Immunity of nanoscale magnetic tunnel junctions with perpendicular magnetic anisotropy to ionizing radiation

Spin transfer torque magnetic random access memory (STT-MRAM) is a promising candidate for next generation memory as it is non-volatile, fast, and has unlimited endurance. Another important aspect of STT-MRAM is that its core component, the nanoscale magnetic tunneling junction (MTJ), is thought to be radiation hard, making it attractive for space and nuclear technology applications. However, studies on the effects of ionizing radiation on the STT-MRAM writing process are lacking for MTJs with perpendicular magnetic anisotropy (pMTJs) required for scalable applications. Particularly, the question of the impact of extreme total ionizing dose on perpendicular magnetic anisotropy, which plays a crucial role on thermal stability and critical writing current, remains open. Here we report measurements of the impact of high doses of gamma and neutron radiation on nanoscale pMTJs used in STT-MRAM. We characterize the tunneling magnetoresistance, the magnetic field switching, and the current-induced switching before and after irradiation. Our results demonstrate that all these key properties of nanoscale MTJs relevant to STT-MRAM applications are robust against ionizing radiation. Additionally, we perform experiments on thermally driven stochastic switching in the gamma ray environment. These results indicate that nanoscale MTJs are promising building blocks for radiation-hard non-von Neumann computing.


S1 Irradiation results
Due to the stochastic, thermally-activated character of both the field-and current-induced switching, the switching fields and switching currents exhibit a statistical distribution of values in the same pMTJ device. In our analysis below, the high and low switching fields (left and right coercive fields), H /2. To visualize and characterize the thermally induced spread in switching fields and switching currents we use "error bars" that are represented by the interquartile range (IQR) of the measured statistical distributions given by where is Q 1 is the first quartile and Q 3 is the third quartile. In the case of n switching events, Q 1 is given by the median of the smallest n/2 switching events and Q 3 is given by the median of the largest n/2 switching events. . The characteristic spread in current switching loop center is δ I 0 = δ I w /2. The total "error bars" plotted in Fig. 2 and Fig. S1-S4 are given by ± the characteristic spread for the corresponding value.
We monitor the effects of irradiation on a set of pMTJ parameters X important for the operation of STT-MRAM, where (X = TMR, H 0 , H w , I 0 , I w ). The irradiation-induced change ∆X in the mean value of the parameter X for a set of pMTJ devices is shown in Table 1. This table also shows standard deviation σ X of irradiation-induced changes in X calculated for this set of pMTJ devices.
Specifically, the irradiation-induced change in the mean value of X is defined as where i is the pMTJ device index and N is total the number of pMTJ devices in the set. For example, a negative value for the change in the width of the field switching loop, given by , would correspond to a general trend for the narrowing of the loop, and thus a reduction in the coercivity, after a particular irradiation.
The standard deviation σ X , defined as gives a measure of the error in detecting irradiation-induced change in parameter X.
As the field and current induced switching processes are thermally assisted, we would additionally like to compare irradiation-induced changes ∆H w and ∆I w to the thermal spread of these parameters δ H w and δ I w . The thermal spread of the parameters δ X are an indication of the experimental error in determining the value of parameter X as the error is dominated by thermal fluctuations.  Specifically, we compare ∆H w to the average of thermal spread in the field loop width over the set of pMTJ devices: where the index j includes individual measurements before and after the irradiation and thus the total measurements is twice the pMTJ device set size. Similarly, we calculate the average thermal spread in the current switching loop width I w as The values for ε H w and ε I w are also tabulated in Table 1.
The before and after irradiation results for TMR, H 0 , H w , I 0 , and I w are shown in Fig. S1 for the gamma irradiated circular pMTJs, Fig. S2 for the gamma irradiated elliptical pMTJs, and Fig. S3 for the TRIGAirradiated circular pMTJs. The before and after 6 months waiting time for the non-irradiated control elliptical pMTJs are shown Fig. S4. For the vast majority of the devices the before/after medians of TMR, H 0 , H w , I 0 , and I w overlap within one IQR, which is a visual indication that the extreme doses of ionizing radiation had negligible permanent effect on the pMTJs. This can also be seen numerically in Table 1 that shows that ensemble averages of irradiation-induced changes in all these parameters do not exceed one standard deviation of these changes over the ensemble in all experiments with the exception of ∆H w for the elliptical pMTJs. In this case, the observed change in ∆H w = −32 Oe is slightly larger than one standard deviation σ H w = 30 Oe; however, we find the apparent change in H w is still less than the experimental error in determining H w due to thermal fluctuations ε H w = 40 Oe.

S2 Tuning superparamagnetic pMTJ to center of bistable region using spin transfer torque
We use random telegraph noise (RTN) measurements to characterize thermally activated switching of pMTJs in the gamma ray environment. Traditionally RTN measurements are performed in an applied external field that is used to compensate for reference layer stray fields and tune the MTJ to the center of the bistable region 1, 2 . In the center of the bistable region, the dwell time in the P and AP states are equal, τ P = τ AP . We define the switching rate w x for a given state (x = P, AP), as the inverse of the dwell time of that state such that w P (≡ 1/τ P ) is the switching rate from P → AP and w AP (≡ 1/τ AP ) from AP → P. In the the center of the bistable region w P = w AP ; therefore, the switching rate for the pMTJ can be defined as w = w P = w AP . Since our gamma irradiation chamber lacks in situ magnetic field tuning capabilities, we use an alternative method of biasing the sample to the center of the bistable region. Specifically, we apply spin transfer torque to tune the pMTJ into the center of the bistable region 3 . Fig. S5 shows the switching rates w P and w AP as a function of applied bias current I. The switching rates display an exponential dependence on the bias current (logarithmic y-axis) 3 . A slower switching rate, longer dwell time, indicates that state is more stable. For example in Fig. S5, reducing the magnitude of I biases the pMTJ into the P state as indicated by reduced w P and increased w AP . Increasing the magnitude of I biases the pMTJ toward the AP state as indicated by increased w P and decreased w AP . We use bias current of I = −4.98 µA for this sample to tune the pMTJ into the center of the bistable region such that w P = w AP . w P (P →AP) w AP (AP →P) Figure S5. Switching rates w P and w AP as a function of bias current I applied to superparamagnetic pMTJ. The condition w P = w AP indicates the center of the bi-stable region where the pMTJ has equal probability of being in the P and AP states.

S3 Estimate of temperature change of pMTJ in gamma chamber for thermally activated switching experiment
The Arrhenius law 4 can be used to estimate the change in temperature required for the observed deviation of the switching rate w of the superparamagnetic MTJ in the random telegraph noise experiment. The Arrhenius law written in terms of switching rate is, where ∆E is the magnetic anisotropy energy barrier and w 0 ≈ 1 GHz is the attempt frequency 5,6 . For the observed switching rate in the absence of gamma irradiation w (300K) = 200 Hz, we can estimate the effective energy barrier expressed as a temperature of ∆E/k B = 4630 K. The average switching rate with the gamma irradiation on, w γ = 182.2 Hz, could be caused by a temperature change of the pMTJ of ∆T < 2 K.