Transition‐Metal Chalcogenide, FeTe: Unveiling Molecular Mechanism of Phase‐Selective Synthesis

ABSTRACT

Phase‐selective chemical vapor deposition synthesis of two‐dimensional (2D) transition‐metal chalcogenides (TMCs) has garnered broad interest, yet its crucial dependence on the growth atmosphere is not understood. The chain of reactions transforming precursors into 2D crystals remains elusive. Focusing on iron telluride—a promising material for quantum and spintronic devices due to its phase‐dependent topological superconductivity and magnetism—our first‐principles calculation elucidates the phase‐selective growth of tetragonal FeTe (t‐FeTe), including its thermodynamic and kinetic shapes. We identify the FeTe₄Cl as the immediate precursor—a molecule reacting directly at the edge of the expanding crystal. Due to the stoichiometric mismatch of this gaseous precursor and the product‐crystal, the growth mechanism through kink propagation requires an additional step—the edge cleaning, eliminating excess atoms after FeTe₄Cl attachment. Based on this, we further demonstrate that the experimental levers, namely Te‐limited condition and H₂ supply, enable edge cleaning, thereby promoting nanosheet lateral, in‐plane expansion and high t‐FeTe phase purity. Conversely, in a Te‐rich environment, off‐plane nuclei become favored, biasing nucleation toward nonlayered phase (hexagonal h‐FeTe). This work explains how the experimental atmosphere affects growth dynamics of t‐FeTe and provides valuable guidelines for optimizing synthesis parameters of other TMCs.

First‐principles calculations and slow‐growth ab initio molecular dynamics reveal CVD phase selectivity in 2D tetragonal FeTe. FeTe₄Cl is identified as the edge‐reactive precursor; its stoichiometry mismatch with FeTe requires H₂‐assisted edge cleaning to sustain lateral growth. Under Te‐rich conditions, edge reactivity is altered to favor off‐plane nuclei, biasing growth toward hexagonal FeTe and explaining experimental bifurcation.

1. Introduction

Atomically thin two‐dimensional (2D) transition‐metal chalcogenides (TMCs) featuring diverse polymorphic phases have garnered significant interest due to their exceptional physicochemical properties and wide‐ranging applications [1, 2, 3, 4, 5, 6, 7]. Precise control over synthesized phases enables tailoring of material properties, leading to enhanced functional performance. Among 2D TMCs, iron chalcogenides exhibit unique electronic structures [8], and are particularly attractive due to their potential in areas of magnetism [9, 10], superconductivity [11, 12], and emergent quantum phenomena [13, 14, 15]. FeTe stands out within this family with exceptional phase‐dependent properties, making it a prime platform for applications in quantum [16, 17, 18, 19] and spintronic devices [20]. FeTe exhibits two polymorphs—layered tetragonal (t‐FeTe) and non‐layered hexagonal (h‐FeTe) phases—with drastically different electronic and magnetic behaviors. Its common tetragonal phase is antiferromagnetic, with superconductivity emerging through heteroatom‐doping or pressure‐modulation [16, 17, 21, 22]. Additionally, reports indicate the electrocatalytic activity of t‐FeTe, attributed to its strong metallic character [23, 24]. The recent discovery of a metastable hexagonal phase with distinct, weak magnetism further underscores the profound influence of crystal polymorphism on its magnetic properties [16]. Experimentally, phase‐controllable growth of t‐FeTe and h‐FeTe has been demonstrated, highlighting that these polymorphs are not merely structural variants but functional building blocks for 2D magnetism and heterostructures [25]. Phase purity is essential because mixed t‐ and h‐domains can complicate the interpretation of magnetic signatures and obscure interfacial phenomena. For example, t‐FeTe heterostructures have been widely explored as potential hosts for topological superconductivity, where interfacial properties depend sensitively on the atomic‐scale structure, stoichiometry, and quality of FeTe layer [26, 27, 28]. Accordingly, controlled synthesis of phase‐pure FeTe is critical for reliably harnessing its functional potential.

Tremendous experimental efforts are devoted to achieving controlled synthesis of FeTe with tunable phases, including mechanical/liquid exfoliation [29, 30], molecular beam epitaxy (MBE) [31, 32], and chemical vapor deposition (CVD) [16, 25, 33, 34, 35, 36]. In particular, the bottom‐up approach of CVD has gained increasing attention due to high tunability, controllability, and scalability [37, 38, 39]. Recent advances in synthesizing FeTe nanosheets by CVD have utilized iron dichloride (FeCl₂) and tellurium (Te) powders as solid precursors (at ∼700−1100 K) with H₂ carrier gas. Experimentally, phase purity was achieved through the precise control of growth atmosphere, yielding t‐FeTe under Te‐limited conditions and h‐FeTe under Te‐rich conditions. Similar tendencies are observed for a variety of modern 2D materials (e.g., S or Te pressure controlling the quality of TMDs [40, 41], H₂ pressure triggering a phase shift between cBN and hBN [42]). Beyond the Te chemical potential per se, growth temperature in multi‐zone CVD also serves as a modulating parameter for phase selectivity, by affecting the partial pressure of precursors: increasing the Fe‐source temperature (at fixed Te supply) shifts the growth toward a higher Fe:Te ratio that favors t‐FeTe, while increasing the Te‐source temperature drives Te‐rich conditions that promote h‐FeTe [34]. While solid precursors are known to sublimate into gaseous reactants, the in situ identification of reaction intermediates and the accurate modeling of high‐temperature reactions under realistic conditions remains challenging [43]. A detailed picture of the growth mechanism is crucial for explaining experimental studies that demonstrate phase selectivity by modulating the growth atmosphere.

By linking high‐temperature gas‐phase chemistry to edge reaction kinetics, we provide an atomistic mechanism for t‐FeTe phase selectivity under Te‐limited conditions and show that the CVD atmosphere—Te and H₂ supply—controls edge reactivity and can bias growth toward off‐plane nucleation under Te‐rich conditions (consistent with h‐FeTe). These insights offer practical guidance to improve phase purity, which is essential for FeTe‐based magnetic and heterostructure devices. The overall gaseous reaction landscape and the nucleation process are explored using first‐principles calculations and slow‐growth constrained ab initio molecular dynamics (SG‐AIMD) simulations which capture the free‐energy profile and the corresponding free‐energy barrier (see the Supporting Information for details).

2. Results and Discussions

2.1. Solid Precursor (FeCl₂ and Te) Sublimation

In the CVD synthesis of FeTe, the FeCl₂ and Te powders serve as solid precursors, with crystal growth occurring on SiO₂/Si substrates with H₂ carrier gas [16]. At high temperatures, these precursors sublime into gaseous species in the reaction chamber. To mimic the CVD environment and identify gaseous reactants, Te (100) and FeCl₂ (001) surfaces were initially chosen to study the sublimation by conducting AIMD simulation at 1500 K. Note that this is not to exhaustively enumerate all the possible gaseous products (which is generally infeasible on AIMD timescales), but to identify the possible molecules that emerge and remain stable under assumption that the most probable reactants would be observed in relatively short simulations. It shows that tellurium sublimates to yield Te₂ as the primary species (Figures 1a and S2a), consistent with experimental measurements of vapor over elemental Te being dominated by Te₂ [44]. While FeCl₂ sublimation generates numerous larger iron‐chloride clusters Fe_xCl_y, and the Fe₂Cl₄ emerges as the smallest stable species at an early stage (Figure 1b). Furthermore, extended simulations on an isolated Fe₂Cl₄ molecule (∼15 ps) confirm its structural integrity (in Figure S2b), aligning with experimental characterization of the gaseous Fe₂Cl₄ as a stable molecule [45, 46]. Thus, Te₂ and Fe₂Cl₄ act as the main species in subsequent high‐temperature reactions.

FIGURE 1.

AIMD simulations of solid‐precursor sublimation and subsequent reactant collisions (initial configuration, left; representative snapshot, right). (a) Te(100) sublimation yielding Te₂. (b) FeCl₂(001) sublimation yielding Fe₂Cl₄. (c) Collision between Te₂ and Fe₂Cl₄ reactants without H₂, leading to rapid Fe–Te bond formation and polymerization into multinuclear Fe_x‐containing Fe_xTe_yCl_z chains. (d) Collision of Te₂ and Fe₂Cl₄ in H₂ atmosphere, where polymerization is suppressed and mononuclear Fe‐containing FeTe₄Cl forms. Brown, yellow, green, and white spheres represent Fe, Te, Cl, and H atoms, respectively; key products are highlighted by orange boxes or arrows.

Following the identification of the main reacting molecules, we performed a preliminary exploration of the gas phase reactions through unrestrained AIMD simulations. While insufficient to resolve the precise, detailed atomistic mechanisms and free‐energy barriers, these simulations readily reveal the high affinity between Te and Fe atoms, leading to rapid Fe–Te bond formation and subsequent aggregation into multinuclear Fe_xTe_yCl_z chain‐like species in the absence of H₂ (Figure 1c). By contrast, the inclusion of H₂ molecules, which was experimentally found to be critical for the t‐FeTe high‐quality growth [33], prevents the formation of larger chains and the formed species shifts toward smaller FeTe₄Cl molecules (Figure 1d), which will later be shown to be an immediate growth precursor—molecule undergoing a direct reaction on the edge of the crystal—in the growth of t‐FeTe. Importantly, within the limited AIMD timescale, we do not observe direct evidence of H₂ chemically participating in the elementary bond‐forming/breaking events, suggesting that its role at this stage may primarily involve suppressing extensive polymerization rather than straightforward dechlorination.

2.2. Gaseous Reaction Network

To identify a detailed reaction network, we subsequently employed the SG‐AIMD approach, which has been shown to enable targeted, computationally feasible exploration of possible reaction pathways [47]. Starting from the gas‐phase Fe‐containing reactant Fe₂Cl₄ (IS, initial state), we systematically evaluate its elementary reactions with the precursor Te₂ and the additive H₂. Figure 2 summarizes the main pathways along with their free energies and kinetic barriers (see the details in Figures S3−S10). For each intermediate, we consider Te₂‐addition pathways (pink and orange curves in Figure 2a) and H₂‐addition pathways (light‐blue curves). The two Te₂‐addition pathways distinguish formation of mononuclear FeTe _x Cl _y clusters (pink curves) from collision‐driven aggregation into multinuclear Fe_xTe_yCl_z species (orange curves). When multinuclear Fe‐containing intermediates form, their monomerization reactions are examined (green curves).

FIGURE 2.

SG‐AIMD free energy profiles of gas‐phase reactions from Fe₂Cl₄ (IS) to the fully‐tellurized FeTe₆ (E1). (a) Pathways are classified by color: H₂ addition (light blue), Te₂ addition forming mononuclear species (pink), Te₂ addition leading to aggregation (orange), and monomerization (green). Solid lines indicate relatively favorable routes, whereas dotted lines denote thermodynamically or kinetically disfavored routes. Insets show the structures for IS, the building unit FeTe₄Cl (B1), and E1; all intermediates are provided in Figure S3. (b) Selected reaction map from A1 to B1 and E1. Reaction byproducts are omitted for clarity. Green, brown, and yellow spheres represent Cl, Fe, and Te atoms, respectively.

Consistent with our unrestrained AIMD simulations (no direct H₂ participation in reaction observed), the quantitative SG‐AIMD results show that all reactions related to chloride removal by H₂ molecules (light‐blue curves in Figure 2a) exhibit much higher barriers (>1.30 eV) and are endothermic, precluding their products from further consideration. Within the first reaction step (IS → A), Te₂ attachment to iron centers of Fe₂Cl₄ is slightly exothermic with a low barrier of 0.12 eV, favoring Fe₂Te₂Cl₄ A1 formation. For the further consecutive reactions (A → B), only the pathway generating Fe‐monomeric species displays the overall‐exothermicity (−0.25 eV, pink curve) with a barrier of 0.14 eV, yielding the stable products of monoiron‐tetratelluride‐monochloride FeTe₄Cl (B1 in Figure 2, see the details in Figure S6a) and FeCl₃. Further tellurization (B → C) of higher energy products also results in forming the FeTe₄Cl molecule, identical to B1 (the difference in free energy originates from varying byproducts, see the detailed structures in Figure S7b,c). At the same time, the tellurization of the FeTe₄Cl B1 structure itself only leads to the isomerization forming a higher‐lying C1 conformer, with the sole chloride ligand coordinating with tellurium (see the details in Figure S7a). Further tellurization of C1 reverses the reaction, leading to the re‐production of the lowest‐lying B1 (Figure S8). As a result, only pathways with significantly higher barriers (through C4) do not lead to the direct formation of B1, indicating it being a dominant product in the gas phase, with the most favorable pathway shown with the pink curves IS→ A1→ B1 → C1 (→ B1).

The arrows in Figure 2b trace the evolution of the main species through the network, indicating that several lower‐lying routes increase the local concentration of FeTe₄Cl molecules (B1, in pink). Notably, previous theoretical studies on crystal growth assumed fully sulfurized or tellurized metal complexes as building units [34, 48, 49]. As shown in Figure 2a, a fully tellurized FeTe₆ E1 in our case can be generated via five subsequent steps; however, forming such clusters is highly improbable due to the significant kinetic barriers. Overall, the above results suggest that FeTe₄Cl emerges as the lowest‐lying and abundant species in the considered network. Moreover, the low barriers for tellurization reactions align with experimental synthesis, where this nanosheet can be readily produced upon annealing under precisely controlled tellurium supply [34].

To further demonstrate that FeTe₄Cl can emerge as a growth‐relevant building unit that transiently forms via multiple collision pathways in the CVD environment, Figure 3 presents the coordination‐number evolution along representative high‐temperature trajectories, where transient Fe–Te–Cl species are identified by bond‐length cutoffs, a common geometric criterion in MD trajectory analysis [50, 51]. Our analysis focuses on the evolution of the coordination number at the selected Fe center (in brown, Figure 3). Generally, Fe–Te–Cl species fragment upon Te₂ addition, yielding varying byproducts (e.g., FeCl₃ and Te₂Cl). For the selected brown Fe center, the dominant pathway of transient clusters is: FeTeCl₃→FeTe₂Cl₃→FeTe₃Cl→FeTe₄Cl, where chloride coordination is progressively converted into tellurium coordination, ultimately leading to FeTe₄Cl formation.

FIGURE 3.

The coordination number (CN) evolution for the selected Fe center (brown) during SG‐AIMD simulations in two reactions: (a) Fe₂Te₂Cl₄ (A1) +Te₂ → FeTe₄Cl + FeCl₃ and (b) Fe₂Te₄Cl₄ (B3) +Te₂ → FeTe₄Cl‐FeTe₂Cl₃. The insets show transient molecular structures of Fe_mTe_nCl_k species along the trajectories. Green, brown, orange, and yellow balls represent Cl, the selected Fe, the other Fe, and Te, respectively. Bond cutoffs for calculating coordination numbers are set to be 3.5 Å for Fe–Te bonds, 2.5 Å for Fe–Cl bonds, and 2.5 Å for Fe–Fe bonds, accounting for molecular thermal fluctuations.

It is important to note that the presented reaction network is obtained through the free energy calculations, while considering only potential energy, as is commonly done in the field, leads to drastically different results (Figure S11). These free energy calculations account for the entropy changes at high temperatures, favoring the early formation of mono‐Fe and overall smaller molecules. Conversely, the potential energy‐based approach produces larger molecules and delays Fe‐monomerization.

2.3. Crystal Edges in CVD Growth

Ultimately, the gas reactions produce a set of molecules with corresponding concentrations that react at the edge of the expanding t‐FeTe nanoflake, leading to the crystal growth. Yet before we can consider this process directly, we determine the structure of the crystal edge using the nanoribbon model and constructing two zigzag nanoribbons (Fe‐terminated zigzag edge, ZZFe; Te‐terminated zigzag edge, ZZTe) and one armchair nanoribbon (armchair edge, AC). Figure 4a illustrates the formation energies of these three pristine edges, shown as solid lines, and their corresponding Wulff‐constructed equilibrium shapes as functions of the tellurium chemical potential Δµ_Te. Polar plots of pristine edge formation energies versus chiral angles are presented in Figure S12. Notably, the nanosheet shape remains predominantly square under a wide range of Δµ_Te, including Fe‐rich (−3.15 eV < Δµ_Te < −3.07 eV), Te‐rich (−0.33 eV < Δµ_Te < 0 eV), and moderate (−2.52 eV < Δµ_Te < −0.88 eV) conditions. Only in narrow Δµ_Te windows (−3.07 eV < Δµ_Te < −2.52 eV; −0.88 eV < Δµ_Te < −0.33 eV), the crystal shape becomes a truncated square due to the energy degeneracy between zigzag and AC edges. Given t‐FeTe synthesis conditions by controlling tellurium supply, the AC configuration is the thermodynamically optimal shape of t‐FeTe islands as it maintains the observed square morphology across the widest effective Δµ_Te ranges in accord with the experimental facts [16, 25, 33, 34, 35, 36].

FIGURE 4.

The edge formation energies and equilibrium shapes of t‐FeTe flakes as a function of tellurium chemical potential Δµ_Te: (a) edge formation energy for various edge terminations: Fe‐terminated zigzag (ZZFe, in green), Te‐terminated zigzag (ZZTe, in pink), pristine armchair (AC, in light blue), and three passivated‐AC edges of Te‐terminated AC (AC‐Te), H‐passivated AC (AC‐H), and H₂‐passivated AC (AC‐H₂) edges. Insets show the corresponding equilibrium flake shapes at selected Δµ_Te. The Te‐rich limit is referenced to bulk trigonal Te (Δµ_Te = 0 eV), and the Te‐poor (Fe‐rich) limit is set by the bulk bcc Fe reference (Δµ_Te = −3.15 eV). (b–d) The optimized nanoribbon structure with AC‐H₂, AC‐H, and AC‐Te edges, respectively. All the snapshots are shown as projections onto the xy plane. Brown, yellow, and white spheres represent Fe, Te, and H atoms, respectively.

However, during the CVD process, diverse gas‐phase species, including additives, interact not only with each other but also with the crystal edge, potentially altering its configuration. In pristine AC t‐FeTe, the presence of exposed, under‐coordinated iron centers on the edge under realistic CVD conditions can lead to new configurations. This prompts us to consider the possibility of edge (self)‐passivation by atoms abundant in the environment, which was shown to play a significant role in the growth of MoS₂ [52], graphene [53], and h‐BN [47]. By considering the presence of additives H₂ and precursors (FeCl₂ and Te) in CVD growth of FeTe, we compare the formation energies of three modified AC edges: AC edges passivated by one or two H atoms (AC‐H and AC‐H₂​) and Te‐terminated AC edge (AC‐Te). The edge formation energies of these three modified AC edges are represented by dashed lines in Figure 4a, and their edge morphologies are shown in Figure 4b−d. It is found that AC‐H₂ or AC‐Te edges yield more stable structures compared to pristine AC edges. Furthermore, we find that pristine edge readily converts to the Te‐terminated configuration upon reaction with the identified molecular precursor FeTe₄Cl (Figure S13). Similarly, the AC‐H₂​ edge under exposure to FeTe₄Cl releases H₂ and transforms to the Te‐termination at 1500 K, even without undergoing a complete t‐FeTe crystal unit attachment. The free energy profile of this process (Figure S14), which includes the temperature effect, reveals that Te connectivity with iron sites is slightly more favorable (−0.12 eV) than H. These findings highlight the importance of accounting for additives and entropy effects when modeling the CVD t‐FeTe growth.

2.4. Crystal Growth Mechanism

Utilizing the identified building unit and edge configurations, we simulate the sequential attachment of the building unit, FeTe₄Cl, to the edge of the expanding t‐FeTe lattice. Here, the substrate effect on growth is assumed to be minimal and is not considered due to the inert and largely non‐catalytic nature of insulating substrate SiO₂/Si, so that the elementary growth steps are primarily dictated by gas‐phase species reacting at t‐FeTe edges (vapor‐solid‐like growth), rather than by substrate‐mediated growth [53]. To represent the growing flake, we employ three edge models—AC‐Te and ZZTe edges (in Figure 5) and AC‐H₂ edge (in Figure S19)—further exploring their reaction kinetics. We exclude Cl‐terminated edges from further consideration during growth because the free‐energy calculations (Figure S15) show that a transiently Cl‐involved edge can be readily dechlorinated during building‐unit incorporation, making a stable Cl‐terminated edge unlikely, being consistent with the absence of detectable Cl residue in experiments [16].

FIGURE 5.

Mechanism of t‐FeTe crystal growth along AC‐Te and ZZTe edges. (a) Free‐energy profiles for crystal growth, with the row‐initiation from edge (E → N) separated from the propagation and cleaning (N* → P → C). These two schematic processes are shown in Scheme S1. The pink and green curves correspond to growth at the AC‐Te and ZZTe edges, respectively. (b,c) Representative snapshots of reactants and intermediates for the AC‐Te and ZZTe pathways, respectively. All snapshots are shown as projections onto the xy plane. For clarity, atoms belonging to the FeTe₄Cl building unit are highlighted with larger spheres. The detailed free‐energy profiles are provided in Figures S16–S18.

The nucleation of a new row of atoms, which is viewed as an energetic bottleneck in the 2D crystal growth [49], begins when the FeTe₄Cl reacts with the clean edge of t‐FeTe monolayer (E→N in Figure 5). Note that the process (E→N) here creates the first unit at the edge initiating a new atomic row, produced by kink propagation. The growth of such an atomic row does not follow classical nucleation theory [54, 55] and should not be, strictly speaking, called a nucleation but better initiation (of the next row). However, the literature mostly follows the analogy and uses the same term, nucleation [49, 56]. The calculated barriers for this step are 0.52 eV for AC‐Te edge and 0.18 eV for ZZTe edge. For the ZZTe edge, while post‐nucleation growth barriers are on par with those for the AC, the nucleation barriers are significantly lower, enabling ZZTe sides to propagate much faster; consequently, ZZTe facets are kinetically eliminated from the crystal shape. Once the kink site forms (configurations N1 and N2), FeTe₄Cl incorporation becomes more exothermic (by 0.1−0.6 eV) with low barriers (0.06−0.17 eV, see the configurations P1 and P2), promoting rapid kink propagation and lateral expansion. Note that growth along the AC‐H₂ edge is excluded because it remains the least favorable pathway both thermodynamically and kinetically (see the details in Figure S19). For the AC‐Te edges, reaction with FeTe₄Cl inserts one FeTe₂ moiety at the kink, while releasing a Te₂Cl byproduct and also leaving a dangling Te atom, an excess, compared to the pre‐reaction configuration. While for the ZZTe edge, the reaction leaves dangling Te₂, Te or Cl species along the edge, with greater exothermicity (−0.6 eV). Importantly in all cases, due to stoichiometry‐misfit of the product‐crystal FeTe (Fe:Te = 1:1) and the FeTe₄Cl (Fe:Te = 1:4) unit, when FeTe₄Cl attaches at the edge, only part of this building unit is incorporated into the lattice, while excess Te and Cl remain at the edge near the kink, producing a kink site significantly altered compared to the previous edge configuration. This recurring “damage” necessitates an additional growth step—the edge cleaning—to ensure seamless kink propagation and defect‐free crystal growth, which is reminiscent of the technique of deliberate inclusion of etching agents in the growth chamber to improve the quality of the grown 2D materials [57, 58]. We have found that the hydrogen assists in removing excess Te as TeH or ClTeH (configurations C1 and C2) to restore the initial kink configuration. This endothermic cleaning displays two sequential barriers, which can be suppressed under sufficient (yet moderate, not to trigger etching) H₂ chemical potential, promoting lateral growth. It appears that for most edges, the excess H can remain attached to Fe centers until it is either consumed in subsequent edge cleaning steps or easily desorbs as H₂ after annealing. Previous experimental findings have reported the role of such hydrogen atoms in suppressing off‐plane growth by passivating the surface active sites and facilitating the lateral expansion along the edge [48].

The experimental evidence emphasizes the necessity of the hydrogen presence, as well as the restriction of the Te amount in the system, for the continuous high‐quality 2D growth of t‐FeTe crystals [16, 25, 34, 35, 36], as opposed to the formation of h‐FeTe when those conditions are not met; yet, the mechanistic reasons remain unknown. Leveraging the growth mechanism described above, we conjecture that an excess of Te and scarcity of H₂ in the chamber can slow down or completely cease the kink cleaning. Using the AC‐Te edge as a representative case, we evaluate the possible effects of this on the growth process. First, the calculations show that out‐of‐plane bonding of the FeTe₄Cl unit with the pristine t‐FeTe surface, away from the edge, is energetically unfavorable (Figure 6a, s), preventing 3D nucleation there despite the abundance of these adsorption sites. Since the s site lies far from the edge on the terrace, its energetics are weakly affected by the change from pristine −FeTe to Te‐saturated −FeTe₂ edge; thus, it is shown only once in Figure 6a as a representative off‐plane terrace site. In contrast, adding FeTe₄Cl to the pristine edge‐kink (k) is energetically favorable, with a low barrier. The precursor can also vertically adsorb on t‐FeTe layer surface in the kink vicinity (sk), with slightly higher energy; however, the bonding is loose, allowing the molecule to eventually tumble to the nearby kink site (Figure S20). It is essential to mention that the release of excess atoms into the gas phase during the kink propagation reaction makes this process effectively irreversible.)

FIGURE 6.

Incorporation of FeTe₄Cl building unit into various t‐FeTe AC edges. (a) SG‐AIMD free energy profiles at 1500 K for pristine—FeTe edge under high Fe:Te ratio (left) and the Te‐saturated—FeTe₂ edge under low Fe:Te ratio (right). The insets show the definitions of the reaction sites: k and k′ correspond to 2D in‐plane incorporation at kink sites, while sk, sk′, se′, and s correspond to 3D‐like off‐plane incorporation at surface or edge sites. The black short segment indicates initial states (IS) with free FeTe₄Cl away from the lattice, blue and pink ones denote 2D‐like and 3D‐like incorporated final states (FS). (b) Representative incorporated structures corresponding to the sites in panel a, shown as projections onto the xz plane. The host‐lattice atoms are shown smaller than those in FeTe₄Cl, and the leaving atoms not bonded to the lattice are omitted for clarity.

As was discussed above, the edge reaction with the precursor can reproduce the kink modified by excess Te atoms (AC‐Te₂, Figure 5b P2). If conditions (excess of FeTe₄Cl or scarce H₂) prohibit efficient edge cleaning, the subsequent FeTe₄Cl would interact with the AC‐Te₂, not the pristine kink k. Such a reaction, while reproducing the AC‐Te₂ kink site (Figure 6a, k′), has a higher barrier and energy rise, in contrast to the exothermicity found for k. Moreover, we find that adding FeTe₄Cl off‐plane, in the direction normal to the layer near such a Te₂‐passivated edge, becomes more energy‐favorable and can produce the off‐plane 3D nucleus‐type configuration under Te‐rich conditions (sk'; see the details in Figure S21). Such advantage of sk' is due to the change of edge structure under Te‐rich conditions: the Te‐saturated—FeTe₂ edge becomes coordinated by excess Te, so there is larger steric hindrance during the in‐plane incorporation at k′, as illustrated in Figure S22a. By contrast, FeTe₄Cl can approach from near the surface at sk′ with much less steric congestion (Figure S22b). To establish a mechanistic insight into the phase selectivity, we compare the coordination of this newly attached Fe center with that in the competing h‐FeTe phase. As shown in Figure S23, their analogous out‐of‐plane structures illustrate the favorable trend of 3D nucleation and off‐plane growth of h‐FeTe consistent with experimental facts [16, 25, 34, 35, 36]. Such 3D‐like structures leading to the nonlayered phase can also be generated in the sites near the edge (se′) under Te‐rich conditions. Note that we only consider the nucleation that may initiate the nonlayered phase; subsequent processes of building‐unit fusion with the crystal lattice, which encompass kink formation and step propagation, are beyond the scope of this work. One concludes that at the higher chemical potential of Te, saturated−FeTe_n (n ≥ 2) moieties near the kink hinder the lateral edge propagation but rather promote FeTe₄Cl off‐plane attachment to the surface, potentially leading to the nucleation of defects or nonlayered structures, such as the h‐FeTe phase observed in experiments. This mechanism underscores the critical importance of a limited tellurium supply and sufficient H₂ flow for the phase‐controlled growth of high‐quality monolayer t‐FeTe.

3. Conclusions

The phase‐controllable growth of t‐FeTe under Te‐limited conditions, an important 2D‐magnetic, also representative of transition‐metal monochalcogenides, is elucidated. The gaseous reaction landscape is traced from the Fe‐dimeric precursor Fe₂Cl₄ to the fully tellurized products. The results reveal that high‐temperature microenvironments promote decomposition of Fe–Te–Cl species into Fe‐monomers, while H₂‐mediated gaseous reactions are thermodynamically and kinetically unfavorable. The FeTe₄Cl molecule emerges as the lowest‐lying species in the considered gas‐phase reaction network and serves as the growth building unit to react at the crystal edge. We have found that armchair edges are favored both thermodynamically and kinetically at CVD temperature, further discovering a preference toward edge termination with Te atoms, akin to the graphene Klein edge. The growth mechanism mostly follows the typical process for 2D materials, including the nucleation of the new row of atoms and its propagation through building block attachment to the kink site. However, due to the composition of the precursor molecule, the attachment leaves extra Te atoms near the kink site, significantly altering its reactivity. In this state, the lateral expansion can be hindered considerably. To prevent this, an additional step in the growth cycle—the edge cleaning—takes place under H₂‐modulation and Te‐moderation, restoring the kink site configuration and enabling continuous lateral growth. By contrast, the nucleus can rapidly switch to off‐plane 3D growth under Te‐rich conditions, and nucleation of the nonlayered h‐FeTe becomes likely, explaining the bifurcation‐events experimentally observed in phase‐selective FeTe synthesis. This discovered mechanism explains the crucial roles of the growth atmosphere in sustaining high‐quality 2D growth, and can extend beyond FeTe to the CVD growth of other transition‐metal chalcogenides on insulating substrates.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: anie72781‐sup‐0001‐SuppMat.pdf.

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article.