Thermodynamic and Transport Study of Electron-and Hole-Doped MGa3 Single Crystals (M = Fe, Co)

M. CABRERA-BAEZ,1 E. THIZAY MAGNAVITA,1 RAQUEL A. RIBEIRO,1

and MARCOS A. AVILA1,2

1.—CCNH, Universidade Federal do ABC (UFABC), Santo Andre, SP 09210-580, Brazil.2.—e-mail: avila@ufabc.edu.br

FeGa3 and related compounds have been subjects of recent investigation fortheir interesting thermoelectric, electronic, and magnetic behaviors. Here,single crystals of FeGa3�yGey were grown by the self-flux technique witheffective y = 0, 0.09(1), 0.11(1), and 0.17(1) in order to investigate the evolu-tion of the diamagnetic semiconducting compound FeGa3 into a ferromagneticmetal, which occurs through the electron doping and band structure modifi-cations that result from substitution of Ge for Ga. Heat capacity and mag-netization measurements reveal non-Fermi liquid behavior in the vicinity ofthe transition from a paramagnetic to ferromagnetic ground state, suggestingthe presence of a ferromagnetic quantum critical point (FMQCP). We alsopresent the first results of hole doping in this system by the growth ofFeGa3�yZny single crystals, and electron- and hole doping of the relatedcompound CoGa3 by CoGa3�yGey and CoGa3�yZny crystal growths, aiming tosearch for further routes to band structure and charge carrier tuning, ther-moelectric optimization, and quantum criticality in this family of compounds.The ability to tune the charge carrier type warrants further investigation ofthe MGa3 system’s thermoelectric properties above room temperature.

Key words: Carrier tuning, non-fermi liquids, quantum criticality

INTRODUCTION

Strongly correlated electron systems have playeda central part in the development of material sci-ence in recent times. These kinds of materials showinteresting physical properties (magnetic, thermal,and electronic) that differ from the usual ones beingused in today’s technological evolution. Phenomenasuch as enhanced thermoelectric power, unconven-tional superconductivity, and quantum criticalityare direct results of the strong electronic correlationin these systems.

FeGa3 is an unusual binary intermetallic com-pound with diamagnetic behavior1,2 which hasshown potential as a low-cost thermoelectric mate-rial2 and has also shown further unexpectedbehavior when tuned through chemical doping. Forexample, substitution of Ge for Ga (FeGa3�yGey)

promptly induces evolution from a diamagneticsemiconductor to a ferromagnetic metal at effectivey = 0.13, passing through a paramagnetic metallicstate with non-Fermi liquid behavior due to theproximity of a quantum critical point (QCP).3,4 Thesame is not observed when substituting Co for Fe(Fe1�xCoxGa3),3,5 although in principle both types ofsubstitution represent electron doping of the sys-tem. Pure CoGa3 is also diamagnetic, but showsmetallic behavior with n-type charge carriers asexpected.3,5 In addition to a new set of single crystalgrowths and measurements on FeGa3�yGey con-ducted at UFABC, here, we present our first resultson the further exploration of these pseudo-binarysystems, aimed at hole doping of FeGa3 by thegrowth of FeGa3�yZny single crystals, and electron-and hole doping of CoGa3 by the growth ofCoGa3�yGey and CoGa3�yZny single crystals,respectively. Mapping of heat capacity behaviors,and more specifically the evolution of the Sommer-feld coefficients c, indicates that hole doping in(Received June 28, 2013; accepted November 20, 2013)

Journal of ELECTRONIC MATERIALS

DOI: 10.1007/s11664-013-2932-1� 2013 TMS

FeGa3 and electron-doping in CoGa3 do very little interms of affecting their respective electronic struc-tures and properties. In contrast, hole doping inCoGa3 produces a steady increase in c, which iscontrary to what would be conventionally expectedfrom a depopulation of the system’s electron carrierdensity. This is a clear indication that theCoGa3�yZny system should also evolve towards non-Fermi liquid behavior and quantum criticality,although likely at a much more gradual (doping)rate than that observed in FeGa3�yGey. Preliminarytransport measurements show that Zn doping inCoGa3�yZny does indeed evolve towards lower car-rier densities and, consequently, improved thermo-electric properties.

EXPERIMENTAL DETAILS

Polyhedral single crystals of FeGa3�yGey,FeGa3�yZny, CoGa3�yGey, and CoGa3�yZny withtypical sizes of 1–3 mm and effective composi-tions 0 £ y £ 0.25(1) were grown by a standard Gaself-flux method.6,7 High purity elements in a1:3(3 � y):3y proportion were sealed inside evacu-ated silica ampoules. These ampoules were heatedto 1,100 �C and cooled to 550 �C over 150 h, thenremoved from the furnace and quickly centrifugedfor separation of the molten flux. The effectivedoping level of the crystals is typically muchsmaller than the nominal y of the initial mix. ForFeGa3�yGey, the effective doping of each crystal wasestimated by comparing the Sommerfeld coefficient,transition temperature, and effective moment withthe results by Umeo et al.3 For the other crystals,the effective doping was estimated by electron-probemicroanalysis (EPMA) on a JEOL JXA-8200 ana-lyzer at Hiroshima University. x-Ray diffractionpatterns of powdered crystals were obtained on aBruker D8 Focus machine and are all consistentwith the FeGa3 type structure, with refined latticeparameter values for the pure compounds inagreement with published works.1,3

Specific heat was measured on all samples from 2to 30 K on a Quantum Design PPMS using thesystem’s standard relaxation method. For theFeGa3�yGey samples, magnetic isotherms and dcmagnetic susceptibility measurements between 2and 300 K were done on the PPMS VSM option inapplied fields up to 9 T. Electrical resistance mea-surements were performed from 2 to 300 K onCoGa3�yZny samples with y = 0.14 and 0.17, usingthe conventional dc four probe method on thePPMS. Growth of larger single crystals will berequired in order to measure the full thermoelectricproperties on the PPMS TTO option.

RESULTS AND DISCUSSION

Magnetic isotherms M(H) of FeGa3�xGex singlecrystals for three concentrations of Ge have beenmeasured at several temperatures between 5 and300 K, and the results for T = 5 K are shown in

Fig. 1. Recalling the diamagnetic state of the purecompound,2 it is seen that the Ge0.09(1) sample hasreadily evolved into a paramagnetic state with iso-therm that follows the Brillouin function, and theGe0.11(1) and Ge0.17(1) samples have acquired a fer-romagnetic ground state. The saturation magneticmoment increases with the Ge concentration.

Figure 2 presents the temperature dependence ofthe dc magnetic susceptibility, v ¼M=H, for thesame Ge-doped single crystals. Once again, theincrease in effective magnetic moment is clearlyseen to grow with the Ge doping, and the twosamples with higher Ge concentration show curveswith inflection points at increasing temperatures,indicating evolution towards ferromagnetic order-ing. The inset of Fig. 2 shows the inverse suscepti-bility, v�1 ¼ H=M, which follows the Curie–Weisslaw, v�1 ¼ ðT �HÞ=C; at higher temperatures. Theslopes of the linear regions decrease with Ge doping.From the fitted Curie constants C and usingleff ¼

ffiffiffiffiffiffiffi

8Cp

, we extract the value of the effectivemoment in number of Bohr magnetons associatedwith each sample, whose magnitude grows accord-ing to the substitution of Ga for Ge. The Ge0.09(1)

sample displays paramagnetic behavior with fittedWeiss temperature H = 0 within error. For Ge0.11(1)

and Ge0.17(1), the ferromagnetic interaction andordering are manifested by the positive andincreasing Weiss temperatures. Table I summarizesthe above results.

The combined results of these magnetic mea-surements clearly indicate that Ge doping stronglyinduces magnetic moments on the Fe atoms. How-ever, the exact mechanism of this induction remainsan open question, and will require microscopicinvestigations to determine whether a Ge impurityinduces more localized moments around its Fevicinity or a more widespread distribution ofitinerant moments. The latter scenario has beenfavored by very recent DFT simulations;8 however,it has become apparent that modeling thesecompounds is not a trivial task, since there has even

0 10 20 30 40 500.00

0.02

0.04

0.06

0.08

0.10

0 10 20 30 40 500.00

0.02

0.04

0.06

0.08

0.10

Ge 0.09(1)

Ge0.17(1)

T = 5 K Ge 0.11(1)

FeGa3-yGey

M (μ

B/f.

u.)

H (kOe)

μ)

Fig. 1. Isothermal magnetization curves for FeGa3�yGey at 5 K.

Cabrera-Baez, Magnavita, Ribeiro, and Avila

been some debate in the literature over straight-forward matters such as whether or not the pre-dicted ground state of pure FeGa3 is magnetic.9,10

Heat capacity measurements bring furtherinsights towards understanding the evolution ofelectronic properties through doping of these com-pounds. The specific heat divided by temperature,Cp/T, as a function of T2 for FeGa3�yGey samples are

shown in Fig. 3. Considering that pure FeGa3 has aSommerfeld coefficient c = 0.03 mJ/mol K2,3 thereis a drastic enhancement in the extrapolated c(electronic contribution in the specific heat) for theGe0.11(1) and Ge0.17(1) samples. Also, as shown inTable II, c (Ge0.11(1)) is larger than c (Ge0.17(1)), sothe trend in c does not follow a monotonic increaseas seen in the magnetization behaviors. Theseanomalies demonstrate a perturbation of the bandstructure that goes beyond the simple addition ofelectrons near the Fermi surface when Ge substi-tutes Ga. The inset evidences a logarithmic depen-dence CP=T � �lnðTÞ at low temperatures, which ischaracteristic of non-Fermi liquid behavior in thevicinity of a QCP. Since both samples are in themagnetically ordered state, the Ge0.17(1) sample hasmoved away from the QCP compared to Ge0.11(1).

Figure 3 also presents measurements onFeGa3�yZny samples with effective doping y =0.003(1) and 0.005(1), which resulted from nominalinitial mixes with y = 0.2 and 0.4, respectively. Thesubstitution of Zn for Ga implies hole doping in thesystem, but does not significantly affect the elec-tronic contribution to specific heat at these effectivedoping levels, since the overall behavior and Som-merfeld coefficient remain close to that of the purecompound, as listed in Table II.

We now focus on the low-temperature heatcapacity of the Co-based system. Figure 4 showsCp/T as a function of T2 for pure and doped CoGa3

single crystals. Pure CoGa3 has a Sommerfeldcoefficient c = 2.70 mJ/mol K2, significantly higherthan that of FeGa3 and consistent with the presenceof electron charge carriers in its metallic groundstate, as previously reported.3 Further electrondoping in this system is expected by the growth ofCoGa3�yGey crystals. However, as in the case ofFeGa3�yZny, nominal initial mixes of Ge0.4 and Ge0.6

resulted in samples with low effective dopings,Ge0.020(1) and Ge0.022(1), and at least in terms of theheat capacity, the behavior remains essentiallyunmodified.

In contrast, the measurements on CoGa3�yZny

crystals show that the hole doping of CoGa3 repre-sented by substitution of Zn for Ga is effective andclearly affects the electronic properties. The overallheat capacity and Sommerfeld coefficient steadilyincrease for samples Zn0.11(1), Zn0.17(1), and Zn0.25(1).The trend, graphically represented in Fig. 5,contradicts the conventional expectation that a

0 10 20 300.00

0.01

0.02

0.03

0.04

0.05

0 10 20 30 40 500

400

800

1200

Ge0.09(1)

Ge0.11(1)

Ge0.17(1)

χ (e

mu/

mol

)

T (K)

Ge0.09(1)

Ge0.11(1)

Ge0.17(1)

χ−1 (

mol

/em

u)T (K)

FeGa3-yGey

Fig. 2. Magnetic susceptibility versus temperature in in H = 1 T forFeGa3�yGey. Inset shows the inverse susceptibility for the samesamples.

Table I. Effective moments and Weiss temperaturesfor FeGa32yGey

y leff lB=Feð Þ TC Kð Þ

0.09(1) 0.36 �0.30.11(1) 0.61 7.60.17(1) 0.77 18.5

0 10 20 30 400

20

40

60

80

100

2 3 4 5 6 7 8 9 100

20

40

60

80

100FeGa3-yXyGe0.11(1)

Ge0.17(1)

Zn0.003(1) Zn0.001(1)

CP

/T (

mJ/

mol

K2 )

T2 (K2)

CP

/T (m

J/m

ol K

2 )

T(K)

Fig. 3. Heat capacity of FeGa3�yGey and FeGa3�yZny presented asspecific heat divided by temperature as a function of T2. Inset showsthe same data for the Ge-doped samples as a function of logT.

Table II. Sommerfeld coefficients for FeGa32yXy

(X = Ge, Zn)

Compound c (mJ/mol K2)

FeGa2.89(1)Ge0.11(1) 87.28FeGa2.83(1)Ge0.17(1) 46.50FeGa2.94(1)Zn0.003(1) 0.085FeGa2.94(1)Zn0.005(1) 0.025

Thermodynamic and Transport Study of Electron- and Hole-Doped MGa3 Single Crystals (M = Fe, Co)

removal of electrons from the Fermi surface woulddecrease the value of c. The removal of electrons isevidenced in Fig. 6, showing normalized resistanceR/R300K that systematically decreases in slopecompared to that of pure CoGa3.3 This anomalousbehavior can be explained by the idea that the bandstructure is being affected by this substitution.11

The density of states at the Fermi level is expressedas D EFð Þ ¼ 4�m�e � kF=h

2, which leads to the con-clusion that, although kF may decrease, the effectivemass of the electrons is being renormalized anddominates the behavior of the Sommerfeld coeffi-cient. Similarity with the previously documentedcase of FeGa3�yGey is close enough for an expecta-tion that non-Fermi liquid behavior and a QCP mayalso be found in the CoGa3�yZny system; however,the evolution of the system with Zn doping appearsto be much more gradual.

In terms of exploring these systems as thermo-electric materials, our results point to the possibilityof tuning the MGa3 system with both n-type and

p-type carriers, which may be achievable in con-junction with the tuning of the QCP proximity, forexample by exploring a generalized pseudo-quinarysystem with a mixture of dopants, summarized asFe1�xCoxGa3�y�zGeyZnz.

CONCLUSION

We have reported a survey of the thermal proper-ties of FeGa3�yGey, FeGa3�yZny, CoGa3�yGey, andCoGa3�yZny single crystals, manifested in the evo-lution of the heat capacity with increasing dopinglevels. In terms of magnetic properties, we report themagnetic characterization of the paramagnetic andferromagnetic phases of FeGa3�yGey. Analysis of theSommerfeld coefficient trends shows that FeGa3�yGey

has a drastic enhancement over that of the diamag-netic semiconductor FeGa3 in the vicinity of thequantum critical point, whereas that of the ‘‘hole’’system FeGa3�yZny remains mostly unmodified. Forthe diamagnetic CoGa3 compound, which has n-typemetallic behavior, adding further electrons(CoGa3�yGey) does not significantly affect the heatcapacity. However, removing conducting electronsthrough hole doping (CoGa3�yZny) results in anom-alous enhancement of the Sommerfeld coefficient,most likely related to a perturbation of the density ofthe states near the Fermi level and a renormalizationof the effective electron mass. The anomalies point toa possible existence of another quantum critical pointin the Co system, and a demonstrated overall abilityto tune the carrier density and type in these com-pounds which may allow further developments interms of their thermoelectric performance.

ACKNOWLEDGEMENTS

We are thankful to Y. Shibata and T. Takabatakefor the EPMA measurements and acknowledge thefinancial support of CNPq (Grant # 473287/2011-0)and FAPESP (Grants # 2012/17562-9, 2011/19924-2and 2011/23795-3).

0 4 8 12 16 200

2

4

6

8

10

CoGa3-yXy

Zn0.25(1)Zn0.17(1)Zn0.11(1)

CP

/T (m

J/m

ol K

2)

Ge0.022(1)Ge0.020(1)CoGa

3

T2 (K2)

Fig. 4. Heat capacity of CoGa3�yGey and CoGa3�yZny presented asspecific heat divided by temperature as a function of T2.

0.0 0.1 0.2 0.30

1

2

3

4

5

6

7

CoGa3-yZnyγ (m

J/m

ol K

2 )

effective y

Fig. 5. Evolution of the Sommerfeld coefficient with effective dopingin CoGa3�yZny.

0 50 100 150 200 250 3000

1

0.17CoGa3-yZny

0.14

R/R

300K

T (K)

Fig. 6. Temperature dependence of electrical resistance R normal-ized by the value at 300 K for CoGa3�yZny.

Cabrera-Baez, Magnavita, Ribeiro, and Avila

REFERENCES

1. U. Haussermann, M. Bostrom, P. Viklund, O. Rapp, T.Bjornangen, R.W. Hu, V.F. Mitrovic, and C. Petrovic,J. Solid State Chem. 165, 94 (2002).

2. Y. Hadano, S. Narazu, M.A. Avila, T. Onimaru, and T.Takabatake, J. Phys. Soc. Jpn. 78, 013702 (2009).

3. K. Umeo, Y. Hadano, S. Narazu, T. Onimaru, M.A. Avila,and T. Takabatake, Phys. Rev. B 86, 144421 (2012).

4. N. Haldolaarachchige, J. Prestigiacomo, W.A. Phelan, Y.M.Xiong, G. McCandless, J.Y. Chan, J.F. DiTusa, I. Vekhter, S.Stadler, D.E. Sheehy, P.W. Adams, and D.P. Young, cond-mat/1304.1897.

5. E.M. Bittar, C. Capan, G. Seyfarth, P.G. Pagliuso, andZ. Fisk, J. Phys. Conf. Ser. 200, 012014 (2010).

6. P.C. Canfield and Z. Fisk, Phil. Mag. B 65, 1117(1992).

7. R.A. Ribeiro and M.A. Avila, Phil. Mag. 92, 2492(2012).

8. D.J. Singh, Phys. Rev. B 88, 064422 (2013).9. Z.P. Yin and W.E. Pickett, Phys. Rev. B 82, 155202

(2010).10. J.M. Osorio-Guillen, Y.D. Larrauri-Pizarro, and G.M.

Dalpian, Phys. Rev. B 86, 235202 (2012).11. P. Viklund, S. Lidin, P. Berastegui, and U. Haussermann,

J. Solid State Chem. 164, 100 (2002).

Thermodynamic and Transport Study of Electron- and Hole-Doped MGa3 Single Crystals (M = Fe, Co)