Magnetic field cycling effect on the non-linear current-voltage characteristics and magnetic field induced negative differential resistance in α−Fe1.64Ga0.36O3 oxide R N Bhowmik and G Vijayasri Citation: AIP Advances 5, 067126 (2015); doi: 10.1063/1.4922511 View online: http://dx.doi.org/10.1063/1.4922511 View Table of Contents: http://aip.scitation.org/toc/adv/5/6 Published by the American Institute of Physics AIP ADVANCES 5, 067126 (2015) Magnetic field cycling effect on the non-linear current-voltage characteristics and magnetic field induced negative differential resistance in α-Fe1.64Ga0.36O3 oxide R N Bhowmika and G Vijayasri Department of Physics, Pondicherry University, R.Venkataraman Nagar, Kalapet, Puducherry - 605 014, India (Received 13 April 2015; accepted 31 May 2015; published online 10 June 2015) We have studied current-voltage (I-V) characteristics of α-Fe1.64Ga0.36O3, a typical canted ferromagnetic semiconductor The sample showed a transformation of the I-V curves from linear to non-linear character with the increase of bias voltage The I-V curves showed irreversible features with hysteresis loop and bi-stable electronic states for up and down modes of voltage sweep We report positive magnetoresistance and magnetic field induced negative differential resistance as the first time observed phenomena in metal doped hematite system The magnitudes of critical voltage at which I-V curve showed peak and corresponding peak current are affected by magnetic field cycling The shift of the peak voltage with magnetic field showed a step-wise jump between two discrete voltage levels with least gap (∆VP) 0.345(± 0.001) V The magnetic spin dependent electronic charge transport in this new class of magnetic semiconductor opens a wide scope for tuning large electroresistance (∼500-700%), magnetoresistance (70-135 %) and charge-spin dependent conductivity under suitable control of electric and magnetic fields The electric and magnetic field controlled charge-spin transport is interesting for applications of the magnetic materials in spintronics, e.g., magnetic sensor, memory devices and digital switching C 2015 Author(s) All article content, except where otherwise noted, is licensed under a Creative Commons Attribution 3.0 Unported License [http://dx.doi.org/10.1063/1.4922511] I INTRODUCTION Development of a new class of ferromagnetic semiconductor, based on metal (Al, Ti, Ga) doped hematite (α-Fe2O3) system, is matter of recent interest for understanding novel magneto-electric phenomena and their potential applications in spintronic devices.1–5 The hematite system stabilizes ¯ The spins of Fe3+ ions in each rhombohedral in rhombohedral structure with space group R3C planes (say, two alternating planes A and B) from ferromagnetic (FM) order, but spins of Fe3+ ions between A and B planes are coupled by antiferromagnetic (AFM) superexchange interactions (Fe3+A–O2−–Fe3+B).3 A weak (canted) ferromagnetic state appears in hematite system due to canting of spins between Fe3+ ions in A and B planes, where spin canting is controlled by → −−−→ ⇀ − − anisotropic Dzyaloshinsky-Morya (DM) interactions [∼ D ( Sn × Sn+1)] The interesting aspect is that doping of non-magnetic Ga atoms in α-Fe2O3 system enhances ferromagnetic properties.5–7 The ferromagnetic enhancement in Ga doped α-Fe2O3 occurs due to modified spin structure and superexchange interactions in rhombohedral planes of the crystal structure Electrically, hematite is a charge-transfer type semiconductor with band gap ∼ 2.2 eV and non-suitable for spintronics applications due to its low electrical conductivity.8 Recent reports3,4,8,9 predicted enhancement of electrical conductivity in metal doped hematite There is not much progress to understand either the mechanism of enhanced conductivity or the hidden magneto-transport properties in metal doped a Corresponding author: Tel.: +91-9944064547; Fax: +91-413-2655734 E-mail: rnbhowmik.phy@pondiuni.edu.in 2158-3226/2015/5(6)/067126/15 5, 067126-1 © Author(s) 2015 067126-2 R N Bhowmik and G Vijayasri AIP Advances 5, 067126 (2015) hematite Many exciting magneto-transport properties can be expected due to coupling between ferromagnetism and conductivity in metal doped hematite system C Papaioannoua et al.10 suggested a possible charge-spin coupling in hematite Our recent work11 has highlighted the effect of magnetic spins flipping on magneto-electric coupling in α-Fe1.6Ga0.4O3 system The ferromagnetic semiconductor properties of Ga doped α-Fe2O3 system are not like conventional III-V ions based diluted ferromagnetic semiconductor.1 More experimental evidences are needed to establish charge-spin coupling in this new class of ferromagnetic semiconductor The coupling between electronic charge and spin is essential for the spintronics applications of magnetic semiconductors.2,12 The current-voltage (I-V) curve correlates the charge flow through a ferromagnetic semiconductor under test (FMSUT) with applied voltages and it is an important tool to determine suitability of the material in electronic devices The I-V characteristics can be used to estimate electrical conductivity, charge accumulation and charge injection efficiency at the interfaces of FMSUT and electrodes.13–15 The I-V curves of a metal/FMSUT/metal junction device are controlled by the injection rate of charge carriers from electrode into FMSUT, and transport of the charge carriers through FMSUT The transport of charge carriers is either injection limited due to a mismatch between the electrode work function and the electronic energy level of the FMSUT or bulk transport limited due to intrinsic charge mobility of the FMSUT or space charge limited at the interfaces and grain boundaries, respectively.16 The control of space charge limited current (SCLC) and charge-spin coupling in metal electrode-semiconductor junction has become a key issue for device applications and understanding the electronic charge transport in FMSUT.14,17 The magnetic field controlled I-V curve has emerged as an effective route for testing SCLC and charge-spin coupling in a FMSUT.13–15,18 The I-V characteristics under magnetic field are also useful to extract the information of the effects of magnetic domain wall motion and magnetic domain rotation on the current flow mechanism in FMSUT.1,19 To our knowledge, there is no work which dealt magnetic field controlled electronic properties for metal doped hematite system, despite the fact that such system has a bright future for the applications in multifunctional devices.20 Our earlier works5,6 have discussed the canted ferromagnetic properties in Ga doped hematite system In this work, we report a detailed study of I-V characteristics of α-Fe1.64Ga0.36O3 sample and demonstrate the magnetic field cycling effect on its I-V characteristics to explore hidden magneto-transport properties in Ga doped hematite system II EXPERIMENTAL ¯ has been preThe α-Fe1.64Ga0.36O3 compound in rhombohedral structure (space group R3C) pared by the coprecipitation of Fe and Ga hydroxides in alkaline medium (pH ∼ 11) at 80 oC The precipitated powder after washing has been annealed under vacuum (10−5 mbar) at 800 OC and formation of single phased structure has been confirmed from synchrotron X-ray diffraction pattern Details of the material preparation and characterization have been discussed elsewhere.6 In this work, we describe the I-V characteristics in the absence and presence of dc magnetic field A disc shaped (Ø13 mm and thickness ∼ mm) sample has been sandwiched between two platinum (Pt) electrodes for I-V curve measurement The contact between Pt electrodes and the sample has been made by adjusting pressure from both sides of the Pt plates using homemade sample holder The sample holder has been placed in pole gap of an electromagnet (MicroSense, USA) The bias voltage (within ± 10 V) across the sample has been applied by connecting the Pt electrodes to Keithley Meter (Model: 2410-C) The current flow and magnetic field are directed perpendicular to the surface area of the disc The current through the sample has been measured by sweeping the dc voltage across the sample in up (0 → ± 10 V) and down (± 10 V → 0) modes We have extracted the static resistance (R = V/I) at each point and dynamic resistance (rd = ∆V/∆I) over a range of I-V curves The magnetic field (H) cycling effect on the I-V curves has been recorded in the presence of selected magnetic fields in five segments, i.e., first segment (S1): increase of H from to + 15 kOe, second segment (S2): reducing the field from + 15 kOe to 0, third segment (S3): reduction of field from to -15 kOe, fourth segment (S4): increase of field from -15 kOe to 0, and fifth segment (S5): increase of field from to +15 kOe These five segments have been decided based on the field dependent magnetization [M(H)] loop of the sample However, I-V curves for 067126-3 R N Bhowmik and G Vijayasri AIP Advances 5, 067126 (2015) all the applied magnetic fields are not shown in figures for better representation of the data The measurement has been designed for the delay time 100 ms between two consecutive bias voltage points The sample has been waited (tw) for s and 60 s, respectively between the application of magnetic field and starting of I-V curve measurement In order to confirm the intrinsic nature of the non-linear I-V curves of the sample, we have repeated some of the measurements in the presence and absence of magnetic field by silver coating on both sides of the sample disc and sandwiching the silvered disc between two pressure-controlled Pt electrodes We have noted the reproducibility of some important features, leaving aside small differences due to silver coating For example, magnetic field induced negative differential resistance (NDR) and positive magnetoresistance (MR) have been confirmed for the sample disc with and without silver coating However, I-V curves exhibited nearly two order increase of measured current when the sample surface has been silver coated We noted similar increase of current after silver coating on hematite sample, but hematite does not show NDR effect We believe that current flow in silver coated sample is increased due to diffusion of some silver atoms at the sample surface We focus on the following aspects, (1) magnetic field cycling effect on I-V curves, (2) magnetic field induced NDR effect, (3) space charge controlled current flow, and (4) charge-spin controlled electronic properties by sandwiching the disc-shaped sample without silver coating between two Pt electrodes III EXPERIMENTAL RESULTS A I-V curves measurement in the presence of magnetic field with zero waiting time The current has been measured in positive (0 to +10 V) and negative (0 to -10 V) bias of the sweeping voltage in the absence and presence of set magnetic field The measurement started immediately without giving any waiting time (tw = 0) for applied magnetic field before recording the I-V curve Fig shows the I-V curves The absolute current values on positive and negative bias voltages are nearly identical, suggesting good electrical contact on both sides of the sample The I-V curves showed nearly linear character with increase of voltage up to ± V The rate of current increment decreases above ± V, implying a non-linear behavior at higher voltage The vertical arrows in Fig represent the directions of magnetic field (H) and current (I) during S1-S5 segments The nature of I-V curves during different segments of magnetic field cycling has been discussed later in terms of the R(V) curves The linear behavior of I-V curves follows a power law: I(V) ∼ Vm , where the power factor (m) has been obtained by m = ∂lnI/∂lnV As shown in Fig 1(f), the power factor (m) in the absence of magnetic field is ∼ 1.4 and stabilized in the range 1.0 ± 0.2 under magnetic field Fig shows the I-V curves during down modes of the sweeping voltage (±10 to V) under constant magnetic fields The I-V curves in the down modes not follow the paths of up modes (Fig 1) This shows irreversible character (hysteresis loop) and two distinct resistance states, i.e., low resistive state (LRS) during voltage up mode and high resistive state (HRS) during voltage down mode in I-V curves The ferromagnetic-semiconductor with bi-stable electronic states is interesting for applications in magnetic memory devices.21 In the lower voltage regime of I-V curves during down modes of voltage sweep, the power factor (m) has stabilized in the range 1.3-1.7 under magnetic field The non-linear I-V curves suggest a good amount of change of electrical resistance with bias voltage, a phenomenon known as electroresistance (ER).22 Fig 3(a)-3(h) shows the variation of static resistance (R) with bias voltage (V), derived from up and down modes of I-V curves Some important features can be noted from R(V) curves First, R(V) curves depend on magnetic fields Second, the nature of R(V) curves for down modes (sweeping voltage ± 10 V to 0) is different from the nature of R(V) curves for up modes (sweeping voltage to ± 10 V) The R(V) curves during up modes are in LR state and weakly dependent on voltage sweep up to ± V, followed by an electric field induced increment above ± V On the other hand, R(V) curves for the down modes are in HR state that increases rapidly for voltage swept from ± 10 V to We estimated )−R(VL ) ] x100 Considering an uncertainty in the change of ER using the formula E R (%) = [ R(VHR(V L) the accuracy of measured current for sweeping voltage approaching to zero volt, we have taken R(VH) and R(VL) as the resistances at 10 V and 1V, respectively Fig shows the variation of ER(%) during all five segments of the magnetic field cycling in the up and down modes of voltage 067126-4 R N Bhowmik and G Vijayasri AIP Advances 5, 067126 (2015) FIG (a-e) I-V curves (voltage increasing mode) at selected magnetic fields and without wating time after application of fields (f) the power factor obtained from fitting of I-V curves at low voltage regime Vertical up arrows represent positive direction and vertical down arrows represent negative directions for magnetic field and current flow through the disc shaped sample sweep We observed that ER (%) is positive for the up mode and negative for the down mode of the voltage sweep The ER values in down mode (- 90 (±10) % for positive and - 92 (±8) % for negative bias voltage) are larger in comparison to the values in up mode (∼ + 50 (±20) % for positive and + 30 (±15) % for negative bias voltage) It is not good to draw any decreasing or increasing pattern of the variation of ER (%) in different segments of magnetic fields, because the ER(%) data are a bit scattered due to low accuracy of R(VL) data at V or due to insufficient waiting time for magnetic field to achieve a quasi-equilibrium magnetic spin ordered state.23 The magnetic field effect on R(V) curves has been confirmed from the magnetoresistance (MR), a phenomenon where resistance at constant bias voltage changes magnetic field variation We used the I-V curves ( with ) )−R(0) x100 R(H) is the resistance at magnetic field at selected voltages to extract M R (%) = R(HR(0) H and R(0) is the resistance at magnetic field zero Fig 5(a)-5(e) shows the variation of MR (%) with magnetic field cycling at positive bias voltages The behavior of MR (%) with magnetic field at negative voltages is identical to that found for positive bias voltages (not shown) We have also directly measured the MR(%) at selected magnetic fields and the data are shown in Fig 5(f)-5(j) 067126-5 R N Bhowmik and G Vijayasri AIP Advances 5, 067126 (2015) FIG (a-e) I-V curves (return paths) shown at selected magnetic fields and without wating time after application of fields (f) power factor obtained by fitting the I-V curves at low voltage regime In direct measurement, the current has been measured at selected magnetic fields under constant voltage The measurement of current at each magnetic field has been repeated 20 times The average current at each point of magnetic field cycling has been used to calculate resistance (R = V/I) and MR(%) Fig shows the MR(%) vs H plots The MR(%) data with H from direct measurement and extracted from I-V curves are well consistent to each other For example, MR(H) curves followed different paths during increase and decreasing modes of magnetic field variation The plots showed a butterfly wings shaped loop, which is associated with bi-stable electronic states and controlled by magnetic field cycling The variation of ER(%) with H and the variation of MR(%) with V indicate a strong magneto-electric coupling in the sample The increasing MR(%) at lower magnetic field regime to achieve a peak and subsequent decrease of MR(%) at higher magnetic field has been found in different ferromagnetic systems, where spin polarization of electrons controls the transport properties.24,25 We estimated the MR(%) maximum (∼ 105 %, 135 %, 95 %, 90 %, and 70 % for the bias voltage at V, V, V, V, and 10 V, respectively) using the MR(%) curves from direct measurement The maximum MR(%) initially increases with V up to Volts and then decreases on further increase of V This feature can be corroborated with the nature of I-V curves, where current has rapidly increased for bias voltage up to ∼ V and then slowed down at higher voltage This shows a critical value of bias voltage for achieving maximum MR(%) The positive MR(%) 067126-6 R N Bhowmik and G Vijayasri AIP Advances 5, 067126 (2015) FIG Resistance vs sweeping voltage at selected magnetic fields at different segments for the initial paths (0 to ± 10 V) in (a-d) and for return paths (± 10V to 0) in (e-h), extracted from Fig and Fig 2, respectively indicates that resistance of the sample increases under magnetic field Such increase of resistance can be attributed to spins scattering effect at the interfaces or grain-boundaries of the magnetic material.16,19 Space charge also exhibited a great impact on the large positive MR in non-magnetic material.17 However, the nature of the MR(%) curves in our sample with H at selected voltages is different from that observed due to space charge effect We have discussed later that space charge effect is not the main mechanism in our sample for controlling magneto-transport properties It may be noted by comparing the MR(%) vs H data from I-V curves and direct measurement that the data extracted from I-V curves showed a bit fluctuation Although current measurement started immediately after setting the magnet field, the repetition of the current measurement for 20 times at each magnetic field takes nearly 20-30 s We understand that the waiting time of magnetic field before current measurement plays a significant role to suppress thermal induced noise and increase of electron spin ordering at the grain boundaries.25 We elaborate the magnetic spin controlled charge transport process by increasing the waiting time of magnetic field so that spins structure can achieve a quasi-equilibrium state before I-V curve measurement 067126-7 R N Bhowmik and G Vijayasri AIP Advances 5, 067126 (2015) FIG Variation of electroresistance (%) with applied magnetic fields for the initial (a-b) and return (c-d) paths of the positive and negative biasing of voltage B I-V curves measurement in the presence of magnetic field with 60 s waiting time In this measurement, the sample has been waited for nearly 60 s under set magnetic field before recording the I-V characteristics Fig shows the I-V curves recorded during up modes of voltage sweep at constant magnetic fields The I-V curve in the absence of magnetic field is obviously identical to that shown in Fig However, the I-V curves measured after 60 s waiting time at set magnetic field have brought remarkable changes We found three distinct regimes, as indicated in Fig 6(d) In low voltage ohmic type regime (LOR), current increases almost linearly up to a critical voltage (VP) The most remarkable feature is the negative differential resistance (NDR: ∂V/∂I < 0) effect above VP In NDR regime, the I-V curve showed non-ohmic character where current decreases with increase of voltage to achieve a current valley at voltage Vm In the high voltage ohmic regime (HOR) at V > Vm, the current once again increases with the increase of voltage up to 10 V The NDR effect is useful for applying materials in spintronics devices,26–29 especially in memory and switching devices.30,31 We have shown in Fig 6(f) that the power factor (m = ∂lnI/∂lnV) in the LOR regime varied in the range 0.90-1.20 These values of m matched with the values obtained from I-V curves measured at zero waiting time for magnetic field We found the critical voltage (VP) in the range ∼ 2- V and ∼ - (2.5 - 4.5 V) for positive and negative 067126-8 R N Bhowmik and G Vijayasri AIP Advances 5, 067126 (2015) FIG Variation of the change of magnetoresistance obtained from I-V curves (a-e) and direct measurements (f-j), respectively bias voltages, respectively We plotted the values of VP and corresponding peak current (IP) at each applied magnetic field in Fig 7(a)-7(b) and Fig 7(c)-7(d), respectively We noted that VP corresponding to the peak current shifts with applied magnetic field The shift of VP with magnetic field showed a step-wise jump between two discrete voltage levels, e.g., + 2.06912 V, + 3.44870 V, +3.79284 V for positive bias and -2.75925 V, -3.79297 V, -4.48322 V for negative bias The least gap (∆VP) between two discrete VP levels has been found 0.345(± 0.001) V The discrete levels have been found as the integer multiple of ∆VP Fig 7(c)-7(d) shows that the variation of peak current (IP) with magnetic field cycling is nearly identical for both positive and negative bias voltages, except the direction of current is opposite The IP path with magnetic field cycling is irreversible The IP varied within a limited range (2.5±1.5 µA) in the S1-S3 segments of magnetic field cycling It increases rapidly in S4 segment to reach the highest value ∼ 8.5±0.5 µA as the magnetic field increases from -15 kOe to +3 kOe On further increase of magnetic field up to +15 kOe in S5 067126-9 R N Bhowmik and G Vijayasri AIP Advances 5, 067126 (2015) FIG I-V curves measured at selected points of magnetic field cycling (a-e) and power factor from I-V curves at different points of magnetic field cycling (f) Three distinct regimes in I-V curves under magnetic field have been shown in (d) segment, the IP decreases down to the level that has been achieved at +15 kOe during S1 segment, and completes the IP loop If we compare the IP(H) loop with ferromagnetic M(H) loop,6 it is realized that there is no remarkable change of peak current during domain wall motion in the S1 segment (magnetization is orienting parallel to magnetic field) and also during magnetic domains rotation from up (parallel to magnetization) to down (opposite to magnetization) directions However, a spectacular increase of IP is started as the magnetic domains started rotating from down to up directions when magnetic field was increasing from -15 kOe to positive field direction An estimation of the negative differential resistance (rd = ∆V/∆I) from NDR regime (Fig 7(e)-7(f)) shows relatively large value at higher magnetic fields This means the rate of decrease of current in the NDR regime is affected by the magnetic field cycling effect We have also estimated static resistance (R) from the I-V curves (Fig 6) and plotted the data in Fig 8(a)-8(e) In the absence of magnetic field (S1 segment), the resistance initially decreases to reach a minimum value at bias voltage ± 1.72 V, followed by an increasing trend with further increase of voltage in both positive and negative bias The features of R(V) curves drastically changed when magnetic field is applied The resistance is nearly independent or weakly dependent up to the bias voltage ∼ ± 3.79 V, which is consistent to ohmic character of the sample at lower voltage Then, static resistance increases 067126-10 R N Bhowmik and G Vijayasri AIP Advances 5, 067126 (2015) FIG Magnetic field cyclic effect on peak voltage (a-b) and peak current (c-d), and negative differential resistance (rd) (e-f) rapidly in the NDR regime and above with the increase of voltage up to ± 10 V Such a characteristic jump of static resistance after a typical bias voltage under magnetic field is useful for applying the material in magnetic sensors and electronic switches Noting a sharp jump of the resistance at bias )−R(VL ) voltage ∼ ± 3.79 V, we have calculated the ER(%) using the formula ER(%) = R(VHR(V x100 L) We have taken VL (at V ≤ VP) as the bias voltage where R(V) curve showed either minimum at about VP or as the bias voltage below which R(V) curve showed a plateau like behavior R(VH) is the resistance at ± 10 V We noted VL ∼ ± 1.72 V at magnetic field zero and ∼ ± 3.79 V at 15 kOe Fig 8(f)-8(g) shows the variation of ER(%) at different points of the magnetic field cycling We noted that magnetic field cycling effect on the ER(%) curves is identical to that of IP(H) loop, and the maximum of IP(H) curves corresponds to the maximum of ER(%) curves This result suggests a coupling between peak current IP and ER(%), and the electrical charge transport is affected by magnetic field cycling We mention that the ER(%) extracted from I-V curves without waiting time for magnetic field has been found less than 100% In contrast, the ER(%) extracted from I-V curves with 60 s waiting time for magnetic field has increased up to ∼ 500 % and ∼ 700 % for positive and negative bias voltage, respectively The differences in ER(%) values by reversing the current 067126-11 R N Bhowmik and G Vijayasri AIP Advances 5, 067126 (2015) FIG (a-e) Bias voltage dependence of resistance at selected points of magnetic field cycling (f-g) ER(%) extracted from I-V curves at selected magnetic fields direction are largely due to bi-stable electronic states and grain boundary heterogeneity in the sample Fig 9(a)-9(c) highlights the hysteresis nature of I-V curves, which we recorded by completing the bias voltage cycling (0 → +10 V → → -10 V → 0) The I-V curves under magnetic field showed a wide hysteresis loop upon reversing the direction of the positive and negative bias voltages back to zero The magnitude and direction of magnetic field have affected the nature of I-V hysteresis loop The I-V loop at +15 kOe is nearly symmetric for negative and positive voltage cycling, whereas the I-V loops at kOe and -15 kOe are asymmetric in nature The loop area represents the electrical energy loss due to rotation of strongly interacting electric/ferroelectric domains The sample offered different resistance or barrier height for rotation of electric domains during up and down modes of voltage sweep The loop area appears to be smaller for negative (-15 kOe) magnetic field The features of R(V) loops in Fig 9(d)-9(f) are similar to that observed in Fig 3, i.e., the resistance in up mode of bias voltages is LRS and the resistance in down mode is HRS 067126-12 R N Bhowmik and G Vijayasri AIP Advances 5, 067126 (2015) FIG I-V hysteresis loops under magnetic fields (a-c) and corresponding loops in R(V) curves of the sample (d-f) The arrows guide the direction of bias voltage The gap between HR and LR states increases as the bias voltage decreases from ± 10 V to zero The irreversible bi-stable electronic states under the applications of both electric and magnetic field is encouraging for applying the material in magnetic and electric field controlled switching and memory devices.25 IV SUMMARY OF THE RESULTS AND DISCUSSION Our experimental data confirm that electrical charge transport in the sample is non-linear in character and strongly affected by the applications of both electric field bias and magnetic field The sample exhibited substantially large ER and MR, bi-stable electronic states with I(V) hysteresis loop, electric and magnetic fields controlled I-V curves The system is a new class of ferromagnetic semiconductor In the absence of much literature reports on magneto-transport properties of metal doped hematite system, we understand the mechanism of magneto-electric field 067126-13 R N Bhowmik and G Vijayasri AIP Advances 5, 067126 (2015) induced properties in our sample using the existing experimental and theoretical works available for different types of magnetic semiconductors Most striking feature is the magnetic field induced NDR effect In literature, various models have been discussed to realize the origin of NDR effect, e.g., Joule heating induced phase segregation21,32 and charge order state,28,29 space charge limited current (SCLC) mechanism,14,16,17,22 spin filter and spin valve effect,25 magnetic domain effect,19 and charge-spin accumulation and transfer across non-magnetic metal electrode-magnetic semiconductor junctions.13–15,33 The ferromagnetic loop in our sample does not show any meta-magnetic phase transition,6 which generally marks for the occurrence of charge ordered structural phase segregation in manganite systems.21,22 However, the coexistence of magnetically soft and hard phases has been indicated in the M(H) loop in our sample It is due to random doping of Ga ions in rhombohedral planes containing Fe ions and distribution of magnetic exchange interactions B Kundys et al.27 attributed similar magnetic field induced NDR effect in Bi0.75Sr0.25FeO3 system to coexistence of antiferromagnetic and weak ferromagnetic interactions In Ga doped hematite system, we noted a good signature of ferroelectricity in the nearby composition (α-Fe1.6Ga0.4O3) and prepared by mechanical alloying.7 However, some differences in the properties of chemical routed and mechanical alloyed samples may be pointed out The mechanical alloyed sample is nearly 2-3 orders less conductive than the chemical routed sample Mechanical alloyed sample does not show magnetic field induced NDR effect On the other hand, chemical routed sample is more conductive, and showed relatively large leakage of polarization and better soft ferromagnetism (higher magnetization and smaller coercivity) than the mechanical alloyed sample These comparative features indicate that magneto-conductivity effect is more prominent in chemical routed sample, whereas ferroelectric properties dominate in mechanically alloyed sample In addition to preparation technique and Ga doping effect in hematite structure, we believe that grain-boundary structure and heterogeneity due to nano-sized grains may have some affects on the I-V characteristics of the chemical routed sample The high quality synchrotron X-ray diffraction (SXRD) data have used to ¯ in the studied sample.6 We could detect an confirm the rhombohedral structure (space group R3C) extra phase whose total contribution is less than 2% of the rhombohedral and the contribution from such extremely minor phase, if really non-negligible, can be included in the contribution from grain boundary disorder and free surface, whose effect surface has been discussed for various alloyed compounds.34,35 In fact many reports highlighted the role of surface defects at the interfaces of metal electrodes and magnetic semiconductor for the NDR effect, e.g., in Pt/TiO2/Pt device.26 In such interfaces, oxygen vacancy and Schottky barrier determine the trapping and de-trapping of charge carriers, and space charge limited current (SCLC) mechanism controls I-V characteristics In a typical vacuum diode, the space charge limited current (SCLC) mechanism obeys the child’s law: I(V) ∼ Vm with power exponent m = 1.5 In a solid material where SCLC mechanism dominates, the I-V curve is expected to exhibit a sharp increase with voltage and m value is expected to be 2.17,36 Although m is appreciably affected by the magnetic fields, but m values are well below of in our sample The obtained values of m suggest that SCLC is not the dominant mechanism in our sample, although the space charge effect is not completely ruled out for the observed I-V characteristics The NDR effect is pronounced only for long waiting time of magnetic field before recording the I-V curves It states that the origin of NDR effect in our sample could be ascribed mainly to magnetic related phenomena We demonstrate that enhanced conductivity and soft ferromagnetism develop a strong magneto-electric coupling in chemical routed sample At this point, we refer two facts related to the electronic charge transfer process in hematite system First one is the electronic charge transport between magnetic Fe3+ ions in hematite structure Roso et al.37 has modeled small polaron hopping of charge carriers (electrons and holes) via superexchange paths between Fe3+ ions (Fe(3±δ)+ ↔ O ↔ Fe(3∓δ)+) The charge transfer process is anisotropic, i.e., charge hopping along in-plane direction is higher than the off-plane direction Second fact is the mechanism of electronic transitions in Ga doped hematite system under light, which is consisting of electric and magnetic field components Choi et al.38 explained the electronic spectra among different energy levels in terms of the ligand field (d-d) transitions between t2g and eg energy levels of Fe3+ ions at lower energy ( 3V) region Although the electronic transitions between different sub-levels of d shell are forbidden according to the 067126-14 R N Bhowmik and G Vijayasri AIP Advances 5, 067126 (2015) selection rule (∆l = ±1, ∆S = 0), but these ligand field (d-d) transitions in α-Fe2O3 system have been possible due to magnetic coupling of the spins of Fe3+ ions in channels A and B In Ga doped hematite system, the LMCT transitions are theoretically predicted between uppermost valence band (consisting of Fe 3d and O 2p states) and conduction band (consisting of Fe 3d, Ga 4s, Ga 4p and O 2p states) Considering the closeness of the discrete values of VP (least gap 0.345(± 0.001) V) in our sample and the energy in different electronic levels with least gap of the order of 0.38-0.40 V, we suggest that there are many possibilities of electronic transition among different energy levels during voltage sweeping and such electronic transitions are affected by the external magnetic field As the bias voltage crosses the critical value VP, there may be a substantial suppression of spin dependent electronic transitions, where spins polarization may play a significant role In rhombohedral structure of the hematite, one can model two polarized spins channels (A and B) for Fe3+ ions with spins up direction in channel A (one rhombohedral plane) and spins down direction in channel B (neighbouring plane) There is a finite canting angle between the spins of these two channels which can orient in the presence of magnetic field.3 Hence, there are two possibilities of decreasing conductivity (current) in the sample; (1) spins flipping from in-plane direction to out-of plane direction in channel A and B, (2) canting between spins from channels A and B The effect of spins polarized electrons can be realized in terms of a metal(non-magetic)-magnetic semiconductor-metal (non-magetic) tunnel junctions,25,33 where space-charge layer at the interfaces determines charge ejection, accumulation, depletion and drain out In this mechanism, bias voltage acts upon Fermi level of the Pt electrodes to force the electrons out of equilibrium and triggers the flow of unpolarized electrons from higher energy electrode to the electrode at lower energy through the magnetic semiconductor Applying the charge-spin transport model,14,33 the total current density (J) through a ferromagnetic semiconductor having conduction electrons density n and spins density s can be written as J↑↓ = Jn↑↓ + Js↑↓, Jn (x) = JnE + JnD(x), JnE = neµn E and JnD(x) = + eDn dn dx , ds Js (x) = JsE + JsD(x), JsE = seµs E and JsD(x) = + eDs dx In these expressions, D is the diffusion coefficient, and µ is the mobility of electrons, E is the electric field (voltage/thickness of the present sample), e is the absolute electronic charge, ↑ represents spin up direction and ↓ represents spin down direction, n = n ↑+n ↓ and s = n ↑-n ↓ with up and down spins density n ↑ and n ↓, respectively The electron and spin current densities can be written as Jn↑↓ = Jn↑ + Jn↓ and Js↑↓ = Js↑ - Js↓, respectively Considering the charge and spin components of current, and the process of spins orientation under simultaneous application of electric and magnetic fields, we suggest that charge and spin current contributions increased during increase of bias voltage up to the critical value VP where spin current component is saturated On further increase of the bias voltage greater than VP, the spin current may decrease due to decrease of spins polarization under the following two situations;14,33 (1) formation of more unpolarized space charge layer due to selective spin filtering effect of the ferromagnetic semiconductor on the injected unpolarised electrons from electrode, (2) formation of spins depletion region at the interface where the diffusion of spin-up electrons decreases and diffusion of spin-down electrons increases along the bias electric voltage Finally, spin diffusion is either minimized or stopped at a greater bias voltage than the value at which spin current showed a maximum value This promotes NDR effect in the material The charge drifting mechanism gives rise to an effective increase of current at higher voltage V CONCLUSIONS The studied sample is a typical canted ferromagnetic semiconductor at room temperature The I-V curve shows non-linear behaviour The non-linear I-V curves are controlled by the magnitude and direction of electric voltage, up and down modes of the variation of bias voltage and presence of magnetic field The space charge limited current (SCLC) affects the I-V characteristics up to certain extent, but the I-V characteristics are largely controlled by the charge flow and spin ordering of electrons under electric and magnetic fields In the absence of detailed experimental and theoretical works in literature on similar system, we explained the NDR effect in terms of a metal(non-magetic)-magnetic semiconductor-metal (non-magetic) tunnel junctions, where space-charge accumulation and depletion determine the spin current contribution and I-V 067126-15 R N Bhowmik and G Vijayasri AIP Advances 5, 067126 (2015) characteristics However, the observations of magnetic field controlled conductivity, electric and magnetic field controlled bi-stable electronic states, electrical hysteresis, large MR and ER in this non-conventional ferromagnetic semiconductor are attractive for the design of spintronic devices We have observed that magnetic field cycling affect on the MR, ER, peak current and voltage of the I-V curves This work will provide a platform for carrying out more experimental works in future to explore many hidden magneto-electronic features in metal doped hematite system, which emerges as a promising ferromagnetic semiconductor/insulator ACKNOWLEDGMENT We acknowledge the research supports from DST (NO SR/S2/CMP-0025/2011) and CSIR (No 03(1222)/12/EMR_II), Govt of India to continue the work M Tanaka, S Ohya, and P.N Hai, Appl Phys Rev 1, 011102 (2014) T Birol, N.A Benedek, H Das, A.L Wysocki, A.T Mulder, B.M Abbett, E.H Smith, S Ghosh, and C.J Fennie, Current Opinion in Solid State and Materials Science 16, 227 (2012) W H Butler, A Bandyopadhyay, and R Srinivasan, J Appl Phys 93, 7882 (2003) P Liao, M C Toroker, and E.A Carter, Nano Lett 11, 1775 (2011) N Naresh and R N Bhowmik, AIP Advances 1, 032121 (2011) R N Bhowmik, G Vijayasri, and R Ranganathan, J Appl Phys 116, 123905 (2014) A.G Lone and R.N Bhowmik, J Magn Magn Mater 379, 244 (2015) B Zhao, T C Kaspar, T.C Droubay, J McCloy, M.E Bowden, V Shutthanandan, S M Heald, and S A Chambers, Phys Rev B 84, 245325 (2011) A K Shwarsctein, M N Huda, A Walsh, Y Yan, G D Stucky, Y S Hu, M M Al-Jassim, and E W McFarland, J Chem Mater 22, 510 (2010) 10 J C Papaioannoua, G S Patermarakisb, and H S Karayianni, J Phys Chem Solids 66, 839 (2005) 11 A.G Lone and R.N Bhowmik, AIP Advances (accepted, 2015) 12 L Machala, J Tucek, and R Zboril, J Chem Mater 23, 3255 (2011) 13 P Chureemart, R Cuadrado, I D’Amico, and R W Chantrell, Phys Rev B 87, 195310 (2013) 14 J Ghosh, T Windbacher, V Sverdlov, and S Selberherr, Solid-State Electronics 101, 116 (2014) 15 Matthew R Sears and Wayne M Saslow, Phys Rev B 85, 014404 (2012) 16 G E Pike and C H Seager, J App Phys 50, 3414 (1979) 17 M.P Delmo, S Yamamoto, S Kasai, T Ono, and K Kobayashi, Nature 457, 1112 (2009) 18 S Takahashi and S Maekawa, Sci Technol Adv Mater 9, 014105 (2008) 19 Q He, C.-H Yeh, J.-C Yang, G Singh-Bhalla, C.-W Liang, P.-W Chiu, G Catalan, L W Martin, Y.-H Chu, J F Scott, and R Ramesh, Phys Rev Lett 108, 067203 (2012) 20 H Yang, W Mi, H Bai, and Y Cheng, RSC Adv 2, 10708 (2012) 21 X Wu, Z Xu, Z Yu, T Zhang, F Zhao, T Sun, Z Ma, Z Li, and S Wang, J Phys D: Appl Phys 48, 115101 (2015) 22 J.C Knott, D.C Pond, and R.R Lewis, PMC Phys B 1, (2008) 23 A G Volkov and A A Povzner, Physics of the Solid State 54, 2351 (2012) 24 C W Chong, D Hsu, W C Chen, C C Li, J G Lin, L C Chen, K H Chen, and Y F Chen, J Phys Chem C 116, 21132 (2012) 25 G.-X Miao, M Muăller, and J.S Moodera, Phys Rev Lett 102, 076601 (2009); G.-X Miao, M Muăller, and J.S Moodera, Phys Chem Chem Phys 17, 751 (2015) 26 K.J Yoon, M.H Lee, G.H Kim, S.J Song, J.Y Seok, S Han, J.H Yoon, K.M Kim, and C.S Hwang, Nanotechnology 23, 185202 (2012) 27 B Kundys, A Maignan, C Martin, N Nguyen, and C Simon, Appl Phys Lett 92, 112905 (2008) 28 T Wu and J F Mitchell, Appl Phys Lett 86, 252505 (2005) 29 Y S Xiao, X P Zhang, and Y G Zhao, Appl Phys Lett 90, 013509 (2007) 30 K M Kim, B J Choi, Y C Shin, S Choi, and C S Hwang, Appl Phys Lett 91, 012907 (2007) 31 H Beneking, High Speed Semiconductor Devices: Circuit aspects and fundamental behaviour (Springer, 1994), pp 114–117 32 Y S Xiao, X P Zhang, and Y G Zhao, Appl Phys Lett 90, 013509 (2007) 33 V Zayets, Phys Rev B 86, 174415 (2012) 34 B.B Straumal, A.A Mazilkin, S.G Protasova, A.A Myatiev, P.B Straumal, G Schütz, P.A van Aken, E Goering, and B Baretzky, Phys Rev B 79, 205206 (2009) 35 S G Protasova, B B Straumal, A A Mazilkin, S V Stakhanova, P B Straumal, and B Baretzky, J Mater Sci 49, 4490 (2014) 36 X.M Shen, D.G Zhao, Z.S Liu, Z.F Hu, H Yang, and J.W Liang, Solid-State Electronics 49, 847 (2005) 37 K M Rosso, D M A Smith, and M Dupuis, J Chem Phys 118, 6455 (2003); N Iordanova, M Dupuis, and K M Rosso, J Chem Phys 122, 144305 (2005) 38 S Choi, C Lefe`vre, F Roulland, C Me´ny, N Viart, B To, D.E Shafer, R Shin, J Lee, and W Jo, J Vacuum Science and Tech B 30, 041204 (2012) ... (2015) Magnetic field cycling effect on the non- linear current- voltage characteristics and magnetic field induced negative differential resistance in α -Fe1. 64Ga0. 36O3 oxide R N Bhowmika and G... temperature The I-V curve shows non- linear behaviour The non- linear I-V curves are controlled by the magnitude and direction of electric voltage, up and down modes of the variation of bias voltage and. .. random doping of Ga ions in rhombohedral planes containing Fe ions and distribution of magnetic exchange interactions B Kundys et al.27 attributed similar magnetic field induced NDR effect in