Nanoengineered Curie Temperature in Laterally-patterned Ferromagnetic Semiconductor Heterostructures K. F. Eid, B. L. Sheu, O. Maksimov, M. B. Stone,∗ P. Schiffer, and N. Samarth† Dept. of Physics and Materials Research Institute, The Pennsylvania State University, University Park PA 16802 5 0 Wedemonstratethemanipulation oftheCurietemperatureofburiedlayersof theferromagnetic 0 semiconductor(Ga,Mn)Asusingnanolithographytoenhancetheeffectofannealing. Patterningthe 2 GaAs-capped ferromagnetic layers into nanowires exposes free surfaces at the sidewalls of the pat- terned(Ga,Mn)AslayersandthusallowstheremovalofMninterstitialsusingannealing. Thisleads n toanenhancedCurietemperatureandreducedresistivitycomparedtounpatternedsamples. Fora a fixedannealing time, theenhancement of theCurie temperature is larger for narrower nanowires. J 3 PACSnumbers: 75.50Pp,75.75.+a,81.16.-c 1 ] i Heterostructuresderivedfromthe ferromagneticsemi- followed by lift off and dry etching. c conductor (Ga,Mn)As are important for studying spin- s For the patterning process, the samples are first spin - dependent device concepts1,2,3,4,5 relevant to semicon- coated with a bilayer of about 400-nm-thick electron- rl ductor spintronics.6,7,8 In contrast to metallic ferromag- beam resist P(MMA-MAA) copolymer/PMMA with mt nets where crystalline defects rarely affect the magnetic molecularweightof950. Thedesiredpatternsaredefined ordering temperature, it is now well established that on the sample using direct-write electron-beam lithogra- . t Mn interstitial defects – double donors that compensate phy at electron energy of 100 keV. After development a holes – play a key role in limiting the hole-mediated m of the resist in MIBK:IPA 1:1 solution, a metallic layer Curie temperature (TC) of (Ga,Mn)As to below 110 K of either 45 nm Al or Al/Au is deposited on the sam- - in as-grown samples.9 In thin (Ga,Mn)As epitaxial lay- d ple using thermal evaporation. Using standard lift off n ers, post-growthannealing10,11 drives Mn interstitials to techniques, we then obtain metal wires that serve as a o benign regions at the free surface where they are pas- hard mask for the subsequent chlorine-based dry etch- c sivated, resulting in significant increases in both the ing process. After dry etching, for samples with an Al [ hole density (p) and TC (which can be as high as 160 mask, the metal layer is dissolved in CD-26 photoresist 2 K).12,13,14 Such values of TC should in principle allow developer, leaving just the semiconductor nanowire; for v the fabricationofproof-of-conceptspintronicdevices op- samples with an Al/Au mask, the metal is retained on 8 erating above liquid nitrogen temperatures. Unfortu- portions of the sample. We note that in the latter case 9 nately, the annealing-induced enhancement of T is al- C the undoped GaAs capping layer serves as an insulator 2 most completely suppressed in heterostructure devices 1 that prevents the shunting of current through the metal 0 containing buried (Ga,Mn)As layers.15,16 In this Letter, during the measurement.17 Figures 1 (a) and (b) show 5 we describe a nanoengineered solution to this problem: plan view SEM images of patterned nanowires. Three 0 lithographic patterning of (Ga,Mn)As heterostructures identical rows of wires are patterned on each wafer: one t/ into nanowires. Such patterning opens up new path- set is measured as-grownand patterned, while the other a waysfor defectdiffusion to free surfaces atthe sidewalls, two sets are annealed after patterning at 190◦C for 5 m and thus restores the annealing-enhanced TC to buried hours. - (Ga,Mn)Aslayers. Theseresultssuggestnovelroutesto- d We probe the ferromagnetic phase transition in single wards the flexible fabrication of (Ga,Mn)As-based spin- n wiresusingfourprobemeasurementsofthetemperature- tronic devices with relatively high T . o C dependent resistivity ρ(T); it is well-established7,8 :c Laterally-patterned wires are fabricated from that ρ(T) shows a well-defined peak close to TC in v GaAs/(Ga,Mn)As/GaAs heterostructures grown by (Ga,Mn)As, especiallyforsampleswithTC <100K.For Xi molecular beam epitaxy on epiready semi-insulating higher TC, the peak is less well-defined but still yields a reasonableestimate of the ordering temperature (typi- GaAs (100) substrates. Crystal growth conditions are ar identical to those described in an earlier publication.15 callyanoverestimateof∼10K).Inaddition,wemeasure the temperature-dependence of both the magnetization The samples consist of a 5 nm thick low temperature (using a commercial superconducting quantum interfer- GaAs buffer layer, followed by a (Ga,Mn)As layer (∼ 6%Mn, thickness of 15 nm or 50 nm) and finally a ence device (SQUID) magnetometer) and the resistivity inmacroscopicpiecesoftheparentwafers(usingthevan low temperature grown 10 nm GaAs capping layer. The der Pauw method). data shown are for 50 nm thick magnetic layers,but the results were qualitatively the same for the 15 nm thick Figure 2 (a) shows the magnetization as a function of samples. We pattern these samples into wires that are temperature for both as-grown and annealed (∼ 5mm2) approximately 4 µm long and with two different widths pieces of a GaAs/50 nm (Ga,Mn)As/10 nm GaAs het- (1 µm and 70 nm) using electron beam lithography erostructure. As found in earlier studies,15 the cap layer 2 30 (a) [110] (a) 0] 3m) 25 11 c 20 H = 50 Oe [ u/ (In-plane) m 15 Annealed M (e 10 As-grown 5 300 nm 0 5.0 (b) (b) m) 4.5 Annealed c As-grown [110] -(cid:49)4.0 m 10] 300 nm ((cid:108)3.5 1 [ 3.0 0 50 100 150 200 250 300 T (K) FIG. 2: (a) Magnetization versus temperature for macro- scopic pieces (area of ∼ 5mm2) of as-grown and annealed FIG.1: SEMimagesofnanowirespatternedalongtwodiffer- GaAs/(Ga,Mn)As (50 nm)/ (10 nm) GaAs heterostructure. ent crystallographic directions: (a) [110] and (b) [010]. The data are measured in a magnetic field of 50 Oe in-plane upon warming after field cooling at 10 kOe. The sample is annealed at 190◦C in ultrahigh purity nitrogen gas (5N) for completely suppresses any annealing-induced enhance- 5 hours. The data show that TC is not affected by anneal- ing. (b) Resistivity versustemperature(measured in vander ment of TC. This suppression of the annealing effect has Pauw geometry) for a 1 mm × 2 mm mesa patterned from been attributed to the formation of a p-n junction at thesame wafer as in (a). eachofthe twoGaAs/(Ga,Mn)AsinterfacesassomeMn interstitials (double donors) initially diffuse across the boundaries into GaAs.14,15 The resulting Coulomb bar- effect has been seen in previous studies of long anneals, rier is expected to strongly inhibit the further diffusion albeit at higher annealing temperatures.11 Figure 3 (b) of Mn interstitials from the bulk of the (Ga,Mn)As layer shows the effect of annealing for a 70 nm wide nanowire towards the interfaces. The temperature-dependence of (also patterned along [110]). Here, annealing produces a the sample resistance is in complete agreement with the striking increase in T of almost 50 K, accompanied by magnetization measurements. Figure 2(b) shows Van C a correspondingly significant decrease in the resistivity der Pauw measurements of ρ(T) for a macroscopic mesa (1mm×2mm); the peak of ρ(T) does not change upon ofthe wire. Boththeseobservationsstronglysuggestthe successfulremovaloftheMninterstitialsfromthebulkof annealing. In addition, there is no significant reduction the(Ga,Mn)Asnanowire. LateraldiffusionoftheMnin- intheresistivityofthesamplewithannealing,indicating terstitials provides the most likely explanation for these that the hole density has not changed. observations. We note that the time and length scales Lateralpatterningproducesdistinctlydifferentbehav- for diffusion observed in our experiments (∼ 35 nm in ior. Figure 3(a) shows the temperature-dependent resis- 5 hours) are consistent with earlier low-temperature an- tivityfora1µmwidewirepatternedalongthe[110]direc- nealing studies of (Ga,Mn)As epilayers.14 tion. The data are shown for both an as-grownwire and one annealed at 190◦C for five hours. The data indicate It is important to examine whether defect-diffusion thatannealingresultsinaslightincreaseinT ,accompa- may depend on the crystalline direction and/or the na- C nied by a slight increase in the resistivity of the sample. ture of the free surface where the interstitials eventu- These observations suggest that annealing induces mod- ally reside. We hence pattern four nanowires (each with est diffusion of Mn interstitials to the free surfaces at 70 nm width), along the principal cubic directions ( the sidewalls of the wire. The diffusion coefficient (esti- [110],[100],[110] and [010]). All four nanowires are an- matedtobeD ∼100nm2/hrat∼190◦C14)issimplynot nealed under identical conditions (190◦C for 5 hours). large enough to allow significant removal of Mn intersti- The results are shown in Fig. 3(c) and indicate that at tials from the bulk of the 1µm wide wire within the five least for the processing and annealing protocol followed hours annealing time. There is however some alteration in this study defect diffusion does not vary significantly of the defect states, as indicated by the small increase with crystalline direction. in the high temperature resistivity; we do not have an In summary, we have shown that nanolithography al- explanation for this observation but note that a similar lows the engineering of defect diffusion pathways, hence 3 providing a means of tailoring impurity-controlled mag- netism in ferromagnetic semiconductor heterostructures. 7 (a) 1µm wire Areas with different T and resistivity can be made on Annealed C 6 As-grown the same wafer simply by having different feature sizes. 5 Smaller features have a higher Curie temperature and 4 lower resistivity. Although we do not observe any obvi- 3 ouscrystallineanisotropyinthediffusionconstantofthe Mninterstitials,suchdirectionaldependencemaystillex- 12 (b) 70 nm wire m) 3.6 m) ist and may be revealed by in situ resistance monitoring m.c(cid:49) 10 Annealed 3.2 m.c(cid:49) wfohrilperoasnpneecatlsinogftdheesinganninowgi(rGesa.,MOunr)Afisn-bdainsgesdasupginutrrowneilcl ((cid:108) 8 As-grown 2.8 ((cid:108) device heterostructures such as magnetic tunnel junc- tions with T well above 77 K. In addition, systematic 6 2.4 C 4 (c) 70 nm wires studies of annealing of nanowires of differing width may providenewinsightsintoongoingcontroversiesaboutthe 3 microscopicdetailsofthediffusionandpassivationofMn I//[100] interstitials in (Ga,Mn)As.18 I//[110] 2 I//[010] I//[110] 0 50 100 150 200 250 300 T (K) FIG. 3: Temperature dependence of the (four-probe) resis- tivity for as-grown and annealed single wires of (a) 1 micron This research has been supported by grant numbers widthand(b)70nmwidth. Both setsofwires arepatterned ONR N0014-05-1-0107, DARPA/ONR N00014-99-1093, along the [110] direction. Plot (c) shows the same measure- -00-1-0951, University of California-Santa Barbara sub- mentsfor fourindividual70 nm wires patterned along differ- contractKK4131,andNSFDMR-0305238and-0401486. ent crystalline orientations. All wires are patterned from the This work was performed in part at the Penn State same wafer used in Fig. 2, and the annealing conditions are NanofabricationFacility, a member of the NSF National identical to those used in Fig. 2. Nanofabrication Users Network. ∗ Condensed Matter Sciences Division, Oak Ridge National 10 T. Hayashi, Y. Hashimoto, S. Katsumoto, Y. Iye, Appl. Laboratory, Oak Ridge TN 37831-6430. Phys. Lett.78, 1691 (2001). † Electronic address: [email protected] 11 S. J. Potashnik, K. C. Ku, S.H. Chun, J. J. Berry, N. 1 Y. Ohno, D. K. Young, B. Beschoten, F. Matsukura, H. Samarth, P. Schiffer,Appl. Phys.Lett. 79, 1495 (2001). Ohno, and D. D.Awschalom, Nature 402, 790 (1999). 12 K.W.Edmonds,K.Y.Wang,R.P.Campion, A.C.Neu- 2 D.Chiba,M.Yamounichi,F.Matsukura,H.Ohno,Science mann, C.T. Foxon, B. L. Gallagher, P. C. Main, Appl. 301, 943 (2003). Phys. Lett.81, 3010 (2002). 3 M. Tanaka and Y. Higo, Phys. Rev. Lett. 87, 026602 13 K. C. Ku, S. J. Potashnik, R. F. Wang, S. H. Chun, P. (2001). Schiffer, N. Samarth, M. J. Seong, A. Mascarenhas, E. 4 S.H.Chun,S.J. Potashnik,K.C.Ku,P.Schiffer,andN. Johnston-Halperin, R. C. Myers, A. C. Gossard, and D. Samarth, Phys. Rev.B 66, 100408(R) (2002). D. Awschalom, Appl.Phys,Lett. 82, 2302 (2003). 5 D.Chiba,Y.Sato,T.Kita,F.Matsukura,H.Ohno,Phys. 14 K.W. Edmonds, P.Boguslawski, K.Y.Wang, R.P. Cam- Rev.Lett. 93, 216602 (2004). pion, S.N. Novikov, N.R. Farley, B.L. Gallagher, C.T. 6 S. A. Wolf, D. D. Awschalom, R. A. Buhrman, J. M. Foxon, M. Sawicki, T. Dietl, M. Buongiorno Nardelli, J. Daughton,S.vonMolnar,M.L.Roukes,A.Y.Chtchelka- Bernholc, Phys.Rev.Lett. 92, 037201 (2004). nova, and D. M. Treger, Science 294, 1488 (2001). 15 M. B. Stone, K.C. Ku, S.J. Potashnik, B.L. Sheu, N. 7 H.OhnoinSemiconductorSpintronicsandQuantumCom- Samarth, P. Schiffer,Appl. PhysLett. 83, 4568 (2003). putation, D. D. Awschalom, D. Loss, and N. Samarth 16 D. Chiba, K. Takamura, F. Matsukura, H. Ohno, Appl. (Eds.), (Springer-Verlag, Berlin, 2002), p.1. Phys. Lett.82, 3020 (2003). 8 N. Samarth in Solid State Physics, Vol. 58, H. Ehrenre- 17 Thisisconfirmedbycomparingtheresistivityofthewires ich and F. Spaepen (Eds.), (Elsevier Academic Press, San with van der Pauw resistivity measured on samples that Diego, 2004), p. 1. had no metal. 9 K. M. Yu, W. Walukiewicz, T. Wojtowicz, I. Kuryliszyn, 18 M. Adell, J. Kanski, L. Ilver, V. Stanciu, P. Svedlindh, X.Liu,Y.Sasaki,J.Furdyna,Phys.Rev.B65,201303(R) cond-mat/0412006. (2002).