Titanium-based potassium-ion battery positive electrode with extraordinarily high redox potential

The rapid progress in mass-market applications of metal-ion batteries intensifies the development of economically feasible electrode materials based on earth-abundant elements. Here, we report on a record-breaking titanium-based positive electrode material, KTiPO4F, exhibiting a superior electrode potential of 3.6 V in a potassium-ion cell, which is extraordinarily high for titanium redox transitions. We hypothesize that such an unexpectedly major boost of the electrode potential benefits from the synergy of the cumulative inductive effect of two anions and charge/vacancy ordering. Carbon-coated electrode materials display no capacity fading when cycled at 5C rate for 100 cycles, which coupled with extremely low energy barriers for potassium-ion migration of 0.2 eV anticipates high-power applications. Our contribution shows that the titanium redox activity traditionally considered as “reducing” can be upshifted to near-4V electrode potentials thus providing a playground to design sustainable and cost-effective titanium-containing positive electrode materials with promising electrochemical characteristics.

be efficiently modulated through materials design to ensure a favorable battery performance, which is helpful to envision new protocols for materials innovation for KIBs. Despite of the large amount of characterizations described in the manuscript, the manuscript shows shortcomings in provide convincing evidence to support the key conclusions claimed by the authors. Therefore I'm not able to suggest an acceptance of this work for its publication in Nat Commun. in consideration of the high standard enforced by the journal. Below are some detailed comments I have during reading the manuscript.
1. Electrochemical characterization is needed to confirm the activity of Ti3+/Ti4+ redox. For example, EELS and EPR data at different charge states, particularly charged to 4.2 V or discharged to 2 V, should be provided to validate the electrochemical transition of Ti3+/ Ti4+ redox. The Ti contribution to the high voltage character seems to be the key argument of this work. Unfortunately, no data related to the charge compensation mechanism is provided, which makes the claim highly suspicious. 2. In fluorophosphates, fluorine: oxygen ratio seems to play a critical role in determining the electrochemical properties. However, the authors failed to provide experimental evidence and necessary discussion about the key factor. Hence, more systematic research may be required for this work, including the influence of fluorine on the operating voltage, galvanostatic curves and reversible capacity. 3. The author considered the charge/discharge during 2.0-3.4V as solid-solution-like process, despite of clearly two oxidation peaks at 2.81 and 3.21 V for charge and two reduction peaks at 2.58 and 2.99 V for discharge, which demonstrate multiple phase transitions during the reaction. A close look in the operando XRD data did show the emergence of tiny peaks during the disappearance of (212) peak of the pristine sample. For example, some tiny ones at 29.7° can be observed when charged to 3.0V. A detailed discussion is needed to facilitate the understanding of structural evolution during the electrochemical reaction. 4. For the rate capability of the tested sample as shown in Figure 4c, the authors used different voltage windows, is there a special reason for such an operation? How about the Coulombic efficiency of the sample? The authors should compare at least the first 2 charge-discharge curves at 0.1 C of the samples, which are critical to understand the structural stability and transition of the electrode materials during cycling. Figure R1. EELS spectra of recovered electrode materials. Ti2O3 and TiO2 EELS spectra are given for reference.
The position of energy offsets and maxima of the Ti-L2-3 edges depend on the titanium valence state in KTiPO4F. Moving from the initial electrode to one charged to 4.2 V a gradual shift towards higher energy loss can be observed which is related to the increasing Ti oxidation state. At 2.0 V the maxima on the EELS spectrum return to the initial positions that are characteristic for Ti 3+ . Ti2O3 and TiO2 spectra are given for reference. Moreover, a closer look at the 3.4 V (half-charged) spectrum unveils a superposition of Ti 3+ and Ti 4+ signals can be seen.
Additionally, the Ti oxidation state was checked with ex situ EPR spectroscopy. The EPR spectrum of the electrode charged to 4.2V contains an anisotropic signal (Fig. R2, A) manifested by residual Ti 3+ ions (according to the Rietveld refinement the chemical composition of the charged sample is K0.18(2)TiPO4F, consequently it might contain up to 20% of Ti 3+ ions). This signal can be simulated using the axial g-tensor with components gxx = gyy= g⟂ = 1.916 and gzz = g|| = 1.833 ( Figure R2, A). The fact that g⟂ > g|| evidences that Ti 3+ are located in distorted octahedral sites, which is in line with the structural data. Worth noting, Ti 4+ ions have a d 0 configuration and hence could not be detected by the EPR spectroscopy. The EPR spectrum of the electrode after a charge(4.2V)-discharge(2.0V) cycle shows a broad signal due to paramagnetic Ti 3+ ions (g ≈ 1.92) -quite similar to that observed in the pristine electrode material (see Fig. R2 Both Figure R1 and R2 are included into the manuscript as a part of supporting information.
The following text was added to the manuscript: "The valence evolution during the extraction/insertion of K + within the crystal structure of KTiPO4F was studied by EELS for recovered electrodes charged to 3.4 V, 4.2 V and discharged to 2.0V after full charge to 4.2 V vs. K + /K. The spectra are presented in Figure S5.
The position of energy offsets and maxima of the Ti-L2-3 edges depend on the titanium valence state in KTiPO4F. Moving from the initial electrode to one charged to 4.2 a gradual shift towards higher energy loss can be observed which is related to the increasing Ti oxidation state. At 2.0 V the maxima on the EELS spectrum return to the initial positions that are characteristic for Ti 3+ . Moreover, a closer look at the 3.4 V (half-charged) spectrum unveils a superposition of Ti 3+ and Ti 4+ signals ( Figure S5).
Additionally, the Ti oxidation state was checked with ex-situ EPR spectroscopy. The EPR spectrum of the electrode sample charged to 4.2V contains an anisotropic signal (Fig. S5, A) manifested by residual Ti 3+ ions (according to the Rietveld refinement the chemical composition of the charged sample is K0.18(2)TiPO4F, consequently it might contain up to 20% of Ti 3+ ions). This signal can be simulated using the axial g-tensor with components gxx = gyy= g⟂ = 1.916 and gzz = g|| = 1.833 ( Figure   S5, A). The fact that g⟂ > g|| evidences that Ti 3+ are located in distorted octahedral sites, which is in line with the structural data. Worth noting, Ti 4+ ions have a d 0 configuration and hence could not be detected by the EPR spectroscopy. The EPR spectrum of the electrode sample after a full charge(4.2V)-discharge(2.0V) cycle shows a broad signal due to paramagnetic Ti 3+ ions (g ≈ 1.92)quite similar to that observed in the pristine electrode material (see Figure S5, B). Thus, results from the ex situ EPR measurements also provide evidence in favor of a reversible Ti 3+ /Ti 4+ redox transition during electrochemical cycling of the KTiPO4F-based electrode material."

Question 2
It can be seen from the CV and charge/discharge profile that there are three main oxidation process and caused two electrochemcial energy storage mechanism, namely, two-phase mechanism (>3.5 V) and solid solution mechanism (2~3.5 V). However, this observation is contradict with the claim given in Figure 2E. The operando XRD of KTiPO4F in PIBs present in Figure 3a is not clear enough to see the new phase growth during the two-phase evolution during the charge/discharge process above 3.5 V. Please clarify the concerns properly.

Response:
To elucidate the concerns regarding the phase transformation behavior of KTiPO4F we performed a deeper technical analysis of the diffraction pattern evolution during operando experiment ( Figure   R3). Indeed, the phase transformation profile of KTiPO4F during de/intercalation might be The following text along with the Figure R3 (now Figure 4) have been added to the manuscript: "several phase transformations likely happen during the charge/discharge around as seen by abrupt changes of the cell volume with co-existence of two phases (Figure 4, top); iv) in the 2.8 -3.1 V region (on charge) a solid solution mechanism is presumably characteristic for K + deintercalation as the cell volume gradually decreases (Figure 4, bottom). The presence of multiple two-phase intercalation mechanisms eventually explains the series of plateaus at the voltage curve centered at 2.7, 3.1 and 3.6 V and reversible oxidation/reduction peaks on the CVs and dQ/dE derivative of the charge/discharge curve ( Figure 2A, B -inset)."

Question 3
Please check the whole manuscript to avoid any similar mistakes found from line 335 in Page 15. The claimed applied titanium sources is titanyl phosphate, but the provided chemical formula was TiOSO4.

Response:
Corrected. "phosphate" appeared to be a typo. For sure, it should be titanyl sulfate, TiOSO4. The manuscript was double-checked to exclude similar errors. Thank you.

Reviewer #2:
Comments to the author: High Ti 4+/3+ redox potential realized in phosphate fluoride system is very astonishing to see its operation at over 2.5 V. The authors also know how to produce trivalent Ti 3+ that is formed in a strong acidic medium.
The material was well characterized by many methods and the authors have confirmed insertion of the large K + into the host structure.

Question 1
Can the authors provide moisture-sensitivity of the KTiPO4F?

Response:
Ti 3+ generally displays poor stability when exposed to moisture or air. That is why dealing with KTiPO4F requires inert (or reducing) conditions at each step to preserve its chemical composition and Ti 3+ oxidation state. A Ti 3+ -to-Ti 4+ conversion in KTiPO4F can take place according to the following possible reactions: KTi 3+ PO4F + y/2H2O + y/4O2 (from air) → K1-yTi (3+y)+ PO4F + yKOH with further catching of CO2 from air by KOH to form K2CO3.
Both presumable reactions lead to formation of Ti 4+ -enriched materials showing reduced cell parameters. In the second case formation of K2CO3 can be easily observed with a time-dependent XRD of samples exposed to air: Figure R4. Time-resolved XRD patterns of the KTiPO4F/C electrode material exposed to air. Insets: the shift of the peaks confirms unit cell shrinkage due to Ti oxidation and K + loss. Asterisk (*) designates the peak of K2CO3. In summary, the overall degradation reaction can be written as follows: KTi 3+ PO4F + (x+y)/2O2 + y/2CO2 (from air) → K1-yTi (3+x+y)+ PO4F1-xOx + xKF + y/2K2CO3 As seen, water may not even take part in the degradation mechanism of KTi 3+ PO4F as evidenced by the KTiPO4F stability under the synthesis conditions (highly acidic medium, pH ~1-2, reducing conditions created by hydrogen from dissolving of metallic Ti) as well as in pure degassed (deoxygenated) water.
A corresponding paragraph has been added to the Discussion section of the manuscript: "It should be particularly noted that dealing with Ti 3+ , which is generally prone to oxidation, requires inert (or reducing) conditions at each step of synthesis and electrode material preparation to preserve the oxidation state. A Ti 3+ -to-Ti 4+ conversion in KTiPO4F might take place when KTiPO4F is exposed to air following with a formation of K2CO3, which is easily observed with a time-dependent XRD of KTiPO4F/C samples exposed to air ( Figure S8)." Figure R4 and additional comments are included in the supporting materials.

Question 2
What is the used carbon sources?

Response:
Since Ti 3+ is sensitive to oxidation especially at elevated temperatures, for carbon-coating we used oxygen-free carbon sources. Conventional oxygen-containing organic additives like glucose, sucrose, ascorbic acid, etc. do not work. In this sense, polyacrylonitrile (PAN) casted on the initial powder from a DMFA solution (under Ar atmosphere) turned out to be an ideal recipe to successfully perform carbon-coating of KTiPO4F.
The following part is added to the experimental section: "Hydrothermally prepared KTiPO4F was carbon-coated using a solution of polyacrylonitrile (PAN) in DMFA casted on the initial powder and then dried under Ar to yield KTiPO4F/C composite by further annealing at 600°C for 2 hours (3K/min heating rate) in the O2-purified Ar atmosphere (titanium powder was used as an oxygen absorber)."

Question 3
Line 399. N should be italic.
Reviewer #3: Comments to the author: The

Question 1
Electrochemical characterization is needed to confirm the activity of Ti 3+ /Ti 4+ redox. For example, EELS and EPR data at different charge states, particularly charged to 4.2 V or discharged to 2 V, should be provided to validate the electrochemical transition of Ti 3+ /Ti 4+ redox. The Ti contribution to the high voltage character seems to be the key argument of this work. Unfortunately, no data related to the charge compensation mechanism is provided, which makes the claim highly suspicious.

Response:
We confirmed that the charge compensation mechanism is based on the reversible activity of Ti 3+ /Ti 4+ redox transition, which was validated by ex-situ EELS and EPR on the recovered electrode materials. Please see the response to Question 1 of Reviewer 1.

Question 2
In fluorophosphates, fluorine: oxygen ratio seems to play a critical role in determining the electrochemical properties. However, the authors failed to provide experimental evidence and necessary discussion about the key factor. Hence, more systematic research may be required for this work, including the influence of fluorine on the operating voltage, galvanostatic curves and reversible capacity.

Response:
Indeed  For this purpose, we synthesized a single-phase KCrPO4F material isostructural to KTiPO4F. Cr 3+ is known to be a stable oxidation state, which along with the absence of OH-groups (confirmed by FTIR and TG-DSC+MS) guarantees near stoichiometric O:F ratio. The SEM-EDX data are presented in Table R1.   (8). Given the average value, the material seems to be slightly oxidized.
However, it can be considered close to stoichiometric within the error window of determination." The discussion part has been updated with the following text: "It was demonstrated that in many fluoride phosphates the fluorine-to-oxygen ratio in fluoridephosphates plays a significant role in governing the electrochemical performance, which was observed and thoroughly studied for many vanadium-containing fluoride-phosphates. Overall, a more detailed investigation of the influence of the F:O ratio on the electrochemical performance might present a separate large and self-consistent study and thus can be published elsewhere."

Question 3
The author considered the charge/discharge during 2.0-3.4V as solid-solution-like process, despite of clearly two oxidation peaks at 2.81 and 3.21 V for charge and two reduction peaks at 2.58 and 2.99 V for discharge, which demonstrate multiple phase transitions during the reaction. A close look in the operando XRD data did show the emergence of tiny peaks during the disappearance of (212) peak of the pristine sample. For example, some tiny ones at 29.7° can be observed when charged to 3.0V. A detailed discussion is needed to facilitate the understanding of structural evolution during the electrochemical reaction.

Response:
Please see the response to Question 2 of Reviewer 1, where detailed discussion on operando XRD is given.

Question 4
For the rate capability of the tested sample as shown in Figure 4c

Response:
Since with increasing discharge currents the polarization of electrode potentials grows (which becomes really significant at high C-rates), to deliver more capacity on discharge the cathodic cutoff potentials at 5C and 10C measurements were slightly shifted to 1.8 V. The anodic potentials for all measurements were set to 4.2 V vs. K + /K. The coulombic efficiency of the sample at 2C and 5C rates are close to 99.5% as presented in Figure R6. The first 2 charge-discharge curves at 0.1 C compared in Figure. The discharge curve at second cycle coincides with that obtained at the first cycle confirming structural stability of our electrode. Figure R6. Left: The first and second charge-discharge cycles at a C/20 rate. Inset: a dQ/dE differential plot for the second galvanostatic cycle at C/20. Right: Discharge capacities during extended cycling at 2C and 5C charge/discharge rate for 100 cycles and coulomb efficiency. Figure R6 was added to the manuscript as a part of the modified Figure 2 with related comments throughout the text. Thank you.
The manuscript is well revised according to reviewers' comments. After including the EELS and EPR results, the electrochemical mechanisms of the KTiPO4F is clearly revealed and confirmed. Therefore, I recommend to accept this manuscript for publication in current stage.
Reviewer #2 (Remarks to the Author): Well done. I believe that the present work will be one of the milestone to excavate a new compound for KIBs.
Reviewer #3 (Remarks to the Author): I appreciate the efforts the authors have taken to address the questions I have raised. Here I still have a couple of major concerns before I can recommend an acceptance of this work in Nat Commun.
1. The fluorine-to-oxygen ratio turned out to play a critical role in determining the electrochemical properties in the prepared fluorophosphates. To make the data more convincing, it become necessary that the fluorine-to-oxygen ration be changed to show the composition-dependent performance.
2. The operando XRD data showed complex phase transitions during the electrochemical reaction. A detailed analysis on the multiple two-phase intercalations is needed so as to provide a clear understanding on the storage behavior of the KTiPO4F.
Comments to the author: The manuscript is well revised according to reviewers' comments. After including the EELS and EPR results, the electrochemical mechanisms of the KTiPO4F is clearly revealed and confirmed. Therefore, I recommend to accept this manuscript for publication in current stage.

Response:
We are really grateful to the reviewer for his/her final opinion on the manuscript. Thank you.

Reviewer #2:
Comments to the author: Well done. I believe that the present work will be one of the milestone to excavate a new compound for KIBs.

Response:
We really appreciate such a high assessment of our work. Thank you.
Reviewer #3: Comments to the author: I appreciate the efforts the authors have taken to address the questions I have raised. Here I still have a couple of major concerns before I can recommend an acceptance of this work in Nat Commun.

Question 1
The fluorine-to-oxygen ratio turned out to play a critical role in determining the electrochemical properties in the prepared fluorophosphates. To make the data more convincing, it become necessary that the fluorine-to-oxygen ration be changed to show the composition-dependent performance.

Response:
We agree that the fluorine-to-oxygen ratio (F:O) in fluoride-phosphates plays a significant role in governing the electrochemical performance. However, to correlate influence of the precise ratio of  Therefore, the fluorine content should be kept closest to the KTiPO4F stoichiometry for realizing high-voltage performance of the material and it is experimentally difficult to synthesize and characterize such small substitutions of F for O to reflect the change in materials behavior. We have updated the manuscript and supporting information with the data given above to detail the sensitivity to F:O ratio.

Question 2
The operando XRD data showed complex phase transitions during the electrochemical reaction. A detailed analysis on the multiple two-phase intercalations is needed so as to provide a clear understanding on the storage behavior of the KTiPO4F.

Response:
To address this comment, we have provided additional synchrotron X-ray powder diffraction (SXPD) data to shed more light on the multiple two-phase intercalations, which are interesting to many readers from the storage behavior mechanism standpoint.
We note that a general view of the diffraction patterns evolution ( Figure R3, A) clearly shows multiple regions with distinct two-phase or solid-solution transitions and indicates that these processes are symmetric and fully reversible within one charge-discharge cycle as also demonstrated by ex situ XRD patterns ( Figure R3, C). At a closer look at the intensity map of selected two-theta sections ( Figure R4, A) one can observe three two-phase transitions accompanied by appearance/disappearance of specific reflections which is connected with changes in the unit cell symmetry, and one solid-solution region with no distinct changes in symmetry. In addition it is worth noting that the orthorhombic symmetry (SG # 33 Pna21) of the initial KTiPO4F material was validated by electron diffraction. The phase cell volume evolution associated with the above-mentioned transitions and corresponding charge curve are given in Figure R4, B. It should be noted that the two-phase transitions are also clearly seen on the dQ/dE curve ( Figure R4, C) as reversible peaks vs. the solid-solution region with no discernable peaks. Figure R3. A) Diffraction patterns of the in operando SXPD of KTiPO4F in K cell in the 6-17° 2Θ range (λ = 0.7239 Å) and corresponding charge-discharge profile (B). Indexed reflections are denoted. The highlighted dark-red XRD patterns show tentative boundaries of the two-phase (2P) or solid solution (SS) deintercalation regimes on charge. The discharge behavior of KTiPO4F can be considered symmetric. The asterisk sign (*) designates steady reflections belonging to cell components. C) Selected regions of the ex situ XPD patterns of electrodes recovered at various potentials (charge/discharge at C/20 rate, hold for 20 hrs. at every potential). To make this all the more clear we have updated the manuscript and SI to detail this complicated mechanistic behavior.