Electromagnetic Radiation Effects on MgO-Based Magnetic Tunnel Junctions: A Review

Magnetic tunnel junctions (MTJs) have been widely utilized in sensitive sensors, magnetic memory, and logic gates due to their tunneling magnetoresistance. Moreover, these MTJ devices have promising potential for renewable energy generation and storage. Compared with Si-based devices, MTJs are more tolerant to electromagnetic radiation. In this review, we summarize the functionalities of MgO-based MTJ devices under different electromagnetic irradiation environments, with a focus on gamma-ray radiation. We explore the effects of these radiation exposures on the MgO tunnel barriers, magnetic layers, and interfaces to understand the origin of their tolerance. This review enhances our knowledge of the radiation tolerance of MgO-based MTJs, improves the design of these MgO-based MTJ devices with better tolerances, and provides information to minimize the risks of irradiation under various irradiation environments. This review starts with an introduction to MTJs and irradiation backgrounds, followed by the fundamental properties of MTJ materials, such as the MgO barrier and magnetic layers. Then, we review and discuss the MTJ materials and devices’ radiation tolerances under different irradiation environments, including high-energy cosmic radiation, gamma-ray radiation, and lower-energy electromagnetic radiation (X-ray, UV–vis, infrared, microwave, and radiofrequency electromagnetic radiation). In conclusion, we summarize the radiation effects based on the published literature, which might benefit material design and protection.


Tunnel Magnetoresistance
The phenomenon of tunnel magnetoresistance (TMR) has gained enormous attention in recent decades because of its essential applications in nonvolatile magnetoresistive random-access memory (RAM) and next-generation magnetic field sensors [1][2][3][4][5][6][7][8]. This interest follows the emergence and success of related magnetoresistance, such as anisotropic magnetoresistance (AMR) and giant magnetoresistance (GMR). Tunneling, as a foundational principle of TMR, arises from the quantum mechanical wave nature of particles and the non-zero probability of particles occupying classical forbidden regions.
The phenomenon of magnetoresistance (MR) was first discovered in 1856 [9] in nickel and iron sheets subjected to parallel or perpendicular magnetic fields; this is known as anisotropic magnetoresistance (AMR). The magnitude of electric resistance changed by about 2% at room temperature for alloy AMR materials [10]. The AMR effect was attributed to there being a higher probability of s − d scattering of electrons traveling along the

Oscillatory TMR effect with respect to MgO barrier
thickness It is difficult to investigate the detailed mechanisms of TMR by using the amorphous Al-O barrier because of its structural uncertainty, and there has not been significant progress in understanding the physics of the TMR effect since Julliere proposed the phenomenological model. 1) Unlike Al-O-based MTJs, epitaxial MTJs with a singlecrystal MgO(001) barrier are a model system for studying the physics of spin-dependent tunneling because of their well-defined crystalline structure with atomically flat interfaces. Other than the giant TMR effect, several interesting phenomena have been observed in epitaxial MgO-based MTJs. For example, epitaxial Fe(001)/MgO(001)/Fe(001) MTJs were observed to exhibit an interlayer exchange coupling mediated by spin-polarized tunneling electrons, 31,32) complex spin-dependent tunneling spectra, 33) and oscillatory TMR effect with respect to MgO barrier thickness. 27,34) It should be noted that these phenomena has never been observed in conventional MTJs with an amorphous Al-O barrier. Clarifying their mechanisms will lead to a deeper understanding of the physics of spin-dependent tunneling. Among the novel phenomena observed in epitaxial MgO-based MTJs, we discuss the oscillatory TMR effect in this subsection.
The MR ratio of epitaxial Fe(001)/MgO(001)/Fe(001) MTJs has been found to oscillate with respect to MgO barrier thickness, t MgO , as described below. 27,34) We made a high-quality fully epitaxial Fe(001)/MgO(001)/Fe(001) MTJ film with a wedge-shaped MgO barrier layer by using MBE growth. The film with a wedge-shaped MgO layer was made into MTJs with a junction area A of 36 mm 2 by using high-precision microfabrication processes 34) that suppressed sample-to-sample variation in junction area A and thus resulted in small sample-to-sample variation in the transport properties. Figure 12 shows the t MgO dependence of a resistance-area (RA) product: tunneling resistance for a unit junction area. The tunneling resistance increases exponen-ratio is seen to oscillate as a functi the solid gray line in Fig. 13(a), th relatively well by a single-period simple cosine curve) with an oscilla plus a background curve. The small observed MR ratios and the fitting c should be noted that such an osci dependence of MR has never been an amorphous Al-O barrier. It mi oscillation is a result of the layer-b of MgO(001), in which the grow almost completed before the grow begins. This growth could cause a ratio because the interface morpholo changes periodically, layer by lay origin of the observed oscillation oscillation period (3.17 Å ) is no monoatomic MgO(001) layer (2.1 Å Butler et al. proposed a model tunneling states. 15) Interference betw states at E F in MgO that correspond could cause an oscillation of tunn function of t MgO as follows. These s vectors, the perpendicular compon which are expressed as k 1 ¼ k r 1 þ Nonzero real parts of the complex essential for quantum-interferencenomena. When k k Á Áz > 0:59 [Áz of MgO(001)], k r 1 6 ¼ k r 2 (6 ¼ 0) and transmittance T can then be expres The tunneling transmittance for lates as a function of t MgO with a

Electronics
One of the most well-known applications of MTJs is their use in data storage, particularly in MTJ-based memory devices [17,[32][33][34], including dynamic random-access memory, flash memory, and hard disk drives. Data can be stored without the need for external magnetic fields [35]. Many review articles have been published on this topic, as listed in the preceding section. MgO-based TMJs exhibit high MR ratios at room temperature and have been utilized in hard disk drives (HDD) with a high density [17,33,[36][37][38][39][40].
Additionally, an MTJ is comprised of two distinct states, namely, parallel and antiparallel. Consequently, a single MTJ has the capability to store data in four different states [41]. Therefore, stacked MTJs are suitable for use in magnetic random-access memory (MRAM) applications [2,7,17,31,[42][43][44][45][46][47][48][49][50]. These nonvolatile MRAMs demand high MR ratios of >150% at room temperature. Figure 3 shows a typical structure of an MRAM element. The element's primary component is an MTJ, which comprises a ferromagnetic free layer, an insulating tunnel barrier, and a ferromagnetic fixed layer. These sandwiched layers exhibit TMRs as a result of spin-dependent electron tunneling [31,42,43,51]. A recent review of the structure of MRAMs can be found in reference [49]. MRAM has the potential to replace all existing memory devices because of it capability of combining the speed of static random-access memory (SRAM) and the density of dynamic random-access memory (DRAM), while also being nonvolatile like hard disk drives (HDD). Therefore, MRAM is a highly desirable form of memory [16,20,32]. As a result, MTJ-based MRAM devices have been extensively investigated in the past two decades.
Compared with other kinds of random-access memory, MTJ-based MRAMs are a type of nonvolatile memory that can work in irradiation environments, such as in outer space applications [52]. The radiation tolerance of these MRAMs is critical to their effectiveness in such harsh environments.
In addition, MTJs have the potential to be employed as magneto-electric spin logic devices, which are capable of converting analog signals to digital ones. Various designs of analog-to-digital converters (ADCs) have been proposed [53][54][55], including sigma-delta (Σ − ∆) ADCs with high bit resolution [55]. As compared to traditional ADCs, the energy consumption of these MTJ-based ADCs is very low, down to 66 fJ for 4-bit MTJ-based ADCs and 37 fJ for 3-bit MTJ-based ADCs [54].

Energy Harvesting
In addition to their conventional applications in memory and sensors, MTJ devices hold promise for renewable energy generation and storage. While relatively new in the field, MTJ-based energy devices have attracted considerable research attention. Although their device efficiencies are lower than those of traditional energy devices, the novel heterostructures of MTJs have the potential to significantly impact the area. Many research groups have explored the fundamentals and future prospects of energy applications involving the spin of electrons. Various kinds of energy, ranging from heat and light to mechanical vibrations, have been successfully converted to electricity through spin conversion [34,61,62].

Heat:
Based on EU data, a considerable amount of industrial energy consumption, ranging from 20% to 50%, is lost as waste heat. In the United States, up to 1734 trillion Btu of waste heat went unrecovered in 2008 [63]. The Seebeck effect, which involves electromotive force (emf) generation under a temperature gradient, has been widely investigated in the past decades. Thermoelectric (TE) devices have the capability to convert heat to electricity based on the Seebeck effect. Because of their unique characteristics, such as having no moving parts, quiet operation, a low environmental impact, and high reliability, TE devices have attracted widespread interest since their discovery. In the past decades, semiconductor TE materials, especially ceramic nanocomposite bulks [64][65][66][67][68], have been developed for this purpose. Up to now, various TE nanocomposites have been investigated in the absence of magnetic fields [69][70][71][72][73][74][75][76][77][78][79][80]. Device efficiencies of up to 10% can be achieved using semiconductor TE devices.
Spin caloritronics, the combination of spintronics and thermoelectrics, is an emerging field [61,81]. Electron spin waves interact with heat in insulating ferromagnets under magnetic fields through the magneto-Seebeck effect, also referred as the spin Seebeck effect or magneto-thermopower effect. A thermal gradient can lead to the production of magnetothermopower and magneto-thermocurrent [82]. Therefore, spin caloritronic devices can serve as waste heat recyclers and heat sensors under magnetic fields.
MTJ devices, comprised of insulating barriers and ferromagnetic layers, can utilize spin caloritronics to generate pure spin currents via magnetization dynamics induced by a temperature gradient. These MTJ devices have a unique potential in harvesting thermal energy and there have been many studies focused on MTJ-based heat recycling in the past decade. The spin Seebeck coefficients of various MTJs have been measured under magnetic fields [81][82][83][84][85][86][87]. For example, CoFeB/MgO/CoFeB MTJs have been integrated with resistance thermometers to recycle waste heat from the spin Seebeck effect [61]. A Seebeck coefficient of Al 2 O)3-based MTJs was measured up to 1 mV/K [88]. A large spin-dependent Seebeck coefficient of 100 µV/K was observed in CoFeB/MgO/CoFeB MTJs [89]. However, due to their nanoscale thickness, the output power of MTJs is much lower than that of their semiconductor TE bulk counterparts (up to kW). It was reported that the output TE power of a CoFeB/MgO/CoFeB MTJ device was only 10 pW per 12.6 cm 2 (∼10 nW/m 2 ) [90]. Even compared with that of semiconductor TE film devices (up to several hundred W/m 2 ), the output power is very low for the state-of-art MTJ devices. Although the power output of present MTJ devices is unsatisfactorily low for industrial heat recyclers, MTJ devices are one kind of emerging energy-harvesting device.

Solar energy:
In addition to their capacity for heat recycling, MTJs can generate electricity through the utilization of solar energy. A phonon can couple to the electron spin and magnon, which enables the generation of spin currents from solar energy [34]. More recently, photoinduced spin currents were observed [62]. Furthermore, the potential of MTJs was explored for spin photovoltaic applications [91]. Recently, a new research field known as spinmechanics or spinmechatronics has emerged, combining spin currents with mechanical motion [62]. Spin currents can be generated from mechanical energy such as vibrations and sounds [92,93].
In brief, MTJs have the capacity to convert different kinds of energy into electricity through the amalgamation of electron spin with established energy conversion techniques. These research areas are relatively nascent and are expected to find many applications in the forthcoming decades.

Electromagnetic energy:
It was reported that MgO-based MTJs could produce significant DC voltage when exposed to microwave radiation [94]. A DC voltage was generated under microwave irradiation with a frequency of 1 MHz to 40 GHz and power density of 10-10 × 10 6 mW/m 2 , with a sensitivity of up to 5000 mV/mW. A similar phenomenon was also observed in AlO xbased MTJs exposed to microwave radiation with a power of 1-100 mW and frequency of 1.5-2.5 GHz [58].

Energy Storage
MTJs have potential applications in the field of energy storage as well, particularly with respect to batteries and capacitors, which are two kinds of popular devices to store energy. Recent work on MTJ-based energy storage devices is highlighted below.

Capacitors:
The magnetic capacitance of MTJs was first investigated in Co/Al 2 O 3 /Co MTJs in the 2000s, and their potential application as supercapacitors for energy storage was explored [95]. The tunneling magnetocapacitance (TMC) of Co 40 Fe 40 B 20 /MgO/Co 40 Fe 40 B 2 MTJs was measured at room temperature about two decades later [96,97]. The voltageinduced TMC ratio reached 1000% due to the emergence of spin capacitance. An inverse of TMC was observed in Fe/AlO x /Fe 3 O 4 MTJs [98]. The inverse TMC reached up to 11.4% at room temperature and could potentially reach 150%. It is believed that the spin accumulation in anti-parallel configurations of MTJs leads to a difference in spin-up and spin-down diffusion lengths, creating a charge dipole that acts as an extra serial capacitance and gives rise to the observed TMC effect [99].
In a recent study, it was reported that MgO-based (001)-textured MTJs exhibited a significant TMC, of 332% at room temperature [100]. Subsequently, an even higher TMC, of over 420% at room temperature, was achieved using epitaxial MTJs with MgAl 2 O 4 (001) barriers possessing a cation-disordered spinel structure [100].
There findings highlight the potential of MTJs in the development of capacitors and related technologies.

Batteries:
MTJ devices have also been employed as spin batteries for the conversion of the magnetic energy of superparamagnetic nanomagnets into electricity [101]. The examined MTJs contained MnAs nanomagnets with a zinc-blende structure. These nanomagnets were chargeable under magnetic fields, providing evidence for the existence of spin batteries. The resulting electromotive force (em f ) was found to operate on a timescale of approximately 10 2 -10 3 s. The em f should result from the conversion of the magnetic energy of the superparamagnetic nanomagnets into electrical energy during their magnetic quantum tunneling.
MTJ devices have diverse applications, such as data storage, sensors, energy generation and storage, even under irradiation. Consequently, it is crucial to evaluate their capacity to withstand irradiation. Here, we focus on the radiation tolerance of MgO-based MTJs. This information provides valuable insights into their stability, and benefit error-free operation and protection of MgO-based MTJ devices in irradiation environments.

Irradiation
MgO-based MTJs may work in various irradiation environments. Therefore, it is necessary to review both natural and artificial sources of radiation prior to reviewing the radiation tolerance of MgO-based MTJs.

Natural Radiation Sources
The Sun is the major natural radiation source in our life [102,103]. Nuclear fusion processes within the Sun produce cosmic rays that consist of high-energy atomic nuclei and electromagnetic waves, which spread through the solar system. These primary cosmic rays are composed primarily of 99% nuclei (protons accounting for 90% , alpha particles accounting for 9%, and heavier element nuclei making up 1%) and approximately 1% solitary electrons and the electromagnetic component (gamma rays, X-rays, UV-visible light, and IR light). The energy of the primary cosmic rays is high, up to 10 20 eV. Owing to the Earth's magnetic field, the energetic particles are deflected and trapped within the Van Allen radiation belts. The belts extend from an altitude of about 640 km to 58,000 km above the Earth's surface, as shown in Figure 4a. Various spacecraft components, including MTJ devices, can be exposed to primary cosmic rays in the Van Allen belts.  Upon entering the Earth's atmosphere, the primary cosmic rays collide with atoms and molecules present in the atmospheric layers [104]. These collisions produce secondary cosmic particles with lower energy and electromagnetic waves. The secondary particles and electromagnetic waves include low-energy neutrons, protons, electrons, alpha particles, γ-rays, and X-rays. The energy of the secondary cosmic particles and electromagnetic waves is much lower than that of the primary cosmic rays, but is still considerable. For instance, the energy of the secondary γ-rays can be 50 MeV on the Earth. Due to their sufficiently high energy, these secondary cosmic particles and electromagnetic waves can potentially damage MTJ devices, leading to soft errors in MTJ-based electronic integrated circuits. There are natural radioactive minerals on the Earth, such as compounds containing uranium-238 (U-238) and thorium-232 (Th-232) radionuclides. These radioactive elements emit high-energy particles or rays in the natural environment. As such, these minerals are another kind of natural radiation source on the Earth. Figure 4b shows the full spectrum of electromagnetic radiation on the Earth. Table 1 lists the wavelength, frequency, and energy of various types of electromagnetic wave. Thus, it is necessary to investigate the potential radiation effects on microelectronic devices that are exposed in outer space or on the Earth. This is particularly important in the case of MTJ devices that are deployed in spacecraft, satellites, and airplanes, which operate in an irradiation environment filled with high-energy particles and high-energy electromagnetic waves.

Artificial Radiation Sources
Besides the natural radiation sources, various artificial sources of radiation exist on the Earth, including nuclear weapons, nuclear power plants, television transmitting towers, microwave ovens, and wireless phones. These artificial radiation sources are also omnipresent in our surroundings, as shown in Figure 4b. For instance, modern microwave ovens used in kitchens can produce microwaves with a frequency of 2450 MHz [106]. Cellphone towers can emit electromagnetic radiation with frequencies of 800 MHz and 1900 MHz for 3G cellphone communications [107,108], with frequencies of 24-47 GHz for high-band 5G phones [108,109]. Moreover, even human bodies can emit infrared radiation [110]. Although the energy of this artificial radiation is significantly lower than that of cosmic rays, it is still required to know if this artificial radiation damages MTJ devices or degrades MTJ device performance. Therefore, this review paper comprehensively examines the radiation impacts of various electromagnetic waves, including γ-rays, X-rays, UV-visible light, microwaves, and even infrared radiation.
To date, various artificial radiation sources have been utilized in laboratories to quantitatively investigate the radiation effects on MgO-based MTJ devices. Most of the data reviewed here were collected using these radiation sources. Table 2 lists some typical radiation sources utilized in the cited literature here.
These artificial radiation sources can produce controllable electromagnetic particles in laboratories. Particle accelerators and synchrotron radiation, for instance, can generate high-energy particles, including neutrons and electrons, with energies ranging from 0.1 MeV to 1.0 MeV and a high flux. γ-rays are usually generated from radioisotopes in laboratories, with energies from several keV to MeV. Some specialized devices, such as electron microscopes, can produce middle-energy particles, of 5-200 keV. Commercial X-ray tubes can emit low-energy X-rays, of tens of electronvolts. Various light sources, such as xenon and halogen bulbs, can generate UV-visible light with energies in the electronvolt range. Additionally, infrared radiation below 1 eV can be generated from electric furnaces in laboratories.

Radiation Units
The impacts of radiation are generally categorized into three types [103,104,122]: (1) total ionizing dose (TID), which is quantified in rad or gray units. TID effects can change the threshold voltages of electronic devices due to trapping of charges during radiation exposure. TID may cause leakages of electric currents. (2) Single event effects (SEE), which are not cumulative but result from individual interactions. SEE may cause soft errors and hard errors of devices. (3) Displacement damage dose (DDD), which can generate lattice defects. Sufficient displacement may change the device or material's performance properties over time. TID and SEE are examples of ionizing radiation effects, while DDD is an instance of a non-ionizing radiation effect. TID and DDD can lead to lasting damage to electronics over an extended period, showing long-term effects, whereas SEE typically results in immediate short-term effects. However, both short-term and long-term effects can potentially have permanent consequences.
To facilitate comprehension of the impact units, a brief summary is provided here. There are four kinds of ionizing radiation quantities: (1) Activity quantity, with units of becquerel (Bq), curie (Ci), and rutherford (Rd); (2) exposure quantity, with units of coulomb per kilogram (C/kg) and röntgen (R); (3) absorbed dose quantity, with units of gray (Gy), erg per gram, and radiation absorbed dose (rad); and (4) equivalent dose quantity, with units of sievert (Sv) and röntgen equivalent man (rem). The definitions of these radiation quantities are also listed in Table 3 for readers without a background in radiation. Table 3. Radiation unit and Terms.

Category Unit Definition
Activity Becquerel (Bq) * activity of a quantity of radioactive material in which one nucleus decays per second (1/s) Curie (Ci) quantity or mass of radium emanation in equilibrium with one gram of radium (element), 1 Ci = 3.7 × 10 10 Bq Rutherford (Rd) activity of a quantity of radioactive material in which one million nuclei decay per second, 1 Rd = 1,000,000 Bq Exposure Röntgen (R) quantity of radiation which liberates by ionization one esu (3.33564 × 10 10 C) of electricity per cm 3 of air under normal conditions of temperature and pressure, 1 R = 2.58 × 10 −4 C/kg

Properties of MTJ Materials
MgO-based MTJs are composited of MgO insulating barriers and ferromagnetic layers. The ferromagnetic layers consist of free layers and fixed layers, typically made of ferromagnetic Fe and CoFeB. In order to understand the radiation tolerance of MgO-based MTJs, the physical properties of MgO and Fe/COFeB are first summarized below. The radiation tolerance of MgO-based MTJs is related to these properties.

Magnesium Oxide Barrier
Magnesium oxide (MgO) possesses an ionic bonding structure, consisting of Mg 2+ and O 2− , with a crystallographic structure of rock salt (NaCl). Figure 5 shows its crystallographic structure. Figure 6 shows its monolayer structure. Magnesium and oxygen atoms alternately stack in the lattice.  MgO is an excellent electrical insulator, exhibiting a conductivity of 10 −14 µS/m at room temperature. Additionally, it is a soft magnetic material, with a magnetic susceptibility of −10.2 × 10 −6 cm 3 /mol. The compound is also a refractory material, with physical and chemical stability up to 2500 • C. Its physical properties are listed in Table 4. Ferromagnetic materials are utilized as free/fixed layers in MgO-based MTJs. Crystalline (001) iron films were initially used as free/fixed layers in MgO-based MTJs to achieve an MR ratio of 220% [26,124] at the beginning of the 2000s. Subsequently, crystalline Co(001) films were employed as free/fixed layers of MgO-based MTJs, achieving an MR ratio of 410% [28]. Currently, CoFeB is extensively used in MgO-based MTJs, and the MR ratio has been enhanced to 500-600% at room temperature [15,16,30]. The structural, thermal, and magnetic properties of these three ferromagnetic materials are listed in Table 5.

Theoretical Radiation Tolerance of MTJs
Radiation-induced damage to electronic circuits has been known since the 1950s. In the 1970s, memory and logic perturbations were detected in satellite electronic devices as a result of heavy-ion radiation within the solar wind [129]. Subsequently, soft errors caused by cosmic rays were reported in Si-based DRAM memory chips at the end of the 1990s [130]. Serving as a counterpart to Si-based devices, the stability of MTJ devices has also been investigated. In this subsection, the theoretical work will be discussed, while the experimental research will be covered in the subsequent section.
Theoretical investigations of the radiation effects on MTJs were initially carried out using the Julliére model [18] and the theory of electron tunneling, both of which established TMR models. In this subsection, the Julliére model will be first discussed, followed by the electron tunneling model.
According to a report in 1997 [131], the Julliére model is more suitable for amorphous barriers, not a precise representation of the magnetoconductance exhibited by free electrons tunneling through a crystalline barrier. Instead, in the case of thick barriers, Slonczewski's model may offer a more accurate approximation. Ionizing radiation, such as γ-rays, can displace atoms and create local lattice disorder, leading to the formation of amorphous regions in barrier layers. Therefore, despite this limitation, the Julliére model is employed here to illustrate the effect of an amorphous state in barrier layers, which is induced by irradiation. The model offers a simplified visual representation of the degradation caused by irradiation.
In non-magnetic materials, the populations of spin-up electrons and spin-down electrons are equal, and are randomly distributed in an equilibrium state. Conversely, in ferromagnetic materials, electron spins are aligned spontaneously, resulting in unequal numbers of spin-up and spin-down electrons. The unequal spin-up and spin-down electrons can tunnel into the empty states of the initial spin channel, which affects electrical resistance under magnetic fields, resulting in non-zero MR ratios. The MR ratios of an MTJ can be expressed in terms of the conduction electron spin polarization P i of the ferromagnetic layers [18,132]. where Here, i = 1, 2. D i,↑ (E F ) and D i,↓ (E F ) are the spin-dependent densities of states of the free/fixed layers at the Fermi energy (E F ) for the majority-spin and minority-spin bands. The spin polarization of the free/fixed layers P i (i = 1, 2) is affected by the free/fixed layer materials. Based on the Julliére model, any factors changing the Bloch states (such as momentum and coherency) within the free/fixed layer can affect the tunneling probabilities and change the TMR ratios. The concept of electron tunneling can explain MTJ too, with a particular focus on crystalline barrier MgO-based MTJs [133,134]. It is generally accepted that the effectiveness of MgO-based MTJs is highly dependent on the crystallinity of the insulating MgO barrier. Figure 7 schematically illustrates coherent tunneling transport in MgO(001)-based MTJs. As illustrated in the schematic, there are three kinds of evanescent states (also known as tunneling states) for ideal coherent tunneling in the band-gap of MgO(001): ∆ 1 , ∆ 2 , and ∆ 5 . ∆ 1 Bloch states are highly spin polarized in the ferromagnetic layers, and tunneling probability is a function of κ wave vectors. Theoretical studies suggested that the ferromagnetic ∆ 1 states dominate the tunneling process through the MgO(001) barrier [133,134]. When the symmetries of tunneling wave functions are conserved, ferromagnetic ∆ 1 Bloch states can couple with MgO ∆ 1 evanescent states, which have the slowest decay and highest tunneling probability [133] along the [001] direction. The dominant tunneling channel for the parallel magnetic state is free layer ∆ 1 ↔ MgO ∆ 1 ↔ fixed layer ∆ 1 . In the parallel magnetic states, the majority-spin conductance occurs dominantly at κ = 0 because of the coherent tunneling of majority-spin ∆ 1 states. In contrast, for the minority-spin conductance in the parallel magnetic state and the conductance in the anti-parallel magnetic state, spikes of tunneling probability would appear at the finite κ points. Although a finite tunneling current flows in the anti-parallel magnetic state, the tunneling conductance of the parallel magnetic state is much higher than that in the anti-parallel magnetic state, leading to a very high MR ratio.  According to the theory of electron tunneling, any modification to the symmetry of the MgO barriers and ferromagnetic free/fixed layers would affect the MR ratio of MgO-based MTJs. This means that the symmetry of both the propagating states in the magnetic layers and the evanescent state in the MgO barrier is critical in determining the tunneling conductance. The symmetry matching of the Bloch actively controls the tunneling conductance and MR states in both the free/fixed layers and the evanescent states in the barrier. Any changes to the symmetry of the MgO barrier and magnetic layers would affect the effective ∆ 1 states between the MgO barrier and ferromagnetic layers, thereby changing the MR ratios.
As discussed above, several essential factors, including the crystallinity and crystallographic orientation of both the barrier and ferromagnetic layers, play essential roles in the MR ratio. The presence of disorders, such as surface roughness, interface inter-diffusion, and impurities, as well as defects such as grain boundaries, stacking faults, and vacancies, would significantly affect the spin polarization and tunneling conductance.
Irradiation is a source of defects in MTJs and potentially affects the MR effects. Various types of ionizing radiation, such as α-particles, β-particles, and high-energy ions, as well as non-ionizing radiation, including neutrons, electromagnetic radiation such as γ-rays and X-rays, and thermal radiation, could degrade MR performance if any microstructures of MTJs are modified.
The radiation tolerance of AMR and GMR sensors has been experimentally investigated [135][136][137]. It was found that these sensors are generally somewhat resistant to radiation. The radiation tolerance of MTJ devices has also been experimentally studied. It was believed that polarization of the conduction currents and MR ratios of MTJs would be reduced if the interfaces between the tunneling oxide barrier and the ferromagnetic layers were damaged by radiation, which results in spin scattering defects [138]. Any permanent damage to the oxide barrier, usually caused by high-energy radiation, would cause leakage paths and reduce the tunneling resistance of MTJs. Low-energy radiation would cause cumulative degradation of MTJs.
A recent review analyzed the effects of radiation on Al 2 O 3 -based MTJs [139]. Highenergy heavy-ion irradiation usually caused the most displacement damage in this kind of MTJ, leading to a deterioration of magnetotransport properties with increasing radiation dose. High-energy protons and γ-ray radiation have minimal effects on the magnetic properties of AlO-based MTJs, suggesting that AlO-based MTJs may be promising candidates for radiation applications.
Compared with the oxide barriers in AlO-based MTJs, the MgO barriers in MgO-based MTJs are thinner, usually 1-2 nm thick. The thinner crystalline layers would be more sensitive to irradiation, as observed in other two-dimensional materials [112,140], affecting the performance of MgO-based MTJs significantly.
Here, we review the literature on the effects of irradiation on MgO-based MTJs, summarize the published experimental data, and evaluate the resulting irradiation effects. This review will highlight the state-of-the-art findings of the effects of electromagnetic radiation on MTJs with MgO barriers.

Effects of Cosmic Radiation
Primary cosmic rays and secondary high-energy cosmic rays include high-energy protons, alpha particles, nuclei, electrons, and various electromagnetic waves. Cosmic rays can be classified into four types: heavy ions, mid-mass subatomic particles (proton and neutron), light-mass subatomic particles (electron), and massless electromagnetic waves. The effects of the first three kinds of cosmic radiation are briefly reviewed in this section. The effects of electromagnetic radiation will be reviewed in the subsequent section.
At sea level, the average annual cosmic ray dose is about 0.27 mSv (27 mrem). The radiation dose is about 0.10 µSv/h. The cosmic radiation dose increases rapidly with increasing altitude, reaching about 2.0 µSv/h at 9 km altitude and about 9 µSv/h at 18 km above the Earth's surface. Therefore, it is necessary to examine the effect of cosmic radi-ation on TMR-based MTJs utilized in daily life, especially in spacecraft and satellites. It is generally accepted that the high-energy particle radiation, such as high-energy ions, neutrons, protons, and electrons, can degrade the performance of MgO-based MTJ devices.

High-Energy Heavy-Ion Irradiation
Insulating oxide barriers can be degraded by heavy-ion radiation. It was reported that ultra-thin aluminum oxide tunnel barriers were damaged by highly-charged ions (such as Xe ions with 19-42 keV) [141,141]). The conductance of AlO-based MTJs linearly increases with radiation flux [113]. Furthermore, high-energy light ions (such as carbon and oxygen ions) and heavy ions (such as nickel ions) within 10 MeV decreased the MR ratio of AlO-based MTJs irreversibly as the ion flux increased [142].
MTJs' MgO dielectric barriers are susceptible to radiation too. Typically, ionizing radiation usually generates charge trap centers in MgO barriers and the interfaces between MgO barriers and ferromagnetic layers. The produced charge trap centers can lead to extra noise of MTJs [143] and reduce the MR ratios of MTJs [144] by perturbing tunneling processes.
MTJ ferromagnetic materials are also susceptible to radiation. It has been well known for decades that ion radiation can damage the crystallographic structures of MTJ ferromagnetic layer materials and change their physical properties [145]. Generally, ion radiation would cause displacement damage, which affects the microstructure and properties linked to displacement damage [139]. It was reported that high-energy argon ions, with energies of 44 MeV, and krypton ions, with energies of 35 MeV, created amorphous zones [146] or defects [147] in BaFe 12 O 19 magnetic materials, changing their magnetic properties and microstructures. High-energy helium ion radiation can create He nanobubbles at ion implantation regions [148] and induce up to a 36% change in the crystal anisotropy [149] of ferroelectric LiNbO 3 materials.
The radiation-induced damage of oxide barrier materials and ferromagnetic layer materials would affect the behavior of MTJ devices. It was reported that CoFeB/MgO/CoFeB MTJs were degraded by high-energy oxygen ion (O − ) radiation during RF sputtering [150]. Table 6 lists some ion irradiation effects on MgO-based MTJs. It is generally accepted that high-energy irradiation usually degrades the TMR behavior of MgO-based MTJs. CoFeB/MgO/CoFeB $ 60 Co, γ-ray, 247-475 Mrad, 220 rad/s, room temperature magnetism was destroyed [120] CoFeB/MgO/CoFeB neutron, 0.1 eV-10 MeV, 5 × 10 10 particles /cm 2 /s, 2.9 × 10 15 particles cm 2 insensitive [152] Numbers in parentheses are nominal thicknesses in nm. † Ta (5) It was also reported that MgO-based MTJ devices exhibit radiation tolerance. NASA conducted a test of an MTJ-based MRAM (MR2A16A from Freescale Semiconductor Inc.) under a heavy ion single event [111]. The tested MRAM was exposed to 3.0 GeV Kr ions, 1.6 GeV Ar ions, and 3.2 GeV Xe ions. The test results indicated that the MRAM device was sensitive to single-event latchup (SEL), which was attributed to the complementary metaloxide-semiconductor (CMOS) portion of the device. However, there was no indication that MTJ elements were damaged from heavy ions.
Radiation tolerance of MTJ devices was also reported by other research groups. Kobayashi et al. exposed CoFeB/MgO/CoFeB MTJs to high-energy Si ion radiation with energies of 15 MeV [114]. The MTJs (consisting of Mg(1.3 nm) and CoFeB (1.5 nm) were sandwiched between 200 µm additional electrodes. Only minimal degradation (∼1%) was observed in their electrical resistance. However, no significant changes were detected in the retention states before and after the irradiation.

High-Energy Neutron Irradiation
High-energy neutron irradiation usually alters atomic arrangements and damages crystalline structures of many materials. The radiation can also create nanoscale amorphous regions within crystal lattices [154]. While metals are relatively immune to ionizing radiation due to their ionic bonds, fast neutrons can still enter metals and cause significant structural damage. For instance, neutron-radiation-induced defect clusters and cavities in copper [155], decreased magnetic remanences of NdFeB permanent magnets [115], and changed the Curie temperature of FeNiCrMoSiB amorphous alloys [156].
High-energy neutron irradiation can also damage the ferromagnetic layers of MgObased MTJs. High-energy neutrons can travel in the crystalline lattice of free/fixed layers and displace these atoms from their initial lattice positions through kinetic energy transfer. These kind of displaced atoms are termed primary knock-on atoms (PKAs). The PKAs can continuously displace other lattice atoms that are named secondary knock-on atoms (SKAs). This series of displacements can generate numerous defects in the crystalline free/fixed layers, ultimately affecting the performance of TMJs. Table 6 lists one case of neutron irradiation, which is generally understood to degrade MTJ devices.

High-Energy Electron Irradiation
High-energy electron irradiation affects MTJ component materials. In one study, amorphous CoFeB thin films (which are used as free/fixed layers of TMJs) were exposed to an electron beam with an energy of 200 keV in a transmission electron microscope [116]. The electron radiation modified the phase and microstructure of the films. Another study examined the thermoluminescent properties of ultrafine MgO particles, with sizes of 250-500 nm, under high-dose electron radiation [157]. A pulsed electron beam with an energy of 130 keV was employed at room temperature, with a pulse duration of 2 ns and current density of 60 A/cm 2 . The absorbed dose was 1.5 kGy/pulse. Figure 8 shows the dose-dependent thermoluminescent (TL) intensity of the electron-irradiated MgO nanomaterials. Clearly, MgO's structure should be modified by the electron radiation.
Unfortunately, there have been few studies on the irradiation effects of high-energy electrons on MgO-based MTJs. Metal layers are usually deposited over ferromagnetic layers of MTJs, preventing electrons from penetrating into MgO barriers and magnetic layers of MTJs. Therefore, high-energy electrons should not affect MTJs due to the screening effect of metal layers. efore, the observed absence of correlae and temperature ( Fig. 4) testifies in eling mechanism. Earlier Kortov et al., s TL mechanism in thermochemically ls. t the horizontal parts of glow curves combination are observed in the therh were irradiated by high doses only. In L glow curves do not contain horizontal mally treated in identical conditions and Gy of beta source ( 90 Sr/ 90 Y). In this case perposition of normal general order TL appearance of tunneling recombination irradiation (more than 1 kGy) is due to f trapping and recombination centers the distance between them. It is known en vacancies can be created after a high particular in a-Al 2 O 3 (Kortov et al., 2015, that such processes take place in MgO ell. An additional factor contributing to form of MgO is a high concentration of rs, which is observed in materials with centers, in particular, can serve as such and Uchino, 2011) discussed the relag and vacancy complexes in MgO. They lly assisted tunneling in PL of MgO at ue to creation of weakly coupled F-type hallow traps. In our case this effect is We can assume that F 2 -type centers can . It is known that the excited levels of O 3 usually lie deeper in band gap in ancies (Evans et al., 1994). It is possible previously (Uenaka and Uchino, 2011) ectron traps in MgO too, causing TL at lt of our paper shows that the tunneling gO luminescence associated with not ith deep trapping centers as well. The nesium oxide require additional study.
nces of TL intensity in peak A at 110 С for assessment of possibility of using MgO samples in high dose dosimetry. The samples were irradiated by an electron beam. The initial compacts which are not thermally treated along with low TL sensitivity have strongly non-linear dose response. It shows their unsuitability for practical application in TL dosimetry. TL dose response of the samples treated at 1400 С is presented in Fig. 6. This dependence is sublinear because the TL intensity is proportional to D 0.49 , where D is the absorbed dose in the range of 1.5e80 kGy. The obtained results show the potential use of ultrafine MgO ceramics thermally treated in reducing conditions for high dose dosimetry of pulse electron radiation.

Conclusions
The ceramics on the base of nanostructured MgO powder are synthesized in strongly reducing conditions. It is found that high temperature treatment of the material under study in vacuum in the presence of carbon leads to growth in PCL and PL intensity at 2.0e3.5 eV. The TL output in the peak at 110 С also increases. It is established that these changes are associated with the concentration growth of anion defect F-type centers. We hypothesized that the PL band at 2.9 eV is due to different types of defects associated with oxygen vacancies. It is shown that TL at Т ¼ 250e400 С of synthesized ceramics irradiated by high doses may be due to tunneling transitions of charge carriers from deep trapping centers. The aggregate F 2 -type centers can act as deep traps. The TL dose response of thermally treated samples is sublinear in the dose range of 1.5e80 kGy after irradiation by an electron beam (130 keV). It indicates the possible future use of ultrafine MgO ceramics in high dose dosimetry of ionizing radiation.

Effects of γ-ray Irradiation
γ-rays are a kind of electromagnetic radiation with wavelengths ranging from 3 × 10 −13 m to 3 × 10 −11 m (approximately 40 keV to 4.0 MeV), and being ionizing radiation. The electromagnetic wave can penetrate materials deeply and interact with matter through three kinds of primary processes: the photoelectric effect, Compton scattering, and electronpositron pair production, depending on the energy of the incident γ-ray. When the energy of the γ-ray is higher than 1.02 MeV, it may spontaneously produce an electron and positron pair. Compton scattering is the principal mechanism when the energy of γ-ray is between 40 keV and 4.0 MeV. The photoelectric effect dominates when the energy of the γ-ray is below 50 keV, whereby an electron absorbs the incident γ-ray and is excited to conduction bands. In all three kinds of processes, the γ-ray collides inelastically with electrons, losing energy and continuously moving with a longer wavelength. Furthermore, γ-rays can directly ionize atoms through the photoelectric effect and the Compton effect and indirectly through secondary ionization. These processes occur when MTJs are exposed to γ-rays.
Depending on the γ-ray's energy and the properties of the MTJ materials, a γ-ray can induce displacements of atoms within the lattice, termed defects. These defects can remain for a long time at room temperature and can be investigated through the Hall effect and electrical measurements. This kind of radiation-produced defect would affect the performance of MTJs. In fact, most studies on MTJ degradation were performed under this kind of radiation interaction.
In contrast to the above interactions, γ-rays may only disturb the atoms of MTJ materials temporarily or transiently. The produced disturbances of atoms may disappear shortly once the γ-ray is removed. This kind of radiation-induced degradation can be only detected in-situ in real-time, while under irradiation.
Experimental investigations have supported these two kinds of γ-ray interactions. Several groups have reported that MgO-based MTJs are highly tolerant of γ-ray radiation up to a dose of 10 Mrad [152,158]. In their work, MTJs were irradiated and then measured ex situ. Their results indicated that γ-ray irradiation did not noticeably change the TMR ratio, coercivity, or magnetostatic coupling of low-frequency noise. As such, MgO-based MTJ devices are expected to operate reliably in a γ-ray irradiation environment, especially at doses below a few hundred Rad [159]. Other scientists hold the view that γ-ray radiation should degrade MgO-based MTJ devices, because γ-rays have been shown to change the microstructure of MTJ materials [160]. Additionally, others suggested that γ-ray radiation may affect the peripheral circuits of MgO-based MTJ devices (not the MTJs themselves) during the read/write operation, leading to soft errors [161]. Table 7 lists some results of γ-ray irradiation. In the following subsections, these published data will be analyzed in detail with respect to the MTJ structures and experimental conditions, including the conditions of γ-ray irradiation and the measurement methods. First, the physical properties of the γ-ray-irradiated MTJ material, including MgO crystals (used as barriers in MTJs), fixed layers and free layers, and MgO/ferromagnetic layer interfaces, will be reviewed. Next, the review will focus on the physical properties of γ-ray-irradiated MgO-based MTJs. Finally, the tolerance ability of MgO-based MTJs will be discussed from γ-ray penetration in MTJs and MTJ devices, to explore potential explanations of MgO-based MTJs' radiation tolerance. CoFeB films γ-ray, 1.2 MeV, 2.58 × 10 5 C/kg, 60 • C sensitive to γ-ray irradiation [118] MgO/CoFeB § γ-ray, 100 kRad no noticeable change in magnetic properties [166] CoFeB/MgO/CoFeB 60 Co, γ-ray, 1 Mrad no effect [119] CoFeB/MgO/CoFeB ¶ 60 Co, γ-ray, 10 Mrad, 9.78 rad/min highly tolerant of γ-ray radiation [152] CoFeB/MgO/CoFeB ‡ 60 Co, γ-ray, below 20 Mrad, 220 rad/s, RT coercivity increased with irradiation while saturation magnetization was not affected [120] Numbers in parentheses are nominal thicknesses in nm.

MTJ Materials under γ-ray Irradiation
MgO-based MTJs consist of MgO barriers, ferromagnetic free/fixed layers, and metal electrodes. The performance of MTJs is influenced by their microstructures, physical properties, and the interfaces of these materials. Therefore, the characteristics of these MTJ materials with respect to γ-ray irradiation are discussed first.

MgO Crystals under γ-ray Irradiation
There is a limited amount of literature available on the effects of γ-ray damage to MgO barriers with nanometer thickness [139]. To ensure adequate information on γ-ray-irradiated MgO materials, the properties of γ-ray-irradiated MgO bulk and thick films are reviewed here. It is expected that the properties of γ-ray-irradiated MgO barriers of nanometer thickness will exhibit similar behavior to that observed in MgO bulk and thick films.
The properties of irradiated MgO have been investigated since the 1960s to explore the potential of MgO for γ-ray dosimetry by studying its response to γ-ray irradiation. MgO crystals were cleaved from ingots and exposed to a 60 Co source with a radiation intensity of 3.0 × 10 6 R/h [162]. The thermoluminescence indicated γ-ray irradiation-induced defects in the MgO crystals.
MgO powder has also recently been irradiated by γ-rays. Kiesh et al. exposed commercial MgO powder to 60 Co γ-ray radiation with a total dose of 20 Mrads [164]. The irradiated powder was then kept at room temperature for over a year before measurements. The results showed that γ-ray irradiation caused a shift in the powder's thermoluminescence peaks. In another study, MgO powder with a purity of 99.9% was exposed to γ-ray radiation, using a 60 Co source with a dose rate of 8.33 mGy/s [165]. Figure 9a shows the thermoluminescence (TL) of the γ-ray-irradiated MgO powder. Low-dose γ-ray irradiation induced a peak around 280 • C, while a high γ-ray radiation dose (above 300 Gy) resulted in a peak at 150 • C, which became dominant after exposure to a dose above 1 kGy. It was believed that the radiation dose affected the recombination centers and caused the shift in the TL peaks. Figure 9b shows the relationship between γ-ray dose and TL response integrated across the entire TL curve over the dose range. The TL response changed linearly with radiation dose at intermediate dose levels of 1-100 Gy, and sub-linearly at higher dose levels of 0.5-50 kGy. of 1 Gy-200 kGy ( Figure 3). The measured TL features are as follows: one finds a strictly linear region of constant TL efficiency from the lowest dose levels measurable of ∼1 up to 100 Gy (I = 7.3228D + 6406.9), where I represents TL intensity and D is the γ dose in Gy followed by a sublinear region (I = 126.98D 0.67 ). The error in TL measurements is ±3% and the combined uncertainty of absorbed dose is 1 and 5% in the linear and sublinear parts, respectively. The linear region of dose response at low doses is believed to arise from recombination between locally trapped electron-hole pairs in spatially correlated trapping centers and luminescent centers (36). A careful observation of glow curves beyond an applied dose of 100 Gy (Figure 2) reveals the presence of a small shoulder located at 150 • C in addition to the main TL peak at 280 • C. With a further increase in radiation dose, the 150 • C peak predominates the glow curve with the gradual disappearance of the 280 • C peak. The interpretation of this observation is that the traps responsible for the TL peak at 150 • C compete with the traps responsible for the 280 • C peak which leads to sublinear dose response.

UV irradiation
In addition to the use of the phototransfer thermoluminescence (PTTL) for dose reassessment in dosimetry and dating, the study of PTTL is one of the sources of information in understanding the luminescence processes in the material of interest. A selection of PTTL glow curves obtained from 0.05 to 10 Gy γ -irradiated MgO is shown in Figure 4. After depletion of the TL signal (heating at 400 • C for 1 h), the aliquots were given an energy of 4.88 eV from an UV lamp with wavelength 254 nm. The glow curves obtained after the illumination contain a peak near 150 • C in addition to the main dosimetry peak located near 280 • C (Figure 4). When the γ dose increased to 10 Gy, an additional high temperature peak appeared at 305 • C. Therefore, there should be competition between the increase of the charge in the dosimetry trap due to phototransfer from deep traps and photostimulated decay of charges from the dosimetry trap. The dose dependence of the PTTL (b)  [121,167]. The grain size of the MgO particles was 0.5-1.0 • C. These MgO capacitors were exposed to γ-ray radiation with a maximum dose of 32.55 mGy and an energy of 0.662 MeV. Figure 10 shows the real-time capacitance of the MgO capacitors as a function of γ-ray radiation dose. The capacitance increased continuously with γ-ray dose, being reversible and less susceptible to γ-ray radiation. γ-rays damaged the MgO particles and produced structural defects (such as color centers or oxygen vacancies) in MgO, changing the density of charge carriers in the MgO films.
Steinike et al. exposed mechanically cleaved MgO samples to γ-rays emitted from a 60 Co source [168]. The irradiation was carried out at a rate of 3.4-4.5 MRad/h and energy 1.25 MeV at −196 • C. γ-ray irradiation generated F + centers and V − centers in the MgO crystals. The concentrations of F + and V − centers increased linearly with radiation doses up to 1-3 MRad, followed by saturation at higher doses. Additionally, the concentration of the F + defect centers decreased with increasing annealing temperature and the F + centers could be removed by annealing at 600 • C.
Clement et al. studied the absorption and luminescence spectra of MgO crystals under γ-ray irradiation in-situ [169]. MgO crystals with 99.99% purity were exposed to γ-rays at a flux of 3.5 × 10 4 rad/h for 7 h at 20 • C and 120 • C, in a vacuum of less than 2 × 10 −6 Torr. The real-time absorption increased with increasing radiation dose at both temperatures. It was also reported that subsequent annealing at 600 • C canceled the irradiation effect.
Based on the results, it was concluded that impurities of Fe (less than 300 ppm) and Cr (less than 100 ppm) played a significant role in the degradation caused by irradiation.
dose of a 32.55 mGy. Moreover, the response of MgO sensor was monotonic over the whole range, which is preferable for dosimetry applications.

Circuit
The circuit design required a system capable of measuring capacitances changes with high precision. For this the AD7746 24-bit capacitance to digital converter was chosen for its inherent high-resolution architecture. The AD7746 allows 19-bit effective resolution at a 16.6 Hz data rate, high linearity (± 01%) and high accuracy (±2 fF factory calibrated). The AD7746 capacitance input range is ± 4 pF (changing), while it can accept up to 17 pF absolute capacitance (not changing), which is compensated by an on-chip digital to capacitance converter (CAPDAC).
simultaneously. Figure 6 show the AD7746. In Figure 7 a block diagr microprocessor is presented. required from this tightly pac I 2 C bus, interrupt line input p port p3.4 to show that the Although this processor p requirements of this applicatio regards to I 2 C routines and RS it a suitable candidate for the in Abramishvili et al. studied γ-ray-irradiated MgO crystals too [163]. The total impurity content in the crystals did not exceed 245 ppm. The irradiation was carried out at room temperature, with a dose of 1.6 × 10 6 rd/h and a maximum γ-ray energy of 2.1 MeV. in-situ measurements were performed at low temperatures. It was observed that the irradiation significantly changed the thermal conductivity of the MgO crystals, as shown in Figure 11a. The thermal conductivity was partially reversed by annealing the irradiated crystals at 515 • C for one hour, which led to the recovery of the crystals' heat conductivity to their initial state. The observed reversal is consistent with other reports [169]. Additionally, their optical absorption was changed after irradiation, as shown in Figure 11b. Upon further analysis, it was believed that γ-ray radiation caused the formation of Frenkel pair defects, which changed both the thermal conductivity and optical absorption. Frenkel pair defects can be eliminated through annealing, which leads to the restoration of the original thermal conductivity.
Kvatchadze et al. measured, ex-situ, the thermo-stimulated luminescence (TSL) of nominally pure MgO single crystals containing minor impurities (Cr 3+ : 12-26 ppm; Mn 2+ : 35-72 ppm; V 2+ : 24-60 ppm; and OH − : 0-4.9 ×10 17 cm −3 ) under γ-ray irradiation (0.8 Gy/s and 1.25 MeV) over a temperature range of 300 K to 775 K [160]. It was reported that in MgO crystals containing OH − impurities, the TSL intensity steadily increased with increasing γ-ray radiation dose at 450 K, as shown in Figure 12a. Additionally, the TSL intensity at 450 K increased linearly with the γ-ray radiation dose (Figure 12b). However, in MgO crystals without OH − impurities, the TSL intensity at 450 K showed extremely low sensitivity to γ-ray irradiation (Figure 12b). It was proposed that foreign hydroxyl ions trapped charges in γ-ray-irradiated MgO crystals, inducing the accumulation of hole centers to change the optical properties.
Lynch et al. investigated the photoconductivities of MgO polycrystalline bulks under γ-ray irradiation fields in-situ, over a temperature range of 300 K to 600 K [170], as shown in Figure 13. The γ-rays were emitted from a 60 Co source, with an energy of 1.17 MeV or 1.33 MeV. It was reported that the photoconductivity of the MgO bulks increased linearly with γ-ray radiation dose. The γ-ray-induced conductivity showed a linear dependence on the radiation dose rate up to 4.0 × 10 5 rad/h. Additionally, the photoconductance of the MgO bulks increased by about three orders of magnitude when exposed to γ-ray radiation with a flux of (2.9-3.7) × 10 5 rad/h. The studies described above demonstrate the sensitivity of MgO materials (single crystals, polycrystalline bulks, mechanically exfoliated layers, and powder) to γ-ray radiation. in-situ work indicated that radiation-induced defects could be restored to their initial pre-irradiated states , especially after being annealed at high temperatures. Fig. 1 shows the plots of heat conduction versus tempera before (curve 1) and after (curves 2 to 4) irradiation (curves 5 to 7). Irradiation changes strongly the heat co but x ( T ) of crystals irradiated with different doses of y coincided in the studied temperature range.

Measurements
Isochronous annealin crystals at I',,,, < 350 conduction at all (cu perature annealing (5 x(T) was observednealing (1325 ") led to so in comparison with x crystal (curve 7). Heat conduction of irradiated MgO crystals (Fig. 1)) unlike th sorption bands 212 and 288 nm (Fig. 3) does not depend upon th iscussion -irradiated n.p. MgO(2) crystals are ver the whole temperature range tensive TSL peaks are observed in ith samples irradiated in reactor [1, nd out: one low-temperature (~ 450 mperature (> 550 K). Intensity of teadily goes up with radiation dose irradiated MgO(3) and MgO(1) tively the same as TSL curves of l, given in Fig. 1 (the number and n of peaks are meant), but differ by . It is especially true for MgO(1) doesn't show considerable rise of with radiation dose increase. ell seen on the plot of this peak ce on radiation dose (curve 1, Fig.  1) crystals TSL peak at 450 K low sensitivity to γ-irradiation. ndences for the rest n.p. crystals of 3) groups given in the same figure respectively, Fig. 2), show linear en intensity of the peak and ose with high rate of rise, which sibility of magnesium oxide dosimetric material for etectors.
l distinction of results obtained for ferent groups [MgO(1) on the one and MgO(3) on the other] must be p distinction between these groups that is displayed in Table 1: though the amount of metallic ions in MgO(1) group and in the other one are of the same order of magnitude, however the first group does not contain impurity of hydroxyl ions at all. These ions usually form hole center structures, so    The thickness of MgO crystalline films employed as barrier layers in MgO-based MTJ devices is only several nanometers. Such kinds of thin layers are expected to be more sensitive to γ-ray radiation than their bulk counterparts, as observed in other twodimensional nanolayers [112]. As a result, MgO-based MTJs should be sensitive to γ-rays if MgO layers are exposed to γ-ray radiation.
Different from free-standing MgO films, MgO barrier layers are sandwiched between ferromagnetic free/fixed layers in MTJs. The ferromagnetic layers could potentially reduce the radiation dose into the MgO barrier layers and provide some level of protection. Furthermore, defects induced by γ-ray radiation may be temporary and disappear shortly after radiation exposure or thermal annealing. Therefore, the irradiation effects on MgO barrier layers are more complex than discussed above. in-situ and real-time measurements are required to examine the effect of γ-ray radiation on MgO barrier layers. thus used to derive current values at appropriate temperatures for each applied voltage. The conductance was then determined, for each temperature, from the slope of the line joining the two points on a current vs. voltage plot. For the measurements in the Gammacell 220 the graph of In (IT) vs. 1/T was fitted to a polynomial of order five for each voltage. Current vs. voltage plots were thus produced to yield the insulator conductance at various temperatures. Figure 3 shows the results obtained from this type of analysis for the conductance of the insulation as a function of temperature before, during, and after irradiation. It can be seen that, at room temperature, the c MgO changes by about thre tude when subjected to the y The cable was then remove and the photoconductance determined at room tempe Gammacell 220 and the Ga results are as shown in Table  in the Gammacell 220 with t Also shown in this table radiation dose rates under th mental conditions used. The dose in this work is the rad an energy deposition of subjected to the radiation flu

Ionic Conduction
The migration of ions thro the influence of an electric fi presence of intrinsic lattice d structure. The simplest imp one most common in MgO, (Lidiard 1967 For personal use only.  (5) (where the numbers in parentheses denote the thickness in nanometers) [171]. Unlike MgO barrier layers with a thickness of 1-2 nanometers, the ferromagnetic layers, such as CoFeB, are thick, and the performance of the MTJs is closely related to their magnetic properties.
Wang et al. [120] investigated CoFeB/MgO's perpendicular-anisotropy magnetic tunnel junction and found that the magnetism was destroyed if the radiation dose was sufficiently high.
Shkapa et al. exposed FeCoB metallic ribbons to γ-ray radiation and examined their magnetic properties using nuclear magnetic resonance and the Móssbauer effect [118]. The (Co, Fe) 85 B 15 metal glasses were irradiated by 1.2 MeV γ-rays at 60 • C. It was reported that Co 85−x Fe x B 15 (x = 12 − 25) magnetic glasses were sensitive to γ-ray radiation, changing the atomic short-ordering of FeCoB ribbons.
Other ferromagnetic materials, such as Fe, Co, and FeCo alloys, should be similar to CoFeB materials under γ-ray irradiation. To ensure the similarity, the displacements per atom cross-section of Fe films with sizes of 100 nm × 3 µm × 12 µm were calculated using the Monte Carlo simulation method [172] under γ-rays with energies of 1.3 MeV and a source activity of 1000 Ci. The displacement cross-section was 0.1 barns. The calculation indicated that the atomic displacement rate was about 0.6/s. Furthermore, the γ-rayinduced displacement cross-sections were very low for γ-ray radiation with energy >1 MeV.
In addition to MgO barrier layers and ferromagnetic layers, non-magnetic metal films in MTJs, such as Ta and Ru layers, can protect MgO layers and ferromagnetic films from γ-ray radiation. However, the consequences of γ-ray irradiation of metal films are not the subject of this review. Although not discussed here, there is literature available on this topic [173].

Interfaces of MgO Barrier/Ferromagnetic Layers
The performance of MTJs is influenced by interfaces between the MgO barriers and ferromagnetic layers. Recent investigations have shown that CoFeB can form Co(Fe)-O bonds and bond to MgO epitaxial grains after annealing [171]. Conversion electron Móssbauer spectroscopy studies indicated that interfaces between MgO(001) and Fe(001) layers were partially oxidized over 60%, and Fe diffused into MgO barriers from both ferromagnetic interfaces [174]. It has been suggested that these interfaces may be more sensitive to γ-ray radiation, similar to Al 2 O 3 -based MTJs, whose physical properties were significantly affected by irradiation [139].
Recent in-situ experiments discovered that the uniaxial magnetic anisotropy decreased systematically with increasing annealing temperature [175]. Specifically, the MgO/FeCoB/MgO layers become isotropic after annealing at 450 • C. The asymmetry at the interfaces was explained by the diffusion of boron from the FeCoB interface layer into the adjacent MgO layer. The electronic structures of MgO/Fe interfaces have been investigated [176]. It is believed that Fe3d-O2p hybridization and distortion of the Fe film play important roles in magnetic anisotropy at the MgO/Fe interface.
Thermal annealing also affects the interfaces between MgO barriers and ferromagnetic layers. The details are discussed in the section on infrared radiation and thermal annealing.

MTJs under γ-ray Irradiation
Until this point in time, there have been two distinct viewpoints regarding the impact of γ-ray radiation on MgO-based MTJs. Some scientists believe that MgO-based MTJs are susceptible to γ-ray radiation and are likely to sustain damage as a result. Other scientists argue that MgO-based MTJs are resilient to γ-ray radiation. In the following subsections, each of these viewpoints will be reviewed in detail.

Sensitivity Results
Considering the reported properties of MgO barrier materials and the discussion on the ferromagnetic layer materials of MgO-based MTJs above, it can be inferred that MgO-based MTJs would be affected by γ-ray radiation. However, there are limited reports on the degradation of MgO-based MTJs under γ-ray irradiation. Two sensitivity cases are reviewed below. Wang  . The CoFeB/MgO p-MTJs were exposed to a Cobalt-60 γ-ray radiation source at room temperature with a dose rate of 220 rad/s. The results showed that the coercivity of the γ-ray-irradiated p-MTJs increased gradually with increasing doses of up to 20 Mrad, as shown in Figure 14. However, there was no observed variation in the saturation magnetization.
It was reported that the magnetism of MgO-based MTJs was destroyed by γ-ray radiation when the dose was sufficiently high, such as 247 Mrad [120]. It was hypothesized that the destruction of magnetism was caused by radiation-induced thermal stress. Figure 15 shows the surfaces of the MTJs after γ-ray irradiation, the observed effects were caused by differences in the thermal expansion coefficients between the MTJ films and the substrate. n/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.    This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. (TMR) ratio, lower resistance-area product and better thermal endurance [13]. Thus double-interface CoFeB/MgO p-MTJs are good candidates for constructing the high-density MRAM and magnetic very-large-scale integrated circuit (VLSI) [14], [15]. Considering the future space application, it is of great importance to evaluate the radiation tolerance of doubleinterface CoFeB/MgO p-MTJs.
In deep-space missions, such as Jupiter, Saturn and Mars explorations, the radiation environment includes galactic cosmic ray (GCR) and solar protons. Especially, the particle energy of Jupiter is ten times that of the Earth's radiation belt as Jupiter has the strongest magnetic field [16]. Fortunately, the TID can be partially mitigated with shielding [17]. In this paper, we study the effects of the gamma irradiation on the double-interface CoFeB/MgO multi-films at various TIDs. It is worth noting that our experimental results presented in this work do not apply to the single-interface MTJ, which is mainly attributed to the differences in structure between our double-interface film stacks and previous single-interface devices.
The rest of this paper is organized as follows. In section II, we will briefly introduce the structure of our double-interface CoFeB/MgO multi-films and the experimental procedure. Then gamma irradiation results and discussions are demonstrated in section III. Finally, this work is summarized in section IV.

A. Sample Description
The schematic structure of the double-interface CoFeB/MgO multi-films is shown as Fig.1 The RL includes CoFeB, W bridge layer and [Co/Pt]/Ru/Co multilayers as synthetic antiferromagnetic (SAF) layers. The SAF layers show a robust perpendicular magnetic anisotropy (PMA), which is coupled with the W bridging layer [18]. Thus the magnetization of the RL is pinned. W was used for both cap and bridge layers in order to improve significantly TMR and E, thanks to the induced scattering state in the CoFeB/W interface. W thickness was determined through a precise time control [13]. Moreover, replacing Ta with W suppresses the degradation of TMR during annealing since W is more resistant to high temperature. Top Ru/Ta layers were used as a protective cap to avoid oxidation. The multi-films were subjected to vacuum annealing from 350 °C to 430 °C for an hour. Fig. 2 illustrates typical M-H hysteresis loops of our multifilms in the out-of-plane (easy axis) and in-plane magnetic (hard axis) fields. These results were measured by a physical properties measurement system-vibrating sample magnetometer (PPMS-VSM) with a magnetic field step of 1 Oe. There are two flips in the out-of-plane magnetic field for estimating high and low coercivity (Hc). We are concerned with the lower one since it determines the switching performance of the FL.

B. Experimental Procedure
The TID experiments were designed to test effects of the gamma irradiation on the primary magnetic properties of our double-interface CoFeB/MgO multi-films. Five samples, 3 mm× 3 mm, were exposed to irradiation on a Cobalt-60 gamma facility at the room temperature. The samples were irradiated in a Pb/Al lined container to avoid dose enhancement effects caused by low energy scattered radiation [19]. The dose rate was 220 rad (Si)/s. Three of the samples named #1, #2 and #3 were exposed with TIDs of 5 Mrad (Si), 10 Mrad (Si) and 20 Mrad (Si), respectively. The maximum TID was twice that of [8], [9]. The other two samples, named #4 and #5, were exposed at the same dose rate in one dose step to approximately 247 and 475 Mrad(Si), respectively. We chose large TIDs to simulate the worst case in the deep-space missions [16].
Before irradiation, the samples were measured by VSM in a 5000 Oe out-of-plane magnetic field to characterize Ms and Hc   measured with the range from -5000 Oe to 5000 Oe before and after irradiation. Table I illustrates numerical variations of low Hc extracted from Fig. 3. It can be seen that Hc shows an increasing trend as the cumulative dose increases. On the other hand, no variation is observed from the distribution of Ms shown in Fig. 4, regardless of TIDs. The measurement of #3 sample with in-plane magnetic field from -8000 Oe to 8000 Oe was also performed, and corresponding results are shown in Fig. 5. The discrepancy between two hysteresis loops means the variation of magnetic anisotropy field (Hk). This can be explained by the fact that Hc is proportional to Hk [20]. In addition, the comparison of the results between Fig. 3 and Fig. 4 demonstrates that Hc and Ms are independent of each other at various TIDs. In other words, the Hc is mainly determined by the quality of CoFeB/MgO interface, while the Ms depends on the property of FM bulk [21]. Therefore, the above results indicate that CoFeB/MgO interfaces are more sensitive to the gamma irradiation, but the bulk FM is hardly influenced by the gamma irradiation when the TID is less than 20 Mrad (Si).
The measured M-H hysteresis loops of #4 and #5 samples are completely different from those of #1~#3, as shown in Fig. 6.
Compared with the pre-rad value of 2.5  10 -5 emu, the saturation magnetizations of #4 and #5 samples after irradiation are approximately 3.5  10 -6 emu and 2.5  10 -6 emu, respectively. The observed degradation is up to 90%. In addition, Hc is almost zero. These results demonstrate that the magnetism of film stacks is destroyed by the excessive TID.
Generally, Hc is a result of the competition among various factors, such as Hk, domain structure and defects of the FL [20]. For the TIDs ranging from 5 to 20 Mrad (Si), the change of Hc is attributed to the combination of various factors as follows.
First, the magnetic Compton scattering (MCS) induced by gamma ray gives rise to the energy transfer between the photons and electrons. Those excited electrons lead to the increase of high-energy free spin electrons. The function is described by the Compton profiles (Jmag (pz)) as (2): free spin electrons interact with the lattice and interface such that the anisotropy is changed [23]. Especially since double-interface CoFeB/W stack leads to a scattering state, the structure of multifilms is more vulnerable to MCS than the single-interface p-MTJ. Second, as TIDs are increased from 5 to 20 Mrad (Si), Ru layer between two (Co/Pt)n multilayers may be also destroyed by the gamma irradiation, leading to the change of SAF coupling. Thus the stray field of RL cannot be offset by SAF coupling [24]. As a result, the exchange bias field (Hex) was changed after irradiation (see TABLE I).
Third, the double-interface structure increases the effective thickness of the FL, which is influenced by gamma irradiation. Thus the Gilbert damping constant is changed [25], [26].
Finally, ionizing energy deposition induced by the gamma irradiation can prevent the domain wall motion [27] and oppose the switching direction of the magnetic moment, which leads to an increasing Hc.
The almost disappearance of magnetism under 247 or 475 Mrad (Si) irradiation may be attributed to the thermal stress and internal stress [28], [29]. The film thermal stress is a result of the differences of the thermal expansion coefficients between multifilms and the substrate. In addition, ionizing energy deposition or uneven distribution of temperature, which are induced by the gamma irradiation, can also result in the thermal stress influencing the magnetic properties [30], [31]. Thus radiationinduced thermal stress and internal stress may destroy the magnetism. The microscope images shown in Fig. 7 indicate that the multi-films suffer from micron-sized cracks induced by the thermal stress and internal stress.
The above-mentioned competing factors explain the TID effects of the gamma irradiation on the double-interface CoFeB/ MgO multi-films.

IV. CONCLUSION
In this work, we firstly performed the gamma irradiation experiments of the double-interface CoFeB/MgO multi-films in a series of TIDs from 5 Mrad (Si) to 20 Mrad (Si) and even up

Tolerance Results
Numerous research groups have reported a high tolerance of MgO-based MTJs to γ-ray radiation, with no observed impacts on the magnetic or electrical properties of the MgO-based MTJs.
Nguyen et al. exposed a bare MTJ to γ-ray radiation (1.25 MeV) for a total ionizing dose of 100 kRad [166]. The MTJ consisted of Ru (7) . Ex-situ measurements revealed that the dose of γ-ray radiation did not cause any noticeable changes in the magnetic properties of the MTJ, as shown in Figure 16. The MTJ exhibited no noticeable changes in either coercivity or magnetostatic coupling. the MTJs, as discussed above, although it is a matter of a circular area with diam of micrometres. The vibration test for surviving a launch into space would not be meaningf they are solid state sensor elements. Therefore, a sensor system was tested ins sensor elements were connected to the electronic circuit by wire bonding. A diameter of 25 micrometres were used for this. Some wires were left as bond two-component epoxy resin, Fig. 9 right. Not even the two additional tests of T damage on the system. A final, full-fledged version of the sensor system for sp employ so called flip-chip mounting, which would eliminate the need for bond w proven not sensitive to the vibration, but maybe to the manual handling, an electromagnetic compatibility (EMC) problems. For additional informatio magnetometer was capable to resolve a magnetic signal of about a 100 pT at MH Fig. 9. Left: The magnetometer with three orthogonal sensor chips integrated onto the circuit for 3-D magnetic field measurement. Right: Some bond wires were, and others

RESULTS AND DISCUSSION
The physical etching process in the ESCA system, especially when resulting in a tilted sidewall of the MTJs, gave rise to a question concerning re-deposition of the materials on the sidewalls possibly causing short-circuiting of the electrodes. It was also of importance to confirm the precision of the etch stop at the barrier. The TEM image to the left in Fig. 3 is taken from a Series B MTJ and shows a slope of the MTJ sidewall going from the left to the right. The right image, taken safely away from the slope, shows that the etch process did stop at the MgO barrier. As also seen in the left image, and in a magnified part of the slope, there are signs of material re-deposition. Energy Dispersive Spectroscopy (EDS) could be used to analyze what materials were re-deposited. Awaiting this, a discussion on the effect of the re-deposited materials on the MTJ performance is difficult to carry out. However, analysis of the MTJ performance, presented later in this section, will shed some light on this matter.

Fig. 3. TEM images of a sample from Series B.
To study the relation between resistance and area of the MTJs, samples from experiment Series A were used, and TMR loops of the circular MTJs of different diameters were acquired. Figure 4 shows that the resistance of the MTJs decreased with the increase of their area. For the TMR ratio, no clear dependency on the areas could be confirmed. This suggests an advantage of using small MTJs.

Fig. 4. TMR loops of circular MTJs of different size (left), and their TMR ratio as a function of area (right).
On the other hand, the MTJs with larger areas (also from Series A) exhibited lower low-frequency noise, Fig. 5. The striving for high resistance and low noise of the MTJs was somewhat complicated by the fact that the inherent noise of the signal amplifier set the noise floor for the system. A high sensor noise would of course drown the weak signal to be measured, but if the sensor noise was lower than that of the amplifier, the tiny signal would be at least partly lost between the two noise levels. Further discussion on noise optimization for the system would go beyond the scope of this paper. However, Figure 16. TMR of a single MgO-based MTJ before and after irradiation. Inset: Cross-sectional TEM image [166]. Copyright 2010, International Training Institute for Materials Science. Reproduced with permission [158]. Copyright 2011, IOP Publishing Ltd.

Ren et al. investigated
MgO-based MTJs exposed to γ-ray radiation [152]. The MTJs had a full structure of Si/Ru(6)/IrMn(11)/CoFeB(6)/MgO(1.4)/CoFeB(5) (numbers in parentheses are thickness in nanometers), as shown in the inset of Figure 17a. The tunnel barrier was made of MgO with (001) crystalline orientation. The junction was exposed to 60 Co γ-ray radiation at a dose rate of 9.78 rad/min. Figure 17a shows the hysteresis loop of a single MTJ before and after exposure to the γ-ray radiation. A 10 MRad dose of radiation had a very weak effect on the electrical resistance. Figure 17b shows the coercive field H c and TMR of other individual MTJs with the same structure that were tested under the same irradiation. The measured coercive field H c and TMR were almost the same before and after γ-ray irradiation. Neither the electrical nor the magnetic properties of the MTJs were affected by the radiation. Therefore, the study concluded that MgO-based MTJs were highly tolerant of γ-ray radiation with a dose of 10 MRad at 1.25 MeV.
Hughes et al. exposed MgO-based MTJs to Co 60 γ-ray radiation with a dose of up to 1 Mrad (Si) [119]. It was reported that γ-ray irradiation did not affect the state retention and switching characteristics of MgO-based MTJs.
Most of the experimental measurements discussed above were carried out after γ-ray radiation exposure, and it remains unclear whether the after-exposure status was equivalent to the exposure status. Nonetheless, it can be inferred that MgO-based MTJs are capable of retaining their non-irradiated initial status after γ-ray irradiation.
In addition to experimental studies, some theoretical research has been reported in support of the radiation tolerance of MgO-based MTJs. For instance, Kang  st. mal cycles. Then the were compared to the radiation had any efmagnetoresistive hyses. Four contact pads allowed four-terminal conductor Device Annce of TMR in all the ts of temperature varie characterized before n in Gammacell 220, was a constant 9.78 lly received a dose of characterized. Irradiaose of 10 Mrad, which ally induces failure of e re-measured electrinducted at the Oregon rch reactor. The extrated in Fig. 4. The erized in situ in a cadal epithermal neutron ined, with the flux at energies ranging from SSIONS s of a single MTJ beiation. The hysteresis   loop and voltage bias dependence of resistance are shown respectively. TMR could be calculated as the resistance difference between the two states normalized by the TMR low resistance. The measured coercive field and TMR of the complete series of experimental groups are shown in Figs. 8 and 10, while the control groups are shown in Figs. 9 and 11. The difference in measured before and after exposure to the radiation is  subjected to identical handling and thermal cycles. Then the measurements from the control groups were compared to the experimental groups to determine if the radiation had any effect. The devices were characterized by magnetoresistive hysteresis loops and resistance-voltage curves. Four contact pads in a single MTJ die, as shown in Fig. 3, allowed four-terminal measurement of resistance using the Semiconductor Device Analyzer B1500A. The temperature dependence of TMR in all the devices was measured to rule out the effects of temperature variations during the experiments. The devices in experimental group were characterized before and after exposure to the gamma radiation in Gammacell 220, containing a source. The dose rate was a constant 9.78 rad/min. The experimental samples initially received a dose of 5.9 Mrad (Si) after which they were again characterized. Irradiation was then continued for a cumulative dose of 10 Mrad, which is significantly greater than the dose typically induces failure of CMOS devices [2], [3]. The devices were re-measured electrically and magnetically.
Neutron radiation experiments were conducted at the Oregon State University TRIGA Mk. II research reactor. The experimental setup is schematically illustrated in Fig. 4. The experimental group devices were characterized in situ in a cadmium-lined in-core irradiation tube. Total epithermal neutron fluence up to was obtained, with the flux at at 50 kW and neutron energies ranging from 0.1 eV to 10 MeV, as shown in Fig. 5.

IV. RESULTS AND DISCUSSIONS
Figs. 6 and 7 show the characterizations of a single MTJ before and after exposure to the gamma radiation. The hysteresis   loop and voltage bias dependence of resistance are shown respectively. TMR could be calculated as the resistance difference between the two states normalized by the TMR low resistance. The measured coercive field and TMR of the complete series of experimental groups are shown in Figs

Discussion of γ-ray Irradiation of MTJs
As mentioned above, certain research groups have claimed that MgO-based MTJs are sensitive to γ-ray irradiation in-situ due to the sensitivity of MgO barriers to γ-ray irradiation. On the other hand, most laboratories have reported that MgO-based MTJs are tolerant to γ-ray irradiation. In order to explain the discrepancy in the response of MgO-based MTJs to γ-ray irradiation, the effects of γ-rays are discussed below with respect to γ-ray penetration, the dynamic behavior of MTJ materials, and tunneling tolerance. The discrepancy may come from different experimental conditions.

γ-ray Penetration in MTJs
MTJs consist of MgO barrier layers sandwiched between ferromagnetic free/fixed layers, and metal electrodes, as well as electrodes made from high atomic number (high Z) materials with high density, such as Ta and Au. Electromagnetic waves, including γ-rays, can pass through these metal and ferromagnetic layers to reach MgO barriers. In order to analyze this penetration under different irradiation conditions, electromagnetic penetration is calculated. The intensity of electromagnetic radiation inside MTJs decreases exponentially from the MTJ's surface, as described by the equation based on the Beer-Lambert law [177,178]: where I is the intensity of electromagnetic radiation transmitted over a distance z, I 0 is the incident electromagnetic wave intensity, µ is the linear attenuation coefficient in cm −1 , µ = nσ = n(σ photoelectric + σ Compton + σ Pair ) (n: the number of atoms/cm 3 ; σ: proportionality constant that reflects the probability of an electromagnetic wave photon being scattered or absorbed), and z is the distance traveled by the radiation in cm. For multilayered films, the electromagnetic intensity is proportional to both the attenuation coefficient and the thickness of each layer through which it passes [179]. The calculation of electromagnetic radiation transmission through an MTJ is based on a typical MTJ structure consisting of Ta(5)/Ru(10)/Ta(5)/Co 20 Fe 60 B 20 (5)/MgO(2)/Co 20 Fe 60 B 20 (3)/Ta(5)/Ru(5)/Cr(10,000)/Au(10,000) (with numbers indicating nominal thicknesses in nanometers), as described in the published literature [171,180]. The linear attenuation coefficients of each film material are obtained from published data and used in the calculation. Equation (4) is then applied to calculate the transmission of electromagnetic radiation through the MTJ device. Figure 18 shows the calculated electromagnetic radiation intensity in a typical MTJ structure. The used electromagnetic radiation spans from 4.950 keV to 1 MeV in energy, covering both γ-ray (with energy greater than 124 keV) and X-ray (with energies of 125 eV-125 keV) radiation. According to the theoretical calculations, γ-rays could penetrate the entire MTJ structure without undergoing significant absorption. Some MTJs may contain thick metal electrodes, which can affect the penetration of γ-rays through the devices. Figure 19 shows the transmission of γ-rays through iron, a ferromagnetic material used in some MgO-based MTJs [2,174,181]. γ-rays can penetrate through iron for several centimeters, consistent with other reports [182]. Thus, γ-rays with various energy levels can easily penetrate entire MTJs, which consist of metal nano-films and thick electrodes, after passing through the top electrodes. This suggests that MTJs can be penetrated by γ-rays and their metal layers cannot shield all γ-rays, especially those with high energies.

Possible Explanations of Radiation Degradation
Based on the discussion of the radiation penetration above, it can be concluded that when exposed to γ-rays, MgO barriers should undergo interactions with γ-rays, i.e., photoelectric effects, Compton scattering, and electron-positron pair production, as discussed in previous sections. These interactions would cause displacement of Mg atoms or O atoms within lattices, resulting in defects or amorphizations of MgO barriers. As a consequence, MTJs would experience γ-ray-induced degradation according to the Julliére model. Figure 20a illustrates the Julliére coherent tunneling in an MTJ with a crystalline MgO barrier and two ferromagnetic layers. The tunneling process involves three kinds of Bloch states with different wave function symmetries existing in the free/fixed layers, which pass through the MgO barrier. The high MR ratio of the Fe/MgO/Fe sandwich structure primarily depends on the coherent spin-dependent tunneling that occurs in the crystalline MgO(001) tunnel barrier. detected the radiation penetrating the barrier. Correction for air-scattered radiation was established by observing the difference in counting rates with and without two inches of lead protecting the sides and bottom of the detector, Figure 1 shows the comparison between experimental results and theoretical predictions of the polynomial method as a function of penetration depth. The solid curves represent theoretical values computed for plane parallel radiation incident upon an iron medium by integrating the spectral intensity, 5 weighted by the detector response, over all energies. Corrections have been made to the experimental points for air-scattered radiation which is larger and more difficult to evaluate for lower energy radiation. The experimental values for Co 60 and Na 24 radiation are in good agreement with the theoretical curves and the slight deviations at large penetration depths are due to incomplete correction for air scattering.
The experimental points for Cs 137 radiation fall below the theoretical curve because less radiation is incident on the detector in our slab geometry than if it was imbedded in an infinite medium. This effect is most important for Cs 137 since large angle scattering is more probable for low-energy radiation.
This experiment extends the verification of the polynomial method to intermediate weights and verifies that the polynomial method contains the proper energy dependence in that the build-up factors increase as the primary energy decreases.
The authors wish to express their appreciation to Dr. E. H. Krause for his aid and support. 5 Curves were obtained from L. V. Spencer prior to publication. sandwiched with sewere made from high rials) and high-density can penetrate through f γ-radiation through in some MgO-based through Fe for several t with other reports 110 n centimeters. Therean easily penetrate the l nano-films after γ-ray In other words, γ-ray tal layers of MTJs can (µ 0 : permeability in free space). The saturation magnetization is 1.58 T. The perpendicular-anisotropy energy density K at this CoFeB thickness, which determines the thermal stability, is 2.1 × 10 5 J m −3 , a value comparable to that of the Co-Pd perpendicular multilayers 25 and high enough to secure good thermal stability at reduced dimensions (40 nm diameter). To separate the bulk and interfacial contribution of the anisotropy, the t CoFeB dependence of K = K b − M S 2 /2 µ 0 + K i /t CoFeB is measured, as shown in the inset of Fig. 2b. Here, K b is the bulk crystalline anisotropy and K i the interfacial anisotropy. From the intercept, K i is determined to be 1.3 mJ m −2 . The bulk contribution is consistent with the demagnetization (−M S 2 /2 µ 0 ), indicating that K b is negligible, that is, that the perpendicular anisotropy in this system is entirely due to the CoFeB-MgO interfacial anisotropy 26 .
From FMR measurements, the information of H K and α can be obtained. We have measured FMR spectra at a microwave frequency of 9.0 GHz for annealed CoFeB/MgO samples at T a = 300 • C as a function of the angle θ between H and the normal axis to the sample surface as shown in Fig. 3a. The θ dependencies of resonant field H R and linewidth (full-width at half-maximum, FWHM) are summarized in Fig. 3a,b, from which we can determine H K and α (ref. 19). Figure 3d,e shows the t CoFeB dependence of the obtained H K and α. The H K increases as thickness reduces and changes its sign reflecting the change of magnetic-easy-axis direction around t CoFeB = 1.5 nm. The t CoFeB dependence of K ·t CoFeB is plotted together with that obtained from M -H curves in the inset of Fig. 2, showing good correspondence between the two measurements. Although the magnitude of α steeply increases as thickness decreases below 2 nm, it is still smaller than those for materials including noble metals 18 . Full understanding of the origin of the increase is important to further reduce I C0 .
The interfacial perpendicular anisotropy between oxide and ferromagnetic metal (Fe/MgO) has been predicted by firstprinciples calculation and attributed to hybridization of Fe 3d and O 2p orbitals 27 . Although earlier experimental studies also indicated the presence of perpendicular anisotropy at the interface in Pt/Co/MO x (M = Al, Mg, Ta and Ru) trilayer structures 28,29 and in MgO/CoFeB/Pt (ref. 30), these structures always contained Pt in direct contact with ferromagnetic transition metals to stabilize the perpendicular anisotropy, which made the origin of the anisotropy ambiguous. As demonstrated in the following, the interfacial anisotropy between MgO and CoFeB is large enough to realize highperformance perpendicular CoFeB-MgO MTJs and no addition of noble metal is necessary. The coercivity H C is much larger than those shown taken on a millimetre-size sample, most probably ow suppression of domain-structure formation. The obt however, is smaller than 2K /M S , suggesting that there contribution of domain nucleation to H -induced mag reversal in these MTJ structures. The H C difference b nominally identical electrodes may be due to different intermixing at the two Ta-CoFeB interfaces during s and/or different areas of the two electrodes because of MTJ pillar introduced during ion milling; the t CoFeB d of M S indicates an approximately 0.5-nm-thick magnet layer in the CoFeB/Ta interface (corresponding to the in the MTJ) and no signature of a dead layer for Ta/C (corresponding to the bottom layer) (not shown). Per magnetic anisotropy with a clear R-H hysteresis is obta greater than 250 • C, and the TMR ratio increases mon with increasing T a and reaches 121% after annealing at T as shown in Fig. 4c. It should be noted that 350 • C is required for integration with complementary me semiconductor transistors. Further increase of T a leads of the TMR ratio.
Next, to show the potential of this material system at r mensions, circular 40-nm-diameter MTJs are fabricated shows an R-H curve of such an MTJ. The MTJ has t and 1.7 nm for bottom and top CoFeB layers, respec t MgO = 0.85 nm. The MTJ is annealed at 300 • C. The is 124% with resistance-area product RA = 18 µm 2 . loop of the top free layer (the free layer is identified CIMS measurement; see below) is shifted by 37 mT w 722 NATURE MATERIALS | VOL 9 | SEPTEMBER 2010 | www.nature.com/n properties. Likewise, to design optimal stacks for the fabrication process, clarifying how each species of atom diffuses and structure changes via annealing is both timely and relevant. So far, although a few first-principles calculations have been conducted, they draw contradictory conclusions regarding structure of the formed MTJs following annealing. 15−17 Experimentally, although the structural and chemical issues in the CoFeB−MgO MTJs have been addressed utilizing transmission electron microscopy (TEM), X-ray photoemission spectroscopy, and secondary ion mass spectroscopy, 18−21 a consistent picture has not emerged yet, largely due to the intricacy of defect structure, the challenge of resolving such a highly complicated heterojunction, and the difficulty of probing local atomic details of the MTJs with a confined dimensionality. In particular, there have been many contradictory results on the fate of B following annealing including diffusion into Ta, 22 segregation at the CoFe/MgO interface, 16,18 B diffusion into MgO forming a magnesium boride phase, 17,19 and diffusion of B into MgO as a dilute substitutional impurity. 20,21,23 As a consequence, most of the theoretical models have been put forward based upon assumed structures only. 24 Here, we combine aberration (Cs)-corrected TEM, highangle annular dark-field 25−29 (HAADF) and annular brightfield 30 (ABF) imaging in Cs-corrected scanning TEM (STEM), electron energy-loss spectroscopy (EELS), 31 and first-principles calculation to uncover atomic-scale structure and local chemistry of confined stacking layers in the CoFeB−MgObased MTJ exhibiting a high TMR and offer direct evidence of how B diffuses in the MTJs via annealing. Such combined techniques allow us to demonstrate that B diffuses out of the crystalline CoFeB into Ta interstitial sites rather than into the MgO tunnel barrier after annealing, and CoFe atomically bonds to the textured MgO grains with an epitaxial orientation relationship by forming Fe TEM Specimen Preparation and Imaging Condition. Cross-section thin-foil specimens for TEM and STEM observations were prepared by cutting, grinding, and dimpling samples down to ∼20 nm. In the argon ion-beam thinning process, a gun voltage of 1−4 kV and an incident beam angle of 4−6°were used to decrease radiation damage. The SADP and HRTEM images were taken using the JEOL JEM-2010F conventional microscope and the FEI TITAN 80-300 aberration corrected microscope (CEOS GmbH) that were operated at 200 and 300 kV, respectively. HAADF and ABF images were taken using the ARM-200F (Cold FEG) STEM operated at 200 kV, which was equipped with a probe corrector (CEOS GmbH), providing an unprecedented opportunity to probe structures with a sub-angstrom resolution. For the HAADF STEM imaging, a probe convergence angle of ∼22 mrad and a detector with an inner semiangle of over 60 mrad were adopted. ABF STEM images were taken using a detector of 11−23 mrad, and EELS was recorded using a Gatan Enfina system equipped on the STEM with an energy resolution at full width at half-maximum of ∼0.3 eV. HAADF STEM image simulations were performed using the WinHREM STEM package (HREM Research Inc.) based on multislice method. Debye−Waller factors were considered for every element involved, yet were averaged over spatial directions, namely, anisotropy of the absorption of thermal diffuse scattering factors was ignored. nomials. This method has given theoretical predictions in good agreement with experimental results 2 " 4 for Co 60 radiation in lead and water. Experimental results of plane parallel Cs 137 , Co 60 , and Na 24 gamma radiation diffusing through iron are com- pared with computations of Spencer. The shield used in these experiments was constructed of J-inch plates four feet square which were secured in a perpendicular position ten feet above the ground to insure a minimum contribution to the counting rate by ground-scattered Irradiation can have an impact on tunneling. Figure 20b demonstrates the tunneling through an amorphous barrier. When the MgO(001) tunnel barrier becomes amorphous due to irradiation, the crystallographic symmetry of the tunnel barrier is lost; Bloch states with various symmetries can couple with the MgO tunneling states, resulting in finite tunneling probabilities. In 3d ferromagnetic metals and alloys, Bloch states with ∆ 1 symmetry (s-p-d hybridized states) generally exhibit a large positive spin polarization P at the Fermi energy E F , while those with ∆ 2 symmetry (d states) tend to have a negative spin polarization P at E F [20,183]. All Bloch states in the ferromagnetic free/fixed layers contribute to the tunneling current, affecting the net spin polarization of the ferromagnetic layers and degrading the functionalities of MTJ devices. In other words, after γ-ray irradiation, the momentum of tunneling electrons is no longer conserved due to local disorder scattering. This would destroy the coherence or symmetry of conducting electrons and changes the coherent tunneling process to incoherent tunneling through the displacement of atoms, degrading MTJs. It was experimentally proved that defects in MgO barriers impacted polarized tunneling, localized states of spin, and polarized symmetry tunneling across MgO barriers [184]. The electronic properties of MgO grain boundaries in MTJs are symmetry dependent [185].
In addition, the energy of γ-rays can be transferred to electrons, resulting in an increase in the number of high-energy free spin electrons that interact with the lattices and interfaces. This increase can change the spin polarization: here, N ↑ (E F ) and N ↓ (E F ) are the density of state at the Fermi energy (E F ) for spin-up electrons and spin-down electrons, respectively. γ-rays can change the density of states at E F , affecting spin-and polarized-symmetry tunneling. Additionally, γ-rays can penetrate through the free/fixed layers, modifying their electrical and magnetic properties through the photoelectric effect and the Compton effect, as well as by indirect ionization, which can intermittently or permanently degrade MTJ performance.
According to the Julliére model and the penetration analysis, MgO-based MTJs are expected to degrade under γ-ray irradiation, as reported in some literature. Briefly, MTJs should be sensitive to γ-ray radiation.

Possible Explanations of Tunneling Tolerance
Most research groups have reported that MgO-based MTJs are highly tolerant to γ-ray radiation and not degraded by γ-ray radiation at all. There are three possible reasons for this.
One possible explanation of their tolerance to γ-rays is that the unique TMR mechanism of MTJs enables MgO-based MTJs to be tolerant. The tunneling mechanism is the most popular explanation. The magnetic properties of MTJs originate from spin charges, which makes MTJs resistant to radiation. While γ-ray radiation can amorphize MgO barriers and ferromagnetic fixed/free layers, the resulting partial amorphous status of MTJ layers has only a slight effect on the magnetic characteristics of the fixed/free layers [171]. Therefore, the degradation caused by radiation in MgO-based MTJs is negligible.
Secondly, the degradation of MgO-based MTJs is limited to specific conditions, such as exposure to extremely high doses of radiation, which can result in complete or partial destruction of the crystallographic structures of the MTJ layers and cause MTJs to loss their functionality. Fortunately, such critical conditions are rare in γ-ray irradiation, although they can occur in neutron irradiation and high-energy ion irradiation. Therefore, the degradation of MTJs under γ-ray irradiation is expected to be minimal and maybe not to be detected.
Thirdly, many reported measurements have been carried out ex-situ. As discussed below, the damage caused by irradiation may diminish over time, and the physical properties of MTJs may be restored by the time measurements are taken.

Possible Explanations for Divergence
It is evident that γ-rays can modify MgO barriers and ferromagnetic layers, and some research groups have reported degradation of MgO-based MTJs as a result. Additionally, the fact that most MTJs cannot operate at high temperatures suggests that MTJs are susceptible to infrared electromagnetic waves, which have lower energy than γ-rays. The temperature-induced degradation indirectly indicates that MTJs should be susceptible to γ-rays, with higher energy. However, most research reports have indicated that MgO-based MTJs are tolerant to γ-rays. The divergence among these reports may be explained in the dynamic properties of γ-ray-induced damage, which can account for the divergence among the literature.

Temporary Defects
Like high-energy ion radiation, γ-rays may only induce temporary defects that do not persist for a long time at room temperature and vanish after exposure to radiation. The excited electrons and ionization can quickly return to the initial state due to thermal motion at room temperature. Figure 13 shows one case where γ-rays changed a physical property (photoconductivity) of MgO, which was restored to its initial state after γ-ray irradiation due to thermal motion. Figure 21 shows another case. The TL intensity of MgO powder irradiated by γ-rays was measured immediately or 75 days after irradiation [165]. The signals induced by the γ-ray radiation diminished over time. Recent calculations have also demonstrated that thermal motion at room temperature can eliminate the impact of γ-ray irradiation.

Radiation Effects & Defects in Solids 263
to its trapping level giving rise to a Cr 2+ ion, leaving a free hole, which, at 150 • C could be trapped at an Fe 2+ ion, resulting in the Fe 3+ ion.

Storage effects
TL fading was performed as follows: annealed MgO aliquots (400 • C for 1 h) were divided into two parts: the first part was irradiated with two doses of 1 and 10 Gy and then stored for varying lengths of time in dark at room temperature. Successive measurements were monitored over a Figure 6. The effect of room temperature storage on laboratory-induced glow curves of magnesium oxide, for the delay period t = 75 days and without delay (t = 0), for a gamma dose of: (a) 1 Gy and (b) 10 Gy. Figure 7. Relative thermoluminescence as a function of storage time for magnesium oxide exposed to 1 and 10 Gy gamma doses.

Gy
TL fading was performed as follows: annealed MgO aliquots (400 • C for 1 h) were divided in two parts: the first part was irradiated with two doses of 1 and 10 Gy and then stored for varyin lengths of time in dark at room temperature. Successive measurements were monitored over  Due to the time-dependent dynamic nature of the impacts of γ-rays, only in-situ measurements can detect the transiently degraded performance of MTJs with excited states. Some studies have reported soft errors of MgO-based MTJs in in-situ measurements under irradiation [161], which were consistent with the assumption of temporary impacts over time.
Until now, most state-of-the-art measurements have been carried out ex-situ, without the presence of γ-rays. Some measurements were performed immediately after γ-ray irradiation, such as within 2 min after removal of the radiation sources [162], or long after irradiation, such as after one year of storage at room temperature [164]. The impact of the irradiation may have diminished prior to making the measurements. Impact information under γ-rays may decay over time and become undetectable. This may be one explanation for why MgO-based MTJs have been reported to be resistant to γ-ray radiation in certain instances.
γ-ray irradiation may only transiently change the physical properties of MgO-based MTJ layers during irradiation procedures and not cause permanent damage. The properties of MTJ layers can be restored reversibly after exposure to γ-ray radiation, and therefore, MTJs can return to their initial state before irradiation. The temporary degradation of MgO-based MTJs induced by γ-ray radiation is not detectable in ex-situ measurements.

Irradiation Annealing
Irradiation annealing may eliminate the impacts of irradiation. High-energy γ-ray radiation can produce permanent defects in MgO barriers and ferromagnetic layers, changing their crystallographic structures and the physical properties of the layers, thereby degrading the performances of γ-ray-irradiated MTJs. However, these defects may revert to their initial equilibrium state over time at high temperatures. High-dose-rate γ-ray radiation can generate such high temperatures in MgO-based MTJs. The radiation-induced heat can self-anneal MTJs, erasing the effects of γ-ray irradiation and preventing the degradation of γ-ray-irradiated MTJs.
Regrettably, there are few experimental reports on irradiation annealing. The temperature of MgO-barriers and free/fixed layers is rarely mentioned in the literature, and the time interval between γ-ray irradiation and physical measurements is also unknown. More comprehensive in-situ and real-time investigations on the interactions between γ-rays and materials are required.

Effects of Lower-Energy Irradiation
Electromagnetic waves with wavelengths longer than gamma rays are commonly known as lower-energy waves, such as X-rays, ultraviolet radiation (UV), visible light, infrared radiation, microwaves, and radio waves. These electromagnetic waves have less energy compared to gamma rays, and are generally classified as non-ionizing radiation, with the exception of X-rays.

X-ray Irradiation
The energy of X-rays ranges from several tens of electron volts to hundreds of kiloelectron volts. The intensity of X-rays decreases exponentially from the surfaces of MTJs, as described by the Beer-Lambert law in Equation (4). X-ray radiation typically only penetrates a few microns into materials, depending on its energy and the material's composition. MgO-based MTJs are typically sandwiched by electric electrodes made of materials such as gold or tantalum. These metal electrodes are usually thick enough to prevent X-rays from penetrating through to the MgO barriers and ferromagnetic layers of the MTJs. The detailed screening effect can be calculated. Figure 18 shows the calculated penetration intensity of X-ray radiation with an energy above 4.950 keV (energy of X-ray: 124.8 eV-124.8 keV). Hard X-rays can fully penetrate MgO-based MTJs with weak absorption, therefore affecting the physical and chemical properties of both the MgO barriers as well as the ferromagnetic layers. The MgO barrier layers should be affected by X-ray radiation in a similar way to two-dimensional MoS 2 monolayers [112,140]. In this case, the effects of X-ray radiation on MgO-based MTJs are very similar to those of γ-ray radiation. These X-ray effects may also be temporary and only detectable through real-time measurements. Soft X-rays with energies of ten kiloelectron volts or less would be strongly screened by metal electrodes, with penetration through to the MgO barriers and ferromagnetic layers of MTJs being prevented. Consequently, the effects of soft X-ray irradiation can be disregarded. Up to now, there have been few studies of X-ray radiation on MgO-based MTJs.

UV-Vis Irradiation
The energy of ultraviolet-visible (UV-vis) electromagnetic waves ranges from 1 eV to several tens of electron volts, with wavelengths of 10-400 nm. As shown in Figure 18, UV and visible electromagnetic waves cannot penetrate through metal layers to reach ferromagnetic and MgO layers. Additionally, metallic electrodes reflect UV-vis radiation, making MgO-based MTJs highly resistant to such radiation. However, heat produced by UV-vis radiation may degrade MgO-based MTJs. Doped MgO materials have been studied as a potential material for UV dosimetry to detect ultraviolet radiation [186,187]. Figure 22 shows the thermoluminescent (TL) response of UVirradiated MgO crystals. Studies have shown that the thermoluminescent peaks of doped MgO crystals depend significantly on the dose of ultraviolet radiation with wavelengths such as 295 nm [186], 289 nm [187], and 249 nm [187]. Even pure MgO crystals are affected by ultraviolet radiation with wavelengths such as 295 nm [186] and 337 nm [188]. Similar behavior was reported at other ultraviolet wavelengths [187]. These studies demonstrate that UV radiation changes the microstructure of MgO materials. However, the specific physical processes underlying the UV-vis radiation and MgO materials have not been not well described in the literature. The most likely explanation, is that UV-vis radiation causes an increase in temperature in the MgO materials, leading to their degradation. It is noteworthy that the changes in the TL signals induced by UV-vis irradiation decreased over time. It was reported that the TL intensity of some irradiated crystals restored up to 95% of its initial value after being stored at room temperature for four days [187].
In theory, UV-vis radiation should degrade MgO-based MTJs, because MgO is sensitive to these electromagnetic waves. However, this degradation should only be temporary and result from radiation-induced heating. If heating effects are avoided, MgO-based MTJs should be highly tolerant to UV-vis radiation. To date, there is no literature available on the subject of the effects of UV-vis radiation on MgO-based MTJs.

Infrared Radiation and Thermal Annealing
Heat radiation or thermal radiation is a well-known term for infrared radiation. Pulsed thermal radiation, with a long wavelength of 1-20 microns and energy of 1-24 eV, can be efficiently screened by metallic electrodes. However, continuous thermal radiation, also known as heat, can penetrate MTJ devices during prolonged exposure to high temperatures, resulting in thermal annealing and thermal equilibrium. Thus, infrared radiation is somewhat different to other types of radiation.
There are reports on the annealing effect on MTJ component materials. Nikiforov et al. studied the pulse cathodoluminescence (PCL) excitation of MgO nanomaterials with a size of 250-500 nm [157]. It was reported that the PCL intensity in the 2.0-3.5 eV band increased by an order of magnitude with increased annealing temperatures, attributed to the relaxation of F-type centers (oxygen vacancies with two captured electrons). Shen et al. investigated the impact of thermal annealing on ferromagnetic CoFeB layers [189]. Their investigation indicated that thermal annealing enhanced the crystallization of CoFeB at the interfaces with MgO, affecting the magnetoresistance of MgO-based MTJs. Yuasa et al. reviewed the annealing effect on CoFeB electrodes [31], and interested readers are referred to the literature cited therein.
Liu et al. investigated the thermal stability of MTJs with MgO barriers at temperatures up to 500 • C [191]. The MTJs consisted of Ta (30) (5). The study observed the irreversible loss of magnetoresistance at high temperatures.
Typically, thermal annealing (using infrared radiation) has a positive effect on the crystallization of MgO barriers, which enhances the performance of MTJs. However, thermal annealing also accelerates interface diffusion between MgO barriers and ferromagnetic layers, leading to degradation of MTJ performance [192]. Xu et al. employed transmission electron microscopy and electron energy loss spectroscopy to investigate the microstructures of the MgO-CoFeB interfaces of MTJs [193]. Figure 23 shows HRTEM images, STEM images, and EELS mapping of the interfaces after thermal annealing. Thermal annealing indeed crystallized MTJ layers, as shown by the HRTEM images, and caused boron diffusion. Boron diffusion led to the growth of CoFe nanocrystals from CoFeB layers under annealing, while the crystallization did not significantly affect the MR properties. Instead, the MR ratio was predominantly determined by grain boundary transport caused by boron distribution. If boron diffused to metallic underlayers from the inside to the outside (as shown in Figure 23e,f), the MR ratio would be improved. Conversely, annealing may result in boron diffusing into grain boundaries of the MgO barriers from the outside to the inside (shown in Figure 23g,h), leading to a decrease in the MR ratio. The interfacial properties of MTJs regulated the diffusion of boron and affected the effect of thermal annealing. Thus, the effect of thermal radiation on MTJ devices depends on the annealing temperature, the duration, and the structure of the MTJs. Thermal irradiation can either benefit or degrade MTJs' performance. Figure 23. Cross-sectional HRTEM images (a-d) and ADF-STEM images and corresponding elemental EELS mappings (e-h) using O-K, Fe-L 3,2 , Co-L 3,2 and B-K ionization edges taken from the Ta/CoFeB/MgO/CoFeB/Ta MTJ (a,b,e,f) and W/ CoFeB/MgO/CoFeB MTJ (c,d,g,h) at 300 • C (a,c,e,g) and 400 • C (b,d,f,h). Reproduced with permission [193] with copyright 2018, Elsevier.
It is important to note that radiation other than infrared radiation can also produce heat, particularly at high-dose rates, which can lead to an increase in the temperatures of MTJs and produce similar annealing effects. Under such circumstances, high-energy radiation, such as γ-ray and hard X-ray radiation, may cause additional annealing effects. To study the effects of irradiation, it is crucial to investigate MTJs at constant temperatures or monitor the internal temperatures of MTJs, particularly the temperatures of the MgO and ferromagnetic layers.

Microwave Irradiation
The penetration depth of microwaves into conductive metal surfaces is typically less than one micron [194]. Therefore, the metallic electrodes of MTJs can efficiently reflect microwaves. In other words, microwaves should not penetrate through the electrodes to irradiate the MgO barriers and ferromagnetic layers. Therefore, the microwave irradiation effect can ignored, and microwave radiation should not have any significant impact on the performance of MTJs.
Although microwave radiation is not expected to penetrate through the electrodes of MTJs to affect the MgO barriers and ferromagnetic layers, it can cause a significant increase in the temperature of metal layers. Research has shown that microwave irradiation can produce a high temperature, of up to 500 • C, in Au films in less than 10 s [195]. Therefore, microwave irradiation can generate a high temperature locally in ferromagnetic fixed-/freelayes of MTJs, which can have a significant impact on the performance of MTJs.
Up to now, there have been limited reports on the impact of microwave radiation on MgO-based MTJs. Some groups have investigated the behavior of MgO-based MTJs under microwave irradiation [94]. Unfortunately, it was not stated whether the MgO-based MTJs were damaged under microwave irradiation.

Radiofrequency Electromagnetic Irradiation
Radiofrequency (RF) electromagnetic radiation can be shielded by conductive or magnetic materials, which is known as RF shielding. Since MTJs have metal electrodes, these electrodes can block RF radiation and therefore MTJs should not be affected. The theoretical calculation shown in Figure 18 also predicts that electromagnetic waves with energy lower than four kiloelectron volts would not penetrate through the electrodes of MTJs. As listed in Table 1, the energy of radiofrequency radiation is typically less than a few milli-electronvolts, so RF radiation should be totally shielded and not affect MTJ performance.
Similar to microwaves, RF irradiation can also induce heating in metals, leading to high temperatures locally in MTJ electrodes. However, the induced temperature is expected to be low due to the extremely low energy of RF radiation.
Therefore, the effects of radiofrequency and other electromagnetic irradiation with longer wavelengths can be ignored. MgO-based MTJs should be highly tolerant to this radiation.

Outlook
MgO-based MTJs are promising for various applications, such as MRAM in quantum computers, logic gates, ultra-sensitive sensors, and energy harvesting and storage. These devices can be utilized in space technology, and therefore, the impact of radiation is crucial. With advancements in super-large-scale integration (SLSI) technology for central processing units (CPUs) and graphics processing units (GPUs) and programming languages such as the open source Python programming language, as well as professional packages/libraries for programming languages, it is possible to simulate complex interactions between radiation and MTJ components at the atomic level. Dynamic simulations at the atomic level can be employed to investigate individual atomic motion and nanoscale displacement under irradiation, and to calculate MR, providing insights into the dynamic behavior of atoms during irradiation. Additionally, the development of artificial intelligence (AI), including machine learning and deep learning, makes it possible to collect most research data on the irradiation of MgO-based MTJs and to systematically analyze the radiation's impacts. Various parameters, such as radiation energy, duration, dose, and dose rate, can be simulated, investigated, and compared with experimental data to understand the electromagnetic-material interaction. Safe operation of MgO-based MTJs can be predicted in various irradiation environments.

Conclusions
The effects of radiation on MgO-based magnetic tunnel junctions have been reviewed and analyzed in various irradiation environments, including high-energy cosmic radiation, gamma-ray, X-ray, UV-vis, infrared , microwave, radiofrequency, and long-wavelength electromagnetic radiation. The examination considered both the material properties and device performance. In general, cosmic radiation (including ions and protons) can damage MTJs due to permanent atom displacements in the MTJ layers. While some groups have reported that γ-ray irradiation degrades the performance of MgO-based MTJs, the majority of scientists have claimed that MgO-based MTJs are tolerant to γ-rays without significant degradation in their performance. The impact of hard X-ray irradiation is comparable to that of γ-ray irradiation. Soft X-ray, UV-vis, infrared, and microwave radiation can be screened or shielded by the metal electrodes of MTJs, and these types of electromagnetic radiation should not significantly affect MTJ devices. Nonetheless, these types of radiation may induce heat or annealing, especially for infrared and microwave radiation, which can affect MTJ performance by causing crystallization of the MgO barriers and ferromagnetic layers as well as interfacial diffusion. There is no strong evidence that the present MgObased MTJ devices are susceptible to radiation. The effects of radiation on MgO-based MTJs are discussed with respect to electromagnetic penetration, the Julliére model, the TMR mechanism, and annealing, to explore the physics behind these reported experimental data. Further in-situ and real-time investigations are necessary to fully understand the radiation tolerance of MgO-based MTJ devices under various types of electromagnetic radiation.