A. HTXRD/GC

Details regarding the system setup have been documented elsewhere (Rodriguez et al., Reference Rodriguez, Coker, Griego, Mowry, Pimentel and Anderson2016). A short description concerning the HTXRD/GC measurements follows. Cathode powders were ground to a fine powder under methanol and coated onto alumina substrates for loading onto the Pt-heating-strip within the HTXRD furnace chamber on a Scintag PAD X₁ diffractometer (Cu K α radiation, solid-state Ge detector). We employed an Agilent CP-4900 micro-GC (or μ-GC) and Ametek CG1000 oxygen meter to monitor the gas flow out of the furnace chamber. The μ-GC was configured to sample this flow in ~2-min increments via a PoraPLOT U (PPU) column (column temperature = 160 °C, helium carrier gas). This timing worked well to monitor the gas conditions during the collection of HTXRD scans which occurred at increments of ~6 min per scan at a given hold temperature. XRD scan parameters were as follows: angular range = 26–41° 2θ, step-size = 0.04° 2θ, count time = 1 s, heating rate between scans = 50 °C min⁻¹, temperature increment between scans = 10 °C. Using these scan parameters for HTXRD data collection, the effective heating rate mimics an ~2 °C min⁻¹ heating rate from the start of the experiment to the maximum temperature of 550 °C. Flowing air was employed for the oxidizing environment, and the inert environment employed flowing ultra-high-purity helium gas.

B. STA/MS

A Netzsch STA 449 F3 Jupiter TGA/DSC system configured with a Hidden Analytical HPR-20 Mass Spectrometer was employed for thermal analysis. Details regarding the system have been documented previously (Rodriguez et al., Reference Rodriguez, Coker, Griego, Mowry, Pimentel and Anderson2016) with a brief description given here. Cathode material was loaded into alumina crucibles for analysis. The heating rate employed for the STA/MS was 5 °C min⁻¹ with a maximum temperature of 550 °C. Comparison of results from STA/MS with those of HTXRD/GC shed light on the breakdown of FeS₂ during heat treatment in an oxidizing environment (flowing air). Flowing argon gas was employed for the inert gas environment in the STA/MS experiment and compared with the inert gas (helium) HTXRD/GC experiment.

A. FeS₂ analysis under inert atmosphere

It is worth documenting the behavior of an FeS₂ cathode processed under inert conditions so that a comparison can be made regarding the proper and expected performance of the cathode, which is not exposed to oxygen during typical operating conditions. Figure 1 shows a waterfall plot of the HTXRD results for the FeS₂ cathode material heated under flowing helium gas ($p_{{\rm O}_2}$ between 2 and 10 ppm). The as-received cathode material is confirmed to be dominated by three distinct phases: pyrite-structured FeS₂, PDF 04-003-1989 with salt phases KCl, PDF 00-041-8317 and LiCl, PDF 00-004-0664. The FeS₂ remains present over the entire temperature range without breakdown as indicated by the continuous presence of the FeS₂ peaks. This is a desirable outcome because any thermal battery made of such a cathode will likely perform properly if the FeS₂ phase persists to sufficiently high temperatures without oxidation. Another important observation is seen in Figure 1. The KCl and LiCl peaks persist well above 300 °C, and the melting behavior of the KCl and LiCl is consistent with expected KCl–LiCl phase equilibria behavior (see Yasuda et al., Reference Yasuda, Nohira, Ogata and Ito2005). Note: The presence of the Al₂O₃ peak in the HTXRD data is from the underlying substrate.

Figure 1.

HTXRD results for the FeS₂ cathode material under flowing helium gas.

Figure 2 shows the μ-GC data collected during the HTXRD experiment under inert conditions. This graph shows no evidence of SO₂ gas evolution at any temperature up to 550 °C. However it does show the evolution of physisorbed water upon initial heating. This gives additional supporting evidence of a stable FeS₂ phase with temperature under these inert gas conditions.

Figure 2.

μ-GC results for the FeS₂ cathode material under flowing helium gas as collected during the HTXRD run (Figure 1).

Figure 3 shows the STA/MS traces for the inert gas run showing a very important observation of an endothermic peak at ~350 °C. This observation is consistent with melting of the LiCl salt phase. Once melted, the presence of LiCl liquid will cause gradual melting of the KCl until the liquidus temperature for the LiCl-KCl binary (see Yasuda et al., Reference Yasuda, Nohira, Ogata and Ito2005) observed in the HTXRD data at ~460 °C (Figure 1). It is also notable that there are changes in both the TGA and DSC curves between room temperature and ~120 °C for the FeS₂ sample heated under inert gas. There is a clear weight loss which occurs up to 100 °C, as well as an endothermic peak at ~100 °C. These thermal events can be traced to the loss of physisorbed water (see μ-GC results in Figure 2) and appear to be associated with the pick-up of water by LiCl as noted by the absence of LiCl peaks in the room temperature scan in the HTXRD data (Figure 1) and their reappearance upon further heating above 60 °C. These changes may look significant, but actually occur in a limited mass range (<2 wt% loss) and do not cause an observed phase change of the FeS₂. Another important point to make from these data is the absence of any observed SO₂ species from the MS output during the STA/MS run. The plot of m/z = 64 shows essentially a background level signal over the entire heating range from room temperature to 550 °C. These results are consistent with the μ-GC results and confirm the stability of the FeS₂ phase over the entire temperature range. Based on these results, a simplified reaction sequence is given in Figure 4 that illustrates the cathode thermal behavior during the desired inert gas conditions.

Figure 3.

STA/MS results for the FeS₂ cathode material processed under an inert (Ar gas) atmosphere.

Figure 4.

FeS₂ inert atmosphere reaction sequence.

B. FeS₂ analysis under an air atmosphere

Heating the FeS₂ powder within the HTXRD reaction chamber in flowing air shows a much more dynamic behavior as shown in Figure 5. There are multiple reactions occurring over the 25–550 °C temperature range as the FeS₂ phase breaks down. At low temperature (between 25 and ~220 °C), the cathode shows the expected FeS₂, KCl, and LiCl phases as observed in the inert experiment (Figure 1). These initial phases decompose at ~230 °C with the formation of short-lived intermediate reaction products: Li₂(SO₄), PDF entry 00-020-0640; KFe₂S₃, PDF 04-014-2712; KFe(SO₄)₂, PDF 04-014-2712; and K₃Fe(SO₄)₃, PDF 00-030-0943. Note, the progression of the intermediate phases appears to depend on the initial formation of sulfates followed by incremental reaction of these sulfate phases with potassium. The initial reaction between ~230 and ~300 °C appears to show LiCl, KCl, and FeS₂ reacting with oxygen to form Li₂(SO₄) and KFe₂S₃. This is followed by the observation of KFe(SO₄)₂ beginning at ~260 °C. The KFe(SO₄)₂ phase appears to tie up all the iron until ~350 °C when it begins to decompose to form K₃Fe(SO₄)₃ and β-Fe₂O₃, PDF 01-083-8470. The K₃Fe(SO₄)₃ phase begins to break down at ~480 °C to form additional β-Fe₂O₃ along with K₂(SO₄), PDF 04-006-8317. Ultimately, the final phases persisting through 550 °C are the K₂(SO₄) phase and β-Fe₂O₃, an interesting form of Fe(III)-oxide similar to cubic ZrO₂ (Danno et al., Reference Danno, Nakatsuka, Kusano, Asaoka, Nakanishi, Fujii, Ikeda and Takada2013). There is no distinct phase of lithium in the final decomposition products. This may be because of the Li atoms remaining in any high-temperature melt that has formed or possibly substituting within the β-Fe₂O₃. This cubic crystal structure (space group Ia3) has demonstrated the ability to readily accept both substitutional and interstitial defects (Tuček et al., Reference Tuček, Machala, Ono, Namai, Yoshikiyo, Imoto, Tokoro, Ohkishi and Zboril2015). Evaluation of the cathodes after sample cool down showed that the β-Fe₂O₃ phase persisted and was stable at room temperature.

Figure 5.

HTXRD results for FeS₂ cathode under a flowing air atmosphere.

Concurrent GC analysis was performed during this experiment and is shown in Figure 6. This figure clearly shows two major gas releases, both occurring at ~70 s (retention time). This signal was confirmed to be SO₂ gas as determined by calibration standards run separately on the PPU column. The SO₂ gas appears to come off at two distinct temperatures (~250 and ~300 °C), waning at 320 °C, with the majority gas loss occurring during the first output stage. A limited signal was also observed around 67 s (retention time) and was determined to be physisorbed water from the initial powder, identical to the previous inert atmosphere testing series (see Figure 2).

Figure 6.

μ-GC results for the FeS₂ cathode heated in flowing air showing the formation of SO₂ gas between 250 and 310 °C. A trace of physisorbed water was also detected below 100 °C.

The thermal analysis experiment revealed a great deal of information regarding the changes detected in the HTXRD/GC experiment. Figure 7 shows the STA/MS results when FeS₂ cathode material was heated in flowing air. What is immediately obvious from this graph is that a major reaction initiates at ~250 °C with a maximum heat release shown in the DSC trace at ~280 °C. Concurrently, the TGA trace shows a weight increase above ~260 °C. The HTXRD data indicate that the FeS₂ phase has begun to break down to form Li₂SO₄ and KFe₂S₃ at approximately these temperatures. This exothermic reaction is in stark contrast to the expected behavior for an oxygen-free environment (namely melting of the LiCl to form a liquid electrolyte) and the oxidation reaction initiates at a significantly lower temperature when compared with the observed LiCl melting as shown in Figure 1. This first oxidizing reaction sets the stage for other reactions and lends to the total decomposition of the initial phases. Hence, the initial formation of Li₂(SO₄) and KFe₂S₃ may serve as a catalyst for subsequent reactions.

Figure 7.

STA/MS results for the FeS₂ cathode processed in an air atmosphere.

The STA/MS data in Figure 7 also show other activities at higher temperatures. The mass gain, predominantly because of sulfate formation, achieves a maximum of ~110% of the initial sample weight at ~425 °C. This weight gain apparently contradicts the expected mass loss owing to off-gassing of SO₂ as observed in the μ-GC data in Figure 6. SO₂ loss because of FeS₂ decomposition with simultaneous weight gain as observed in TGA can be understood in terms of a very fast-paced oxidation reaction behavior where mass loss of SO₂ is more than compensated by the weight gain because of oxidation and sulfate formation, i.e. Li₂(SO₄), KFe(SO₄)₂, and K₃Fe(SO₄)₃. Additional support for off-gassing of SO₂ observed in HTXRD/GC analysis was observed from the MS data collected during the STA/MS measurement. The MS results in Figure 7 (red) show a broad doublet peak for a gaseous species with a mass/charge ratio of 64 amu (SO₂). Just below 280 °C, the first peak of the doublet begins to grow. At the same time, breakdown of FeS₂ is observed in the DSC data. The first peak in the SO₂ doublet looks to achieve a maximum at ~280 °C, which coincides with the largest exothermic release in the DSC trace. Then the MS data show the SO₂ signal dropping a bit in intensity near 300 °C, followed by a second peak at ~310–320 °C. This second SO₂ release corresponds to another exotherm in the DSC trace that has been identified as the decomposition/oxidation of the KFe₂S₃ phase. Hence, the two releases of SO₂ gas are because of separate sulfide decompositions. When comparing the MS data in Figure 7 with that of the μ-GC results in Figure 6, the results are self-consistent in that the μ-GC confirms two distinct SO₂ release maxima (~250 and ~300 °C). Note that there is a temperature difference observed between HTXRD/GC and STA/MS data in terms of these gas releases. This can be attributed to experimental heating rate and sampling frequency. On average, the heating rate was ~2 °C min⁻¹ for the HTXRD/GC experiments whereas the STA/MS used a 5 °C min⁻¹ heating rate. Faster heating rates tend to push the observed reaction temperatures higher. It should be noted that there is an observed mass loss above 500 °C. This can be attributed to vaporization of Cl from any melt present at these high temperatures.

To summarize, the reaction sequence can be shown using the schematic diagram given in Figure 8. This figure illustrates the reactions that occur when FeS₂ is heated under an air atmosphere along with the presence of potassium and lithium salts. The clear demarcation of this reaction sequence occurs with the conversion of FeS₂ to KFe₂S₃ and Li₂(SO₄) via the reaction with LiCl and KCl. It appears that the formation of the short-lived Li₂(SO₄) and KFe₂S₃ phases, along with the corresponding decomposition of the KCl and LiCl phases, serve as the critical first-steps in the decomposition process. From the onset of this reaction, a sequence of increasing K-based sulfates form: KFe(SO₄)₂, then K₃Fe(SO₄)₃, and finally K₂(SO₄). Because of the consumption of the FeS₂ phase, the evolution of SO₂ gas is detected at the onset of the Li₂(SO₄) and KFe₂S₃ phase formations at ~250 °C. With the presence of the initial reaction product KFe₂S₃, a further oxidation reaction occurs forming the KFe(SO₄)₂ phase, again with significant loss of SO₂ gas between ~300 and 320 °C. Above this temperature, no additional SO₂ release is observed as all the sulfide phases have been consumed. Oxidation of the sample continues as the KFe(SO₄)₂ breaks down to form β-Fe₂O₃ and K₃Fe(SO₄)₃. Further oxidation of the K₃Fe(SO₄)₃ yields β-Fe₂O₃ and K₂(SO₄), both of which persist after the samples were cooled to room temperature. Both these reactions correspond to weight gain for the sample but without SO₂ release as the sulfate is stable at these temperatures.

Figure 8.

The reaction sequence for the breakdown of the FeS₂ cathode processed in air.