 6. The Mg# of olivine grains changed along temperature gradient inside of graphite capsule from experiments heated at 1 400 ℃. The largest Mg# from these three experiments agree well with each other, showing consistent with output of pMELTS program. Temperature difference between the cold end and the hot end of sample from 1-minute heating could be 180 ℃, which drop to 70 ℃ for experiment heated for 30 min because compositional homogeneity caused through Soret diffusion.
3 DISCUSSION 3.1 Effect of Thermal Gradient on Partial Melting of KLB-1

The chemical composition variation along the short traverse at the hot part of the sample from experiment KLB-1_43 (Fig. 3) was presented in Fig. 7 and Table S1, showing the distribution of major elements between olivine grains and the coexisting melt. Considering the fact that the heating time of this experiment is short (1 min) and the relic olivine grains are smaller than 5 µm, the shoulders of the peaks in Fig. 7 may represent the reaction rim between the liquid phase and the relic olivine grains or the effect of the neighboring phases on the compositional measurement. The temperature estimated from the Mg# of olivine grains (~90.5) is about 1 500 ℃, which agrees well with the heating temperature calibrated from thermocouple. However, the chemical composition of the liquid phase is rich in SiO2 and depleted in MgO than those calculated from pMELTS program at ~1 500 ℃, on the other hand, the contents of Al2O3, CaO, TiO2 and MnO of liquid phase are comparable with output of pMELTS program. In addition, the scatter of FeO content in olivine indicates that the experiment has not approached chemical equilibrium, although the liquid pools separated by olivine grains show internal consistence on composition because chemical diffusion in liquid is much faster than that in solid phases. Therefore, the homogenized liquid composition can only represent the composition of melt at the hot end of the graphite heater but not the equilibrated liquid composition of partial melting of KLB-1 peridotite at 1 500 ℃.

Table S1 Chemical composition of olivine grains from the quenched sample KLB-1_43. The X and Y values are the stage coordinates showing the position of the grains along the traverse from the cold end to the hot end as showed in Fig. 4.

The microprobe traverse at the hot end of the quenched sample presents the diffusional exchange profile across the liquid pools and olivine grains (Fig. 7). With no qualitative surprises, we observed that Na, Ti, Ca and Al are strongly incompatible in olivine; Cr and Mn are moderately incompatible. However, we do observe difference between the partial melting experiments and the calculated result from pMELTS program. The chemical analysis of olivine grains through the long traverse lines show that olivine can accommodate small amount of Al2O3 in addition to CaO, both of which increase with Mg# of olivine grains, thus heating temperature (Fig. 8a). And the 0.1 wt.%–0.5 wt.% of Al2O3 could be a compensation for the depleted in CaO in the KLB-1 sample compared with pMELTS program during partial melting (Fig. 8b). On the other hand, pMELTS program reports no aluminum content in olivine at 1.5 GPa. The possible reason for this discrepancy is that the chemical composition of the olivine grain boundary was modified considerably by liquid phase, which is supported by the fact that both Al2O3 and CaO contents increase from the center to the edge of olivine grains (Table S2). However, it is hard to determine whether this modification happened through diffusion during heating or reaction during quenching. With the short run duration in the present study, the incompatible element Al dissolved into liquid phase during melting of spinel and clinopyroxene and accumulated at grain boundaries or triple junctions, as shown by study with X-ray synchrotron microtomography (Zhu et al., 2011). Al3+ and Ca2+ cations then further diffuse into olivine grains from the liquid phase. This process is consistent with the observation that partitioning of Al2O3 between olivine and silicate liquid increases with temperature (Agee and Walker, 1990). Therefore, the 0.1 wt.%–1 wt.% of Al2O3 in olivine grains may represent the partitioning of Al2O3 into olivine grain from coexisting liquid phase at high temperature. By increasing heating temperature or heating duration, melt fraction increases, liquid phase starts to migrate, and the chemical composition of olivine will be modified continually by the liquid phase through Soret diffusion. Thus, olivine grains saturated with liquid phase show different composition on incompatible elements comparing with that dynamically change composition with coexisting liquid phase through Soret diffusion, showing rich in Al2O3 and CaO content.

Figure 8. (a) Chemical composition changes of olivine grains along traverse from experiment KLB-1_43. The CaO content of olivine is lower than output of pMELT program. From the cold end to the hot end, Mg# of olivine grains increase, indicating the larger degree of partial melting at higher temperature. And with increasing temperature, we observed slightly incensement on CaO content of olivine. (b) The CaO (wt.%) and Al2O3 (wt.%) contents of olivine increase with temperature, suggesting that olivine saturated with liquid phase can take some about 0.5 wt.% Al2O3.
Table S2 Chemical composition along the short traverse in the quenched sample KLB-1_43 (Figure 4B). The X and Y values are the stage coordinates showing the position of the grains along the traverse.
3.2 Time Scale of Partial Melting at 1.5 GPa

It is difficult to determine the composition of the quenched liquid phase if the heating temperature is just above subsolidus and the small amount of melt stayed at grain interfaces. Here we show that the partial melting KLB-1 sample quenched after 1-minute heating at high temperature developed large thermal gradient inside the heater. The Mg# of olivine grains increased gradually with temperature along the thermal gradient, indicating that chemical composition of melt and the solid residuals change simultaneously with temperature. Compared with the staring material, the slightly larger Mg# of olivine grains at the cold end indicates that partial melting started spontaneously when temperature is higher than subsolidus. In another word, we can get information whether the small degree (< 5 vol.%) of partial melting has happened by checking the Mg# of olivine grains.

The Mg# of olivine grains from the cold end to the hot end of quenched samples from experiment KLB-1_24, KLB-1_25, and KLB-1_26 were plotted in the model and the temperature was determined simultaneously (Fig. 6). The similarities of the largest Mg# from all the three experiments indicate the same degree of partial melting at the hottest place inside the heaters, suggesting that (1) non-diffusion related melting occurred at time scale as short as 1-minute; (2) local chemical equilibrium achieved if the surrounding material can supply the eutectic composition of the melt at the heating temperature and pressure condition, although longer heating time is necessary for chemical equilibrium between the liquid phase and the residual solid phases through diffusion. As soon as the partial melting was initiated, tangible changes on the volume of the sample can be detected by X-ray imaging because the different thermal expansion of the solid and the liquid phase (Li and Weidner, 2014). The short time and non-equilibrium related partial melting will impact the effective thermoelastic properties of the system, for example 2% melt is sufficient to double the thermal expansion and cut the bulk modulus in half (Weidner and Li, 2015).


(1) The 1-minute partial melting experiment on KLB-1 spinel peridotite at 1.5 GPa and 1 500 ℃ showed a well-developed temperature gradient within the cell assembly used in many D-DIA experiments. Local chemical equilibrium achieved within 1 min after partial melting started, but longer heating time is necessary to homogenize the sample and decease the thermal gradient.

(2) The Mg# of olivine grains at the hot place inside the heater from the 1-minute heating experiment show consistent result with those from longer heating time (30 min), indicating that partial melting happened in seconds and the chemical distribution of melting in a local region can be accurately predicted by a thermodynamic equilibrium models even though the heating duration is as short as 60 s.


KLB-1 samples used in this study were generously denoted by Prof. Claude Herzberg from Rutgers University. We thank Christopher A. Vidito from Rutgers University for his support with all the electron microprobe measurements and Jim Quinn from Stony Brook University for the SEM measurement. The authors acknowledge support by the National Natural Science Foundation of China (No. 41773052), and the National Science Foundation of USA (Nos. EAR 1141895, EAR 1045629, and EAR 0968823). The final publication is available at Springer via

Electronic Supplementary Materials: Supplementary materials (Table S1S2) are available in the online version of this article at

Agee C. B., Walker D., 1990. Aluminum Partitioning between Olivine and Ultrabasic Silicate Liquid to 6 GPa. Contributions to Mineralogy and Petrology, 105(3): 243-254. DOI:10.1007/bf00306537
Anderson D. L., Sammis C., 1970. Partial Melting in the Upper Mantle. Physics of the Earth and Planetary Interiors, 3: 41-50. DOI:10.1016/0031-9201(70)90042-7
Asimow P. D., Ghiorso M. S., 1998. Algorithmic Modifications Extending MELTS to Calculate Subsolidus Phase Relations. American Mineralogist, 83(9/10): 1127-1132. DOI:10.2138/am-1998-9-1022
Dasgupta R., Hirschmann M. M., Smith N. D., 2007. Partial Melting Experiments of Peridotite+CO2 at 3 GPa and Genesis of Alkalic Ocean Island Basalts. Journal of Petrology, 48(11): 2093-2124. DOI:10.1093/petrology/egm053
Davis F. A., Hirschmann M. M., Humayun M., 2011. The Composition of the Incipient Partial Melt of Garnet Peridotite at 3 GPa and the Origin of OIB. Earth and Planetary Science Letters, 308(3/4): 380-390. DOI:10.1016/j.epsl.2011.06.008
Davis F. A., Tangeman J. A., Tenner T. J., et al., 2009. The Composition of KLB-1 Peridotite. American Mineralogist, 94(1): 176-180. DOI:10.2138/am.2009.2984
Donovan, J. J., 2012. Probe for EPMA: Acquisition, Automation and Analysis. Enterprise Edition Probe Software Inc., Eugene
Du W., Li L., Weidner D. J., 2014. Experimental Observation on Grain Boundaries Affected by Partial Melting and Garnet Forming Phase Transition in KLB-1 Peridotite. Physics of the Earth and Planetary Interiors, 228: 287-293. DOI:10.1016/j.pepi.2013.11.011
Ghiorso M. S., Hirschmann M. M., Reiners P. W., et al., 2002. The pMELTS: A Revision of MELTS for Improved Calculation of Phase Relations and Major Element Partitioning Related to Partial Melting of the Mantle to 3 GPa. Geochemistry, Geophysics, Geosystems, 3(5): 1-35. DOI:10.1029/2001gc000217
Ghiorso M. S., Sack R. O., 1995. Chemical Mass Transfer in Magmatic Processes Ⅳ. A Revised and Internally Consistent Thermodynamic Model for the Interpolation and Extrapolation of Liquid-Solid Equilibria in Magmatic Systems at Elevated Temperatures and Pressures. Contributions to Mineralogy and Petrology, 119(2/3): 197-212. DOI:10.1007/s004100050036
Harmon N., Forsyth D. W., Weeraratne D. S., 2009. Thickening of Young Pacific Lithosphere from High-Resolution Rayleigh Wave Tomography: A Test of the Conductive Cooling Model. Earth and Planetary Science Letters, 278(1/2): 96-106. DOI:10.1016/j.epsl.2008.11.025
Herzberg C., Gasparik T., Sawamoto H., 1990. Origin of Mantle Peridotite: Constraints from Melting Experiments to 16.5 GPa. Journal of Geophysical Research, 95(B10): 15779-15803. DOI:10.1029/jb095ib10p15779
Herzberg C., Raterron P., Zhang J. Z., 2000. New Experimental Observations on the Anhydrous Solidus for Peridotite KLB-1. Geochemistry, Geophysics, Geosystems, 1(11): 1-15. DOI:10.1029/2000gc000089
Herzberg C., Zhang J. Z., 1996. Melting Experiments on Anhydrous Peridotite KLB-1: Compositions of Magmas in the Upper Mantle and Transition Zone. Journal of Geophysical Research: Solid Earth, 101(B4): 8271-8295. DOI:10.1029/96jb00170
Hirose K., 1997. Melting Experiments on Lherzolite KLB-1 under Hydrous Conditions and Generation of High-Magnesian Andesitic Melts. Geology, 25(1): 42-44. DOI:10.1130/0091-7613(1997)025<0042:meolku>;2
Hirose K., Fei Y. W., 2002. Subsolidus and Melting Phase Relations of Basaltic Composition in the Uppermost Lower Mantle. Geochimica et Cosmochimica Acta, 66(12): 2099-2108. DOI:10.1016/s0016-7037(02)00847-5
Hirose K., Kushiro I., 1993. Partial Melting of Dry Peridotites at High Pressures: Determination of Compositions of Melts Segregated from Peridotite Using Aggregates of Diamond. Earth and Planetary Science Letters, 114(4): 477-489. DOI:10.1016/0012-821X(93)90077-M
Hirschmann M. M., 2000. Mantle Solidus: Experimental Constraints and the Effects of Peridotite Composition. Geochemistry, Geophysics, Geosystems, 1(10): 1042. DOI:10.1029/2000gc000070
Hirschmann M. M., 2010. Partial Melt in the Oceanic Low Velocity Zone. Physics of the Earth and Planetary Interiors, 179(1/2): 60-71. DOI:10.1016/j.pepi.2009.12.003
Hirschmann M. M., Ghiorso M. S., Wasylenki L. E., et al., 1998. Calculation of Peridotite Partial Melting from Thermodynamic Models of Minerals and Melts. I. Review of Methods and Comparison with Experiments. Journal of Petrology, 39(6): 1091-1115. DOI:10.1093/petroj/39.6.1091
Ito K., Kennedy G. C., 1967. Melting and Phase Relations in a Natural Peridotite to 40 Kilobars. American Journal of Science, 265(6): 519-538. DOI:10.2475/ajs.265.6.519
Kato T., Ringwood A. E., Irifune T., 1988. Constraints on Element Partition Coefficients between MgSiO3 Perovskite and Liquid Determined by Direct Measurements. Earth and Planetary Science Letters, 90(1): 65-68. DOI:10.1016/0012-821X(88)90111-2
Lesher C. E., Pickering-Witter J., Baxter G., et al., 2003. Melting of Garnet Peridotite: Effects of Capsules and Thermocouples, and Implications for the High-Pressure Mantle Solidus. American Mineralogist, 88(8/9): 1181-1189. DOI:10.2138/am-2003-8-901
Lesher C. E., Walker D., 1988. Cumulate Maturation and Melt Migration in a Temperature Gradient. Journal of Geophysical Research: Solid Earth, 93(B9): 10295-10311. DOI:10.1029/jb093ib09p10295
Li L., 2009. Studies of Mineral Properties at Mantle Condition Using Deformation Multi-Anvil Apparatus. Progress in Natural Science, 19(11): 1467-1475. DOI:10.1016/j.pnsc.2009.06.001
Li L., Weidner D. J., 2013. Effect of Dynamic Melting on Acoustic Velocities in a Partially Molten Peridotite. Physics of the Earth and Planetary Interiors, 222: 1-7. DOI:10.1016/j.pepi.2013.06.009
Li L., Weidner D. J., 2014. Detection of Melting by X-Ray Imaging at High Pressure. Review of Scientific Instruments, 85(6): 065104. DOI:10.13039/100000001
Munro R. G., 1997. Evaluated Material Properties for a Sintered Alpha-Alumina. Journal of the American Ceramic Society, 80(8): 1919-1928. DOI:10.1111/j.1151-2916.1997.tb03074.x
Ohtani E., 1979. Melting Relation of Fe2SiO4 up to about 200 Kbar. Journal of Physics of the Earth, 27(3): 189-208. DOI:10.4294/jpe1952.27.189
Raterron P., Merkel S., Holyoke C. W. III, 2013. Axial Temperature Gradient and Stress Measurements in the Deformation-DIA Cell Using Alumina Pistons. Review of Scientific Instruments, 84(4): 043906. DOI:10.13039/100000015
Smith P. M., Asimow P. D., 2005. Adiabat_1ph: A New Public Front-End to the MELTS, PMELTS, and PHMELTS Models. Geochemistry, Geophysics, Geosystems, 6(2): 1-8. DOI:10.1029/2004gc000816
Takahashi E., 1986. Melting of a Dry Peridotite KLB-1 up to 14 GPa: Implications on the Origin of Peridotitic Upper Mantle. Journal of Geophysical Research, 91(B9): 9367-9382. DOI:10.1029/jb091ib09p09367
Takahashi E., Shimazaki T., Tsuzaki Y., et al., 1993. Melting Study of a Peridotite KLB-1 to 6.5 GPa, and the Origin of Basaltic Magmas. Mathematical, Physical and Engineering Sciences, 342(1663): 105-120. DOI:10.1098/rsta.1993.0008
Walker D., DeLong S. E., 1982. Soret Separation of Mid-Ocean Ridge Basalt Magma. Contributions to Mineralogy and Petrology, 79(3): 231-240. DOI:10.1007/bf00371514
Walter M., 1998. Melting of Garnet Peridotite and the Origin of Komatiite and Depleted Lithosphere. Journal of Petrology, 39(1): 29-60. DOI:10.1093/petrology/39.1.29
Weidner D. J., Li L., 2015. Kinetics of Melting in Peridotite from Volume Strain Measurements. Physics of the Earth and Planetary Interiors, 246: 25-30. DOI:10.13039/100000015
Yoshino T., Takei Y., Wark D. A., et al., 2005. Grain Boundary Wetness of Texturally Equilibrated Rocks, with Implications for Seismic Properties of the Upper Mantle. Journal of Geophysical Research, 110(B8): 1-16. DOI:10.1029/2004jb003544
Zhang J. Z., Herzberg C., 1994. Melting Experiments on Anhydrous Peridotite KLB-1 from 5.0 to 22.5 GPa. Journal of Geophysical Research: Solid Earth, 99(B9): 17729-17742. DOI:10.1029/94jb01406
Zhu W., Gaetani G. A., Fusseis F., et al., 2011. Microtomography of Partially Molten Rocks: Three-Dimensional Melt Distribution in Mantle Peridotite. Science, 332(6025): 88-91. DOI:10.13039/100006151