Nano granular metallic Fe - oxygen deficient TiO 2−δ composite films: A room temperature, highly carrier polarized magnetic semiconductor
S. D. Yoon, C. Vittoria, V. G. Harris, A. Widom, K. E. Miller, M. E. McHenry
aa r X i v : . [ c ond - m a t . m t r l - s c i ] J a n Nano granular metallic Fe - oxygen deficient TiO − δ composite films: A roomtemperature, highly carrier polarized magnetic semiconductor S. D. Yoon, ∗ C. Vittoria, † and V. G. Harris ‡ Center for Microwave Magnetic Materials and Integrated Circuits,Department of Electrical and Computer Engineering,Northeastern University, Boston, MA. 02115 USA
A. Widom § Department of Physics, Northeastern University, Boston, MA. 02115 USA.
K. E. Miller and M. E. McHenry
Department of Material Science and Engineering,Carnegie Mellon University, Pittsburg, PA. 15213, USA.
Nano granular metallic iron (Fe) and titanium dioxide (TiO − δ ) were co-deposited on (100) lan-thanum aluminate (LaAlO ) substrates in a low oxygen chamber pressure using a pulsed laserablation deposition (PLD) technique. The co-deposition of Fe and TiO resulted in ≈
10 nm metal-lic Fe spherical grains suspended within a TiO − δ matrix. The films show ferromagnetic behaviorwith a saturation magnetization of 3100 Gauss at room temperature. Our estimate of the saturationmagnetization based on the size and distribution of the Fe spheres agreed well with the measuredvalue. The film composite structure was characterized as p-type magnetic semiconductor at 300K with a carrier density of the order of 10 / cm . The hole carriers were excited at the interfacebetween the nano granular Fe and TiO − δ matrix similar to holes excited in the metal/n-type semi-conductor interface commonly observed in Metal-Oxide-Semiconductor (MOS) devices. From thelarge anomalous Hall effect directly observed in these films it follows that the holes at the interfacewere strongly spin polarized. Structure and magneto transport properties suggested that these PLDfilms have potential nano spintronics applications. PACS numbers: 75.50.Pp, 71.30.+h, 72.20.-i, 71.70.Gm, 72.25.-b, 73.40.Qv
I. INTRODUCTION
The search for semiconductors exhibiting magnetismat room temperature has been long and unyielding. How-ever, recently much progress has been made toward thisgoal . By doping a host semiconductor mate-rial with transition metal ferromagnetic atoms, diluteferromagnetic semiconductors have been produced withCurie temperatures ( T c ) as high as 100 K . Hall effectmeasurements below T c showed evidence for carriers be-ing spin polarized raising hopes for spintronics applica-tions. Specifically, metallic manganese (Mn) was dopedinto gallium arsenide (GaAs) whereby the carriers werestrongly spin polarized .We have previously reported the magnetic and dc-transport properties of magnetic semiconductor films ofTiO − δ , where δ indicates the degree of oxygen defi-ciency or defects in the film . The Curie tempera-ture, T c ≈
880 K, was well above room temperature witha saturation magnetization, 4 πM S ≈
400 Gauss. Tita-nium dioxide, TiO , is a well known wideband gap oxidesemiconductor, belonging to the group IV-VI semicon-ductors, described in terms of an ionic model of Ti and O − . Its intriguing dielectric propertiesallow its use as a gate insulator material in the Field-Effect-Transistor (FET) . Also, TiO is characterizedto be an n-type semiconductor with an energy gap vary-ing in the range 3 volt < ∆ e < . Films of TiO − δ on sub-strates of (100) lanthanum aluminate (i.e. LaAlO ) weredeposited by a pulsed laser ablation deposition (PLD)technique at various oxygen chamber pressures rangingfrom 0.3 to 400 mtorr. The origin of the presence of Ti and Ti (as well as Ti ) ions was postulated as a re-sult of the low oxygen chamber pressure during the filmsgrowth . Oxygen defects gave rise to valence states ofTi and Ti (in the background of Ti ) whereby dou-ble exchange between these sites dominated. The samecarriers involved in double exchange also gave rise to im-purity donor levels accounting for the transport proper-ties of the film. The dilute number of carriers were polar-ized in external applied magnetic field, yet the magneticmoment was still rather small . For example, normalHall resistivity was measured to be much bigger thanthe anomalous Hall resistivity . In order to increase theanomalous contribution to the Hall effect and therebyincrease the number of polarized carriers, we have fabri-cated a nano-granular (NG) metallic iron (Fe) in semi-conducting TiO − δ matrix. The intent was to introducea substantial magnetization component internally to thesemiconductor TiO − δ .The basic difference between our film composite andthe previously reported magnetic semiconductors by oth-ers is that in our films the NG of metallic Fe co-exist ina metallic state, whereas magnetic semiconductors pre-pared by others the transition metal co-exists as metaloxides and often oxide clusters . This major differ-ence is important in terms of the magnetic and trans-port properties of the magnetic semiconductor presentlyproduced by us and that of others . For example,NG metallic Fe contains a significant higher moment andCurie temperature than any other transition metal ox-ides. Also, the presence of NG metallic Fe allows forthe creation of a reservoir of conduction electrons in theconduction band and, therefore, holes in the TiO − δ ma-trix. As electrons from the conduction band of TiO − δ are thermally introduced into the metallic Fe conductionband, holes are created in TiO − δ much like in junctionsof NG metal/semiconductor interfaces or in Metal OxideSemiconductor (MOS) devices . Conduction of holesoccurs in the TiO − δ host. This mechanism gives rise tolower resistivity at high temperature in contrast to thepure TiO − δ reported earlier where the carriers wereonly electrons. No such mechanism is possible in mag-netic semiconductors doped with transition metal oxides.We report here that in our composite films of NG metal-lic Fe in anatase TiO − δ the magnetization, 4 πM S ≈ ≈ . µ B /Fe), at room temperature, T c above 800 K, the carriers are strongly spin polarized andthe room temperature resistivity is lowered by a factor of ≈ − relative to films of the undoped TiO − δ semicon-ductors previously produced . Experimental results,discussions and concluding remarks follow. II. EXPERIMENTAL RESULTS AND ANALYSIS
Thin films consisting of nano granular (NG) metallicFe and TiO − δ were deposited by a pulsed laser abla-tion deposition (PLD) technique from binary targets ofTiO and metallic Fe on (100) LaAlO substrates. Tar-gets of TiO and Fe were mounted on a target rotatordriven by a servomotor and synchronized with the trig-ger of the pulsed excimer laser λ = 248 nm. In eachdeposition cycle, the ratio of laser pulses incident uponthe TiO target to those upon the Fe target was 6:1, thistechnique denoted as alternating targets-pulsed laser ab-lation deposition (AT-PLD) technique. The substratetemperature, laser energy density, and pulse repetitionrate were maintained at 700 o C , ≈ . / cm , and 1 Hz,respectively. The deposition was carried out in a pureoxygen background of around 10 − torr in order to in-duce defects in the TiO host. There were a total of4,206 laser pulses (3,606 pulses on TiO target and 600pulses on Fe target) for each film resulting in a thicknessof approximately 200 nm as measured by a Dek-Tek stepprofilometer.Crystallographic and micrographic properties of theAT-PLD films were measured using x-ray diffractome-try (XRD) and transmission electron microscopy (TEM).Results indicate phases of anatase TiO , metallic bodycentered cubic (bcc) Fe. (001) plane of anatase TiO − δ and (110) plane of bcc Fe phase are clearly exhibited inthe xrd pattern shown in FIG.1, whereas peak ap-
20 40 60 80 100 120 b cc ( ) b cc ( ) S ( ) S ( ) S ( ) S ( ) T i O ( ) T i O ( ) Log ( C oun t s ) FIG. 1: X-ray diffraction pattern shows anatase of (001)
TiO − δ and (110) bcc phase. peared at 2 θ = 32 . o could not be readily indexed.From our speculation, the peak may be originated by pos-sible presence of iron oxide (FeO or Fe O ), or ilmenite(FeTiO ) phase.In FIG.2, reflection electron diffraction (see upper in-set) supports the existence of the epitaxial TiO − δ filmhost while the TEM image (and lower inset) reveal thepresence of nano granular (NG) metallic Fe particles sus-pended within the TiO − δ host. A JOEL 2000EX highresolution transmission electron microscope operating at200 keV and 400 keV was used in the analysis. A HRTEMimage shown in FIG.2 illustrates that NG metallic Fespheres were formed at the interface between TiO − δ layers and LaAlO substrate. Average diameter of NGspheres was measured to be in the order of ≈
10 to 15 nmas shown in the TEM image. The image suggests thatthe AT-PLD process may lead implantation of NG metal-
FIG. 2: TEM images for a representative film of nano granularFe in TiO − δ . -4 -3 -2 -1 pure TiO (Fe)TiO Log [ ( T ) ( c m ) ] T (K)
FIG. 3: Resistivity, ρ as a function of temperature, T , for thenano granular Fe in TiO − δ film (dashed line) and the pureTiO − δ (solid line) from the reference . lic Fe spheres into host TiO − δ . Average atomic ratioof Fe/Ti ions of the AT-PLD films was obtained by en-ergy dispersive x-ray spectroscopy (EDXS) within an ul-tra high resolution scanning electron microscope (UHR-SEM) column of Hitachi S-4800 that ≈ ± .
5% of Feare possibly presented in the 1 × area of the films.For steady currents and with the magnetic intensity H directed normal to the film, the resistance matrix R inthe plane of the film may be written as R = (cid:18) R xx R xy R yx R yy (cid:19) = 1 t (cid:18) ρ − ρ H ρ H ρ (cid:19) (1)wherein ρ and ρ H represent, respectively, the normal-resistivity and the Hall resistivity. A ρ as a functionof temperature for the film was measured in an appliedfield of H = 0 Oe and shown in FIG.3. We note thatthe value of ρ ( T ) was quite small at high temperature.The ρ ( T ) behavior for pure TiO − δ film (solid line) isalso shown in FIG.3 exhibiting a typical metal-insulatortransition in the temperature range of 4 K and 300 K. Incontrast to the ρ for pure TiO − δ , ρ for the compositeNG metallic Fe and TiO − δ film is a factor of 1000 lowerand is constant for temperatures between 225 K and 400K. The temperature variations are of the characteristicform expected from metal semiconductor interfaces. The ρ value at 300 K was measured to be 183 µ Ω − cm, and isabout a factor of 20 larger than resistivity of pure Fe .Presence of these unique transport properties in theNG metallic Fe spheres embedded in TiO − δ films havebeen modeled as coexisting electronic structures of bothsemiconducting TiO − δ and NG metallic Fe. We havemodeled the mechanism for transport in this compos-ite film in a sketch shown in FIG.(4)a . A com-mon chemical potential energy or Fermi energy in bothTiO − δ and metallic Fe implies that the conduction bandof TiO − δ is degenerate with the metallic Fe conduction FIG. 4: (a) Energy band structure of nano granular metallicFe and semiconducting TiO − δ , where it is similar to metaloxide semiconductor. (b) Contact potential model. band. Since the Fermi energy in metallic Fe is about4.6 eV above the conduction energy level it implies thatthe energy band gap (3 eV) of TiO − δ is degeneratewith the iron conduction band. This means that thedonor levels in TiO − δ are also degenerate, and thereby,electrons hopping between Ti and Ti sites wouldfind a conduit into the metallic conduction band leav-ing behind holes in TiO − δ . Therefore, a small potentialbarrier must exist at the interface. This is illustratedschematically in FIG.4a. As noted in FIG.2, the NGmetallic Fe sphere particles are isolated or disconnected,implying that conduction in the TiO − δ is via hole con-duction. This can be also explained by contact poten-tial at the interface between NG metallic Fe sphere andanatase TiO − δ as described in Landau Lifshitz . In or-der to remove electron trough the surface of a metallicFe, work must be done thermodynamically. Accordingto the reference , Poisson’s equation for potential Φ( x )along the x − axis normal to the interface between NGmetallic Fe / TiO − δ implies − Φ ′′ ( x ) = 4 πρ ( x ) ,̟ = Z ∞−∞ xρ ( x ) dx = − π Z ∞−∞ x Φ ′′ ( x ) dx, Φ( ∞ ) − Φ( −∞ ) = V c = 4 π̟, (2)
100 150 200 250 300010203040 p ( x / c m ) T (K) H ( H ) ( x - c m ) T (K) H (H=0 kOe) H (H = 50 kOe) FIG. 5: (Hall resistivity, ρ H ( H ), versus temperature for H= 0 (circle symbol) and H = 50.0 kOe (dashed line) in thenano granular Fe in TiO − δ film. Inset shows relation carrierdensity in function of temperature.. wherein ̟ is defined to be dipole moment per unit con-tact area and V c is contact potential. The schematic ofcontact potential at the interface is shown in FIG.4b.Electron conduction between metallic spheres may notbe possible, see FIG.2. In FIG.5, Hall resistivity is plot-ted as a function of temperature that at high tempera-tures carriers are holes which also confirmed by Seebeckmeasurements. Furthermore, the number of carriers rel-ative to pure TiO − δ has increased by factor 1,000 andthe mass of holes is approximately about 10 times largerthan the electron mass . Also, in FIG.5 the carrier holedensity p is plotted as a function of temperature. In orderto calculate p , we employ ρ H = R B + 4 πM R S and R = 1 pec , (3)wherein R and R S are the normal and anomalous Hallcoefficients, respectively. In FIG.6, ρ H ( H ) scales linearlywith magnetic field for temperatures below 300 K. Theslope of ρ H ( H ) vs . H gives rise to the normal Hall coeffi-cient which corresponds to the first term in Eq.(3), wherethe hole carrier density, p , may be deduced. For temper-atures well above 250 K, p is governed by the secondterm in Eq.(3), the anomalous Hall coefficient. Eq.(3)may re-written as ρ H = Hpec + (cid:18) pec + R S (cid:19) (4 πM ) where R S ≫ R . (4)The anomalous Hall effect is dominant at high temper-atures ( T >
250 K) and the second term in Eq.(4) isconstant for saturation magnetization M = M S .The spontaneous magnetization, M ( H, T ), was mea-sured to be nearly constant as a function of temperatureas shown in FIG.7. The measurement was performedwith an external dc-magnetic field of 10.0 kOe applied
FIG. 6: (a) Magneto resistivity, ρ ( H ) versus H in differenttemperatures. (b) Hall resistivity, ρ H ( H ) versus H in differenttemperatures. normal to the film plane (out-of-plane measurement).The film can be fully saturated with a field of ≈ M ( H, T )data in FIG.7 shows no difference which implies that nospin glass effects are present in the films. There aretwo sources for magnetism in this composite structure:(1) ferromagnetism in TiO − δ and (2) ferromagnetism inmetallic Fe spheres. The ferromagnetism in TiO − δ isdue to double exchange as calculated by us in a previouscalculation and it gives rise to a relatively small satura-tion magnetization at room temperature ( ≈
400 Gauss).Based on the data presented in FIG.2, we estimate asphere density of 0.125 to 0 . × spheres / cm andsphere diameter between 10 and 15 nm. This implies thatthe loading factor of metallic Fe in the composite film isin the order of 0.05 to 0.25 which correspondsto 1,100 to 5,100 Gauss for the saturation magneti-zation of the composite. This compares with a mea-sured value of 3,100 Gauss. We have measured thetransverse magnetoresistivity, ρ ( H ), and Hall resistiv- M ( H ) ( x G ) T (K)
H= 10 kOe FC H = 10 kOe ZFC
FIG. 7: Static magnetization, M ( H, T ), versus temperaturewith H = 10.0 kOe normal to the film plane. ity, ρ H ( H ), as a function of magnetic field sweeps be-tween - 90 and + 90 kOe as shown in FIGS.6a and6b, and at different temperatures. In FIG.6a, the nor-mal resistivity exhibits negative magnetoresistivity de-fined as, ∆ f = ( ρ ( H ) − ρ (0)) /ρ (0) for temperatures100 K < T < K . ∆ f is negligible for temperaturesabove 300 K. However, plots of ρ H ( H ) shown in FIG.6b,clearly demonstrate that saturation effects at tempera-tures near room temperature as a result of the hole carrierdensity increasing toward saturation levels, see FIG.3.Notice that plots of ρ H ( H ) at T ≥ K exhibit clearhysteresis loop behavior similar to ferromagnetic hystere-sis loops. Interestingly, saturation for both ρ H ( H ) and( M/M S ), normalized magnetization, occurred at exactlythe same external magnetic field value of ≈ π/
3) to magnetically satu-rate a sample of spherical shape in order to overcome a -20 -15 -10 -5 0 5 10 15 20-0.70-0.350.000.350.70 -1.5-1.0-0.50.00.51.01.5 at 300K H ( x - - c m ) H (kOe) at 300 K M / M S FIG. 8: Symbol (X) shows a typical anomalous Hall hysteresisloop, ρ H ( H ), at T = 300 K with magnetic hysteresis loop(symbol (o)) at 300 K. magnetic field of 6.8 kOe in Fe spheres at (say) roomtemperature. These data also show relatively smallercoercive field for the ( M/M S ) hysteresis loop than forthe ρ H ( H ) hysteresis loop. The difference in hysteresisloops is due to the fact that the NG Fe spherical samplesgive rise to dipole internal magnetic fields. In order toreverse the magnetization in each sphere it requires anexternal field to overcome this interactive internal field.Hence, the coercive field is approximately constant withtemperature as it scales with magnetization. In the Hallresistivity measurement the coercive field is strongly tem-perature dependent, since there are two contributions tothe resistivity measurement: (1) Normal Hall resistivitycontribution which is not hysteretic with respect to H and (2) the anomalous Hall effects which is hystereticwith H , since this effect is proportional to the magne-tization. At high temperatures the AHE contributiondominates the resistivity measurement in contrast to lowtemperatures where the normal Hall contribution domi-nates. Thus, spin polarization as reflected in ρ H ( H ), seeFIG.8, correlates extremely well with the magnetic hys-teresis loop implying that the magnetized Fe spheres arepolarizing the carriers. III. DISCUSSION AND CONCLUSIONS
The plots of ρ H ( H ) at T < K exhibit paramag-netic hysteresis behavior, see FIG.6b. This transitionbetween paramagnetic and ferromagnetic ρ H ( H ) hys-teresis behavior was also reported and correlated withhole carrier density related with RKKY theory in pre-vious research . Their estimate of the threshold holedensity for the carrier spin polarization transition was p = 3 × / cm from paramagnetic p < × / cm to ferromagnetic p ≥ × / cm behavior . Ourestimate of hole carrier density of the films was p ≈ . × / cm and p ≈ × / cm at T = 100 K and200 K, respectively. The film exhibited strong polariza-tion and hysteresis behavior as a function of applied fieldat 300 K. Since anomalous Hall effects may be observedin the presence of spontaneous magnetization , we in-fer that about ≈ / cm hole carriers are polarizedby an applied field H . This indicates that nearly all ofthe carriers were polarized by the spontaneous magneti-zation at T = 300 K. According to FIG.8, carrier polar-ization correlated very well with the magnetic hysteresisloop of NG Fe spheres. Also, FIG.8 indicates that thecarrier polarization is not affected by external fields upto 3.0 kOe, which can be an advantage for memory de-vice applications However, if the shape of the par-ticles embedded in the composite could be shaped intoneedles, it would imply spin polarization of carriers inexternal magnetic fields below 0.1 kOe which is ideallysuited for spintronics applications.In summary, magnetic and magneto-transport data forfilms of nano granular metallic Fe and oxygen defectedTiO − δ are reported. The essence of this paper showedthat conduction carriers of the films were strongly cou-pled to residual magnetic moments of metallic Fe grainsin the nano composite structure. The dramatic reductionof normal resistivity ( ρ ( T )) of the films is a consequenceof two factors: (1) oxygen defects in the TiO − δ host in-duced electron hopping; (2) electrons from the TiO − δ were introduced into the conduction band of Fe to createholes in TiO − δ similar to a Metal Oxide Semiconduc-tor (MOS) structure. As a result the number of carriersincreased at room temperature. The holes in TiO − δ were polarized due to the presence of ferromagnetic nanogranular metallic Fe, where the carrier polarized density was measured to be near ≥ . × / cm . Therefore,spintronics and spin dependent memory applications canbe based upon the results presented here. IV. ACKNOWLEDGMENT
This research was supported by the National ScienceFoundation (DMR 0400676) and the Office of Naval Re-search (N00014-07-1-0701). ∗ Corresponding author e-mail: [email protected] † Electronic address: [email protected] ‡ Electronic address: [email protected] § Electronic address: [email protected] H. Ohno,
Science , 951 (1998). K. Ueda, H. Tahata, and T. Kawai,
Appl. Phys. Lett. ,988 (2001). Y. Masumoto et al., Science , 854 (2001). S.J. Pearton et al., Mat. Sci. and Engn.
R40 , 137 (2003). S.A. Chambers and R.F.C. Farrow,
MRS Bulletin , 729(2003). I. ˘Zuti´c, J. Fabian, and D. Sarma,
Rev. Mod. Phys. . 32(2004). J.M.D. Coey, M. Venkatesan, and C.B. Fitzgerald,
Nat.Maters. , 173 (2005). N. H. Hong et al., Phys. Rev. , B 73 , 132404 (2006). S.D. Yoon et al., J. Phys.; Condens. Matter. , L355(2006). S.D. Yoon et al., J. Phys.: Condens. Matter.
9, 326202(2007). M. Earle,
Phys. Rev. , 56 (1942). R.G. Breckenridge and W.R. Hosler,
Phys. Rev. , 793(1953). N. Daude, C. Gout, and C. Jouanin,
Phys. Rev.
B 15 ,3229 (1977). J. Pascual, J. Camassel, and H. Mathieu,
Phys. Rev.
B 18 ,5606 (1978). H. Tang et al., Solid State Commun. , 847 (1993). S.A. Campbell et al., IBM J. Res. Develop. , 383 (1999). B.G. Streetman, “Solid State Electronic Devices,” ch. 8,301, Prentice Hall, Inc., Englewood Cliffs, New Jersey(1990). Natl. Bur. Stand. (U.S.) Monogr. , 82 (1969). C.J. Howard, T.M. Sabine, and F. Dickson,
Acta Crystal-lographica, B: Structural Science
B47 , 462 (1991). H.E. Swanson, J.C. Doukhan, and G.M. Ugrinic,
Natl.Bur. Stand. (U.S.), Circ. , 539 (1955). L.D. Landau and E.M. Lifshitz, “Electrodynamics of con-tinuous media,” R.C. Weast and M.J. Astle, “CRC Handbook of chemistryand physics,” section E, 81, CRC press, Inc. Boca Raton,Florida, 63rd edition (1982). M.L. Knotek and P.J. Feibelman,
Phys. Rev. Lett. , 964(1978). Z. Zhang, S.P. Jeng, and V. Henrich,
Phys. Rev.
B 43 ,12004 (1991). C. M. Hurd, “The Hall Effect in Metals and Alloys,” T. Story, et al., Phys. Rev. Lett. , 777 (1986). C.L. Chien and C.R. Westgate, “The Hall effect and its ap-plications,”
Plenum Press, New York and London (1980). R. Karplus and J. M. Luttinger,
Phys. Rev. , 1154(1954). G. Prinz,
Science , 1660 (1998). S.A. Wolf et al., Science294