Multielemental single–atom-thick A layers innanolaminated V2(Sn, A) C (A = Fe, Co, Ni, Mn)for tailoring magnetic propertiesYoubing Lia,b,1, Jun Luc,1, Mian Lia, Keke Changa, Xianhu Zhaa,d, Yiming Zhanga, Ke Chena, Per O. Å. Perssonc,Lars Hultmanc, Per Eklundc, Shiyu Dua, Joseph S. Franciscoe,2, Zhifang Chaia, Zhengren Huanga, and Qing Huanga,2

aEngineering Laboratory of Advanced Energy Materials, Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences, 315201Ningbo, Zhejiang, China; bUniversity of Chinese Academy of Sciences, 100049 Beijing, China; cThin Film Physics Division, Department of Physics, Chemistry,and Biology, Linköping University, SE-581 83 Linköping, Sweden; dCenter for Quantum Computing, Peng Cheng Laboratory, 518055 Shenzhen, China;and eDepartment of Chemistry, University of Pennsylvania, Philadelphia, PA 19104-6316

Edited by C. N. R. Rao, New Chemistry Unit, CSIR Centre of Excellence, International Centre for Materials Science and Sheikh Saqr Laboratory, JawaharlalNehru Centre for Advanced Scientific Research, Bangalore, India, and approved November 27, 2019 (received for review September 18, 2019)

Tailoring of individual single–atom-thick layers in nanolaminatedmaterials offers atomic-level control over material properties.Nonetheless, multielement alloying in individual atomic layers innanolaminates is largely unexplored. Here, we report 15 inher-ently nanolaminated V2(AxSn1-x)C (A = Fe, Co, Ni, Mn, and combi-nations thereof, with x ∼ 1/3) MAX phases synthesized by analloy-guided reaction. The simultaneous occupancy of the 4 mag-netic elements and Sn in the individual single–atom-thick A layersconstitutes high-entropy MAX phase in which multielementalalloying exclusively occurs in the 2-dimensional (2D) A layers.V2(AxSn1-x)C exhibit distinct ferromagnetic behavior that can becompositionally tailored from the multielement A-layer alloying.Density functional theory and phase diagram calculations are per-formed to understand the structure stability of these MAX phases.This 2D multielemental alloying approach provides a structural de-sign route to discover nanolaminated materials and expand theirchemical and physical properties. In fact, the magnetic behavior ofthese multielemental MAX phases shows strong dependency onthe combination of various elements.

MAX phases | high-entropy ceramics | multielement alloys | magnetism

Tailoring individual single–atom-thick layers in nanolaminatedmaterials offers atomic-level control over modifying a desired

property of a material. For example, artificially nanolaminatedmagnetic materials are widely used in storage media and devices.Notably, the demonstration of the giant magnetoresistance (GMR)effect has enabled hard drives and other storage media (1). TheGMR sensitivity is expected to be highest when only single-atomlayers of ferromagnetic materials are sandwiched since alignmentof magnetization vectors of neighboring ferromagnetic atoms hasthe lowest energy cost.Conceptually, this type of structure with single–atom-thick

layers can be correlated to Mn+1AXn phases (or MAX phases),which are a family of inherently nanolaminated ternary com-pounds with hexagonal crystal structure (space group P63/mmc,194), where M is an early transition metal, A is mainly from Agroup elements, X is carbon and/or nitrogen, and n = 1 to 3 (2,3). Their crystal structure can be depicted by alternative stackingof Mn+1Xn sublayer and a single atomic layer of A, usually re-ferred to as 211, 312, and 413 phases according to the value of n.If alloying or replacement of ferromagnetic elements at specificcrystal sites in MAX phases could be realized, their magneticproperties may also be tailored. The key notion here is that the Alayer is just 1-atomic layer thick. However, there are few reportson magnetic MAX phases in which M site contains Mn or Fe,such as (Cr0.75Mn0.25)2GeC (4), Mn2GaC (5), (V,Mn)3GaC2 (6),and (Cr,Fe)2AlC (7), etc.Control of the occupancy of magnetic elements on the A rather

than M sites would be crucial to tune magnetic properties. The

finding of iron in the A site of Mo2(GaAuFe)C (8) is encouraging;however, the presence of secondary iron-containing impurityphases in the Mo2(GAuFe)C films impedes the determination ofmagnetic properties on the A plane. Moreover, theoretical pre-dictions suggest that Ni and Co should tend to occupy the M sitesof the MAX phases (9), but this has not been demonstratedexperimentally, likely due to the thermodynamically preferredformation of competing binary MA alloys or intermetallic pha-ses. Generally, the late transition elements of Fe, Co, Ni, and Mnhave not been considered as possible A-site elements in thedefinition of MAX phases. It would be of great interest to in-troduce these magnetic elements at A sites since the overlapbetween electron clouds of M and A atoms is limited comparedwith that between M and X atoms, which should aid in pre-serving the magnetic properties. A possible approach to realizethis is by alloying with other main group elements in such a waythat the alloying transforms neither the atomic stacking in thecrystal nor their bonding with the parent structure. Fe, Co, andNi have been used as additives for the formation of M′3SnC2(M′ = Ti, Zr, Hf) MAX phases (10), while none of the Fe/Co/Ni

Significance

Mn+1AXn phases are a family of inherently nanolaminatedternary compounds with hexagonal crystal structure (spacegroup P63/mmc, 194). Here, M is vanadium element, and A is Fe,Co, Ni, Mn, or their binary/ternary/quaternary mixtures. Due tothe elemental flexibility at A site, 15 nanolaminated V2(AxSn1-x)C MAX phases are synthesized, including 1 high-entropy MAXphase that all Fe, Co, Ni, Mn, and Sn elements simultaneouslyoccupied A site. Tailoring of individual single–atom-thick layersin nanolaminated MAX phases offers atomic-level control ofmaterial properties, such as their distinct magnetic behaviors.The alloying in 2-dimensional A layer of MAX phases provides aunique route to design their crystal structure and to discoverunexploited properties, which would develop promising func-tional materials for microelectronic device.

Author contributions: S.D., J.S.F., and Q.H. designed research; J.L., M.L., K. Chang, X.Z.,P.O.Å.P., L.H., S.D., and J.S.F. performed research; J.L., Y.Z., K. Chen, L.H., P.E., S.D., J.S.F.,Z.C., and Z.H. analyzed data; and Y.L., S.D., and Q.H. wrote the paper.

The authors declare no competing interest.

This article is a PNAS Direct Submission.

Published under the PNAS license.1Y.L. and J.L. contributed equally to this work.2To whom correspondence may be addressed. Email: frjoseph@sas.upenn.edu orhuangqing@nimte.ac.cn.

This article contains supporting information online at https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1916256117/-/DCSupplemental.

First published December 26, 2019.

820–825 | PNAS | January 14, 2020 | vol. 117 | no. 2 www.pnas.org/cgi/doi/10.1073/pnas.1916256117

Dow

nloa

ded

by g

uest

on

Janu

ary

19, 2

021

were detected in the final target MAX phases. The AxSny (A = Fe,Co, Ni, Mn, or their binary mixtures) phases belong to thesame space group (P63/mmc or 194) as MAX phases. Thissimilarity in crystal structures facilitates the nucleation of MAXphases in saturated alloys by a reaction between a binary car-bide and an intermediate state of AxSny alloys. During such areaction, these complex atoms in molten (or solid) alloys maythermodynamically and coherently arrange with [M6C] octahedralbuilding blocks to form ternary-layer structure. Thus, it would beof great interest to obtain magnetic MAX phases by an alloy-guided reaction.Here, we demonstrate this approach to A-site alloying of Sn

with Fe/Co/Ni/Mn magnetic elements to synthesize a series ofmagnetic MAX phases of V2(AxSn1-x)C (A = Fe, Co, Ni, Mn, ortheir binary/ternary/quaternary combination). The regular M2AXcrystal structure of V2SnC is retained, with a mixture of 2 to 5elements (Sn plus 1 or more of Fe, Co, Ni, and Mn) randomlyoccupying the A sites. The magnetic properties of V2(AxSn1-x)CMAX phases exhibit distinct ferromagnetic behavior that can betailored by their constitutive elements. These results have wide-ranging implications since they not only can be used for tailoringmagnetic properties but most importantly, demonstrate the for-mation of multielement A layers as an analogy to high-entropyalloys (11, 12), 2-dimensional (2D) in the sense that alloying ex-clusively occurs in the single–atom-thick A layers, in layeredtransition metal carbides.

Results and DiscussionMAX Phases Containing 1 Magnetic Element. X-ray diffraction(XRD) pattern of V-Fe-Sn-C system is close to the characteristiccrystal structure of V2AlC MAX phase (13), with characteristicpeaks at 2θ ∼ 13°, 2θ ∼ 26°, and 2θ ∼ 40°, indicating formation ofa 211 MAX phase (shown in Fig. 1A) and also, with some by-products of Sn metal, intermetallic compounds (FeSn2), andnonstoichiometric vanadium carbide (VCx) (SI Appendix, Fig.S1A). The micromorphology of observed particles with terracesis typical layered structure of MAX phases (Fig. 1B). The atomicforce microscopy (AFM) measurement also shows the typicallayer structure in nanoscale (SI Appendix, Fig. S2), and theheight of terrace is about 3 nm, which is close to double length of

unit cell of M2AX in c direction. In previously reported M′3SnC2(M′ = Ti, Zr, Hf) phases (10), Fe was absent in the finalproducts. Here, however, the corresponding energy-dispersiveX-ray spectroscopy (EDS) spectra of these particles showed thepresence of Fe beside V, Sn, and C (Fig. 1C). The relative atomicratio of V:(Fe + Sn) is ∼2:1, consistent with the stoichiometry of211 MAX phases (SI Appendix, Table S1). The high content ofFe (9.4 at.%) and its uniform distribution in elemental mapping(SI Appendix, Fig. S3) further indicate that Fe is incorporated inthe as synthesized MAX phase.To determine the position of Fe, we performed high-

resolution, high-angle annular dark field (HAADF) scanningtransmission electron microscopy (STEM) and lattice-resolvedEDS as shown in Fig. 1 D–G. STEM images acquired with thebeam along the [11�20] and [1�100] zone axes are shown in Fig. 1 Dand E, respectively. One layer of brighter spots (the A atomiclayers) is interleaved by 2 adjacent layers of darker spots (the Matomic layers). Carbon is typically not visible because of its lowcontrast compared with the heavier M and A atoms, and thecharacteristic zigzag stacking of the Mn+1Xn slabs viewed alongthe [11�20] zone axes is seen (14–17). STEM-EDS mapping (Fig.1F) and line scanning (Fig. 1G), which indicate that V is in the Msites following the zigzag stacking, and overlapped Fe (purple)and Sn (green) are proving that both are in the A sites of MAXphase. The compositions of the elements V, Sn, and Fe are 67,22, and 11 at. % (atomic percentage), respectively (SI Appendix,Table S2). The molar ratio is in good agreement with the aboveEDS result. Thus, the chemical formula of the resultant MAXphase is close to V2(Sn2/3Fe1/3)C. To prove the generality ofthis methodology, we used Co, Ni, and Mn instead of Fe in thestarting materials and followed the same chemical synthesisprocess. The comprehensive characterization of MAX phaseswhose A sites contain Co, Ni, and Mn elements are shown in SIAppendix, Figs. S4–S6 and section S1).

MAX Phases Containing 2 Magnetic Elements. If 2 magnetic ele-ments can be simultaneously incorporated into the MAX struc-ture, this would offer additional prospects for tuning magneticproperties since strong spin-electron coupling or interactionbetween different magnetic elements may enhance the magneticproperties as known from, for example, permalloy Ni80Fe20 withhigh magnetic permeability (18). For coincorporation of Fe andCo, the XRD pattern (Fig. 2A) showed that the main product is a211 MAX phase with small amounts of by-products (SI Appendix,Fig. S7). SEM demonstrated the typical terraced laminate mi-crostructure (Fig. 2B). From the EDS results in SEM (Fig. 2C),V, Fe, Co, Sn, and C (5 elements) were detected; the molar ratioof V:(Fe + Co + Sn) is very close to 2:1; and the Sn:(Co + Fe)ratio is 2:1 (SI Appendix, Table S3). Elemental mapping furtherprovided evidence that Fe, Co, and Sn have the same distribution(SI Appendix, Fig. S8). The molar ratio of V:(Fe + Co + Sn) isvery close to 2:1, and the Sn:(Co + Fe) ratio is 2:1. Therefore, theobtained MAX phase has the formula of V2(Fe1/6Co1/6Sn2/3)C.Furthermore, STEM images of the V2(Fe1/6Co1/6Sn2/3)C phasewith the beam along the [11�20] and [1�100] zone axes are shown inFig. 2 D and E, respectively. Furthermore, atomically resolvedEDS mapping analysis (Fig. 2F) and line scan analysis (Fig. 2G)in STEM mode provided direct evidence that Fe, Co, and Snelements all only occupy A sites.Following the same synthesis methodology, we also synthesized

several other MAX phases with 2 magnetic elements on the A site:that is, V2(FexNiySn1-x-y)C, V2(CoxNiySn1-x-y)C, V2(MnxCoySn1-x-y)C,V2(MnxFeySn1-x-y)C, and V2(MnxNiySn1-x-y)C. Phase composition,micromorphology, and elemental distribution are provided in SIAppendix, Figs. S9–S13.

Fig. 1. V2(Fe1/3Sn2/3)C MAX phase: XRD pattern (A) and SEM image (B) afteracid treatment. (C) EDS spectrum indicated that particles contain V, Sn, Fe,and C elements. High-resolution STEM images showing atomic positionsalong [11�20] (D) and [1�100] (E) directions, respectively. The correspondinghigh-resolution image together with EDS mapping (F) and line scanning (G)of V-Kα, Sn-Kα, and Fe-Kα signals are shown. (Scale bars: STEM images, 1 nm.)a.u., arbitrary unit; cps, counts per second.

Li et al. PNAS | January 14, 2020 | vol. 117 | no. 2 | 821

APP

LIED

PHYS

ICAL

SCIENCE

S

Dow

nloa

ded

by g

uest

on

Janu

ary

19, 2

021

Multielement A-Site Phases. A multielemental feature in chemicalcomposition of a single-phase homogeneous material can generallyhave drastic implications for physical and chemical properties asrealized for so-called high-entropy alloys or multiprincipal elementalloys (11, 12). In analogy with this materials design strategy, weposed the hypothesis that, since 1 or 2 magnetic elements fromthese 4 (Fe, Co, Ni, and Mn) can be alloyed with Sn at the A site, itwould be logical that also 3 or even 4 magnetic elements could besimultaneously incorporated. We therefore performed the samesynthesis for all combinations of 3 of these elements. The XRDand SEM-EDS results showed the formation of MAX phases ofV2(AxSn1-x)C, where A is a combination of 3 elements from Fe,Co, Ni, and Mn and x ∼ 1/3. The details are provided in SIAppendix, Figs. S14–S17.Furthermore, we synthesized an MAX phase with all 4 mag-

netic elements as well as Sn simultaneously. The XRD patternshows that the final product is composed of MAX phase andvarious tin alloys (Fig. 3A and SI Appendix, Fig. S18). The lam-inated morphology of the particles is similar to the above-mentioned MAX phases (such as in Figs. 1B and 2B) but withmore rounded edges (Fig. 3B). EDS in SEM detected all con-stitutive elements (V, Sn, Fe, Co, Ni, Mn, and C) in these par-ticles (Fig. 3C), and the molar ratio of all 4 magnetic elements toSn was close to 1:2 (SI Appendix, Table S4). Elemental mappingof Sn, Fe, Co, Ni, and Mn corroborated that all of these 5 ele-ments have the same distribution (Fig. 3D) and certainly occupythe A site. The obtained V2(AxSn1-x)C (where again A is acombination of Fe, Co, Ni, and Mn and x is close to 1/3) thusconstitutes a realization of a multielement (analogous to a high-entropy alloy) nanolaminated material, 2D in the sense that themultielement alloying exclusively occurs on the A layers.

Phase Diagrams. Fig. 4 shows the calculated isothermal sections at1,100 °C of the phase diagrams of V-Sn-C, V-(Sn2/3Fe1/3)-C, andV-Fe-C systems. These results indicate that V2SnC is the onlythermodynamically stable ternary phase in the V-Sn-C system(Fig. 4A) at equilibrium with Sn metal and vanadium carbide. Inthe case of Fe addition (Fig. 4B), the V2(Fe1/3Sn2/3)C phase canalso be at equilibrium with V3C2 and FeSn2, consistent with theexperimental results (Fig. 1A). However, in the V-Fe-C phasediagram, the hypothetical ternary MAX phase V2FeC is not

stable (Fig. 4C). Instead, vanadium carbides (V2C and V3C2) andan Fe-rich V-Fe intermetallic phase are the most competitivephases, corroborated by the experimental results. In fact, all ofour attempts to synthesize V2AC (A = Fe, Co, Ni, and Mn; i.e.,without Sn) phases failed (SI Appendix, Fig. S19).The Gibbs free energy values of V2(Fe1/3Sn2/3)C as well as

intermediate phases during synthesis are shown in Fig. 4D.According to this, V2(Fe1/3Sn2/3)C can appear at temperatures aslow as 400 °C. Below this temperature, the V2C and Fe5Sn3phases are dominant. Above 750 °C, V2C and Fe5Sn3 graduallydisappear and transform into V2(Fe1/3Sn2/3)C phase by a peri-tectic reaction: that is, solid V2C and an intermediate liquidFe5Sn3 transform into V2(Fe1/3Sn2/3)C. Without the presence ofFe (or Co, Ni, and Mn), the formation of V5Sn3 is insteadthermodynamically favored. Fe has higher affinity to Sn andthus, a stronger tendency to form Fe-Sn alloys than V does. Thisshould favor nucleation of VC1-x at low temperature and pro-mote the peritectic reaction between VC1-x and liquid Fe5Sn3alloy to form the final V2(Fe1/3Sn2/3)C phase.In general, the stable 52(AxSn1-x)C (A = Co, Ni, or Mn) MAX

phases follow similar reaction paths as V2(Fe1/3Sn2/3)C becauseof the reduced Gibbs free energy of the phase in the V-A-Sn-Csystem through the addition of A elements. The same is appar-ently true for multielement MAX phases V2(AxSn1-x)C, where Ais 2, 3, or 4 of Fe, Co, Ni, and Mn. The mixing entropy at the Asite must, therefore, account for most of the decrease in Gibbsfree energy and the corresponding thermodynamic stability.

Density Functional Theory Calculations. Here, first-principles den-sity functional theory calculations (19) were performed on theconfigurations V2AC (A = Sn, Fe, Co, Ni, Mn) and V2(A1/3Sn2/3)C(A = Fe, Co, Ni, Mn). The corresponding total energies (SIAppendix, Table S5), lattice parameters and bond lengths (SIAppendix, Table S6), and atomic charges (SI Appendix, Table S7)are provided. Side-view images of V2SnC (Fig. 5A), V2FeC (SIAppendix, Fig. S20A), and V2Sn2/3Fe1/3C (Fig. 5C) are presented,and the corresponding charge density distributions in real spaceare also provided. Based on the total energies, the reaction en-ergies for the possible chemical reactions numbered as ReactionsS1 to S4 in SI Appendix, section S4 were also calculated. Theseresults imply that A-site alloying in V2(A1/3Sn2/3)C (A = Fe, Co,Ni, Mn) is thermodynamically favored. Compared with V2AC(A = Fe, Co, Ni, Mn), V2SnC may show a higher stability due tothe lower atomic charge and fewer valence electrons of Sn

Fig. 2. V2(Fe1/6Co1/6Sn2/3)C MAX phase: (A) XRD pattern and (B) SEM imageafter acid treatment. (C) EDS spectrum indicating particles contain V, Sn, Fe,Co, and C elements. High-resolution STEM images showing atomic positionsalong [11�20] (D) and [1�100] (E) directions, respectively. The correspondinghigh-resolution image together with EDS mapping (F) and line scanning (G)of V-Kα, Sn-Kα, Fe-Kα, and Co-Kα signals are shown. (Scale bars: STEM im-ages, 1 nm.) a.u., arbitrary unit; cps, counts per second.

Fig. 3. V2(A,Sn)C (A = Fe, Co, Ni, Mn) MAX phase: XRD pattern (A) and SEMimage (B) after acid treatment. (C) The corresponding EDS spectrum indicatedthat particles contain V, Sn, Fe, Co, Ni, Mn, and C elements. (D) Elementalmapping on 1 particle clearing proving the uniform distribution of Fe, Co, Ni,Mn, and Sn. (Scale bars: 5 μm.) a.u., arbitrary unit; cps, counts per second.

822 | www.pnas.org/cgi/doi/10.1073/pnas.1916256117 Li et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

19, 2

021

according to the findings of a previous report (20). In fact, thecalculation results of phonon dispersion (SI Appendix, Fig. S21)show that V2SnC is dynamically stable, while V2FeC is not andthus, should not exist in the current space group.The projected densities of states (PDOSs) of V2SnC (Fig. 5B),

V2FeC (SI Appendix, Fig. S20B), and V2(Fe1/3Sn2/3)C (Fig. 5D)were also investigated. Sn showed a small contribution in thevicinity of Fermi level in the PDOS of V2SnC, while Fe pre-sented large density of states around the Fermi level of V2FeC.This is also consistent with a lower stability of V2FeC thanV2SnC due to the higher electron energy in the PDOS. Re-garding the solid solution V2(Fe1/3Sn2/3)C, Fe showed a lowerPDOS in the vicinity of Fermi level compared with that ofV2FeC.

Structure Stability. In earlier work, Fe/Co/Ni have been shown toeffectively promote the formation of M3SnC2 (M = Ti, Zr, Hf)MAX phases (10) without incorporation of Fe, Co, or Ni inthe final MAX phases. In contrast, in these experiments, addingFe/Co/Ni/Mn element into V/Sn/C raw materials did not promotethe formation of V2SnC but a series of V2(AxSn1-x)C (A = Fe,Co, Ni, Mn, or their combination) MAX phases. This also in-dicates that the M element (or the Mn+1Xn layer) plays an im-portant role in the incorporation of Fe, Co, Ni, and Mn elementsinto the A layer. As mentioned by Villars (21), at the microscopiclevel, it is well known that the structural stability is determined by3 pertinent factors, namely 1) the difference in atomic radius, 2)the electronegativity difference, and 3) the electron per atomratio (electron concentration).Vanadium has 1 more outmost d electron but smaller atomic

radius than titanium (also Zr and Hf). Thus, the correspondingMn+1Xn layer should become more compact. Therefore, the spaceavailable for Sn atoms in V2SnC is less than that in M3SnC2 (M =Ti, Zr, Hf) MAX phases. Fe, Co, Ni, and Mn have smaller atomicradii than Sn. Thus, these elements can pack closely to Sn atomsand reduce the final molar volume in V2(AxSn1-x)C MAX phases,meaning that the mixture of A element and Sn would keep the

structure stable in a constrained A layer. That is, the Mn+1Xn

layers provide less room for the A-element layers in V2SnC whencompared with M3SnC2 (M = Ti, Zr, Hf) MAX phases.However, the criterion of atomic radius cannot be the only

factor determining the structure stability; the electronegativity

Fig. 5. The side views (A) and PDOS (B) of V2SnC. The side views (C) andPDOS (D) of V2Sn2/3Fe1/3C. The charge densities in the real space are alsoshown in this figure, and the isosurface is set as 0.1 e Å−3.

Fig. 4. The isothermal sections at 1,100 °C for phase diagrams of (A) V-Sn-C system, (B) V-(Sn,Fe)-C system, and (C) V-Fe-C system. (D) Gibbs free energy ofphases during experimental synthesis of V2(Fe,Sn)C MAX phase.

Li et al. PNAS | January 14, 2020 | vol. 117 | no. 2 | 823

APP

LIED

PHYS

ICAL

SCIENCE

S

Dow

nloa

ded

by g

uest

on

Janu

ary

19, 2

021

also accounts for the bonding structure and strength in alloys.The electronegativity (Allen scale) difference between the pre-sent A elements and Sn is also very small (χMn = 1.75, χFe = 1.8,χCo = 1.84, χNi = 1.88, and χSn = 1.82). Obviously, the compat-ibility of A atoms with Sn atoms will not modify the stackingmode of A element in between Mn+1Xn layers due to similarity inelectron donor–acceptor capability.Moreover, the electron concentration effect is crucial in

complex materials, such as ternary MAX phases, because ofcomplex electronic interactions among the constituent elements.As mentioned by Mizutani (22), the Hume–Rothery rules, whichare guiding principles in the search of new alloys, use e/a as itselectron concentration rule to show a unique e/a-dependent phasestability (here, e is total itinerant electron of all constituent ele-ments in a primitive cell, and a is the corresponding total atomnumbers). Different phases can successively exist at a particular e/arange. The itinerant electron of Fe, Co, Ni, and Mn is the same asthat of Sn, which is 2. Therefore, the coexistence of these A ele-ments with Sn at A site of MAX phases can stabilize the originalcrystal structure of V2SnC.

Magnetic Properties. Temperature-dependent magnetizationM(T) curves under 0 field-cooled (SI Appendix, Fig. S22) andmagnetic hysteresis loops (Fig. 6) of the as synthesized series ofV2(AxSn1-x)C (A = Fe, Co, Ni, Mn, or their combination) MAXphases were studied. For V2(FexSn1-x)C, except at 2 K, all of themagnetic hysteresis loops follow “S-shaped” character (Fig. 6A)with small coercive force and residual magnetization (SI Ap-pendix, Table S8), suggesting that the V2(FexSn1-x)C compoundis a typical soft magnetic material (above 2 K) with saturationmagnetization (Ms) gradually decreasing with increasing tem-perature. At 2 K, the coercive force (Hc), residual magnetization(Mr), and saturation magnetization (Ms) of V2(FexSn1-x)C are150.88 Oe, 0.00038 emu/g, and 0.0806 emu/g, respectively. In thecase of V2(FexCoySn1-x-y)C, the magnetic hysteresis loops werecollected at 2, 50, 100, 200, 300, and 400 K (Fig. 6B and SIAppendix, Table S9). At 2 K, the coercive force (Hc), residualmagnetization (Mr), and saturation magnetization (Ms) ofV2(FexCoySn1-x-y)C are 481.85 Oe, 0.0262 emu/g, and 0.1378emu/g, respectively.Compared with Fe on A sites of V2(FexSn1-x)C, the magneti-

zation with 2 magnetic elements on A sites of V2(FexCoySn1-x-y)Cis stronger. Therefore, it can be concluded that the mag-netic properties can be tuned by adjusting the quantity and

type of magnetic elements on the A sites. For instance, forV2(FexCoyNizSn1-x-y-z)C (Fig. 6C), the residual magnetization(Mr) and saturation magnetization (Ms) at 2 K are 0.1499 and0.7400 emu/g, respectively (SI Appendix, Table S10), much higherthan for V2(FexCoySn1-x-y)C and V2(FexSn1-x)C MAX phase. InV2(MnxFeyCozNinSn1-x-y-z-n)C (Fig. 6D), which contains the anti-ferromagnetic element Mn, at 2 K the coercive force (Hc), re-sidual magnetization (Mr), and saturation magnetization (Ms) are320.4 Oe, 0.0471 emu/g, and 0.5677 emu/g, respectively. Althoughthe ferromagnetic properties of V2(MnxFeyCozNinSn1-x-y-z-n)C areless prominent than those of V2(FexCoyNizSn1-x-y-z)C in the rangeof 2 to 300 K, they are still much stronger than for the phasescontaining 1 or 2 magnetic elements. Thus, this approach providesa route for altering the magnetic properties of MAX phases bychanging the chemical composition and component. The positionsof magnetic elements (Fe, Co, Ni, Mn) in A layer are structurallyidentical but chemically disordered for their close atom massesand radii. If the distribution and proportion of magnetic elementsin A layers can be controlled, the variation of magnetic propertiesof these MAX phases will be well understood. Moreover, onepossible way to reduce the chemical disorder of magnetic elementsis to take advantage of double-A layer in some MAX phases, suchas Mo2Ga2C. Double-A layer in these MAX phases may provideadditional out-of-plane coupling of spin electrons other than in-plane interaction, which would be promising to further enhancethe magnetic properties.

Concluding Remarks and OutlookWe have demonstrated that the series of V2(AxSn1-x)C phases(A = Fe, Co, Ni, Mn, or their binary/ternary/quaternary combi-nations) can be synthesized by A-site element alloying, providinga generally applicable route to introduce 1 or more magneticelements in A sites and tune the resulting properties. A multi-element phase V2(AxSn1-x)C (where again A is a combination ofFe, Co, Ni, and Mn and x is close to 1/3) was realized. The factthat the magnetic properties are greatly enhanced for multiele-ment A layers lends credence to this concept and offers a richchemical space for discovering materials and properties usingA-site multielement alloying strategy.

MethodsPreparation of V2(AxSn1-x)C (A = Fe, Co, Ni, Mn, and Combinations Thereof). Thestarting powders were mixed in a stoichiometric ratio of V, Sn, C, A, NaCl, andKCl; the experimental details are shown in SI Appendix, section S6. Aftergrinding for 10 min, the mixed powders were placed into an aluminumoxide boat. Then, the alumina boat was inserted into a tube furnace andheated to reaction temperature during 3 h with a heating rate of 10 °C/minunder an argon atmosphere. After the end of reaction, residual chloridesalts were washed out from MAX phases in deionized water. Some metaland alloys impurities were further dissolved in HCl acid in order to obtainpure MAX phase powders. The weight change during various steps of theprocess is provided in SI Appendix, Fig. S23 and Table S12.

Computational Details. All the first-principles calculations were performedusing the VASP code (23, 24). Based on the projector-augmented wavepseudopential (25) with a plane-wave cutoff energy of 500 eV, the gener-alized gradient approximation as implemented by Perdew–Burke–Ernzerhof(26) was used for describing the exchange-correlation functional. Regardingatomic charges, a Bader charge analysis (27) based on 180 × 180 × 180 gridwas performed. Phonon dispersion was investigated with Phonopy software(28) and the VASP code based on the density functional perturbation theory(29). All of the structures were visualized in the VESTA3 code (30).

The Calculation of Phase Diagrams (CALPHAD) approach was applied tocalculate the phase diagrams. Due to the lack of experimental data on theternary V-Sn-C, V-Fe-C, and V-Sn-Fe-C compounds, first-principles calculationswere conducted to support the CALPHAD work (31). The Gibbs free energyfunction of V2(Sn,Fe)C was then determined with the Neumann–Kopp ruleand added in the CALPHAD-type dataset of the V-Sn-Fe-C system, whichincluded the thermodynamic parameters of the binary V-Sn, V-Fe, V-C, Sn-C,

Fig. 6. Magnetic hysteresis loops of V2(FexSn1-x)C (A), V2(FexCoySn1-x-y)C (B),V2(FexCoyNizSn1-x-y-z)C (C), and V2(MnxFeyCozNinSn1-x-y-z-n)C (D) at differenttemperatures in the range from −10 to 10 kOe.

824 | www.pnas.org/cgi/doi/10.1073/pnas.1916256117 Li et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

19, 2

021

Sn-Fe, and Fe-C systems (32–36). All of the calculations were performed withthe Thermo-Calc software.

Characterization. Phase composition of the samples was analyzed by XRD (D8Advance; Bruker AXS) with Cu Kα radiation. The microstructure and chemicalcomposition were observed by SEM (QUANTA 250 FEG; FEI) equipped withan energy-dispersive spectrometer. AFM analysis was performed by meansof a Dimension 3100 V system (Veeco) under tapping mode. Scanningtransmission electron microscopy (STEM) imaging and EDS analyzing wereperformed in a double-corrected FEI Titan3 60 to 300 microscope. Themagnetic properties were measured on a Quantum Design superconductingquantum interference device magnetometer.

Data Availability. The data presented in this article are available in SI Ap-pendix. SI Appendix contains experimental conditions, XRD patterns, AFM

topography, SEM images, SEM-EDS mapping, STEM images, mapping andline scanning, density functional theory calculation, and magnetic charac-terization of V2(AxSn1-x)C MAX phases.

ACKNOWLEDGMENTS. This study was supported financially by NationalNatural Science Foundation of China Grants 21671195, 21805295, and21875271 and Chinese Academy of Sciences Grants 2019VEB0008 and174433KYSB20190019. We acknowledge support from the Swedish Govern-ment Strategic Research Area in Materials Science on Functional Materials atLinköping University through Faculty Grant SFO‐Mat‐LiU 2009 00971. TheKnut and Alice Wallenberg Foundation is acknowledged for support of theelectron microscopy laboratory in Linköping through Grant KAW 2015.0043, aScholar grant (to L.H.), and an Academy Fellow grant (to P.E.). P.O.Å.P. alsoacknowledges Swedish Foundation for Strategic Research Project FundingEM16‐0004 and Research Infrastructure Fellow RIF 14‐0074. We thank Cai Shenfor the AFM characterization of MAX phases.

1. M. N. Baibich et al., Giant magnetoresistance of (001)Fe/(001)Cr magnetic super-lattices. Phys. Rev. Lett. 61, 2472–2475 (1988).

2. M. W. Barsoum, The MN+1AXN phases: A new class of solids: Thermodynamically stablenanolaminates. Prog. Solid State Chem. 28, 201–281 (2000).

3. M. Sokol, V. Natu, S. Kota, M. W. Barsoum, On the chemical diversity of the MAXphases. Trends Chem. 1, 210–223 (2019).

4. A. S. Ingason et al., Magnetic self-organized atomic laminate from first principles andthin film synthesis. Phys. Rev. Lett. 110, 195502 (2013).

5. A. Thore, M. Dahlqvist, B. Alling, J. Rosen, Magnetic exchange interactions and criticaltemperature of the nanolaminate Mn2GaC from first-principles supercell methods.Phys. Rev. B 93, 054432 (2016).

6. Q. Tao et al., Thin film synthesis and characterization of a chemically ordered mag-netic nanolaminate (V,Mn)3GaC2. APL Mater. 4, 086109 (2016).

7. C. M. Hamm, J. D. Bocarsly, G. Seward, U. I. Kramm, C. S. Birkel, Non-conventionalsynthesis and magnetic properties of MAX phases (Cr/Mn)2AlC and (Cr/Fe)2AlC. J. Mater.Chem. C 5, 5700–5708 (2017).

8. C.-C. Lai et al., Thermally induced substitutional reaction of Fe into Mo2GaC thin films.Mater. Res. Lett. 5, 533–539 (2017).

9. A. Talapatra et al., High-throughput combinatorial study of the effect of M site alloyingon the solid solution behavior of M2AlC MAX phases. Phys. Rev. B 94, 104106 (2016).

10. T. Lapauw, B. Tunca, T. Cabioc’h, J. Vleugels, K. Lambrinou, Reactive spark plasmasintering of Ti3SnC2, Zr3SnC2 and Hf3SnC2 using Fe, Co or Ni additives. J. Eur. Ceram.Soc. 37, 4539–4545 (2017).

11. M.-H. Tsai, J.-W. Yeh, High-entropy alloys: A critical review. Mater. Res. Lett. 2, 107–123 (2014).

12. E. P. George, D. Raabe, R. O. Ritchie, High-entropy alloys. Nat. Rev. Mater. 1, 515–534(2019).

13. B. Wang, A. Zhou, Q. Hu, L. Wang, Synthesis and oxidation resistance of V2AlCpowders by molten salt method. Int. J. Appl. Ceram. Technol. 14, 873–879 (2017).

14. P. Eklund, M. Beckers, U. Jansson, H. Högberg, L. Hultman, The Mn+1AXn phases:Materials science and thin-film processing. Thin Solid Films 518, 1851–1878 (2010).

15. H. Fashandi et al., Synthesis of Ti3AuC2, Ti3Au2C2 and Ti3IrC2 by noble metal sub-stitution reaction in Ti3SiC2 for high-temperature-stable Ohmic contacts to SiC. Nat.Mater. 16, 814–818 (2017).

16. M. Li et al., Element replacement approach by reaction with Lewis acidic molten saltsto synthesize nanolaminated MAX phases and MXenes. J. Am. Chem. Soc. 141, 4730–4737 (2019).

17. Y. Li et al., Single-atom-thick active layers realized in nanolaminated Ti3(Al x Cu1-x)C2

and its artificial enzyme behavior. ACS Nano 13, 9198–9205 (2019).

18. C. Tsang, N. Heiman, K. Lee, Exchange induced unidirectional anisotropy at FeMn-Ni80Fe20 interfaces. J. Appl. Phys. 52, 2471–2473 (1981).

19. W. Kohn, L. J. Sham, Self-consistent equations including exchange and correlationeffects. Phys. Rev. B 140, A1133–A1138 (1965).

20. D. Ohmer, G. Qiang, I. Opahle, H. K. Singh, H. Zhang, High-throughput design of 211−M2AX compounds. Phys. Rev. Mater. 3, 053803 (2019).

21. P. Villars, “Factors governing crystal structures” in Crystal Structures in IntermetallicCompounds, J. H. Westbrook, R. L. Fleischer, Eds. (John Wiley & Sons, 2000), pp. 1–49.

22. U. Mizutani, Hume-Rothery rules for structurally complex alloy phases. MRS Bull. 37,169 (2012).

23. G. Kresse, J. Furthmüller, Efficiency of ab-initio total energy calculations for metalsand semiconductors using a plane-wave basis set. Comput. Mater. Sci. 6, 15–50 (1996).

24. G. Kresse, J. Furthmüller, Efficient iterative schemes for ab initio total-energy calcu-lations using a plane-wave basis set. Phys. Rev. B 54, 11169–11186 (1996).

25. P. E. Blöchl, Projector augmented-wave method. Phys. Rev. B 50, 17953–17979 (1994).26. J. P. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approximation made

simple. Phys. Rev. Lett. 77, 3865–3868 (1996).27. W. Tang, E. Sanville, G. Henkelman, A grid-based Bader analysis algorithm without

lattice bias. J. Phys. Condens. Matter 21, 084204 (2009).28. A. Togo, F. Oba, I. Tanaka, First-principles calculations of the ferroelastic transition

between rutile-type and CaCl2-type SiO2 at high pressures. Phys. Rev. B 78, 134106(2008).

29. X. Gonze, C. Lee, Dynamical matrices, Born effective charges, dielectric permittivitytensors, and interatomic force constants from density-functional perturbation theory.Phys. Rev. B 55, 10355–10368 (1997).

30. K. Momma, F. Izumi, VESTA3 for three-dimensional visualization of crystal, volumetricand morphology data. J. Appl. Cryst. 44, 1272–1276 (2011).

31. K. Chang, B. Hallstedt, D. Music, Thermodynamic and electrochemical properties ofthe Li–Co–O and Li–Ni–O systems. Chem. Mater. 24, 97–105 (2011).

32. K. H. Kumar, P. Wollants, L. Delaey, Thermodynamic evaluation of Fe-Sn phase diagram.Calphad 20, 139–149 (1996).

33. C. Guo, C. Li, X. Zheng, Z. Du, Thermodynamic modeling of the Fe–Ti–V system. Calphad38, 155–160 (2012).

34. T. Studnitzky, B. Onderka, R. Schmid-Fetzer, Phase formation and reaction kinetics inthe vanadium-tin system. Z. Kristallogr. Cryst. Mater. 93, 48–57 (2002).

35. L. Chen et al., Thermodynamic description of the Fe–Cu–C system. Calphad 64, 225–235(2019).

36. W. Huang, Thermodynamic properties of the Fe-Mn-VC system. Metall. Trans., A, Phys.Metall. Mater. Sci. 22, 1911–1920 (1991).

Li et al. PNAS | January 14, 2020 | vol. 117 | no. 2 | 825

APP

LIED

PHYS

ICAL

SCIENCE

S

Dow

nloa

ded

by g

uest

on

Janu

ary

19, 2

021