Reduced low frequency noise in electron beam evaporated MgO magnetic tunnel junctions

Abstract

We compare low frequency noise in magnetic tunnel junctions with MgO barriers prepared by electron-beam evaporation with those prepared by radiofrequency sputtering, both showing a high tunneling magnetoresistance. The normalized noise parameter in the parallel state of junctions with evaporated barriers is at least one order of magnitude lower than that in junctions with sputtered barriers, and exhibits a weaker bias dependence. The lowest normalized noise is in the 10 −11 m 2 range. A lower density of oxygen vacancies acting as charge trap states in the evaporated MgO is responsible for the lower noise. © 2010 American Institute of Physics. ͓doi:10.1063/1.3431620͔Magnetic tunnel junctions ͑MTJs͒ have attracted a great deal of attention since the demonstration of the tunneling magnetoresistance ͑TMR͒ effect at room temperature. 1,2 Following theoretical predictions, 3,4 a large TMR ratio of up to 200% at room temperature in MTJs with CoFeB electrodes and MgO tunnel barriers was achieved. 5,6 A record room temperature TMR of 604% has since been reported in a pseudospin valve stack, 7 which is close to the theoretical maximum. 3,4 Major advances in MTJs have led to important applications in hard-disk read heads, 8 sensors, 9 and magnetic random access memory. 10 For magnetic field sensing applications, both signal and noise in the MTJ are of equal significance. It has been found previously that 1 / f noise usually dominates other kinds of noise, e.g., thermal noise and shot noise, at frequencies up to a few kilohertz in an MTJ device, and it limits the low-frequency sensitivity. [11][12][13][14][15][16] The 1 / f noise in these devices is normally quantified by the Hooge parameter ␣, defined as AfS V / V 2 , in which A is the junction area, f the frequency, S V the noise power spectrum density, and V the voltage applied to the MTJ. 12,17 In conventional radio frequency ͑rf͒ sputtered MTJs with CoFeB electrodes and MgO tunnel barriers, ␣ was reported to decrease as a function of annealing time and reach a minimum value of 2 ϫ 10 −10 m 2 . 18 The lowest ␣ value ever observed in MTJs, between ͑1-5͒ ϫ 10 −11 m 2 , was achieved in fully epitaxial Fe͑100͒/MgO͑100͒/Fe͑100͒ junctions grown by molecular beam epitaxy ͑MBE͒. 19 Recently, some of us suggested electron-beam evaporation as an alternative to MBE or rf-sputtering for growing high quality MgO tunnel barriers. The junctions fabricated in this way have been shown to possess high TMR ratios of ϳ240% at room temperature and a reduced density of oxygen vacancies, compared to counterparts grown entirely by rf-sputtering. 20 In this work, we report on low frequency noise in the MTJs with electron-beam evaporated MgO barriers of different thicknesses ͑EB-MTJs͒, comparing it to that measured in our MTJs with rf-sputtered MgO tunnel barriers ͑rf-MTJs͒. We show that the Hooge parameter ␣ in the EB-MTJs in their parallel state is at least one order of magnitude lower than that in the rf-MTJs, across a large range of resistance-area ͑RA͒ product 5 ϫ 10 2 Ͻ RAϽ 1 ϫ 10 6 ⍀ m 2 ...

Resistance switching in transparent magnetic MgO films

“…Growth conditions have been previously observed to control the density of oxygen vacancies within the MgO [23]. In particular, adding O2 to the Ar plasma during MgO growth also helps in altering the oxygen population and the barrier performance [24].…”

Tunable linear magnetoresistance in MgO magnetic tunnel junction sensors using two pinned CoFeB electrodes

“…The TMR was measured with the applied field along the original bottompinned layer direction, and an obvious linear response is obtained with a TMR ratio of 219%. It can be improved up to 250% by changing the applied field direction by 30 . Figure 2(a) summarizes the TMR ratios for several junctions against t Ru .…”

1 / f noise in MgO double-barrier magnetic tunnel junctions

“…In contrast to the bias dependence of TMR, ␣ AP in DMTJs and SMTJs varies asymmetrically with bias, reflecting dissimilar CoFeB/MgO interfaces in the MTJ stacks 11,13 and possibly more oxygen vacancies appearing at the top interfaces. 22 Importantly, V 1/2 in both ␣ and TMR for DMTJs is approximately double that for SMTJs. 10,12 The noise-bias dependence was also measured for the DMTJ at -45 mT, where only the magnetization of the bottom pinned CoFeB layer switches.…”

“…High vacuum postannealing was performed in the temperature range of 250-375°C in an applied magnetic field of 800 mT for 1 h. The magnetotransport and noise measurements are conducted as described in our previous publication. 22 Positive bias represents electron flow from top to bottom CoFeB layers. Figure 1 shows the resistance-area product ͑RA͒ versus magnetic field of a DMTJ with the CoFe 2 nm/CoFeB 0.8 nm free layer at T a = 375°C.…”

“…It has been previously reported that 1 / f noise dominates the low frequency response of MTJs. [15][16][17][18][19][20][21][22][23] The 1 / f noise can be characterized by a noise magnitude parameter ␣ = AfS V / V 2 , where A is the junction area, f is the frequency, S V is the noise power spectrum density, and V is the applied bias. 15 Recently, Guerrero et al 24 suggested that 1 / f noise can be greatly reduced in field sensors, when a large number of MTJs are connected either in series or in parallel.…”