Cleavable magnetic materials from van der Waals layered transition metal halides and chalcogenides

The development of 2D magnetic materials produced by exfoliating bulk single crystals opened vast and exciting research areas for solid-state chemists and physicists and materials scientists. Over the past decade and, particularly, the last 5 years or so, the study of 2D materials, monolayers, and heterostructures and the devices they enable that combine magnetic, electronic, and optical functionalities has grown rapidly.

Crystals that can be exfoliated to a few-layer form have enabled much of the progress that has been made. These materials have crystal structures comprising charge neutral 2D layers that are held together by van der Waals bonding. The term van der Waals layered materials can be used to distinguish them from other types of layered phases (where there is charge transfer or covalent bonding between the layers) and from other types of van der Waals compounds (like those made of molecular or 1D structural units).

van der Waals (vdW) layered materials can support local-moment and itinerant magnetism, and those that do, especially those with long-range-ordered ground states, can be referred to as cleavable magnets or vdW layered magnets (magnetic material may be more accurate than magnet in this context). Cleavable here refers specifically to the easy cleavability associated with the van der Waals layered structure. The term 2D magnets is also sometimes used to describe these compounds, but this could lead to some confusion with materials in which the magnetism alone is 2D or quasi-2D, which does not necessarily imply an exfoliatable structure.

Because of the rapid and significant progress that has occurred recently and the high level of interest in magnetism in these materials and devices, many review articles have been published in the last few years. This is not a review article, but it is useful to briefly introduce some of them to the reader here.

In 2017, Duong et al. published a review entitled “van der Waals layered materials: Opportunities and challenges.”¹ The broad scope of this article includes good overviews of specific materials, material families, and their chemistries and bonding, and physical properties including thickness dependent properties, monolayer stability, and heterostructure stacking and growth studies. This review is not limited to magnetic compounds. In the same year, I published “Crystal and magnetic structures in layered, transition metal dihalides and trihalides,” a review of this specific class of cleavable magnetic materials,² with a focus on the crystal chemistry and magnetic properties of binary halides with triangular and honeycomb nets of transition metals. In 2018, Burch et al. published a review entitled “Magnetism in two-dimensional van der Waals materials.”³ The authors briefly review the materials landscape and focus on how optical and electronic probes and devices provide insight into the materials and their physics. They note potential impacts for vdW magnets in the areas of new materials, fundamental physics, devices and applications, and quantum and topological phases.

Many more reviews followed in 2019. Gong and Zhang published “Two-dimensional magnetic crystals and emergent heterostructure devices.”⁴ This review introduces a broad materials library including halides, chalogenides, and others. It covers intrinsic magnetism and how it can be controlled in 2D materials, as well as reported results and ideas for heterostructures and devices exploiting vdW layered magnets. Gilbertini et al. published a review entitled “Magnetic 2D materials and heterostructures.”⁵ There the authors discuss the evolution of magnetism from bulk to monolayers including critical behaviors and magnetic ordering. The materials focus is on $CrI 3$ and $Fe 3 GeTe 2$⁠, covering both intrinsic properties and heterostructure results. Li et al. published “Intrinsic van der Waals magnetic materials from bulk to the 2D limit: New frontiers of spintronics” reviewing crystal growth and exfoliation, measurements to detect magnetism, and stacking effects.⁶ Su et al. published “Van der Waals 2D transition metal tellurides,” a review focusing specifically on tellurium compounds and including many magnetic ones.⁷ The authors present an overview of the growth, exfoliation, physical properties and phase transitions, and potential applications. Mak et al. published a review entitled “Probing and controlling magnetic states in 2D layered magnetic materials.”⁸ After surveying magnetic vdW layered materials and their magnetism, the authors discuss optical and electronic methods of probing magnetic order in the 2D limit as well as methods for manipulating it, in the context of reading and writing magnetic states for potential applications in spintronics. Zhang et al. published a review entitled “Van der Waals magnets: Wonder building blocks for two-dimensional spintronics?”⁹ The authors review vdW layered magnetic materials based on Cr, Fe, Mn, and V and note technological challenges associated with chemical stability, limitations on Curie temperatures, and scalability of manufacturing. With a spintronics point of view, the authors note 2D antiferromagnets deserve as much attention as 2D ferromagnets.

Additional reviews have continued to appear in 2020. Huang et al. published “Recent advances in two-dimensional ferromagnetism: materials synthesis, physical properties, and device applications.”¹⁰ This review focuses specifically on ferromagnetism, how it can be manipulated by doping and strain, and its consequences in the 2D limit for materials and devices. Wang et al. published “Prospects and opportunities of 2D van der Waals magnetic systems.”¹¹ There, the authors review the theoretical underpinnings of magnetic order in 2D as well as experimental results from thin flakes and heterostructures. There are surely more reviews forthcoming as the field continues to expand and innovate, as interest among researchers grows, as breakthroughs occur, and as new materials are introduced.

Tables of materials that have been predicted or realized to be vdW magnets are particularly useful for researchers interested in developing new materials. Examples are found in some of the reviews noted above.^1,4,6–8,11 Inspiration can also be drawn from high throughput theoretical studies that screen candidate materials for cleavability and magnetism, for instance, the report from Mounet et al. in Ref. 12.

The field continues to grow and produce exciting discoveries and is poised to continue growing for the foreseeable future. Currently, only a handful of materials have been explored in detail at the few-layer and monolayer levels, and those materials continue to produce interesting science. There are many more potential candidate materials waiting to be explored. Expansion into new compounds will accelerate as experience and techniques advance for protecting the delicate and often air sensitive samples. Although the field is growing fast and maturing quickly, there is expansive room for growth in many directions. This begins with the development of the individual materials that provide the foundation for the field. Then, the incorporation of these materials into heterostructures combining ferromagnetism, antiferromagnetism, complex spin textures, optoelectronic properties, topological bands, and ferroelectricity provides an enormous space for exploration and discovery. Growth will also be driven by the continued development of clever device designs, new ways of probing material and device properties, and manipulation of behaviors with external fields.

For an experimental materials scientist who grows and studies bulk crystals, this research area is particularly interesting and rewarding because it provides a connection to 2D physics and nanoscience that was previously not possible. It integrates solid-state chemistry and materials development, crystal growth and characterization, heterostructures and devices, fundamental physics, and the potential for important applications. In this Perspective, I will present some of my personal viewpoints on the current field of cleavable magnetic materials, specifically transition metal halides and chalcogenides, with a focus mainly on the crystal chemistry and magnetism in the bulk materials. I will introduce some of the key material families and specific compounds and highlight recent advances and innovations that I find particularly interesting. I will provide my perspective on where I think the field could be headed in terms of growing these families and expanding into “new” materials, and the important role that detailed crystallographic information plays in understanding them.

A major milestone in the development of vdW layered magnetic materials is the experimental demonstration of stable monolayers and comparison of their properties with those of their bulk crystals. A monolayer here is defined as one of the ionically/covalently bonded 2D slabs that is separated by the vdW gap in the crystal. This is usually several atoms thick and sometimes thinner than one unit cell of the bulk crystal structure, which can contain multiple monolayers when the stacking is not simply AA. Several magnetic materials have been exfoliated down to or near this limit. These include T $PS 3$ with T = Mn,^13–15 Fe,^16,17 Ni,^18, $CrGeTe 3$⁠,^19,20 $Fe 3 GeTe 2$⁠,^21,22 $VSe 2$⁠,^23,24 $MnBi 2 Te 4$⁠,^25,26 CrX $3$ with X = Cl,^27,28 Br,^29,30 I,^31,32 and $RuCl 3$⁠.^33,34

The utility of cleavable magnetic materials extends beyond the study and exploitation of their magnetic behavior in the ultrathin limit. In addition, a freshly cleaved atomically smooth surface of such a material provides a substrate for coupling of magnetism to deposited films or flakes of other materials. For example, cleaved $CrI 3$ and $CrBr 3$ crystals have been used to induce strong valley Zeeman splitting in monolayer $WSe 2$ and $MoSe 2$ observed through optoelectronic measurements,^35,36 and $CrGeTe 3$ crystals have been used to induce the anomalous Hall effect in $( Bi x Sb 1 − x ) 2 Te 3$ and Pt.^37,38

The materials mentioned in the preceding paragraphs fall into two chemical families, transition metal halides and transition metal chalcogenides, and there are of course other interesting vdW layered phases in these systems that are currently being investigated. In the following, I will discuss separately these two families, presenting a general overview with emphasis on key observations and developments that I find particularly interesting, and that present opportunities for expanding vdW layered magnetic materials within these families and beyond.

While there are a wide variety of vdW layered magnetic transition metal halides,² much of the recent progress related to exfoliable materials has centered on the trihalides with honeycomb nets of ruthenium, chromium, and more recently vanadium [Fig. 1(a)]. Initial interest in $RuCl 3$ arose not because of the prospect of exfoliation but because of the pseudo-2D magnetism related to its layered structure and promising evidence of a Kitaev quantum spin-liquid state at low temperatures and high magnetic fields.^39–44 Reports of experimental results from ultrathin exfoliated samples are still limited; however, the material is remarkably stable in air³⁴ and appears to present interesting opportunities for suppressing interlayer magnetic interactions in exfoliated samples and also coupling charge to the spin-liquid magnetism in insulating $RuCl 3$ using van der Waals heterostructures. Biswas et al. recently performed a theoretical study of $RuCl 3$ in proximity to graphene.⁴⁵ They report several surprising effects, including charge transfer that electron dopes the insulating $RuCl 3$ to near a metal–insulator transition and an increased Kitaev interaction strength. Strain due to lattice mismatch was also induced in the $RuCl 3$ layer, despite the weak vdW nature of the interlayer interaction. This “epitaxial” strain is responsible for the enhanced Kitaev interactions. This is important because it points out the potential role of small structural changes that may develop in 2D materials when they are incorporated into heterostructures. It is certainly not surprising that the properties are dictated by the crystal structure, but the weak interlayer interactions in vdW materials make it easy to neglect subtle structural responses that may occur due to crystallographic proximity between layers. A recent experimental study detected a surface reconstruction on the surface of cleaved $RuCl 3$ crystals.⁴⁶ Significant buckling was seen in the topmost monolayer. This is somewhat surprising for a van der Waals system and has implications for what should be assumed about the relationship between the structures of crystals and monolayers cleaved from them. The important role of crystal structure, including how it may be different in crystals, few-layer specimens, and monolayers and its consequences on magnetic behavior, will be a recurring theme here, with several other examples described below.

More rapid experimental progress has been made on exfoliated chromium trihalides. The behaviors of mono- and few-layer specimens and physics of devices and heterostructures incorporating these materials are summarized in the review papers and other references mentioned above, and I would note here an aspect that I find particularly interesting and relevant to the field at large. It began with the observation that monolayer $CrI 3$ is ferromagnetic at low temperature, like the bulk material, but the bilayer behaved like an antiferromagnet and was inconsistent with ferromagnetic coupling between the two Cr layers.³¹ This was quite a surprise and led to revaluation of data from bulk crystals looking for any sign of antiferromagnetism. A minor complication was that the crystals do not show any magnetic hysteresis that is often associated with ferromagnetism. But this is not unusual in ferromagnetic single crystals with few defects and no grain boundaries to pin magnetic domain walls. In the end, it turned out that the interpretation of data from the few-layer specimens and bulk single crystals were both correct, but the assumption that the bilayer sample represented two layers from the bulk crystal structure was not. The stacking relationship between the layers is different, resulting in different interlayer magnetic exchange.^47–49 This is illustrated in Fig. 2. The key to understanding this difference was the crystallographic phase transition that crystals of $CrI 3$ undergo when cooled below room temperature.⁵⁰ The layers stack in an ABCABC pattern at both room and low temperature, but the direction that the layers are shifted with respect to one another is different (Fig. 2). At room temperature, this results in a structure with monoclinic symmetry, and at low temperature, the structure has rhombohedral symmetry. When ultrathin specimens are cooled, they do not undergo this stacking transition.^49,51 While bulk crystals are rhombohedral at low temperature, few-layer samples are stacked like the monoclinic structure the crystals have at room temperature.

It is not completely clear yet why the structure change is suppressed in thin samples. Perhaps, it is related to the substrate or to some unknown defect or strain introduced during the exfoliation process. Recent studies suggest that it may simply be a surface effect, uncovering evidence that the surface layers of thicker specimens have antiferromagnetic stacking while the bulk remains ferromagnetic.^52,53 This is almost certainly related to the surprising structure and behavior of few-layers specimens. While the origin of the stacking differences in $CrI 3$ samples is certainly interesting, the consequences are especially dramatic and make exfoliated $CrI 3$ in some ways more interesting than it might be otherwise. In zero magnetic field, the antiferromagnetic stacking gives no net moment for even layer numbers and a large uncompensated ferrimagnetic-like response for odd layer numbers, resulting in metamagnetic transitions and layer by layer switching,^54,55 effects that would not be present if the simple ferromagnetic stacking of the crystal persisted in the thin flakes. Like $CrI 3$⁠, $CrCl 3$ crystals have this transition below room temperature, and it is suppressed in few-layer samples.⁵⁶ Tunneling magnetoresistance studies have proven particularly useful in these materials,^28,54–57 since the antiferromagnetically aligned layers act as switchable spin filters.

Interestingly, $CrBr 3$ undergoes its stacking transition above room temperature and is already rhombohedral when exfoliated giving ferromagnetic few-layer and monolayer specimens when cleaved from bulk crystals.³⁰ However, $CrBr 3$ bilayers grown by molecular beam epitaxy (MBE) can have different stacking arrangements, and a similar correlation between stacking and interlayer magnetic ordering noted above for $CrI 3$ has been directly observed in MBE grown $CrBr 3$⁠.⁵⁸ In the series $CrCl 3 − x Br x$⁠, Abramchuk et al. have shown that the bandgap, magnetic ordering temperature, and magnetic anisotropy of crystals evolve smoothly between the end members.⁵⁹ The composition dependence of the rhombohedral to monoclinic stacking order transition has not been studied, but one expects it would evolve smoothly as well, giving a transition near room temperature for intermediate compositions. It would be interesting to see how such materials behave when the exfoliation is carried out near their stacking transition temperature, bearing in mind that these are first order transitions with some thermal hysteresis.

There is, perhaps, an opportunity to better understand, predict, and control these materials by learning what drives the crystallographic transition that switches the stacking and the sign of the interlayer exchange interaction. Theoretical and experimental studies designed to probe this could add substantially to our understanding of these materials and more importantly allow better control over stacking sequences and resulting magnetic structures in other materials. Perhaps, there are materials that when exfoliated undergo stacking or other crystallographic phase transitions not experienced in bulk samples, or ones that could be pushed through a stacking transition under the right conditions. For example, pressure has been used to induce changes between layer stacking arrangements in $CrI 3$⁠.⁶⁰ Most of our detailed structural knowledge about compounds comes from diffraction studies of bulk materials. Translating this to ultrathin exfoliated samples is likely reasonable in most cases, but we must remember that the details of the structure of exfoliated samples are often simply assumed.

In general, including some level of structural characterization of exfoliated specimens beyond coordinating thickness with layer number is very valuable, especially as new materials emerge as potential mono- and few-layer magnets. X-ray diffraction tools appropriate for thin films are probably precluded by the limited lateral extent of most exfoliated materials. This leaves primarily real space methods with atomic resolution transmission electron microscopy. Such measurements can be very useful for observing in-plane symmetries and deducing stacking order. However, images typically record the projection of the atomic positions within the plane, with limited insight into the out of plane coordinates of the atoms. Surely, these must be sensitive to whether a layer adjoins another layer, a substrate, or vacuum, and the influence of these positions on the electronic structure may be important for accurate theoretical modeling of magnetism and other properties. The importance of crystal structure, specifically stacking sequence, on determining magnetic behavior is highlighted by $CrI 3$ and $CrBr 3$ above, but it has also been noted in $RuCl 3$⁠.^41,61 Details of coupling between magnetism and layer stacking are likely also relevant to vanadium compounds discussed next, as well as some of the chalcogenides described further below.

Vanadium trihalides, with structures similar to chromium trihalides [Fig. 1(a)], present another interesting direction for cleavable magnets among halides. These materials have received some attention over the last couple of years, and although it is agreed that $VI 3$ is ferromagnetic below 50 K , there has been some disparities in reports of crystal structures with monoclinic, trigonal, and rhombohedral space groups reported for the room temperature structure.^62–66 The most recent reports support the rhombohedral structure for both $VI 3$ its antiferromagnetic analog $VBr 3$⁠.^66,67 Symmetry lowering structural changes are seen in $VI 3$ near 79 and 32 K, with the details of the low temperature structures not well settled at this time.

With a structure change below the ferromagnetic ordering temperature with some clear magnetoelastic character,^65,66 $VI 3$ presents a perhaps more complicated and interesting case than $CrI 3$⁠. Isothermal magnetization measurements also tend to show significant hysteresis, suggesting defects or other microstructural features pin ferromagnetic domain walls. Perhaps, twinning at the structural transitions creates structural domain walls that could contribute to pinning. It will be interesting to see how these features, as well as stacking disorder that is apparent in these compounds,^64,67 translate into few-layer samples of vanadium trihalides. Stacking disorder in the crystals may make the behavior of ultrathin samples particularly unpredictable and difficult to understand without complementary studies of the structure/stacking. Will exfoliation suppress the structural transition near 78 K that is most likely related to layer stacking? Will the structural change in the ferromagnetic state, which appears to be most likely an in-plane distortion, persist in monolayer specimens with no interlayer interactions and new interactions with the substrate? With their complex magnetostructural properties only just being explored and understood, we can expect that mono- and few-layer vanadium trihalides will provide behaviors and interactions that are complementary to those currently available and inspire new ideas for measurements to probe both their magnetism and their crystallography as well as clever devices to exploit their physics.

What lies beyond Ru, Cr, and V halides? While those three metals have duly received considerable attention, there are of course many other layered, magnetic, transition metal halides.^2, $CrI 3$ and $VI 3$ were particularly attractive because of their bulk ferromagnetic order. The most natural progression of this area seems to be the extension to materials with more complex magnetic structures in the bulk. As studies of few-layer $CrI 3$ have shown, antiferromagnetic order can be more interesting than ferromagnetism since transitions between spin structures can be induced by applied fields. Studies of antiferromagnetic transition metal halides comprising ferromagnetic layers stacked antiferromagnetically may be expected to produce interesting optical and magnetotransport results analogous to few-layer $CrI 3$⁠. This expectation is being borne out in the most closely related example $CrCl 3$⁠.^28,56 Other possibilities yet to be explored experimentally are found among the triangular lattice dihalides [Fig. 1(b)]. $FeCl 2$⁠, $FeBr 2$⁠, $CoCl 2$⁠, $CoBr 2$⁠, $NiCl 2$⁠, and $NiBr 2$ have ferromagnetic layers stacked antiferromagnetially over some temperature range, with examples shown in Fig. 3. They provide systems with moments in plane in the Co and Ni compounds and out of plane in the Fe compounds. Theory and modeling results of some of these compounds are already available, for example, in Refs. 68–71, and experimental studies of these compounds should accelerate as capabilities and expertise for handling exfoliated air sensitive materials continue to expand.

Finally, there are more complex magnetic structures known among these materials. This section began with some discussion of $RuCl 3$⁠, which has a non-collinear antiferromagnetic structure and likely hosts a quantum spin-liquid in applied fields. Chemically related but vacancy laden $Os 0.55 Cl 2$ also shows spin-liquid-like behavior, which is coupled with short- and long-range vacancy ordering, presenting a structurally and magnetically complex cleavable material that is stable in air.⁷² In terms of complexity, between the simplest ferro- and antiferromagnets described above and disordered spin-liquid-like phases, there are non-collinear magnetic structures. These include the helimagnets $MnCl 2$ and $FeCl 3$⁠; multiferroic $MnI 2$⁠, $CoI 2$⁠, $NiI 2$⁠; and $NiBr 2$⁠,^73–77 and the frustrated in-plane antiferromagnets $VCl 2$ and $VBr 2$ that adopt a 120 $°$ structure. Slightly more complicated but still collinear structures are found in the in-plane antiferromagnets $MnBr 2$ and $FeI 2$ with stripe-like patterns (Fig. 3).^78,79 Understanding how their behaviors evolve toward the 2D limit and exploiting applied magnetic fields and coupling to other order parameters to control the magnetism in these compounds presents exciting prospects for future experimental studies. Clearly, there is still plenty of materials space to be explored among these well known binary transition metal halides, especially when one considers the additional degree of freedom provided by alloying materials of similar structure types but different magnetic behaviors.⁸⁰ 

Expanding the halide family to realize additional structural and magnetic versatility requires moving beyond binary compounds. Multinary transition metal halides with vdW layered crystal structures appear to be somewhat rare. While it is typical to think of multinary compounds in terms of multiple types of cations, current examples among transition metal halides are more common in mixed anions systems. These include the oxyhalides CrOCl and FeOCl, vdW layered antiferromagnetic insulators with trivalent transition metals that have received some recent attention and appear to have some promise for producing magnetic monolayers.^81,82 More conducting analogs of these are realized by replacing oxygen with sulfur in sulfide-halides like CrSBr.⁸³ These oxyhalides and sulfide-halides are isostructural [Fig. 1(c)], and a unique feature of these materials compared to all of the others discussed here is that they are not based on close-packed (triangular) anionic layers; they adopt an orthorhombic structure. Their cations form a puckered rectangular net, and unlike all of the other change-balanced compounds discussed here, the transition metals have strongly distorted coordination environments due to different bond lengths to the different anions, though still approximately octahedral. These materials have been known for some time and are being rediscovered based on interest in their potential application to vdW magnets and heterostructures. Work on these compounds will grow and continue to open up new areas for the study of their structures and magnetism in the bulk and 2D limit. The oxyhalide family, in particular, is quite diverse and can stabilize many different transition metals (3d, 4d, and 5d) in several different oxidation states, forming multiple layered structure types. As an example, $RuOCl 2$ with tetravalent Ru is shown in Fig. 1(d). Their layered nature, distorted coordination, and incorporation of heavy transition metals provide a potential for realizing strong spin–orbit coupling and magnetic anisotropy.

The first experimental evidence for magnetic order in a monolayer cleaved from a van der Waals layered material was reported by Lee et al. in the transition metal chalcogenide $FePS 3$⁠, observed via Raman modes associated with the in-plane antiferromagnetic order.¹⁶ Soon after, ferromagnetism was reported down to bilayers in the isostructural compound $CrGeTe 3$⁠.¹⁹ The structures of these materials are related to, but more complex than, those of the binary halides discussed above. Briefly, they are composed of honeycomb planes of Fe or Cr with P–P or Ge–Ge dimers piercing the planes in the centers of the honeycomb cells and have distorted close-packed layers of S or Te above and below the plane [Fig. 4(a)]. The transition metals are in octahedral coordination and the P and Ge atoms are each bonded to three chalcogens forming ethane-like units, so the compounds can be more instructively written as $Fe 2 P 2 S 6$ and $Cr 2 Ge 2 Te 6$⁠. The study of exfoliated magnetic vdW layered chalcogenides quickly expanded beyond these compounds, and many systems are currently being studied. Several are closely related to $FePS 3$ and $CrGeTe 3$⁠, and those are addressed next.

A chemical understanding of these “113” or “226” materials requires slightly more complicated electron counting than the halides above, and this is key in rationalizing which phases form when exploring related materials. Formally treating these as ionic compounds with complex anions, each $Ge 2 Te 6$ anionic unit has a charge of $−$6 (44 valence electrons from the neutral elements but 50 are required to satisfy all of their octets). Similarly, each $P 2 S 6$ unit has a charge of $−4$ (with P having one more valence electron than Ge). Thus, in the simplest cases, $MGeTe 3$ is expected to form for trivalent M atoms, and $MPS 3$ is expected to form for divalent M atoms. This presents a formula for identifying analogous magnetic materials and is borne out in $NiPS 3$ and $MnPS 3$ with divalent Ni and Mn as already noted above.

Interestingly, the only close analogs of $CrGeTe 3$ expected to be magnetic appear to be constrained to Cr for the trivalent metal, with Si replacing isovalent Ge or Se replacing isovalent Te. This includes $CrSiTe 3$⁠, for which ferromagnetism in several-layer specimens has been reported,⁸⁴ and $CrSiSe 3$⁠, about which little is known. It would be interesting to learn more about the behavior of the latter compound, although magnetocrystalline anisotropy, critical for maintaining long-range order in the 2D limit, is expected to decrease as one moves up the periodic table (Ge to Si, Te to Se) to elements with weaker spin–orbit coupling. Generally, there appears to be less chemical flexibility in the materials with Ge–Ge or Si–Si dimers than those with P–P dimers (discussed next). This likely reflects the fact that the magnetic first row transition metals are more commonly found in a divalent rather than a trivalent state in chalcogenides, but perhaps this is also because solid-state chemists have spent about 20 more years working on the $P 2 S 6$ systems. The recent rediscovery of these materials may lead to some of the versatility described below for the thiophosphates being realized for $CrGeTe 3$⁠.

Recent work on $FePS 3$-related phases has focused on sulfides, but many examples among selenides are known as well (see, for example, Ref. 85). As described above, this family of compounds produced some of the earliest studies of exfoliated van der Waals layered magnetic materials (⁠ $MnPS 3$⁠, $FePS 3$⁠, $NiPS 3$⁠) and the first evidence for magnetic order in few and single layer materials. However, there are still numerous examples of likely interesting analogous that have not been well studied, and this family will surely continue to produce exciting discoveries.

The most obvious and direct analog is $CoPS 3$⁠, which has a zigzag AFM structure within the plane and moments lying nearly in the plane (similar to $NiPS 3$⁠) below about 120 K.⁸⁶ While the XY-like anisotropy may suppress ordering in 2D as in $NiPS 3$⁠, this compound has the honeycomb network and $d 7$ electronic configuration predicted to favor a quantum spin-liquid state.⁸⁷ Thus, there is ample motivation to develop strategies to tune the anisotropy and magnetic exchange interactions in this material using alloying, dimensionality, and stacking approaches. The next closest analogs are the selenide variants $MnPSe 3$⁠, $FePSe 3$⁠, and $NiPSe 3$⁠, in which the magnetic order in the plane is antiferromagnetic.^88,89 The antiferromagnetic nature of all of these $FePS 3$-related phases make them somewhat less promising for strong magnetic proximity effects, but very promising for spintronic applications, as noted in Ref. 9.

It is interesting to note that there are stacking variations among the $FePS 3$-related compounds, as in the halides. The in-plane structures are essentially the same, but $MnPS 3$⁠, $FePS 3$⁠, $CoPS 3$⁠, $NiPS 3$⁠, and $NiPSe 3$ adopt a monoclinic stacking of the layers, while $MnPSe 3$ and $FePSe 3$ have rhombohedral stacking. Temperature induced stacking transitions may not be common among these compounds (one example is non-magnetic $CdPS 3$⁹⁰); however, doping studies and investigation of alloy systems should not presume the stacking behavior with any certainty, and structural studies should not be neglected.

The versatility and potential promise for these layered thiophosphates and selenophosphates really become apparent in the slightly more complicated analogs. These are realized by substituting the two divalent transition metal ions with one that is monovalent (⁠ $Cu +$ or $Ag +$⁠) and one that is trivalent. This allows the incorporation of other magnetic species like $V 3 +$ and $Cr 3 +$⁠. However, it is the behavior of the monovalent ions that makes these materials particularly interesting additions to the field of van der Waals layered compounds. Preferring a lower coordination number, the Cu or Ag ions are displaced away from the center of the octahedron of chalcogen ions toward the van der Waals gap. In $CuCrP 2 S 6$⁠, the Cu ions displace in a stripe-like antiferroelectric pattern shown in Fig. 4(b), and the Cr moments order with ferromagnetically aligned spins lying in the plane and the planes stacked antiferromagnetically.⁹¹ For the (non-magnetic) compound $CuInP 2 S 6$⁠, the Cu atoms all move in the same direction and this results in a net polarization at room temperature.⁹² The ground state is ferrielectric since the In atoms slightly displace in the opposite direction, probably due to the distortion of their octahedral cages caused by the Cu displacement. Spontaneous polarization has been reported in flakes as thin as 4–10  $μ$m,^93,94 and van der Waals heterostructures of $CuInP 2 S 6$ and $MoS 2$ have been used to create field effect transistors.⁹⁵ 

Many of the magnetic examples of $A + B 3 + P 2 S 6$ compounds have antiferroelectric Cu displacements, but the presence of both magnetic order and cooperative ionic displacements within this family of compounds presents promise for realizing multiferroic compounds or heterostructures as well as devices that can be controlled by multiple external fields. Further developments in these materials should provide interesting substrates for coupling to electronic or magnetic materials in van der Waals heterostructures. As these are developed, layer stacking and interlayer interactions in these ferrielectrics should be considered carefully as well, especially since there is evidence of Cu ions displacing into the van der Waals gap in $CuInP 2 S 6$ near room temperature^92,96 and a recent report shows evidence of a change in layer stacking sequence as the sample thickness is reduced below about 100 nm.⁹⁷ 

Two other schemes for replacing Fe while maintaining charge balance can be pointed out: replace the two Fe ions with two monovalent ions and one divalent ion, or replace six Fe ions with four trivalent ions and two vacancies. The former is realized in antiferromagnetic $Ag 2 MnP 2 S 6$⁠, where two Ag ions share a single octahedral cage,⁹⁸ one at the top and one at the bottom, and the latter is realized in, for example, $In 4 / 3 P 2 S 6$⁠. The vacancy ordered compound $In 4 / 3 P 2 S 6$ can be alloyed with ferrielectric $CuInP 2 S 6$ to form in-plane heterostructures by spinodal-like phase separation into submicrometer polar and non-polar domains.⁹⁹ This should also provide an interesting substrate for imposing pseudo-periodic potentials on van der Waals materials or deposited thin films. It remains to be seen if such natural nanostructuring can be achieved by similar routes with the magnetic materials mentioned above.

Materials related to $Fe 3 GeTe 2$ provide another important group of vdW layered magnets. This compound has an intermetallic-like slab of composition $Fe 3 Ge$ between close-packed layers of Te anions.¹⁰⁰ Similar compounds even richer in Fe also form, specifically $Fe 4 GeTe 2$¹⁰¹ and $Fe 5 GeTe 2$^102,103 [Fig. 4(c)], although the Fe content tends to vary around these integer stoichiometric coefficients due to partial occupancy of Fe sites. These materials have several distinguishing features that make them attractive for the study of van der Waals layered magnets and should provide space for continued development of the interplay of magnetism, chemistry, and structure. These features include magnetic ordering transitions that can be near room temperature, metallic conduction important for potential spintronic applications and so that transport can be used to probe the magnetism, and chemical, electronic, and structural flexibility associated with the intermetallic-like Fe–Ge slabs.

This family provides another example of how layer stacking order and disorder in the bulk crystals affect the magnetism and also how this stacking can be controlled. Different stacking sequences and degrees of disorder are seen in crystals grown in different ways and undergoing different thermal histories, and this is reflected in the magnetic behavior.¹⁰⁴ The vacancy-containing, intermetallic-like Fe–Ge slabs are amenable to chemical substitutions and changes in electron count. In addition to tuning Fe vacancy concentration,^105,106 cobalt substitution for Fe has been studied and produces some interesting effects. In $Fe 3 GeTe 2$⁠, cobalt doping produces increased magnetocrystalline anisotropy with a lower Curie temperature,¹⁰⁷ while in $Fe 4 GeTe 2$ and $Fe 5 GeTe 2$⁠, cobalt doping results in a switch from ferromagnetism to antiferromagnetism.^108,109 Importantly, a change in layer stacking sequence (Fig. 5) occurs near the transition between the magnetic ground states, and it appears that both the Co and the stacking change are necessary to stabilize the antiferromagnetic state.¹⁰⁹ 

It seems the opportunities for materials development in the $Fe 3 GeTe 2$-related compounds are only beginning to be explored. In addition to the cobalt doping described above, many other transition metals likely could be substituted for Fe,¹⁰² and substitutions for the other atomic species should also enable control over the magnetic and electrical properties. With the presence of vacancies, split crystallographic sites, chemical substitutions, and different stacking arrangements and stacking disorder, progress in these systems will certainly require careful crystallographic studies if connections between structure and physical properties are to be drawn, or if first principles calculations are to be used for studying these compounds. The electrical/chemical flexibility provides an opportunity for manipulating the magnetism in these materials by controlling electron count, and this is highlighted by the remarkable observation of tripling the Curie temperature of $Fe 3 GeTe 2$ ultrathin flakes, from near 100 to near 300 K, by intercalation of lithium.¹¹⁰ This is consistent with the increase in Curie temperature with increasing iron content in $Fe 3 GeTe 2$ (220 K), $Fe 4 GeTe 2$ (270 K), and $Fe 5 GeTe 2$ (310 K), although this comparison is somewhat crude. This suggests that even higher curie temperatures might be realized in analogs with even more Fe. Although there are still vacancies in $Fe 5 GeTe 2$⁠, it appears that the limiting Fe content is not much larger than 5, and higher iron contents may be realizable only in homologous structures with thicker slabs perhaps incorporating additional Ge layers as well.

A recent and exciting example of combining multiple functionalities within a single vdW material is being realized in naturally heterostructured compounds. A key example is $MnBi 2 Te 4$ [Fig. 4(d)], which combines magnetism in a MnTe layer with topological bands associated with $Bi 2 Te 3$ layers.¹¹² This family includes homologs with additional $Bi 2 Te 3$ layers and larger separation of the magnetic planes as well, like $MnBi 4 Te 7$ [Fig. 6(b)] and $MnBi 6 Te 10$⁠,^113–115 and isostructural compounds also form with Sb.¹¹⁶ Coupling magnetism with topology in these compounds produces the quantum anomalous Hall effect.^25,117 Details of the structure and defects and how these affect the magnetic interactions in these phases are still being sorted out, but it is clear that antisite defects are likely key to understanding both the chemistry and the magnetic structures.^118,119 In fact, some crystals of $MnSb 2 Te 4$ displayed ferromagnetism and others antiferromagnetism depending upon the conditions of their growth. This is being tied directly to subtle differences in the amount of mixing of Sb onto the Mn site and Mn onto the Sb site that are stabilized under different thermodynamic conditions. Figure 6 shows how the alignment of the net moments of neighboring slabs changes from antiferromagnetic to ferromagnetic with small changes in chemical defects.¹¹¹ Such layered vdW compounds with tunable magnetic structures provide an excellent platform for realizing magnetic topological materials, with promise for Weyl semimetals and axion insulators in addition to the quantum anomalous Hall effect, and this provides a promising intersection of vdW and quantum materials, with the potential for controlling magnetic interactions and their impact on quantum transport through careful tuning of chemistry and structure.

In concluding this discussion of transition metal chalcogenides, I will mention briefly the magnetic binary dichalcogenides noted earlier, which have layers like the $FeCl 2$ layers shown in Fig. 1(b). Single layer $MnSe 2$ has been grown by MBE and is reported to be ferromagnetic at room temperature.¹²⁰ The MBE-grown monolayers can be constructed like a single $CdI 2$-type layer [Fig. 1(b)]. However, the compound $MnSe 2$ is not layered; it adopts the cubic pyrite structure. Thus, $MnSe 2$ cannot be exfoliated from bulk crystals. $VSe 2$ does have the $CdI 2$ structure in its bulk crystal form. Monolayers of this phase have also been grown by MBE^23,121 and more recently by exfoliation using chemical rather than mechanical means.²⁴ Ferromagnetic behavior has been reported in the monolayers at room temperature, as predicted by theory;¹²² however, a competing charge density wave ground state is known to be favored in bulk materials and thick flakes^123,124 and has been seen in some monolayers as well.¹²¹ The origin of the ferromagnetism in monolayer $VSe 2$ is still not settled and studies of the charge density wave state in thin samples and crystals are ongoing. It appears that conflicting reports may result at least in part from differences in crystallographic/chemical defects.¹²⁵ The van der Waals layered structure of $CrTe 2$ is metastable and can be formed by chemically removing potassium from between the $CrTe 2$ layers in the stable compound $KCrTe 2$⁠.¹²⁶ Very recently, crystals have been exfoliated to produce ferromagnetic thin flakes.¹²⁷ 

These examples of binary dichalcogenides highlight several examples of opportunities for expanding the palate of magnetic vdW layered compounds and monolayers. This includes “artificial” vdW monolayers grown by interrupted MBE, which is very interesting but less relevant to the current discussions. More relevant are the opportunities to (1) potentially realize magnetic monolayers of materials that are non-magnetic in the bulk (like $VSe 2$⁠) and (2) use non-equilibrium synthesis techniques to produce metastable layered phases in the bulk crystal form (like $CrTe 2$⁠). The first relies on difference in crystallographic structures and instabilities between bulk and ultrathin materials, how they are related to the crystal chemistry, and how they affect magnetic properties. This expands the materials of interest beyond those that show interesting magnetic behavior in the bulk to those whose magnetism may be quenched by structural or electronic instabilities. The second provides prospects for discovering entirely new bulk vdW layered phases that could not be obtained by other means.

I have provided here an overview of several families of materials combining magnetism and van der Waals layered crystal structures along with ideas of how each of these areas may evolve as materials development continues in this chemically, structurally, and magnetically rich space. I have focused on transition metal halides and chalcogenides, because these are the most well developed compounds and ones with which I have the most familiarity. I would note here that this is not intended to be a review, so many important and interesting results are neglected. However, it is clear that many exciting developments have occurred in these materials over the last several years, and it is clear that there is still immense potential for discovery.

Transition metal halides continue to produce interesting results in the 2D limit. As noted above there is a wide selection of candidate compounds known in this area with a wide range of magnetic ground states. A key challenge here is the air sensitivity of many of these materials even in the bulk form and especially in ultrathin specimens. As methods and experience for handling such materials improve, a more diverse set of these compounds should open up for detailed studies. Work on ferromagnetic materials has recently expanded beyond Cr compounds to include $VI 3$⁠, in which the magnetostructural transitions and microstructure seem to be more complex. Still to be explored are those with more complicated in-plane magnetic structures, and the multiferroic behaviors noted among some dihaldes. Mixed anion compounds including oxyhalides and sulfide-halides represent an exciting addition to the family, providing more structural and chemical diversity and potential for smaller bandgaps.

Currently, transition metal chalcogenides represent a broader family than the binary compounds that have been the main focus of research on halides, in the sense that they include a wider variety of structure types as well as multinary compounds. The $MPS 3$-related phases present a relatively large family that includes magnetic ordering and dipolar ordering. This family shows remarkable chemical flexibility with ordered distributions among multiple cations and vacancies, as well as in-plane heterostructures from chemical phase separation. The study of single materials that combine magnetic and dipolar ordering and how these behaviors evolve with thickness provide exciting potential areas for development. As noted, quaternary phases in this family are realized by mixing cations on sites in ternary phases. This provides structural diversity and enables additional functionality like the dipolar order described above. This is somewhat reminiscent of the much more heavily studied oxide double perovskites and should provide plenty of room for the development of new or understudied multifunctional vdW layered compounds. $Fe 3 GeTe 2$-related compounds provide high magnetic ordering temperatures and metallic conduction not likely to be found among halides. These compounds have also been shown to host magnetic skyrmions,¹²⁸ providing an example of how vdW layered materials are impacting a wide range of forefront areas of condensed matter physics. The chemical flexibility of this relatively small group of compounds is only beginning to be explored and should motivate continued interest in these compounds. The multifunctionality and connection to quantum materials makes the natural heterostructured compounds based on $MnBi 2 Te 4$ a particularly active and promising direction for layered vdW magnetic materials for the foreseeable future. The dichalcogenides have provided several exciting surprises including vdW-like monolayers of a non-layered compound, potential magnetism in monolayers of a non-magnetic compound, and the importance of understanding defects and differences between samples of the same material, and a metastable vdW layered material produced by deintercalation. These few examples indicate a huge potential for further development of promising materials yet to be imagined.

A common thread throughout this discussion has been the need for careful crystallographic and structural studies to understand not only the behavior of bulk crystals but also how the structural details may evolve as chemical modifications are made or as thicknesses approach the 2D limit. With weak interlayer interactions, the stacking relationships in these materials are particularly susceptible to perturbations, and they play an important role in determining the magnetic structures. Learning how to control or manipulate these stacking arrangements, as was done with applied pressure in $CrI 3$⁠,⁶⁰ presents a major opportunity in this area. This also points to a key challenge, especially relevant in studying restacked materials as opposed to multilayers produced by exfoliation. The properties of bilayer sample produced by restacking two monolayers of a single material can depend upon how the two layers are translated with respect to one another on the scale of one unit cell. This is, of course, in addition to the rotational degree of freedom that can produce moiré structures currently of great interest and potential.^129,130 Controlling and monitoring these structural relationships will be key to not only understanding observed behaviors, but unlocking new functionality and physics. In addition, the observation of reconstruction on the surface monoloayer of RuCl₃ crystals⁴⁶ highlights the oversimplification of viewing these compounds as essentially non-interacting monolayers, and further motivates careful studies of how structure may evolve in exfoliated materials.

This work was supported by the U.S. Department of Energy, Office of Science, Basic Energy Sciences, Materials Sciences and Engineering Division. The author gratefully acknowledges ongoing discussions and interactions with colleagues and collaborators, in particular, B. C. Sales, D. Mandrus, A. F. May, J.-Q. Yan, V. R. Cooper, X. Xu, and D. Xiao.

This paper has been authored by UT-Battelle, LLC, under Contract No. DE-AC05-00OR22725 with the U.S. Department of Energy (DOE). The U.S. Government retains and the publisher, by accepting the article for publication, acknowledges that the U.S. Government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this paper, or allow others to do so, for U.S. Government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan).

Data sharing is not applicable to this article as no new data were created or analyzed in this study.