betaFeedback

WorksPhysics and Astronomy

Anisotropy driven critical phenomena in a van der Waals ferromagnet

Abstract

We report a detailed investigation into the field-dependent critical behavior of the van der Waals ferromagnet Fe3GeTe2\mathrm{Fe_3GeTe_2} to elucidate how applied magnetic fields (BB), strong magnetocrystalline anisotropy, and emerging other energy scales modulate its critical behavior. Utilizing the recently developed method, viz., the average of normalized slope (ANS), for analysis of critical behavior, we evaluate the critical exponents (ββ, γγ, and δδ) across distinct field windows up to 7 T\rm{7\,T}. For BB applied along the crystal's easy axis (B∥cB\parallel c), the system exhibits long-range interactions dominated by competing double-exchange and itinerant Ruderman-Kittel-Kasuya-Yosida (RKKY) mechanisms, which is systematically suppressed with increasing BB to converge toward the long-range 3D Ising universality class (β∼0.330β\sim0.330, γ∼1.00γ\sim1.00). Conversely, under an in-plane hard-axis configuration (B⊥cB\perp c), the critical behavior study exhibits a sub-unity regime (γ<1γ<1) across the measured range of BB. This behavior is attributed to the intense competitive interplay between the external Zeeman energy and strong out-of-plane magnetocrystalline anisotropy. Our findings demonstrate that critical behavior in nontrivial van der Waals magnets is fundamentally modulated by the orientation of BB and the strength of magnetocrystalline anisotropy, demonstrating the necessity of BB-dependent critical behavior investigation over conventional frameworks.

Introduction

The investigation of critical behavior in low-dimensional van der Waals (vdW) materials provides vital insights into the mechanisms driving the magnetic ordering, the spatial range of exchange coupling, and the underlying magnetic universality class (UC) of a system [1, 2]. A UC formally defines the space-dimensionality (dd) and spin-dimensionality (nn) of the local magnetic moments. In the vicinity of a continuous paramagnetic phase transition (TCT_{\mathrm{C}}), the thermodynamic response functions—such as susceptibility (χ\chi) and specific heat—exhibits power-law scaling governed by a set of critical exponents (CEs) [1–11]. These exponents are uniquely determined by the symmetry of the order parameter, the dimensionality of the system, and the range of the underlying exchange interaction, classifying the phase transition into standard UCs such as the Ising (n=1n = 1), XY (n=2n = 2), Heisenberg (n=3n = 3), or mean-field models [1, 2, 9–11]. Ideally, the Heisenberg UC describes the critical behavior of isotropic magnetic systems, the XY UC explains that of magnetic systems with planar anisotropy, and the Ising UC applies to that of magnetic systems with uniaxial anisotropy [4].

Conventionally, these CE values are extracted from the magnetic field (BB)-dependent magnetization (MM-HH) isotherms near TCT_{\mathrm{C}} using the conventional modified Arrott plot (MAP) analysis or the Kouvel-Fisher (KF) method [12–17]. While these methodologies are strictly applicable to trivial magnetic systems, such as those governed by the Hamiltonian (H=−J∑i=1nS⃗i⋅S⃗jH = -J \sum_{i=1}^{n} \vec{S}_{i}\cdot\vec{S}_{j}), they frequently yield inconsistent results when applied to a nontrivial magnetic system [23, 24, 25, 26, 27] [9, 18, 19, 20, 21, 22]. In 2022, Chauhan et al. [28] proposed a modified iteration method to address the erroneous CE values, obtained from the conventional critical analysis process, in nontrivial magnetic systems by estimating separate sets of CE values using MM-HH isotherms for T≤TCT \leq T_{\mathrm{C}} and T≥TCT \geq T_{\mathrm{C}}. Subsequently, Chauhan et al. [9] introduced intercept analysis of standard short-range (SR) and mean-field UCs alongside the average of normalized slope (ANS) method to evaluate the reliability and dominant exchange interactions of critical behavior analyses, validating these approaches on trivially treated Ni and nontrivial Gd.

The vdW ferromagnet Fe3GeTe2\mathrm{Fe}_{3}\mathrm{GeTe}_{2} (FGT) represents a complex nontrivial magnetic system [29, 30, 31, 32, 33, 34, 35, 36, 37] [10, 11, 18]. It features a rich interplay of localized and itinerant electron magnetism, a strong out-of-plane (OOP) uniaxial magnetocrystalline anisotropy, and an intrinsic competition between interlayer ferromagnetic and low-temperature antiferromagnetic exchange channels. Consequently, prior efforts to investigate critical behavior have led to conflicting conclusions [34] [10, 11]. An early study by Liu et al. [11] reported CE values (β∼0.327\beta\sim0.327, γ∼1.079\gamma\sim1.079, and δ∼4.261\delta\sim4.261) that did not belong to any standard UC, attributing the discrepancy to a three-dimensional (3D) long-range (LR) magnetic coupling that decays with distance rr as J(r)∼r−4.6J(r) \sim r^{-4.6}. Notably, CE values do not seem to have to belong to a standard UC when investigating the critical behavior of a non-trivial magnetic system [9]. Investigations on Fe-deficient variant (Fe3−xGeTe2\mathrm{Fe}_{3-x}\mathrm{GeTe}_{2} with x≈0.36x \approx0.36) yielded heavily altered CE values (β∼0.372\beta\sim0.372, γ∼1.265\gamma\sim1.265, and δ∼4.50\delta\sim4.50), highlighting a strong sensitivity to local vacancies and showing 3D Heisenberg UC with weaker LR magnetic coupling that decays with distance rr as J(r)∼r−4.89J(r) \sim r^{-4.89} [10]. More recently, reports by Anikta et al. [34] argued for a two-dimensional Heisenberg exchange with LR interaction (Table I) and discussed the violation of the HohenbergMermin-Wagner theorem [38, 39]. N. D. Mermin and H. Wagner [38] proved theoretically that it is impossible to stabilize long-range antiferromagnetism or ferromagnetism in one- or two-dimensional systems hosting isotropic Heisenberg models. Alternatively, this implies that long-range antiferromagnetism or ferromagnetism can exist in one- or two-dimensional systems hosting Ising or XY models. P. C. Hohenberg proved that LR order cannot exist in one- and two-dimensional Bose and Fermi systems at finite temperatures [39]. The experimental studies show that the CE values for FGT and its stoichiometric variants vary with Fe content, vacancies, and the chosen BB windows for investigations, thereby failing to demonstrate scaling convergence. To resolve these discrepancies and clarify whether these variations represent intrinsic physics or analytical artifacts, a BB- and BB-direction-dependent critical analysis is required.

In this article, we present a detailed investigation into the critical behavior of FGT using recently developed methodologies [9]. BY analyzing both OOP (B∥cB \parallel c) and in-plane (IP) (B⊥cB \perp c) magnetization isotherms, we quantify BB-driven crossover between dominant LR and SR exchange interactions. Our results show that the CE values of FGT are BB-dependent, shedding light on the fragile balance of competing energy contributions in anisotropic vdW magnets.

Experimental details

The chemical vapor transport (CVT)-grown single crystals of FGT were obtained from commercial sources, viz., HQ graphene (https://www.hqgraphene.com/). The magnetic measurements were carried out using a Magnetic Property Measurement System (MPMS3) (make: Quantum Design). Temperature-dependent magnetization (MM-TT) measurements were conducted under zero-field-cooled warming (ZFCW) and field-cooled warming (FCW) protocols. We recorded two sets of MM-HH isotherms in the vicinity of TCT_\mathrm{C} by applying B∥cB \parallel c and B⊥cB \perp c, where cc represents the FGT crystal’s cc-axis.

Results and Discussion

Magnetic properties of FGT

FGT is predominantly a ferromagnet hosting two types of Fe ions, viz., Fe3+^{3+} and Fe2+^{2+}. Fe3+^{3+} (3d53d^5) is coupled with Fe2+^{2+} (3d63d^6) via Te following double exchange mechanism. There is direct exchange coupling between Fe3+^{3+} ions, resulting in weak intralayer antiferromagnetism. The interlayer magnetic interaction is weak and can be easily modulated by applying a small BB (the Zeeman energy) [18]. The AFM exchange breaks first around 160 K, leading the system into a predominantly FM state near TCT_\mathrm{C}.

We record the ZFC and FC MM-TT Figure 1(a) and Figure 1(b) in the warming mode by applying magnetic field of 8 mT along both parallel (B∥cB \parallel c) and perpendicular (B⊥cB \perp c) to the crystal’s cc-axis. The observed negative magnetization in the ZFCW curve results from dominant AFM exchange at low temperature and, to some extent, the experimental design, which may stabilize the magnetization in the negative direction via a partially induced diamagnetic interaction. The OOP (B∥cB \parallel c) and IP (B⊥cB \perp c) M−TM-T plots confirm that FGT exhibits an easy axis along the crystal’s cc-axis, since the IP magnetic moment at 10 K is ∼50\sim50 times lower than the observed OOP magnetic moment; see Figure 1(a) and Figure 1(b). When B∥cB \parallel c [Fig. 1(b)], a dip in the M−TM-T is clearly visible near 160 K which is the temperature where AFM coupling breaks and leads the system to a predominantly FM phase [29] [18]. The minimum of the first derivative of M−TM-T [inset of Figs. 1(a) and 1(b)] yields TC=198.6 KT_\mathrm{C}=198.6\,K for both B∥cB \parallel c and B⊥cB \perp c. The observed lower TCT_\mathrm{C} from the TCT_\mathrm{C} value of intrinsic FGT suggests that the used crystal may have minor Fe vacancies [18]. Therefore, the existing AFM and FM exchange interactions [18], the coexistence of itinerant and localized electrons [37], and emerging Fe vacancies make FGT a nontrivial magnetic system [28] [9]. We investigate how these competing energies modulate the critical behavior properties of FGT. Additionally, since FGT is a vdW magnet with an easy axis along the crystal’s cc-axis, we explore how the critical behavior properties modulate with applied BB directions.

Temperature-dependent magnetization and magnetization isotherms of FGT

Figure 1. Temperature-dependent magnetization (MM-TT) recorded in zero-field-cooled (ZFC) and field-cooled (FC) warming mode by applying a magnetic field (BB) of 8 mT (a) along the crystal’s c-axis (B∥cB \parallel c) and (b) perpendicular to the crystal’s c-axis (B⊥cB \perp c). The inset of the respective figures shows the first derivative of the MM-TT, whose minima represent the Curie temperature, TC=198.6 KT_\mathrm{C}=198.6\,K. The BB-dependent magnetization (MM-HH) isotherms recorded in the vicinity of TCT_\mathrm{C} by applying (c) B∥cB \parallel c and (d) B⊥cB \perp c.

Order of phase transition

The primary test for the study of critical behavior is to check whether the system shows a second-order phase transition (SOPT) at TCT_{\mathrm{C}}. There are several methods to confirm SOPT at TCT_{\mathrm{C}} in a magnetic system [40, 41, 42]. To determine the order of the phase transition, we apply Landau phenomenological theory [42], which states the following. The free energy of a magnetic system, under BB, can be expressed as the power-law variation of the order parameter MM as

F(M,T)=F(0)+a(T)2M2+b(T)4M4+c(T)6M6+⋯−BM,(1)F(M,T)=F(0)+\frac{a(T)}{2}M^2+\frac{b(T)}{4}M^4+\frac{c(T)}{6}M^6+\cdots-BM, \tag*{(1)}

where a(T)a(T), b(T)b(T), and c(T)c(T) are the Landau coefficients. The variation of a(T)a(T), b(T)b(T), and c(T)c(T) with TT depicts the physical characteristics of the magnetic system under investigation. At equilibrium, dF/dM=0\mathrm{d}F/\mathrm{d}M=0, resulting

μ0H=a(T)M+b(T)M3+c(T)M5+….(2)\mu_0H=a(T)M+b(T)M^3+c(T)M^5+\ldots. \tag*{(2)}

Here, a∝(T−TC)a \propto(T-T_{\mathrm{C}}) and bb changes its sign at T=T0T=T_0. A system will show SOPT if b≥0b \ge0 at T=TCT=T_{\mathrm{C}} and a first-order phase transition if b<0b<0 at T=TCT=T_{\mathrm{C}}. We fit the MM-HH isotherms [Figure 1(c) and Figure 1(d)] using Eq. (2) by restricting it up to M5M^5 [43]. We observe b>0b>0 [Figure 2(a) and Figure 2(b)] at T=TCT=T_{\mathrm{C}} for both B∥cB \parallel c and B⊥cB \perp c, thus confirming SOPT in FGT. Below, we explore how the coexisting itinerant and localized electrons, FM and AFM interactions, anisotropy, and other competing energies dictate the critical behavior properties of FGT with BB.

Variation of the Landau coefficients and log-log plots of the M-H isotherms

Figure 2. The variation of the Landau coefficients a(T)a(T) and b(T)b(T), with temperature, estimated from the fit to the MM-HH isotherms using Eq. (2) for (a) B∥cB \parallel c and (b) B⊥cB \perp c. b(T)>0b(T) > 0 at TCT_{\mathrm{C}}, indicating second-order phase transition. The log-log plot of the MM-HH isotherms at TCT_{\mathrm{C}} for (c) B∥cB \parallel c and (d) B⊥cB \perp c. The dashed black lines, in the respective figures, are the linear fits to the log-log plot of the MM-HH isotherms for 1 T≤B≤2 T1\,\mathrm{T} \le B \le2\,\mathrm{T}. The solid red lines, in the respective figures, are the linear fits to the log-log plot of the MM-HH isotherms for 6 T≤B≤7 T6\,\mathrm{T} \le B \le7\,\mathrm{T}.

Field-dependent critical behavior of FGT

The Necessity of Field-Dependent Critical Analysis

Recently, Chauhan et al. [9] proposed two methodologies to investigate the BB-dependent critical behavior of a magnetic system. The log-log plots of the MM-HH isotherms at T=TCT=T_{\mathrm{C}} [Figure 2(c) and Figure 2(d)] do not show linear variation, confirming the need for a BB-dependent critical behavior study of the FGT. Following the reported methodologies, we take a fixed temperature range of ∼9.4%\sim9.4\% of TCT_{\mathrm{C}} for both the OOP and IP MM-HH isotherms. We estimate BB-dependent δMH\delta_{MH} for both OOP [Figure 3(a)] and IP [Figure 3(a)] MM-HH isotherms, using the relation

Field-dependent critical exponent values

Figure 3. The δMH\delta_{MH} values estimated for different magnetic field (BB) ranges as mentioned when the magnetic BB is applied (a) parallel (B∥cB \parallel c) and (b) perpendicular (B⊥cB \perp c) to the crystal’s cc-axis.

M=A0B1/δ, at T=TC,(3)M=A_0B^{1/\delta},\ \text{at }T=T_{\mathrm{C}}, \tag*{(3)}

where A0A_0 is the proportionality constant. In the measured BB range from 1−7 T1-7\,\mathrm{T}, the estimated δMH\delta_{MH} values vary between 3 and 5 when B∥cB \parallel c [Fig. 3(a)], indicating that the FGT spins should reside in three-dimensional (d=3d = 3) space. Theoretically, 3≤δ<53 \le\delta< 5 for d=3d = 3 and δ=15\delta= 15 for d=2d = 2 [44, 45]. However, when B⊥cB \perp c, the determined δMH\delta_{MH} values at low BB [Fig. 3(b)] violate the theoretical developments. This is evidence that the magnetocrystalline anisotropy plays a key role in defining the critical behavior of a magnetic system. When B∥cB \parallel c [Fig. 3(a)], the δMH\delta_{MH} seems to attain saturation rapidly, though the exact saturation point lies beyond the applied BB range. Conversely, when B⊥cB \perp c [Fig. 3(b)], a much larger BB is required to reach the saturation of δMH\delta_{MH}.

Estimating the range of exchange interaction: the intercept analysis

In 1972, Fisher et al. [46] first proposed the relation of γ\gamma with σ\sigma, dd and nn as

γ=1+4d(n+2n+8)Δσ+(8(n+2)(n−4)d2(n+8)2)[1+2G(d2)(7n+20)(n−4)(n+8)]Δσ2+…(4)\gamma= 1 + \frac{4}{d}\left(\frac{n+2}{n+8}\right)\Delta\sigma+ \left(\frac{8(n+2)(n-4)}{d^2(n+8)^2}\right)\left[1 + \frac{2G\left(\frac{d}{2}\right)(7n+20)}{(n-4)(n+8)}\right]\Delta\sigma^2 + \ldots \tag*{(4)}

where Δσ=(σ−d2)\Delta\sigma= \left(\sigma- \frac{d}{2}\right) and G(d2)=3−14(d2)2G\left(\frac{d}{2}\right) = 3 - \frac{1}{4}\left(\frac{d}{2}\right)^2. Equation 4 is used to estimate σ\sigma, which eventually evaluates J(r)∼r−(d+σ)J(r) \sim r^{-(d+\sigma)} emerging in a magnetic system. The proposed relation is exactly valid for a trivial magnetic system with LR interaction; however, it has been used for systems with SR interaction as well [34, 47] [4, 10, 11]. It is stated that a system will show LR interaction if σ≤1.5\sigma\le1.5 and SR (SR) interaction if σ≥2.0\sigma\ge2.0. However, there is a speculative explanation regarding the range of interaction if σ\sigma lies between 1.5 and 2.0. Several groups treat the magnetic system as weakly LR for 1.5<σ<2.01.5 < \sigma< 2.0 [47, 48] [10, 11]. The recent investigation claims that the Fisher et al. relation (Eq. 4) may not yield the exact value of σ\sigma for a nontrivial magnetic system [28] [9]. In this case, no method exists other than the recently proposed qualitative approach to identify the range of the exchange interaction by estimating the intercepts of the linear fits to modified Arrott plots (MAPs) constructed using the CE values of the standard UCs [9]. The intercept versus temperature plots (see SM [43]) show crossover at temperature TT for a set of CE values belonging to a standard UC. We define T1T_1 as the crossover temperature belonging to SR standard UCs, viz., 3D Heisenberg, 3D XY, and 3D Ising, and T2T_2 as the crossover temperature associated with the LR mean-field UC. Figure 4 shows the variation of T1T_1 and T2T_2 with the BB range that was used to estimate the values of T1T_1 and T2T_2 [43]. We observe an almost constant shift of T1T_1 from TCT_C irrespective of the used BB range for both B∥cB \parallel c [Fig. 4(a)] and B⊥cB \perp c [Fig. 4(b)], however the distance of T2T_2 from TCT_C varies with increasing BB. This is an indication that the range of exchange also depends on BB. At low BB, the FGT exhibits dominant LR exchange interaction, which is suppressed by increasing BB. The variation of T1T_1 and T2T_2, thus, indicates that the LR interactions become weaker with BB when B∥cB \parallel c and remains dominant when B⊥cB \perp c. These observations confirm the role of competing energies in determining the critical behavior of FGT. The following two points justify the observed variation of LR behavior. (i) The itinerant electrons cause LR Ruderman-Kittel-Kasuya-Yosida (RKKY) interaction which suppresses with BB. (ii) The existing double exchange mechanism between Fe3+^{3+} and Fe2+^{2+} ions also leads to LR interaction due to an emerging percolation mechanism, which also suppresses with BB. The BB-direction-dependent qualitative behavior of the existing range of exchange interaction indicates the strong effect of uniaxial magnetocrystalline anisotropy (easy axis) and thus the role of dipolar energy in modulating the critical behavior of a magnetic system. We thus argue that the BB-direction-dependent critical behavior will be strong in vdW magnets. Below, we estimate β\beta, γ\gamma, and subsequently σeff\sigma_{eff} to quantitatively evaluate the range and type of exchange interaction exhibited in FGT as a function of the applied BB direction using the proposed ANS analysis method [9].

Estimating the field-dependent critical exponents: the ANS method

The proposed ANS method [9] states that the estimation of appropriate β\beta and γ\gamma values for T≤TCT \le T_C is sufficient to explore the critical behavior of a magnetic system. Therefore, hereafter we only discuss the critical behavior of FGT using the estimated β\beta and γ\gamma values for T≤TCT \le T_C. The proposed ANS method is as follows. First, estimate δMH\delta_{MH} for a fixed BB-range using Eq. (3). Then, correlate β\beta and γ\gamma with δMH\delta_{MH} using the Widom relation as

γ=β×(δMH−1).\gamma= \beta\times(\delta_{MH} - 1).

Now vary β\beta as

βm=βi±m×βstep,(5)\beta_m = \beta_i \pm m \times\beta_{step}, \tag*{(5)}

where βi\beta_i is the initially selected value, βstep\beta_{step} is the step increment in β\beta for each iteration and mm is the positive integer representing the iteration number. To start with, βi\beta_i can be chosen either as a standard theoretical value corresponding to any standard UC or as a random value, with the condition that the chosen set of β\beta and corresponding γ\gamma leads to linear behavior of the constructed MAPs, see [43]. Thus, for a selected βm\beta_m, determine the corresponding γm\gamma_m using Eq. (5) as

γm=βm×(δMH−1).(6)\gamma_m = \beta_m \times(\delta_{MH} - 1). \tag*{(6)}

Now, construct the MAPs using the above correlated βm\beta_m (Eq. (5)) and γm\gamma_m (Eq. (6)) values. Next, estimate normalized slopes (NSs) using the slopes of the linear fit to the MAPs. Determine the ANS using the estimated NS values as

ANS=∑i=1N(NS)iN,(7)ANS = \frac{\sum_{i=1}^{N}(NS)_i}{N}, \tag*{(7)}

where ii denotes the MM–HH isotherm at a given TT, ∑i=1N(NS)i\sum_{i=1}^{N}(NS)_i is the sum of the NSs, and NN is the total number of isotherms used for the estimation. Thus, for a selected βm\beta_m and corresponding γm\gamma_m, we get an ANS value. Figure 5 (a) and Figure 5(b) show the representative ANS plots with nn for the BB range of 4 − 5 T when B∥cB \parallel c and B⊥cB \perp c, respectively. We vary βm\beta_m and the corresponding γm\gamma_m in such a way that the ANS →1\to1. We stop the increment or decrement of βm\beta_m and the corresponding γm\gamma_m when ANS = 1 within the error limit. The error estimation varies depending on the selected βstep\beta_{step}. For βstep=0.001\beta_{step} = 0.001, ANS = 1.00 ±\pm 0.005 [see shaded regions in Figure 5 (a) and Figure 5(b)] is best to get appropriate values of β\beta and γ\gamma. Following the above-discussed processes, we estimate the BB-dependent β\beta and γ\gamma values (see Table I) for both B∥cB \parallel c and B⊥cB \perp c. The simplest and effective way to test the robustness of the estimated β\beta and γ\gamma values is the estimation of NSs for the MM-HH isotherms recorded in the vicinity of TCT_C. The β\beta and γ\gamma values are considered appropriate if the NS values remain close to 1 irrespective of the temperature. As shown in Figure 5 (c) and Figure 5(d), the NS values remain close to 1 irrespective of the temperature for T≤TCT \leq T_C. The NS ∼1\sim1 in the vicinity of TCT_C confirms that the critical behavior remains the same on either side of TCT_C, however,

ANS plots and normalized slopes for FGT

Figure 5. The average of normalized slope (ANS) plots with temperature for the FGT when (a) B∥cB \parallel c and (b) B⊥cB \perp c. The ANS is estimated using the relation 7 as discussed in the main text. (c) and (d) the estimated normalized slopes (NSs) from the linear fit to modified Arrott plots constructed using the final critical exponent values obtained using the ANS method for B∥cB \parallel c and B⊥cB \perp c, respectively. The T0T_0 in the respective figure is the temperature value up to which the NSs remain close to 1 above TCT_C for the BB range of 6 to 7 T.

Material/ModelsReferenceMethodField range (T)TCT_C (K)δ\delta (δMH\delta_{M H})β\betaγ\gammaσeff\sigma_{eff}σ\sigma (d:nd:n)
Mean-field (MF)[49, 50, 51]Theory3.0000.5001.0001.5001.500
Tricritical MF[49, 51]Theory5.0000.2501.0001.000
3D Ising[49, 50, 51]Theory4.8000.3251.2371.9671.967
3D XY[49, 50, 51]Theory4.8000.3451.3161.9681.963
3D Heisenberg[49, 50, 51]Theory4.8000.3651.3861.9651.965
Fe3_3GeTe2_2[34]MAP0.25–0.75223MT223^{MT}3.0000.5001.001.000 (2 : 3)
Fe3_3GeTe2_2[34]MAP0.75–1.53.2920.4391.011.011 (2 : 3)
Fe3_3GeTe2_2[34]MAP1.5–3.03.6840.3881.051.050 (2 : 3)
Fe3_3GeTe2_2[34]MAP3.0–5.03.8630.3841.091.089 (2 : 3)
Fe3_3GeTe2_2[11]MAP0–8215MT215^{MT}4.261(9)0.327(3)1.079(5)
Fe3_3GeTe2_2[10]MAP0.3–5152MT152^{MT}4.50(1)0.372(4)1.265(15)
Fe3_3GeTe2_2 (B∥cB \parallel c)This workANS1–2198.6MT198.6^{MT}3.237(20)0.432(3)0.967(6)1.584(61)1.420(16) (3 : 1)
2–33.560(13)0.394(3)1.009(8)1.684(79)1.520(17) (3 : 1)
3–43.771(16)0.365(3)1.011(8)1.742(84)1.524(17) (3 : 1)
4–53.921(14)0.354(3)1.034(9)1.781(92)1.572(18) (3 : 1)
5–64.060(15)0.350(3)1.071(11)1.814(106)1.642(20) (3 : 1)
6–74.082(16)0.330(3)1.017(10)1.819(105)1.537(21) (3 : 1)
Fe3_3GeTe2_2 (B⊥cB \perp c)This workANS3–4198.6MT198.6^{MT}2.592(14)0.500(3)0.796(5)1.330(49)NA
4–52.831(11)0.447(3)0.819(5)1.434(54)NA
5–63.003(10)0.411(3)0.823(6)1.501(66)NA
6–73.141(9)0.392(3)0.839(6)1.551(68)NA

Table 1. The list of estimated field-dependent CE values for Fe3_3GeTe2_2 in this work, along with reported literature on the CE values for Fe3_3GeTe2_2 and the defined UCs. The symbols have their usual meaning as discussed in the main text. MAP: modified Arrott plot, ANS: average of normalized slope.

one needs to vary the temperature range for T>TCT > T_C depending on the selected BB range. Below, we discuss the role of the competing energies and the effect of BB-direction on the critical behavior of FGT.

The critical behavior of FGT: the role of competing energies

When B∥cB \parallel c, the β\beta values [Figure 6(a) and Table I] vary from ∼0.43\sim0.43 to ∼0.33\sim0.33 with increasing BB while the γ\gamma values remains close to 1 within the measured BB range. To explain this variation, we delve into the properties of FGT. In FGT, at least five energy parameters, viz., double exchange, direct exchange, RKKY, dipolar, and anisotropy energies, modulate the phase transition and thus the critical behavior [29, 30, 31, 32, 33, 34, 35, 36, 37] [10, 11, 18]. Since FGT is a nontrivial magnetic system, we get the effective CE values using the ANS method. We thus argue that the effective β∼0.43\beta\sim0.43 at low BB belongs to the mean-field UC. With increasing BB, this β\beta value is suppressed due to the modulation of the competing energies and ends up belonging to the 3D Ising UC. Since the estimated β\beta value is effective, it will be hard to observe its convergence to a standard UC. Following the recent investigation [9], we restate the above explanations as follows.

Variation of critical exponents and effective interaction range with field range

Figure 6. The variation of (a) β\beta, (b) γ\gamma and (c) σeff\sigma_{eff} with the applied BB range when B∥cB \parallel c. The β\beta converges 3D Ising universality class at higher BB, while γ\gamma remains close to 1. The variation of σeff\sigma_{eff} indicates the modulation of LR interaction from strong to weak with increasing BB. The variation of (d) β\beta, (e) γ\gamma and (f) σeff\sigma_{eff} with the applied BB range when B⊥cB \perp c. The β\beta gradually decreases below 0.400 and does not converge in the measured BB range. The γ\gamma slowly increases, but does not reach 1 and beyond in the measured BB range, indicating a strong effect of anisotropy energy. σeff\sigma_{eff} gradually increases but remains close to 1.5 indicating dominant LR interaction.

When B∥cB \parallel c, the double exchange and RKKY interactions cause LR ordering, resulting in γ∼1\gamma\sim1 with no sign of a monotonic increase within the measured BB range. The anisotropy and dipolar energies restrict nn, which is clearly visible in the high-BB critical behavior (Table I). Taking into consideration only the BB-dependent β\beta, we find that n=3n=3 (β=0.365\beta=0.365) in the range of 3–4 T and n→1n \to1 (β→0.325\beta\to0.325) in the high-BB range. For β≫0.365\beta\gg0.365, the nn values have no meaning as the system exhibits LR interaction. The values of nn change due to the suppression of LR interactions with BB. These findings directly support the evaluated BB-dependent variation of the range of exchange interactions obtained using intercept analysis; see Figure 4.

Variation of crossover temperatures with field for parallel and perpendicular magnetic field orientations

Figure 4. The variation of crossover temperatures T1T_1 and T2T_2 with field when (a) B∥cB \parallel c and (b) B⊥cB \perp c. The T1T_1 and T2T_2 values are estimated from the crossover of the intercepts, which were estimated from the linear fit to the modified Arrott plots constructed using the critical exponent values associated with SR standard universality classes and the LR mean-field universality class, respectively.

We now attempt to quantitatively determine the range of exchange interactions by estimating σeff\sigma_{eff} and σ\sigma (Eq. (4)). One can estimate σeff\sigma_{eff} by treating the Rushbrooke inequality as an equality, α+2β+γ=2\alpha+2\beta+\gamma=2. Furthermore, γ\gamma is related to σ\sigma and ν\nu by σ=γν\sigma=\frac{\gamma}{\nu}, while α=2−νd\alpha=2-\nu d. Combining these relations, we estimate

σeff=γd2β+γ(8)\sigma_{eff}=\frac{\gamma d}{2\beta+\gamma} \tag*{(8)}

Equation 8 does not involve nn, and the value of dd is also unknown. However, based on the estimated δMH<5.0\delta_{MH}<5.0 (Figure 3(a)), we assume d=3d=3 [49, 50]. Thus, by substituting d=3d=3 along with the β\beta and γ\gamma values, we estimate the BB-dependent σeff\sigma_{eff}, see Table I. As shown in Figure 6(c), at low BB, σeff\sigma_{eff} indicates the stabilization of the dominant LR interaction, which weakens with increasing BB. This behavior is fully justified by the intercept analysis (Figure 4(a)).

Furthermore, one can estimate σ\sigma by substituting the dd, nn and γ\gamma, associated with localized spins, into Eq. (4). Usually, this relation is used to determine dd, nn and σ\sigma. Here, the estimated nn for a particular dd will represent the UC of the magnetic system under investigation [28] [13, 14, 15, 16, 17]. For FGT, the determined σ\sigma (Table I) is closer to σeff\sigma_{eff} for d=3d=3 and n=1n=1, indicating that the critical behavior of the localized spins belongs to 3D Ising UC. This is further supported by the intrinsic anisotropy and dipolar energies, which restrict nn of the local electronic moments.

When the BB is applied perpendicular to the crystal’s c-axis (i.e., in the IP direction), the obtained CE values for FGT [Figure 6(d) to Figure 6(f) and Table I] exhibit unusual characteristics that are not supported by existing theories [1, 2, 9–11]. As shown in Figure 6(d), the β\beta value does not converge within the measured BB range. However, the variation of β\beta from ∼0.5\sim0.5 at low BB to ∼0.4\sim0.4 at high BB indicates the domination of the LR interactions within the measured BB range. The estimated γ\gamma value (for T<TCT < T_C) Figure 6(e) monotonically increases with BB but remains strictly below the theoretical threshold of ≥1\ge1 across the measured window up to 7 T. While standard scaling laws mandate γ≥1\gamma\ge1 for isotropic phase transitions [49, 50, 51], such sub-unity values Figure 6(e) can be understood as follows. We estimate γ\gamma using the MM-HH isotherms for T≤TCT \le T_C, whereas typically reported γ\gamma values are extracted from the initial inverse susceptibility χ0−1(T)\chi_0^{-1}(T) for T≥TCT \ge T_C using conventional analysis methods [12–17]. The magnetization process is dominated by coherent spin rotation against the anisotropy barrier rather than purely thermodynamic spin fluctuations. This introduces an effective BB-dependent constraint (on nn), suppressing the divergence of χ\chi near TCT_C and artificially truncating the scaling of the MAPs. Consequently, γ<1.0\gamma< 1.0 highlights the limitations of equilibrium scaling models when treating strongly anisotropic systems in their hard-axis (IP) orientation, demonstrating that the apparent critical parameters (Table I) are effective rather than intrinsic thermodynamic exponents. Moreover, in an anisotropy-driven magnetic system, γ\gamma can differ on either side of TCT_C, typically taking a lower value for T<TCT < T_C [52]. These discussions indicate that the obtained sub-unity value of γ\gamma Figure 6(e) may actually reflect the intrinsic physical characteristics of χ\chi for T≤TCT \le T_C in highly anisotropic magnetic systems. This behavior is well supported in FGT, which exhibits strong uniaxial magnetocrystalline anisotropy [29, 30, 31, 32, 33, 34, 35] [10, 11, 18].

Moreover, the δMH\delta_{MH} achieves the theoretically predicted value beyond B=5B = 5 T [49, 50, 51]. This critical BB value signifies the threshold where the Zeeman energy overcomes the magnetocrystalline anisotropy energy for B⊥cB \perp c. Nevertheless, the MAPs [43] constructed for the MM-HH isotherms for T≤TCT \le T_C exhibit quasi-linear behavior in this BB range. We further verify the quasi-linear behavior of the MAPs by demonstrating a horizontal collapse of all isotherms for the first derivative of the MAPs [34, 43].

These observations—specifically observing γ<1.0\gamma< 1.0 [9, 20] despite the MAPs showing quasi-linear behavior and NSs ∼1\sim1 pose serious questions about the complete understanding of the critical behavior of nontrivial magnetic systems. The lack of complete convergence in β\beta and γ\gamma values for B⊥cB \perp c within the 1–7 T window suggests that the applied Zeeman energy has not yet fully overcome the intrinsic magnetocrystalline anisotropy energy barrier in FGT. At much higher BB (>7> 7 T), where the Zeeman energy completely dominates over the anisotropy energy, the local spins are expected to align fully along the hard axis. In this high-BB limit, the critical parameters should undergo a crossover, recovering a physically valid regime where β\beta and γ\gamma ultimately converge to standard UC. Future high-BB magnetization experiments are therefore necessary to completely map the BB-dependent phase diagram and confirm the isotropic nature of the intrinsic CEs in this class of vdW ferromagnets.

The estimated σ\sigma Figure 6(f) indicates dominating LR interactions within the measured BB range up to B=7B = 7 T for the B⊥cB \perp c configuration. However, for such low values of γ<1.0\gamma< 1.0, Eq. 4 yields imaginary values for σ\sigma. Specifically, Eq. 4 can only yield real values for σ\sigma when η>0.855\eta> 0.855 (for d=3d = 3, n=3n = 3), η>0.863\eta> 0.863 (for d=3d = 3, n=2n = 2), and γ>0.871\gamma> 0.871 (for d=3d = 3, n=1n = 1). This highlights an additional challenge: the lack of an established methodology to reliably determine nn for a nontrivial magnetic system undergoing a continuous phase transition under these specific critical conditions.

Conclusions

In conclusion, we have presented a comprehensive investigation into the critical phenomena of the vdW ferromagnet Fe3_3GeTe2_2 by employing the recently developed ANS method [9]. Our experimental and analytical results reveal that the critical parameters in this system are fundamentally modulated by both the strength and spatial orientation of BB, challenging conventional critical behavior investigation.

For the easy-axis configuration (B∥cB \parallel c), the system undergoes a continuous BB-driven crossover. At low BB, the critical behavior is governed by LR magnetic interactions mediated by a competitive interplay of double-exchange mechanisms and itinerant electron (RKKY) magnetism, yielding a mean-field-like exponent (β∼0.43\beta\sim0.43, γ∼1.0\gamma\sim1.0, δMH∼3.24\delta_{MH} \sim3.24). With increasing Zeeman energy, these LR interactions are systematically suppressed, driving the system to a clean convergence within LR 3D Ising UC (β→0.330\beta\to0.330, γ∼1.0\gamma\sim1.0) at high BB.

Conversely, the hard-axis configuration (B⊥cB \perp c) uncovers an anomalous critical regime characterized by sub-unity susceptibility exponents (γ<1.0\gamma< 1.0) across the entire measured BB window. Although this behavior is theoretically undefined, we propose that γ\gamma can acquire a sub-unity value in magnetic systems with strong uniaxial magnetocrystalline anisotropy, as observed in FGT. We further show that the obtained CE values mark a clear crossover where the strong OOP magnetocrystalline anisotropy energy barrier strongly constrains conventional thermodynamic spin fluctuations. In this regime, the magnetization profile is heavily driven by a coherent spin rotation toward the hard plane rather than pure thermodynamic phase transition dynamics, rendering the extracted CE values effective rather than intrinsic exponents. The lack of convergence in β\beta and γ\gamma values in the measured BB-range suggests that the intrinsic critical behavior of FGT should remain isotropic irrespective of the applied BB direction.

Ultimately, this study underscores that evaluating CEs across fixed, broad BB windows in strongly anisotropic or low-dimensional magnets can obscure or distort the underlying physical scaling. The success of the BB-dependent ANS approach in mapping the complex magnetic landscape of Fe3_3GeTe2_2 establishes it as a vital diagnostic tool for resolving competing interactions in emerging quantum and topological van der Waals materials.

Acknowledgments

NY is thankful to the Council of Scientific and Industrial Research (CSIR), Government of India, for the research fellowship. NY, HCC and PD acknowledge the Department of Physics, Central Research Facility (CRF) and Nano Research Facility (NRF) of the Indian Institute of Technology (IIT) Delhi, for characterization facilities. PD gratefully acknowledges the partial funding support from the Science and Engineering Research Board (SERB) of the Government of India (grant no. SPR/2021/000762), Indo-French Centre for the Promotion of Advanced Research (IFC-PAR/CEFIPRA) through Grant No. 6508-2/2021 and the National Quantum Mission, an initiative of the Department of Science and Technology, Government of India. HCC gratefully acknowledges the Anusandhan National Research Foundation (ANRF), Government of India, for the partial financial support through the National Post-Doctoral Fellowship (N-PDF) (Grant/Sanction No. PDF/2026/005456).

Data Availability Statement

The data supporting this study’s findings are available from the corresponding author upon reasonable request.

References

References

  1. [23]A. Tozri, E. Dhahri, E. K. Hlil, and M. A. Valente, Critical behavior near the paramagnetic to ferromagnetic phase transition temperature in La₀.₇Pb₀.₀₅Na₀.₂₅MnO₃, Solid State Commun. 151, 315–320 (2011).DOI
  2. [24]S. Dash, E. D. Bagoslov, B. Weise, A. V. Lukoyanov, Y. V. Knyazev, Y. I. Kuz’min, P. Kumar, and A. K. Patra, Impression of magnetic clusters, critical behavior and magnetocaloric effect in Fe₃Al alloys, Phys. Chem. Chem. Phys. 21, 10823–10833 (2019).DOI
  3. [25]R. P. Madhogaria, R. Das, E. M. Clements, V. Kalappattil, M. H. Phan, H. Srikanth, N. T. Dang, D. P. Kozlenko, and N. S. Bingham, Evidence of long-range ferromagnetic order and spin frustration effects in the double perovskite La₂CoMnO₆, Phys. Rev. B 99, 104436 (2019).DOI
  4. [26]A. K. Heilman, Y. Y. Xue, B. Lorenz, M. Gospodinov, S. G. Dobreva, and C. W. Chu, Field-induced crossover in La₀.₇Pb₀.₃MnO₃, Physica C 341–348, 707–710 (2000).DOI
  5. [27]A. Wang, A. Rahman, Z. Du, J. Zhao, F. Meng, W. Liu, J. Fan, C. Ma, M. Ge, L. Pi, L. Zhang, and Y. Zhang, Field-dependent anisotropic room-temperature ferromagnetism in Cr₃Te₄, Phys. Rev. B 108, 094429 (2023).DOI
  6. [28]H. C. Chauhan, B. Kumar, A. Tiwari, J. K. Tiwari, and S. Ghosh, Different Critical Exponents on Two Sides of a Transition: Observation of Crossover from Ising to Heisenberg Exchange in Skyrmion Host Cu₂OSeO₃, Phys. Rev. Lett. 128, 015703 (2022).DOI
  7. [29]J. Yi, H. Zhuang, Q. Zou, Z. Wu, G. Cao, S. Tang, S. A. Calder, P. R. C. Kent, D. Mandrus, and Z. Gai, Competing antiferromagnetism in a quasi-2D itinerant ferromagnet: Fe₃GeTe₂, 2D Mater. 4, 011005 (2016).
  8. [30]Y. Deng, Y. Yu, Y. Song, J. Zhang, N. Z. Wang, Z. Sun, Y. Yi, Z. Wu, S. Wu, J. Zhu, J. Wang, X. H. Chen, and Y. Zhang, Gate-tunable room-temperature ferromagnetism in two-dimensional Fe₃GeTe₂, Nature 563, 94–99 (2018).DOI
  9. [31]C. Lee, T. J. Lee, S.-G. Jung, T. Park, S. Alb arakati, J. Partridge, M. R. Field, D. G. McCulloch, L. Wang, and C. Lee, Hard magnetic properties in nanoflake van der Waals Fe₃GeTe₂, Nat. Commun. 9, 1554 (2018).
  10. [32]Z. Fei, B. Huang, P. Malinowski, W. Wang, T. Song, J. Sanchez, W. Yao, D. Xiao, X. Zhu, A. F. May, W. Wu, D. H. Cobden, J.-H. Chu, and X. Xu, Two-dimensional itinerant ferromagnetism in atomically thin Fe₃GeTe₂, Nat. Mater. 17, 778–782 (2018).DOI
  11. [33]N. León-Brito, E. D. Bauer, F. Ronning, J. D. Thompson, and R. Movshovich, Magnetic microstructure and magnetic properties of uniaxial itinerant ferromagnet Fe₃GeTe₂, J. Appl. Phys. 120, 083903 (2016).
  12. [34]A. Tiwari, H. Ahn, B. Kumar, J. Saini, P. K. Srivastava, B. Singh, C. Lee, and S. Ghosh, Phase transition in the two-dimensional Heisenberg ferromagnet Fe₃GeTe₂ with long-range interaction, Phys. Rev. B 109, L020407 (2024).DOI
  13. [35]H. Kim, K.-X. Zhang, Y.-H. Li, G. Park, R. Cheng, and J.-G. Park, Emergent giant topological Hall effect in twisted Fe₃GeTe₂ metallic system, Nat. Commun. 17, 2931 (2026).DOI
  14. [36]Y. Zhang, H. Lu, X. Zhu, S. Tan, W. Feng, Q. Liu, W. Zhang, Q. Chen, Y. Liu, et al., Emergence of Kondo lattice behavior in a van der Waals itinerant ferromagnet, Fe₃GeTe₂, Sci. Adv. 4, eaao6791 (2018).
  15. [37]P. R. Sharma, T. W. Kim, and H. Noh, Non-monotonic temperature dependence of anomalous Hall effect in a 2D van der Waals ferromagnetic material Fe₃GeTe₂, Mater. Chem. Phys. 302, 127738 (2023).DOI
  16. [38]N. D. Mermin and H. Wagner, Absence of Ferromagnetism or Antiferromagnetism in One- or Two-Dimensional Isotropic Heisenberg Models, Phys. Rev. Lett. 17, 1133–1136 (1966).
  17. [39]P. C. Hohenberg, Existence of Long-Range Order in One and Two Dimensions, Phys. Rev. 158, 383–386 (1967).DOI
  18. [40]J. Y. Law, V. Franco, L. M. Moreno-Ramírez, A. Conde, D. Y. Karpenkov, I. Radulov, K. P. Skokov, and O. Gutfleisch, A quantitative criterion for determining the order of magnetic phase transitions using the magnetocaloric effect, Nat. Commun. 9, 2680 (2018).DOI
  19. [41]V. Franco, J. S. Blázquez, and A. Conde, Field dependence of the magnetocaloric effect in materials with a second-order phase transition: A master curve for the magnetic entropy change, Appl. Phys. Lett. 89, 222512 (2006).DOI
  20. [42]J. C. Toledano and P. Toledano, The Landau Theory of Phase Transitions: Application to Structural, Incommensurate, Magnetic and Liquid Crystal Systems, World Scientific (1987).
  21. [43]See Supplemental Material at www.yyy.com for discussion on the intercept of the MAPs and normalized slope plots, the modified Arrott plots and their derivatives.
  22. [44]J. García and J. A. Gonzalo, Accurate Monte Carlo critical exponents for Ising lattices, Physica A 326, 464–472 (2003).
  23. [45]J. G. García, M. I. Marqués, and J. A. Gonzalo, Monte Carlo study of Ising critical isotherms, Ferroelectrics 426, 166–170 (2012).DOI
  24. [46]M. E. Fisher, S.-k. Ma, and B. G. Nickel, Critical Exponents for Long-Range Interactions, Phys. Rev. Lett. 29, 917–920 (1972).DOI
  25. [47]J. K. Tiwari, B. Kumar, H. C. Chauhan, and S. Ghosh, Critical scaling and magnetic phase diagram of bi-skyrmion-hosting quasi-two-dimensional La₁.₃₇Sr₁.₆₃Mn₂O₇ bi-layer manganite, J. Magn. Magn. Mater. 535, 168020 (2021).DOI
  26. [48]A. Bhattacharya, A. Ahmed, S. DuttaGupta, and I. Das, Critical behavior and phase diagram of skyrmion-hosting material Co₃.₆Fe₄.₄Zn₈Mn₄ probed by anomalous Hall effect, J. Alloys Compd. 960, 170277 (2023).DOI
  27. [49]D. Kim, B. L. Zink, F. Hellman, and J. M. D. Coey, Critical behavior of La₀.₇₅Sr₀.₂₅MnO₃, Phys. Rev. B 65, 214424 (2002).
  28. [50]S. N. Kaul, Static critical phenomena in ferromagnets with quenched disorder, J. Magn. Magn. Mater. 53, 5–53 (1985).DOI
  29. [51]J. C. Le Guillou and J. Zinn-Justin, Critical exponents from field theory, Phys. Rev. B 21, 3976–3998 (1980).
  30. [52]F. Léonard and B. Delamotte, Critical Exponents Can Be Different on the Two Sides of a Transition: A Generic Mechanism, Phys. Rev. Lett. 115, 200601 (2015).DOI

Paper details

vlnc:W_ahe9akb2
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 8

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents