Home Search Collections Journals About Contact us My IOPscience Magneto-transport properties of magnetic tunnelling transistors at low and room temperatures This content has been downloaded from IOPscience Please scroll down to see the full text 2006 Nanotechnology 17 3359 (http://iopscience.iop.org/0957-4484/17/14/004) View the table of contents for this issue, or go to the journal homepage for more Download details: IP Address: 137.189.170.231 This content was downloaded on 02/10/2015 at 07:42 Please note that terms and conditions apply INSTITUTE OF PHYSICS PUBLISHING NANOTECHNOLOGY Nanotechnology 17 (2006) 3359–3365 doi:10.1088/0957-4484/17/14/004 Magneto-transport properties of magnetic tunnelling transistors at low and room temperatures H D Quang1, C X Huu2 , S K Oh1 , V S Dang1 , N H Sinh3 and S C Yu1,4 Applied Physics Laboratory, Department of Physics, Chungbuk National University, 361-763 Cheongju, Korea Institute of Physics, Academia Sinica, Nankang, Taipei, Taiwan, Republic of China Cryogenic Laboratory, College of Natural Science, Vietnam National University, 334 Nguyen Trai, Thanh Xuan, Hanoi, Vietnam E-mail: scyu@chungbuk.ac.kr Received 15 February 2006, in final form May 2006 Published 15 June 2006 Online at stacks.iop.org/Nano/17/3359 Abstract Si(100)/CoFe/AlOx /CoFe/FeMn/Cu/Ta magnetic tunnelling transistors (MTTs) with differing base thicknesses (W ) were investigated The magneto-transport properties of the MTTs were measured at 77 K and room temperature (RT) We obtained magneto-current ratios of 48.3% and 55.9% for emitter–base bias voltages of 1.45 and 2.0 V, respectively, at 77 K The transfer ratios are 2.83 × 10−5 and 1.52 × 10−4 , respectively, corresponding to bias voltages of 1.45 and 2.0 V Moreover, the highest tunnel magneto-resistance (TMR) ratios turned out to be 12% and 20% for a base ˚ at RT and 77 K, respectively These properties raise not thickness of 30 A only some fundamental questions regarding the phenomenon of spin-independent tunnelling at low and room temperatures, but also show some promising aspect for magneto-electronic applications In addition, we attempted to elucidate the reason behind the outstanding TMR effect at low and room temperatures Finally, the origin of the decrease in the mean free path asymmetry (λ↑ /λ↓ ) was clarified by using x-ray photoelectron spectroscopy profile analysis of the elements existing in the interface between Si and the CoFe base (Co, Fe, Al, Si, O) (Some figures in this article are in colour only in the electronic version) Introduction Recently, many researchers have investigated spin valve structures with nano-oxide layers (NOLs) or specular spin valves (SSVs) in the current-in-plane (CIP) configuration [1] But they have found many serious problems in achieving a sufficient quality for read head device applications with a density of over 300 Gbit in−2 In these structures, the giant magneto-resistance (GMR) sensitively depends on either the thickness of the free, pinned and spacer layers or the mean free path (MFP) of the electrons Generally, in a spin valve majority Author to whom any correspondence should be addressed 0957-4484/06/143359+07$30.00 carrier electrons with a long MFP can travel with low resistance through a multilayer in which an applied magnetic field aligns the magnetization of adjacent layers Moreover, if a NOL is inserted on top of the pinned layer to enhance the GMR effect in the SSVs, which induces more frequent scattering, then the magneto-resistance (MR) ratio is seriously degenerated at high temperatures (>250 ◦ C) [2] Thus, there has been a lot of effort made to get rid of this problem by using a currentperpendicular-to-plane (CPP) configuration in many different structures, such as magnetic tunnelling junctions (MTJs), spin-valve transistors (SVTs), magnetic tunnelling transistors (MTTs) and organic spin valves [3–15] The CPP configuration is expected to solve most of the problems because electrons © 2006 IOP Publishing Ltd Printed in the UK 3359 H D Quang et al cross all magnetic layers, whereas in the CIP configuration the MR is diminished by shunting and channelling, and diffusive surface scattering also reduces the MR ratio Previous studies of MTT structures have revealed a number of factors affecting the characteristics of a device, such as collector current IC , the transfer ratio α (which is defined as α = IC /IE , where IE is emitter current), emitter and collector Schottky barrier heights, the choices of material for the nonmagnetic layers, and the thicknesses of the nonmagnetic layers In MTTs, the transfer ratio strongly depends on electron transmission at the metal/semiconductor surface The Schottky barrier height is thought to be an important factor in the determination of the electron transmission coefficient at the interface The difference in the transfer ratio is due to different scattering mechanisms in the base, wherein the spin-dependent scattering can be attributed to bulk and interface scattering Thereby, we can enhance the transfer ratio without affecting the relative change in the collector current, because a small change in the collector barrier height increases the chance of collecting both spin-up and spin-down electrons equally (only if we consider the MTT structures containing ferromagnetic layers but not half-metals) [3, 4] Some parameters, such as choice of material for a nonmagnetic layer, thickness of a non-magnetic layer, change in collector current ( IC = ICP − ICAP ) at a certain thickness of the magnetic layers, a domain wall in the junction area, interfacial and barrier spin scattering, and direct coupling between two ferromagnetic layers, have to be optimized for the purpose of applications [4–7, 9, 10] Another important parameter is the magneto-current (MC) ratio, which can be expressed by (ICP − ICAP )/ICAP , where ICP and ICAP are the collector currents in parallel and anti-parallel configurations, respectively The MC ratio is sensitive to the mean free path asymmetry of majority and minority electrons, λ↑ /λ↓ [4, 6, 7] So far, the MC value is not really high enough for applications to read head elements with a density of over 300 Gbit in−2 Besides, the role of order/disorder in the barrier, electrodes, ferromagnetic/insulator interfaces (FM/I), the dependence of voltage and tunnel magneto-resistance (TMR) value, and their TMR are inadequately understood [5, 7, 9, 14] Neither is the theoretical understanding of the temperature dependence of tunnelling conductance of Al2 O3 insulators complete Nonetheless, for the case of amorphous barriers, all the results have been interpreted successfully [9, 13, 14] In this article we try to elucidate the role played by those parameters including λ↑ /λ↓ , with special emphasis on MTT structures at 77 K and RT, through both simulations and experiments Moreover, the origin of the decrease in λ↑ /λ↓ and the MC ratio are clarified by using x-ray photoelectron spectroscopy (XPS) profile analysis of the elements existing in the interface between Si and the CoFe base (Co, Fe, Al, Si, O) Experimental details The magnetic tunnelling transistors (MTTs), which have ˚ ˚ ˚ (60 A)FeMn the substrate/CoFe (W A)/AlO x (35 A)/CoFe ˚ ˚ ˚ (200 A)/Cu (100 A)/Ta (50 A) structures, were prepared at room temperature by dc and rf magnetron sputtering onto ˚ substrate Here, we used the targets Si(100)/SiO2 (1500 A) 3360 of Ta, Cu, CoFe and FeMn, where CoFe and FeMn stand for Co90 Fe10 and Fe50 Mn50 , respectively The collectors of cm) the MTTs were of prime grade P2 O5 -doped n-Si (50 substrates The background vacuum was better than × 10−4 Pa A native oxide layer on the n-Si surface was removed in a dilute HF solution prior to deposition Three different metal shadow masks were utilized to form the ferromagnetic base layer (CoFe), the tunnel barrier (AlOx ) and the emitter (CoFe/FeMn/Cu/Ta) of each MTT The base layer thickness ˚ The ohmic contacts to the (W ) was varied from 30 to 80 A n-Si substrate were made with thin Pt and thick Al layers, and the largest device dimension is 10 × 12 mm2 The magnetic moment of the CoFe emitter was pinned by an antiferromagnetic FeMn layer, and it could be switched on and off by an external magnetic field of ±800 Oe The properties of the transistor were measured with an emitter–base bias voltage range of less than 2.5 V While the emitter–base voltage ( VEB ) was maintained at zero, the emitter–base current (IEB or IE ) and the base–collector current ( IBC or IC ) were measured as a function of applied magnetic field A schematic illustration of the measurement setup is shown in figure For this, the transfer ratio of the base is related to scattering events in the base Inelastic scattering, for example by electron–electron interaction, will generally lower the hot electron energy below the collector barrier and thereby prevent collection (figure 1(c) upper panel) Elastic scattering on impurities and defects results in a change of moment without any loss of energy So an electron can scatter elastically several times, and only if the electrons within the acceptance cone [11] can contribute to the collector current (figure 1(c) lower panel) In the case of quasielastic scattering, like on phonons and magnons, both energy and momentum are important [5–7, 11, 12] Here IB is the base current, and ICP and ICAP are the collector currents corresponding to the configurations with parallel and anti-parallel magnetizations, respectively We estimated the magneto-current (MC) ratio using the expression (ICP − ICAP )/ICAP Note that hot-electron current operation at RT requires a large emitter current density, which in turn requires a smaller collector barrier area, a higher collector barrier and an enhanced base transport factor, etc [6, 9, 14] Since our MTTs have a large collector area, the leakage current is quite large, and furthermore its magnitude exceeds the MC for an injection current of 100 mA Hence, we measured the magneto-transport properties of MTT structures at 77 K to reduce the leakage current in the Schottky junctions to an acceptable level Theoretical results for the MTTs In MTTs, the transfer ratio α is usually small since only electrons surmounting the collector barrier are observed at the collector In MTTs, hot electrons leaving the emitter are injected into the base and they are spin polarized If we reduce the thickness of the magnetic layer, then the transfer ratio increases However, the MC is simultaneously reduced if the spin-dependent transmission is dominated by a volume scattering process Interestingly, the MC ratio increases when spin polarization in the emitter ( PE ) increases The increases in the spin polarization for the parallel alignment configuration ( PEP ) and in the ratio for the net transmission probabilities Magneto-transport properties of magnetic tunnelling transistors at low and room temperatures acceleration in the x -direction perpendicular to the interface within the metal/semiconductor interface region The collector only accepts electrons within a certain acceptance cone Electrons are reflected back into the base if their direction of motion (k -vector) is not within the acceptance cone And the collector current depends on an elastic scattering mechanism in the base Elastic scattering can remove electrons from the acceptance cone, but it can also scatter electrons which were already outside the acceptance cone back into it [6, 7, 11] In fact, elastic interface scattering that occurs at the metal– metal as well as at the silicon–metal interfaces is introduced by the transport parameter, D , which represents the interface diffusivity From the incident distribution of electrons, a fraction (1- D ) cross the interface without scattering, while a fraction D scatter at the interface The elastic interface scattering is assumed to be isotropic, such that a fraction D/2 of the electrons still moves in the original direction, but the other D/2 is scattered into the opposite direction Physically, this elastic interface scattering is strongly related to the defects, impurities and stacking faults at the interfaces, as well as to the abrupt change in electrical properties (band-structure) If the materials have extremely weak spin–orbit interactions and weak hyperfine interactions, then the electron spin diffusion length is especially long Then, the properties of these materials make them ideal for spin polarized electron injection and transport applications [7, 8] In the developed model, no spin-flip scattering was taken into account, so that the current of spin-up electrons can be calculated independently of the spin-down current Thus, because of the above reasons, many factors influence the magneto-transport properties of an MTT Based on the equations and optimal conditions in both theoretical and experimental results, the collector currents in parallel ( ICP ) and anti-parallel ( ICAP ) configurations can be expressed as follows [6, 7, 11, 12]: ICP = 12 (1 + PE )I E e − λW ↑ ↑ αC + 12 (1 − PE )I E e − λW ↓ ↓ αC + Ileak (1) and ICAP = 12 (1+ PE )I E e ↑ Figure A schematic diagram showing our MTT structure for cases of parallel ferromagnetic moments (a) and anti-parallel ferromagnetic moments (b) The schematic energy diagram (upper panel of (c)) and real space momentum picture (lower panel of (c)) of a spin valve transistor [11] In the energy diagram, emitter and collector Schottky barriers are shown In the momentum picture, three possible electron trajectories in the base are drawn: an electron can scatter inelastically (1) or can scatter several times elastically, after which it can arrive within (2) or outside (3) the acceptance cone of the ferromagnetic layer are very important in making the MC ratio larger [3–7, 9–14] For our MTT structure, due to the use of a spin valve in the emitter and base, scattering can be controlled magnetically When the local magnetization is anti-parallel (parallel), IC has the minimum (maximum) value owing to high (low) scattering rate Moreover, electrons are injected within a very narrow emission cone in the base, as indicated in [11] This forward focusing is due to the − λW ↓ αC↓ + 21 (1− PE )I E e− λ↑ αC↑ + Ileak , (2) W ↓ where αC and αC are the spin-independent electron collection efficiencies and are equal at the metal–semiconductor interface The MC ratio is to be expressed in the form (ICP − ICAP )/ICAP or by the equation below: MC(%) = PE − e − W λ ↑ −1 λ↓ λ↑ (1 − PE ) + (1 + PE ) e − W λ ↑ −1 λ↓ λ↑ × 100(%), (3) where W is the base thickness From this equation, we plotted the MC ratio as a function of λ↑ /λ↓ for various W/λ↑ values when spin polarization in the emitter ( PE ) is 28% (figure 2(a)) We see from this that the MC ratios easily attain the maximum value of 60–70% with high values of W/λ↑ This implies that the MC ratio is higher with a greater base thickness and/or shorter majority mean free path for hot electrons in the base The MC ratio is also plotted as a function of λ↑ /λ↓ at a given W/λ↑ of 0.65 and the polarization ranging from 25 to 45% (figure 2(b)) This plot shows that the higher 3361 H D Quang et al Figure Magneto-current ratio as a function of the mean free path asymmetry for various W/λ↑ when spin polarization in the emitter ( PE ) is 28% (a) and for various spin polarizations in the emitter ( PE ) when W/λ↑ = 0.65 (b) the MC ratio is the higher the polarization is with the same value of λ↑ /λ↓ For PE = 45%, the MC ratio reached the maximum value of 140% with λ↑ /λ↓ = This means that longer majority MFP and/or shorter minority MFP together with high polarization can provide a higher MC ratio Thus, we note that high PE of the ferromagnetic emitter and/or high λ↑ /λ↓ of the ferromagnetic base is very important for large MC ratio In this section, we assumed that (1) the leakage current was negligible, (2) the structure is perfect and (3) spin-independent electron collection efficiency is equal at the metal–semiconductor interface However, in a real experiment, the result can be quite different because all the factors affecting the MC ratio mentioned above need to be included as the following experimental section shows Experimental results and discussion In previous works, other authors showed that MTTs with a base ˚ thick exhibited a lower MC and larger layer less than 20 A switching field [4, 12] In our case, the base thickness range 3362 ˚ and the thickness of the tunnel barrier (AlOx ) was 30–80 A, ˚ The Schottky barrier height is 0.7 eV, which was was 35 A found from the fitting of I –V character curves Moreover, the leakage current ( Ileak ) was 0.2–1.0 nA when the reverse bias voltage was 0.25 V at 77 K, and it was 1.0 µA when the reverse bias voltage was 1.0 V at RT Our results can be compared with the leakage current and Schottky barrier height of 30 µA (0.1 µA) and 0.6–0.7 eV (0.7–0.8 eV) at RT given in the [6, 7], respectively Figures 3(a) and (b) show collector currents as a function of magnetic field, which were measured at given emitter– base bias voltages ( VEB ) of 1.45 and 2.0 V with a CoFe ˚ and figure 3(c) shows the MC and base thickness of 80 A, transfer ratios as a function of bias voltage of the MTTs with differing base thicknesses The MC ratios of the MTTs are 55.9% and 48.3%, whereas the transfer ratios are 2.83 × 10−5 and 1.52 × 10−4 , respectively, for emitter bias voltages of 1.45 V (figure 3(a)) and 2.0 V (figure 3(b)) From figure 3(c), we can see that the MC ratios of the MTTs increase within the bias voltage range of 1.1–1.45 V for base thicknesses ˚ of 60 and of 80 A At bias voltages above 1.45 V, the MC ratios slowly decrease, whereas the MC ratio for a base ˚ monotonically increases and attains the thickness of 30 A maximum MC value of about 55.9% at a bias voltage of 2.0 V With differing base thicknesses, the maximum MC values are not substantially different Besides, the transfer ratios of MTTs monotonically increase when emitter–base bias voltages increase from 1.1 to 2.3 V Although interpretation of these results is controversial so far, we would like to interpret these results in the following way The MC and transfer ratios in MTTs with differing base thicknesses vary with increasing bias voltage This might be due to a variation in the MFP of hot electrons According to [4, 6, 7, 10, 11], the transfer ratio is proportional to W/λ↑ Assuming that λ is constant, a higher transfer ratio should be obtained for MTTs with smaller base thicknesses at a given bias voltage [4, 6] However, rather higher transfer ratios at greater base thicknesses are obtained within the bias voltage range near Vth (threshold voltage for hot-electron collection in a MTT structure) [10, 11] This means that the MFP may decrease substantially with increasing base thickness The λ↑ and λ↓ , obtained by making use of d IBC /dV curves for the parallel and anti-parallel configurations for the MTTs with differing base thicknesses, show that they vary in the ranges ˚ respectively This result means that 30–40 and 14–16 A, λ↑ decreases rapidly with decreasing base thickness in the measured bias voltage range of 1.45–2.3 V Increase in the emitter–base voltage only changes the number of electrons that are injected into the base, but it does not change their energy or momentum distributions Needless to say, the exponential decay varies from layer to layer, and it depends on the hotelectron spins in the ferromagnetic layers Obviously, the spin-dependent exponential decay is one of the factors that lead to the relatively huge magnetic response in the SVT and MTT structures This can be contrasted with the role played by the MFP in the GMR of a magnetic multilayer, where the resistances are linearly proportional to the inverse of the MFP [11] The origin of the decrease in λ↑ in the thin base and interface between the Si substrate and the CoFe layer can be Magneto-transport properties of magnetic tunnelling transistors at low and room temperatures Figure Depth profiles, in atomic percentage, for the elements existing in the structure of Al2 O3 /CoFe/Si(100) (a), and 2p XPS spectra depth profiles of the Co element in this structure (b) at RT Figure Collector current as a function of applied magnetic field for the bias voltages of 1.45 V (a) and of 2.0 V (b) for MTTs with a base ˚ The MC and transfer ratios as a function of VEB thickness of 80 A ˚ at 77 K (c) for the MTTs with differing base thicknesses of 30–80 A clarified by the depth profile analysis of XPS spectra The profiles of Co, Fe, Al, O and Si elements were measured simultaneously (figure 4) The structure of the analysed sample ˚ ˚ ˚ Figure 4(a) shows (80 A)/Si (1500 A) is Al2 O3 (35 A)/CoFe the depth profile of each component, in atomic percentage (%), obtained from XPS spectra We note that CoFe adjacent to Si(100) includes a large Si content, and there is a Co2 Si zone ˚ (depth sputtering time of ∼7.5 min) at the of 16.5 ± 0.5 A CoFe/Si interface Co silicide was identified by 2p XPS spectra depth profiles of Co as shown in figure 4(b) From the 2p3/2 peak of Co, we see that CoFe/Si interface has a Co2 Si peak of 778.6 ± 0.3 eV while the inner CoFe layer has a Co peak of 778.21 eV Co2 Si may be formed by high-energy bombardment of Co atoms onto the Si surface during sputter deposition of CoFe [16] From the Fe XPS spectra, there was no Fe silicide peak and only a Fe 2p peak was found at the CoFe/Si interface From the XPS interface analysis, the CoFe base has an intermediate region of Co2 Si and Fe adjacent to Si(100), and this interface region may be influenced by the reduced λ↑ and/or λ↑ /λ↓ owing to either spin-independent defects or impurity scattering of hot electrons In the case of a thin base ˚ the region of about half the base with a thickness of 30 A, thickness is the intermediate region consisting of Co2 Si and Fe (not shown), and this may also induce the decrease in λ↑ Hence, the interfacial scattering may become more important for MTTs with very thin base layers At RT, in spite of a large leakage current for the MTT with ˚ MC ratios of 6.5% and 7.8% were a base thickness of 80 A, obtained at bias voltages of 2.3 and 2.5 V, respectively (not shown) The low MC ratio is closely related to the high leakage 3363 H D Quang et al Figure TMR curves of emitter–base tunnel junctions measured at 77 K (a) and at RT (b) with differing CoFe base thickness of 30, 60 ˚ All curves are taken for a bias voltage of 2.0 mV and 80 A current of about 1.0 µA The MC change ( IC = ICP − ICAP ) at the bias voltage of 2.3 V at RT was 70 nA, which is smaller than the MC change of 170 nA at the same bias voltage at 77 K This smaller change may be closely related to the decrease in λ↑ and/or λ↑ /λ↓ as well as the decrease in PE at RT because of high spin wave scattering [7, 8] So, our results, measured at 77 K with excellent collector output current and high transfer ratio, are comparable to those for GaAs-based MTTs fabricated by other authors [12, 15, 17, 18] Figure shows the TMR curves measured for the low bias voltage of 2.0 mV at 77 K and at RT with differing base thicknesses Here, the TMR ratio is defined as [(RP − RAP )/RP ] × 100%, where RP and RAP are the resistance of a MTT corresponding to the parallel and anti-parallel orientation of the FM layers, respectively Generally, the two ferromagnetic layers of a MTT have different coercive fields Hc1 and Hc2 , and their magnetization can be either parallel or anti-parallel when we sweep the external magnetic field H (Hc1 < H < Hc2 ) This is essential for us to produce the spin valve effect For our MTT structures, exchange biasing is the other way to realize parallel and anti-parallel magnetization alignments For this, one of the magnetic layers is in direct contact with an anti-ferromagnetic (AFM) material 3364 (FeMn) From the TMR curves for MTTs with differing base thicknesses, the estimated values for the TMR ratio are 8– 12% and 14–20% at RT and 77 K, respectively A TMR ratio of 14–20% was obtained for VEB 1.0 V and of 2– 3% for VEB of 1.0–2.0 V (not shown) Hence, the TMR ratio decreases at high bias voltages and temperatures, which can be explained in terms of the formation of magnons in the tunnel barrier Such magnons have been found in NiO by Tsui et al [9] More specifically, there is little change in Hc for CoFe layers upon cooling to 77 K, thereby bringing the two coercive forces of the two ferromagnetic layers quite close at low temperatures Moreover, Hc for CoFe, NiFe and Co films changed by different amounts In fact, Co shows the maximum increase in Hc at lower temperatures compared with the others [9–11, 14, 17] On the other hand, the TMR ratio can be also expressed by P1 P2 /(1 − P1 P2 ); P = [(n ↑ − n ↓ )/(n ↑ + n ↓ )], where n ↑ , n ↓ are the density of the majority and minority states, respectively, at the Fermi energy, P1 and P2 are the spin polarization of ferromagnetic layers of the emitter and base, respectively, and P = Pu (1 − 1.67b); where b is spin–orbit parameter and Pu is uncorrected polarization value For instance, Pu of Co84 Fe16 has decreased from 63.5% (0.25 K) to 57% (0.4 K) It appears that there is a strong temperature dependence of spin polarization, and the lower TMR ratio is due to higher temperature and dc bias dependence [11] The magnetization of ferromagnetic materials at temperatures far below the critical temperature is affected by thermal excitation of spin waves leading to T 3/2 dependence, i.e M S (T ) = M S (0)[1 − B S T 3/2 ], where M S (0) is the magnetization at K Since the spin polarization P(T ) has been shown to be proportional to M S , by applying this result to our system the spin polarization is P = P0 (1 − γ T 3/2 ), where P0 is the spin polarization at K and γ is a constant (γ is different for bulk and thin films) In addition to P , the factors that influence the TMR value are FM/I interface cleanness, barrier quality and Hc of the ferromagnetic layers It is nontrivial to completely oxidize Al films without oxidizing the surface of the bottom ferromagnetic layer and also achieve a clean FM/I interface to reach the full TMR value Also, spin scattering is negligible in the antiferromagnetically ordered state of the oxides at low temperatures, and impurities in the barrier are detrimental to the TMR [9–11, 14] Conclusion The Si-based MTT structures were fabricated using a magnetron sputtering deposition system MC ratios of 48.3– 55.9% for VEB of 1.45–2.0 V, and transfer ratios of (1.3–2.0) × 10−4 for VEB of 1.8–2.0 V were obtained in the MTT structures ˚ at 77 K MC ratios with differing base thicknesses of 30–80 A of 6.5% and 7.8% were obtained at RT for bias voltages of 2.3 and 2.5 V, respectively A low MC ratio is closely related to a high leakage current and is due to the decrease in λ↑ and/or λ↑ /λ↓ as well as the decrease in PE at room temperature TMR measurement was also carried out at 77 K and RT Moreover, the origin of the decrease in λ↑ in a thin base and interface state between the Si substrate and CoFe base layer could be clarified by the depth profiles analysis of XPS spectra in the ˚ ˚ ˚ structure Moreover, Al2 O3 (35 A)/CoFe (80 A)/Si (1500 A) Magneto-transport properties of magnetic tunnelling transistors at low and room temperatures the (Co2 Si, Fe) intermediate region, formed at the CoFe/Si interface, may have caused the decrease in the value of λ↑ Therefore, by reducing the leakage current, operation of Sibased MTT structures with an MC ratio of over 53.9% will be realized at RT; this is a good candidate for a new read head device for 300 Gbit in−2 that goes beyond the specular spin valve structures Acknowledgments One of the authors (HDQ) would like to thank Dr Phan Thanh Son (Tohoku University, Japan) and Dr Simone Herthfor (Universităat Bielefeld, Germany) for helpful discussions This research, carried out at Chungbuk National University, was supported by the Korea Science and Engineering Foundation Grant (R01-2004-000-10882-0) References [1] Shen F, Xu Q Y, Yu G H, Lai W Y, Zhang Z, Lu Z Q, Pan G and Al-Jibouri A 2002 Appl Phys Lett 80 4410 Takiguchi M, Ishii S, Makino E and Okabe A 2000 J Appl Phys 87 2469 Li H, Freitas P P, Wang Z, Sousa J B, Gogol P and Chapman J 2001 J Appl Phys 89 6904 and references therein [2] Quang H D, Hien N T, Oh S K, Sinh N H and Yu S C 2004 J Phys D: Appl Phys 37 3290 Quang H D, Sinh N H, Oh S K, Hien N T and Yu S C 2005 J Phys D: Appl Phys 38 3560 and references therein [3] Hirose T, Fujiwara Y, Jimbo M, Kobayashi T and Shiomi S 2004 Phys Status Solidi a 241 1502 [4] van’t Erve O M J, Vlutters R, Anil Kumar P S, Kim S D, Postma F M, Jansen R and Lodder J C 2002 Appl Phys Lett 80 3787 Vlutters R, van’t Erve O M J, Kim S D, Jansen R and Lodder J C 2002 Phys Rev Lett 88 027202 [5] van Dijken S, Jiang X and Parkin S S P 2002 Appl Phys Lett 80 3364 van Dijken S, Jiang X and Parkin S S P 2003 Appl Phys Lett 82 775 [6] Monsma D J, Lodder J C, Popma T J A and Dieny B 1995 Phys Rev Lett 74 5260 and the references therein [7] Monsma D J, Vlutters R and Lodder J C 1998 Science 281 407 van Dijken S, Jiang X and Parkin S S P 2002 Phys Rev B 66 094417 and the references therein [8] Xiong Z H, Wu D, Vardeny Z V and Shi J 2004 Nature 427 821 [9] Moodera J S, Kinder L R, Wong T M and Meservey R 1995 Phys Rev Lett 74 3273 Clonczewski J C 1989 Phys Rev B 39 6995 Tsui D C, Dietz R E and Walker L R 1971 Phys Rev Lett 27 1729 [10] Shang C H, Nowak J, Jansen R and Moodera J S 1998 Phys Rev B 58 R2917 Gibson G A and Meservey R 1984 J Appl Phys 58 1584 Zhang S, Levy P M, Marley A C and Parkin S S P 1997 Phys Rev Lett 79 3744 and references therein [11] Vlutters R, van ’t Erve O M J, Kim J S D, Lodder J C, Vedyayev A and Dieny B 2001 Phys Rev B 65 024416 Bell L D and Kaiser W J 1988 Phys Rev Lett 61 2368 O’Shea J J, Reaves C M and Denbaars S P 1996 Appl Phys Lett 69 3022 Vlutters R, Jansen R, van’t Erve O M J, Kim S D and Lodder J C 2001 J Appl Phys 89 7305 and references therein [12] Rippard W H and Buhrman R A 2000 Phys Rev Lett 84 971 [13] Monsma J D and Parkin S S P 2000 Appl Phys Lett 77 720 [14] Moodera J S, Nowak J and van de Veerdonk R J M 1998 Phys Rev Lett 80 2941 [15] Hanbicki A T, Jonker B T, Itskos G, Kioseoglou G and Petrou A 2002 Appl Phys Lett 80 1240 [16] Prabhakaran K and Ogino T 1996 Appl Surf Sci 100/101 518 Prabhakaran K, Sumitomo K and Ogino T 1999 Surf Sci 421 100 [17] De Teresa J M, Barth´el´emy A, Fert A, Contour J P, Lyonnet R, Montaigne F, Seneor P and Vaur`es A 1999 Phys Rev Lett 82 4288 [18] Worledge D C and Geballe T H 2000 Phys Rev Lett 85 5182 Drittler B, Stefanou N, Blăugel S, Zeller R and Dederichs P H 1989 Phys Rev B 40 8203 Oleinik I I, Tsymbal E Y and Pettifor D G 2000 Phys Rev B 62 3952 LeClair P, Hoex B, Wieldraaijer H, Kohlhepp J T, Swagten H J M and de Jonge W J M 2001 Phys Rev B 64 100406 3365 ... between the Si substrate and the CoFe layer can be Magneto-transport properties of magnetic tunnelling transistors at low and room temperatures Figure Depth profiles, in atomic percentage, for... polarization, and the lower TMR ratio is due to higher temperature and dc bias dependence [11] The magnetization of ferromagnetic materials at temperatures far below the critical temperature is... A)/CoFe (80 A)/Si (1500 A) Magneto-transport properties of magnetic tunnelling transistors at low and room temperatures the (Co2 Si, Fe) intermediate region, formed at the CoFe/Si interface, may have