Na-ion batteries (NIBs) represent a cost-effective and sustainable alternative to Li-ion batteries (LIBs), promising for application in large-scale stationary energy storage systems (Vaalma et al., 2018; Hasa et al., 2021). While operating under the same rocking-chair principle as LIBs, the conventional anode materials used for LIBs, graphite and Si, are not suited for the use in NIBs (Moriwake et al., 2017; Zhang et al., 2017). Studies of hard carbon materials (most common anode material) are still in progress, but additional research efforts are focused on the other anode materials that follow different operating mechanisms, such as alloying and conversion (Wu et al., 2018; Usiskin et al., 2021; Zheng et al., 2021).

Conversion-based anode materials (Figure 1A) are mainly composed of transition metal oxides, sulfides, selenides, and phosphides, while elements from groups 14 and 15 (Si, Ge, Sn, P Sb, Bi, or binary alloys of these, e.g., BiSb, etc.) represent the most common alloying-based anode materials (Figure 1B) (Zhang H. et al., 2018). Both the alloying- and the conversion-based materials exhibit high theoretical capacities ranging between 300-2000 mA h g⁻¹ (Wang L. et al., 2019). However, as the chemical bonding structure and morphology constantly are altered during the (de)sodiation process (Figures 1A,B), both the conversion- and alloying-type anodes suffer from large volume expansion/contraction (Yao et al., 2017; Chen et al., 2019). This often leads to poor cycling stability (Tan et al., 2019; Wang L. et al., 2019; Fang et al., 2021; Villevieille, 2022). Moreover, before the (de)sodiation process reaches the final state, the varying intermediate alloy compositions are formed as crystalline (Figure 1C), of which metastability is determined within the energy of 70 meV/atom above the convex hull (ground state) (Sun et al., 2016). In addition, the mechanical stress by extensive volume variation limits the electronic contact and causes the electrode pulverization in the pure conversion and alloying process. Consequently, their thermodynamic change, which is caused by the kinetic factor after cycling, increases the free energy barrier and results in performance loss (Hudak et al., 2014; Huang et al., 2020).

Conversion-alloying materials (CAMs) offer a new strategy paving the diversity of available compounds for an innovative anode material (Wang L. et al., 2019; Fang et al., 2021). CAMs can potentially deliver high specific capacities without the structural degradation mentioned above. Additionally, CAMs can also offer a reduction of the average working potential compared to pure conversion anodes and therefore increase the energy density in a full cell. CAMs combine the conversion and alloying mechanisms (Figure 1, Eq. 1). ( x + y ) N a +   +   ( x + y ) e − + M X →   M +   N a x X   +   y N a +   + y e −   ⇋   N a y M   +   N a x X (1)

The conversion reaction in these materials can be either reversible or irreversible. A reversible conversion reaction provides significantly higher capacities but suffers from the same problems as pure conversion reactions. Therefore, most studies on CAMs are with an irreversible conversion reaction and these will be the focus of this review. The initial irreversible conversion reaction results in the formation of electrochemically active nanoparticles dispersed within a matrix comprised of Na_nX (X = O, S, Se, Te, P, oxometallates). The formed active nanoparticles or clusters then participate in the reaction with Na-ions through a reversible alloying mechanism. The formed matrix is capable of conducting Na-ions, and together with the active nanoparticles, which have high electronic conductivity, they secure a bi-continuous network beneficial for the electron- and Na⁺-transport that translates into enhanced performance at high cycling rates (Kim et al., 2016; Durai et al., 2017; Lu et al., 2019). The beneficial role of entropy stabilization can be expected in enhancing the cycling performance of the anode during the redox phenomena (Wang Q. et al., 2019; Ghigna et al., 2020). The Na_nX matrix can accommodate the volumetric changes occurring during cycling and provide a physical separation between formed nanoparticles/clusters essentially prohibiting agglomeration/electrochemical sintering of active components as well as particle pulverization. A good compromise between high capacity and cycling stability makes CAMs promising for implementation in next-generation NIBs.

CAMs must include an element that should be capable of alloying with Na-ions in its pure form. Therefore, CAMs for NIBs are based on elements from groups 14 or 15 (e.g., Ge, Sn, Sb, or Bi). CAMs require one or more elements that can participate in the conversion reaction, resulting in two classes: binary or ternary. Binary materials, primarily represented by Ge-, Sn-, Sb-, and Bi-based oxides, chalcogenides, and phosphides are the most investigated CAMs (Gu et al., 2013; Sottmann et al., 2016b; Li et al., 2019; Li C. C. et al., 2020; Wu et al., 2020; Shi et al., 2021). The oxides generally adopt a non-layered structure, while the chalcogenides tend to form layered structures due to the increased polarization of the electron cloud of the anions (Tan et al., 2019). In addition to the conversion and alloying reaction, the layered CAMs also include intercalation in the reaction mechanism. The performance of binary CAMs have been extensively reviewed elsewhere (Zhang H. et al., 2018; Fang et al., 2021).

Compared to binary CAMs, ternary materials are significantly less studied, however, several examples based on Bi- and Sb-based oxometallates of transition metals (BiFeO₃, Bi₂(MoO₄)₃, BiVO₄, Sb₂MoO₆, SbVO₄, Bi₂MoO₆) were reported (Durai et al., 2017; Sottmann et al., 2017; Lu et al., 2019; Pan et al., 2019; Brennhagen et al., 2022; Surendran et al., 2022). The choice of the second cation (Fe, Mo, V, etc.) is driven by the ability of the element to form a stable matrix without affecting gravimetric capacity. The large number of transition metals introduces the possibility to produce a variety of ternary CAMs.

The work reported by Sottmann et al., illustrates the benefits of CAMs by comparing three Bi-based anodes: Bi-metal, BiVO_4, and Bi₂(MoO₄)₃ (Sottmann et al., 2016a; Sottmann et al., 2017). Bi-metal had a high initial capacity (∼500 mAh g⁻¹), rapidly decaying after 100 cycles to ∼300 mAh g⁻¹. BiVO₄ and Bi₂(MoO₄)₃ showed stable capacities of ∼350 mAh g⁻¹ during the first 100 cycles. Sb₂MoO₆ exhibited stable capacities of ∼ 600 mAh g⁻¹ for 100 cycles when cycled at 200 mA g⁻¹, and 450 mAh g⁻¹ for 450 cycles at 2 A g⁻¹. It also showed great rate capabilities maintaining the capacity of ∼ 400 mAh g⁻¹ cycled at 5 A g⁻¹, without any signs of significant degradation (Lu et al., 2019). However, recent studies on Sb₂MoO₆ have not been able to reproduce such incredible performances (Yang et al., 2019; Wu and Wang, 2020). Such discrepancies make the comparison between active materials difficult, not only because of the complexity of batteries in general, but also because of the number of choices for selecting other components of the electrodes. This can undermine materials’ development because the materials themselves may have variations in chemical composition and different morphological features, surface chemistry, (nano)particle’s sizes, and surface coatings that can substantially alter the electrochemical performance. The testing procedures of the materials (potential window, current density, temperature, etc.) are also different between studies, complicating the comparison of “equal” materials even more.

Nanostructuring techniques can generally improve the performance of CAMs, often applied in combination with carbon-based additives/coatings (Sottmann et al., 2016b; Sottmann et al., 2017; Wen et al., 2019; Dai et al., 2020), which is driven by the electrical conductivity. Many CAMs in their pristine state do not have high electronic conductivity, nor has the matrix formed through the conversion process. Creating nanosized CAMs mitigates the consequences of the volume change during the (de)sodiation processes and enhances their activity towards the conversion and alloying reaction (Poizot et al., 2000; Xu et al., 2016; Fang et al., 2019). This leads to a considerable reduction of the diffusion distance of Na-ions in the solid phase, which is beneficial for cycling stability and rate capability (Fang et al., 2021).

The rational design of CAMs is practically impossible without understanding the operating mechanism of CAMs. However, CAMs typically undergo a set of complex structural and chemical transformations during cycling (Fang et al., 2021). That often involves transformation through or into amorphous phases, which complicates the characterization of the materials. The “standard set” of electrochemical characterization techniques can aid in the overall understanding of the conversion-alloying mechanism, but other characterization techniques are also required. Due to complexity of the operating mechanism (i.e., formation of new phases, amorphization, etc.), post mortem and ex situ characterization techniques do not reveal the full scope of the reaction mechanisms. Additionally, air sensitivity and general instability of the materials after sodiation or metastable intermediates also represent a significant issue with ex situ characterization (Gao et al., 2018). Techniques such as operando X-ray diffraction (XRD), operando X-ray absorption spectroscopy (XAS), and in situ transmission electron microscopy (TEM) have emerged as suitable characterization techniques providing information complementary to the post mortem techniques. Worth mentioning that data-driven guidance is also considered as powerful tool to acquire in-depth understanding of reaction processes and metastable intermediates (Bai et al., 2018). Mechanistic insights into the conversion-alloying reaction can be obtained by monitoring changes in both crystalline and amorphous phases, oxidation states of active elements, and local structure during (de)sodiation combined with structural modeling (Shadike et al., 2018; Cui et al., 2021a; Brennhagen et al., 2021).

For example, by applying operando XRD and in situ TEM it has been shown that layered Sn-chalcogenides (SnS₂ and SnSe₂) generally undergo through an intercalation step in addition to the conversion and alloying reactions (Eqs 2–4, respectively) (Shi et al., 2019; Liu et al., 2020; Zhang et al., 2020). In situ TEM have also been used to study other binary CAMs, and examples include Sb₂Se₃, Sb₂S₃, Sb₂Te₃/C Bi₂Te₃, SnO₂, and Sb₂Se₃/rGO (Gu et al., 2013; Ou et al., 2017; Yang et al., 2017; Yao et al., 2017; Wu et al., 2020; Cui et al., 2021b). S n X n + x N a + → N a x S n X n (2) N a x S n X 2 + ( 4 − x ) N a + →   S n + 2 N a 2 X (3) S n + 3.75 N a + ⇋ N a 3.75 S n (4) Upon sodiation of SnX_n (X = S, Se), the intercalation of Na-ions 1) occur in the voltage range of ∼ 2.5–1.1 V vs. Na/Na⁺, followed by the conversion reaction 2) in the range 1.1–0.6 V vs. Na/Na⁺, and the 1st cycle is finalized by the alloying reaction 3) in the 0.6–0.1 V region vs. Na/Na⁺. Often, in situ TEM can be complementary to operando XRD, and vice versa (Ou et al., 2017).

Operando XRD (due to its availability) is the most common characterization technique for assessment of the operation mechanism of CAMs (Brennhagen et al., 2021). Operando XRD has been used to elucidate the conversion and alloying reactions for binary CAM anodes (e.g., SnS₂, Bi₂S₃, Sb₂S₃ Sb₂O₃@Sb, Sb₂O₃/rGO, Sb₂S₃@FeS_2, and Sb₂Se₃/rGO) (Sottmann et al., 2016b; Ou et al., 2017; Li et al., 2018; Ma et al., 2018; Shi et al., 2019; Cao et al., 2020). In many cases, (ternary CAMs) the matrix formed through the conversion is amorphous (Durai et al., 2017; Lu et al., 2019; Pan et al., 2019; Brennhagen et al., 2022), which calls for other characterization techniques to obtain an understanding of the operation mechanism. Pair distribution function (PDF) analysis is among such techniques allowing elucidating amorphous phases formed during (de)sodiation. However, to the best of our knowledge, no operando PDF has been performed on CAMs, while bulk Sb and Sn metal (a pure alloying anode) has been studied using operando PDF, showing that these relatively simple alloying reactions also contains amorphous intermediates (Allan et al., 2016; Stratford et al., 2017). This also means that we have more to learn from the alloying reactions in CAMs, even though the cycling mechanisms might seem quite clear when only using operando XRD.

Another useful operando technique, which is used to monitor specific elements in CAMs is operando XAS, where X-ray energies are scanned across the absorption edges of the target elements. This technique has been utilized to gain information on changes in oxidation states and local structures during cycling for Bi₂O₃, Bi₂S₃, BiVO₄, Bi₂(MoO₄)₃, SnO_2, and BiFeO₃ (Kim et al., 2016; Sottmann et al., 2016b; Sottmann et al., 2017; Dixon et al., 2019; Surendran et al., 2022). In the case of Bi₂S_3, a measurement combining operando XRD and XAS was used to reveal important aspects of the cycling mechanism (Figure 2) (Sottmann et al., 2016b). During the first sodiation, the irreversible conversion reaction forms amorphous Bi (wide and diffuse peak, Figure 2A) which further goes through a two-step alloying reaction into crystalline NaBi and Na₃Bi. A sharper peak for the Bi phase was observed after the first desodiation, indicating higher degree of crystallinity. The changes in the oxidation states of Bi are also linked to the edge shift in the XANES data (Figure 2B), showing that the oxidation state varies from Bi³⁺ in Bi₂S₃ to Bi^3- in Na₃Bi. The study further shows that the Na_xS matrix is not stable because it cycles between Na₂S₄ and Na₂S (Figure 2C). The conversion reaction is partially reversible because some Bi₂S₃ is retrieved at the end of the desodiation (Figure 2C), which is beneficial for specific capacity, but not practical for cycling stability (the second cycle demonstrates diminished formation of Bi₂S₃). Similar instabilities in the matrixes of some ternary CAMs, such as Bi₂MoO₆ and BiFeO₃, were observed (Brennhagen et al., 2022; Surendran et al., 2022). However, detailed studies on the behavior of the matrixes linked to the electrochemical performance are difficult to perform due to the complexity of the cycling mechanisms and species formed. As a result, such studies are limited and not always conclusive. This illustrates the importance of analysis of the matrix materials when working and designing CAMs.

There are clear indications that CAMs have significantly better cycling stability than the pure conversion or alloying materials, while still possessing higher capacities than intercalation materials. Variety of chemical compositions makes it possible to meet the desired compromise of high capacity, good cycling stability, cost, and environmental friendliness. There is a large playground for evaluation of alloying elements with transitions metals to produce ternary CAMs. Until present, only a few examples have been explored and there are more to discover in the future. Although the main general cycling mechanism is relatively well understood, the details of the mechanisms have only recently begun to unravel. The stability of the matrix is crucial to maintaining cycling stability and therefore deserves more research efforts. Meanwhile, the non-equilibrium phase of the nanoparticles formed through conversion is also relevant for materials design because co-existing phases (matrix) often exhibit and maintain extreme non-stoichiometry (Yao et al., 2017). There have been high-throughput search reports, along with the accelerated discovery efforts of promising anodes. However, due to the complicated electrochemical reaction with co-existing metastable phases in CAMs, the most high-throughput investigations of anodes have focused yet on single-phase mode within the limited length scale (Ong et al., 2011; Kirklin et al., 2013; Zhang X. et al., 2018; Zhang et al., 2019). This complexity makes CAMs challenging to study and develop. While CAM anodes emerge, it is critical to understand and classify the metastability of intermediate compounds for elements in the active anode’s (de)sodiation reactions to provide data-driven guidance.

One of the biggest challenges with CAMs in full cells is the irreversible capacity loss during the first sodiation due to the irreversible conversion reaction. This demands that the cathode or electrolyte should supply more reaction Na-ions for the first discharge and could lead to a significant amount of dead mass for the subsequent cycles. This problem could be solved by presodiation. Even though this complicates the battery production, it could be viable due to the high capacity, good cycling stability, and rate performance that CAMs could potentially deliver. In addition, the design and optimization of the active element/matrix combination can also assist in mitigation of this complex challenge.

Estimation of the theoretical capacity of CAMs represents a challenge and requires a detailed understanding of the cycling mechanism, which involves multiple chemical and electrochemical transformations. Calculation of the theoretical capacity for CAMs is done by either assuming a fully reversible conversion and alloying reactions or assuming that only the alloying reaction contributes to the capacity. However, for several materials described above partially reversible conversion reactions and/or some electrochemical reactions in the matrix are commonly encountered, which complicate the estimation of theoretical capacities. Such uncertainty, before determination of the operating mechanism makes it difficult to assess which reactions should be included in the calculated theoretical capacities. The 1-st cycle irreversible capacity (from the irreversible conversion reaction) combined with the SEI formation increases the gap between theoretical and measured specific capacity. It can still be useful to compare the theoretical capacity of certain stages of the reactions to the specific capacities of the corresponding plateaus in the (de)sodiation curves to improve the understanding of the reaction mechanisms.

Considering the abovementioned complication, it is very important to provide all the necessary details of the study and do comparative studies to get a clearer picture of the actual performance of the material as was emphasized in a few recent publications (Li J. et al., 2020; Stephan, 2021; Sun, 2021). The experimental uncertainties add to the complexity of the problem for assessment of the specific vs. theoretical capacities of CAMs. That includes proper consideration of other components of the electrodes, potentially uneven distribution of active material on the electrode sheet, proper statistics of the results, and accurate measurement of the mass of the active material. The latter can deliver an error of ≥ 10% to the calculated specific capacity when performed at the laboratory scale (Brennhagen et al., 2022). In addition, specific current densities rather than C-rates should be preferably reported when working with these materials, as a determination of the C-rate requires an accurate estimation of the theoretical capacity.

In conclusion, CAMs, while yet underexplored and poorly understood deliver a large potential for the future of NIBs, however, their challenges and operating mechanisms have to be properly examined. That requires not only synthetic efforts, but also careful consideration of the electrochemical characterization techniques as well as development and selection of the operando tools.