For over five decades, charge-based electronics represented predominantly by the silicon transistor have powered the information technology revolution. However, as complementary metal-oxide-semiconductor (CMOS) devices approach limits in terms of speed, power dissipation, and miniaturization, researchers have turned attention toward alternative state variables such as electron spin polarization to potentially realize improved or even entirely new functionality (Zutic et al. 2004; Žutić et al. 2006).
This approach, termed spintronics, offers several intriguing possibilities:
- Non-volatile logic and memory: Aligned electron “up” or “down” spin states can persist without applied power for extended periods. This phenomenon enables instant-on, near-zero standby power electronic devices (Prejbeanu et al. 2022).
- Increased operational frequencies: Spin polarization orientation can be altered at GHz to THz rates utilizing small magnetic fields. Such high-frequency spin manipulation allows for faster device operation compared to conventional transistors (Kruglyak et al. 2010).
- Reduced power dissipation: Coherent spin state control mechanisms can potentially exceed theoretical minimum energy dissipation limits for CMOS and related charge-based devices (Behin-Aein et al. 2010).
Conventional electronics rely primarily on the modulation of electrical charge carrier density and associated current/voltage manipulation to realize computation and digital logic functionality. A key performance metric is charge storage capacity in capacitors and transistors or the number of electrons transported via an applied current over a given interval. Building upon this, the field of spintronics incorporates the inherent electron spin quantum property as an additional degree of freedom available to transport and store information (Žutić et al. 2004).
Electron spin refers to intrinsic angular momentum quantized into either “up” ↑ or “down” ↓ orientations along some axis conventionally chosen to be the z-axis. Unlike binary logic levels limited to solely 0 and 1 charge density conditions mapped to capacitor/transistor on-off states, spin enables access to new combinations of levels including pure spin state pairs such as (↑, ↓) or superposition states like (↓, ↑) within individual electrons.
This multiplicity empowers the representation of multiple bits or exploitation as qubits for quantum information applications (Awschalom et al. 2013).
The following physical phenomena provide pathways to electrically generate, manipulate and measure spin polarization and associated currents.
Spin Transfer Torque
Applied charge currents featuring definite spin alignment can reversibly reorient the magnetization of an adjacent ferromagnet via transfer of angular momentum, an effect termed spin transfer torque (STT) (Ralph et al. 2008; Brataas et al. 2012).
The Slonczewski torque expression for spin current I_s flowing through volume V, polarized along unit vector m̂ and interacting with local spin magnetization vector S pointing along saturation magnetization M_s is:
τSTT = (ħ I_s / |e|) (S x m̂) / (2 M_s V)
Where ħ is the reduced Planck constant and e is the fundamental electron charge (Slonczewski 1989, 1996). Constructive STT switching of magnetic elements enables scalable, low energy writing of bits within magnetic tunnel junctions (MTJs) organized into magnetoresistive random access memory (MRAM) arrays (Kent & Worledge 2015).
Spin Hall Effect
The spin Hall effect (SHE) generates pure transverse spin current densities within nonmagnetic heavy metals possessing strong spin-orbit coupling (SOC) (Sinova et al. 2015). An applied charge current density Jc through these spin Hall materials yields a spin current density Js given by:
Js = (ħ/2e) ΘSHE Jc x σ
where the spin polarization vector σ points along the spin axis and ΘSHE denotes the material-specific spin Hall angle determining SHE efficiency (Hoffmann 2013). Momentum transfer between spin-orbit coupled lattice ions and conducting electrons deflects electron trajectory depending on spin state. Up and down spin electrons therefore flow oppositely transverse to the electrical current direction due to this effective spin-dependent magnetic field arising from relativistic interplay between the electric field E and electron velocity v.
SHE enables highly efficient electrical injection and detection of spin currents vital for the realization of both spin logic and memory circuitry (Han et al. 2017; Aradhya et al. 2016).
Spin Seebeck Effect
The spin Seebeck effect (SSE) refers to spin voltage generation from an applied temperature gradient ∇T across a ferromagnet (FM) and attached paramagnetic metal (PM) thin film structure (Uchida et al. 2008, 2010). Though the exact mechanism remains under investigation, phonon propagation likely imparts transverse spin polarization subsequently detectable through the inverse spin Hall effect in the PM.
Remarkably, SSE produces spin currents persisting over centimetre length scales, unlike other nanoscale spin transport effects. This long-range spin transport empowers exploration of spin caloritronics harnessing heat to temperature gradients to induce spin polarization (Bauer et al. 2012).
The origins of spintronics research date to early quantum mechanics. In 1924, Wolfgang Pauli proposed the electron spin concept to explain observed atomic spectral splittings (Pauli 1924). However, practical utilization of electron spin polarization in solid-state devices first emerged in the 1980s enabled by advances in thin film growth and nanofabrication capabilities.
A seminal event occurred in 1988 when Albert Fert (Fert 1988) and independently Peter Grünberg (Baibich et al. 1988) reported sizable change in electrical resistance under applied magnetic field within alternating ferromagnetic and nonmagnetic metallic thin film structures. This giant magnetoresistance (GMR) arises from spin-dependent electron scattering rates that depend on the relative magnetization alignment among the ferromagnetic layers.
GMR transducers soon enabled enhanced hard disk drive storage densities (Parkin 2003; Chappert et al. 2007) by converting magnetic bit stray fields into measurable resistance changes. This success sparked intense industrial and academic research into harnessing the electron spin degree of freedom.
Since the initial discovery of GMR, extensive efforts have focused on uncovering new materials and designing innovative devices to generate, transmit, and detect spin polarization (Žutić et al. 2004; Fabian et al. 2007). Such advances have opened possibilities for spintronics not just in data storage but also extended into telecommunication systems, medical diagnostics tools, and other application spaces (Hirohata et al. 2020).
Breakthroughs in nanofabrication of high-performance magnetic tunnel junctions (MTJs) containing thin oxide barriers have enabled the emergence of scalable, non-volatile spintronic devices manufactured with available CMOS infrastructure (Huai 2008; Kent & Worledge 2015).
As depicted in the below figure, MTJs incorporate two ferromagnetic metallic layers separated by an ultra-thin (≈1 nm) insulating barrier. Applying an electrical current induces interaction between spin populations within the two ferromagnets, modulating MTJ resistance – an effect termed tunnelling magnetoresistance (TMR) (Miyazaki & Tezuka 1995; Moodera et al. 1995). This measurable, non-volatile resistance state change supports the usage of MTJs as memory elements or reconfigurable logic gates.
Compatibility with standardized semiconductor manufacturing avoids need for specialized tools and materials, smoothing the integration of spintronics. Rather than necessitating complete replacement, spin-based devices can interoperate with and complement conventional CMOS (Behin-Aein et al. 2010; Manipatruni et al. 2019).
Besides MTJ random access memory (MRAM), various alternate spintronics device concepts presently seek to harness electron spin (Zutic et al. 2022):
- Spin valves: Spin valve structures consisting of dual ferromagnetic layers sandwiching a nonmagnetic metallic spacer enable giant magnetoresistance similar to MTJs. The additional nonmagnetic layer helps suppress spin scattering, improving spin valve sensitivity (Pratt et al. 1991; Dieny et al. 1991).
- Spin transfer torque magnetic random access memory (STT-MRAM): STT-MRAM relies on spin-polarized currents rather than magnetic fields to switch magnetization orientation within the free layer of each MTJ cell. This reduces write energy dissipation by orders of magnitude compared to conventional MRAM while still allowing nanosecond switching times (Kent & Worledge 2015).
- Spin Hall effect devices: These structures exploit spin orbit mediated transverse spin current generation from applied charge currents to electrically detect or generate spin polarization (Sinova et al. 2015).
- Topological insulators: Topological insulators can conduct spins on their surfaces without dissipation due to their unique electronic band structure. Such materials offer tantalizing potential for low-power spin logic or as interconnects (Hasan & Kane 2010).
Here is a continuation of the spintronics scientific document:
While spintronics promises significant benefits over conventional charge-based devices, several major challenges remain to be addressed before widespread commercial adoption becomes viable:
Efficient Electrical Spin Injection and Detection
Achieving highly efficient interconversion between charge and spin currents represents a key obstacle, particularly for room-temperature spintronic systems (Zutic et al. 2004). At ferromagnet/nonmagnet interfaces, conductivity mismatch hampers the injection of spin polarization from a metallic ferromagnet into a semiconductor, limiting maximum attainable injection efficiency values to around 50% for typical device structures (Schmidt et al. 2000).
Additionally, detecting spun currents via inverse spin Hall effect measurements suffers from inaccuracies depending on measurement geometry and materials selections. Interface and device engineering approaches to enhance spin injection/detection efficiencies closer to the ideal >90% mark across operating temperature ranges continue as active research areas (Dash et al. 2009; Jeon et al. 2018).
Identifying and optimizing ferromagnetic materials compatible with silicon CMOS technology and operable at room temperature remains an ongoing pursuit. While metallic ferromagnets exhibit adequate characteristics, their integration with existing fabrication flows proves complicated. Alternatively, ferromagnetic semiconductor candidates allowing simplified CMOS integration have so far only demonstrated functionality well below room temperature.
Recent research has focused on materials doping strategies to potentially attain high-temperature magnetic ordering (Zutic et al. 2022). However, sizable improvements must still occur before spintronic devices with both ambient operation and back-end-of-line CMOS manufacturability can be realized.
Spin Coherence Time
Preserving spin polarization coherence and lifetime against perturbations represents another key challenge. Defects, impurities, lattice dynamics, and environmental noise all facilitate the loss of spin information via scattering (Zutic et al. 2004). Cryogenics extends spin coherence substantially, but this nullifies advantages over conventional electronics.
While recent experimental measurements have achieved spin lifetimes of around 10 ns at room temperature (Dash et al. 2009), even longer intervals on the order of microseconds will likely prove necessary for complex spin logic or quantum computing aims (Behin-Aein et al. 2010). Further materials science and spin transport physics advances are required to minimize interactions disrupting spin coherence.
Circuit and System Design
Architectures and circuits tailored to optimally exploit emerging spin-based devices also require significant development (Manipatruni et al. 2019). To fully harness the unique capabilities of spintronics, innovation of beyond-CMOS computing paradigms incorporating spin-based logic, memory, and interconnects will be essential. This entails codesigning devices, circuits, and software stacks in a cross-disciplinary, beyond-CMOS fashion.
Potential Technological Impacts
Despite facing hard fundamental and implementation obstacles, spintronics promises revolutionary impacts on specialized computing applications (Žutić et al. 2006; Mahfouzi et al. 2018):
Non-Volatile Embedded/Main Memory
Spintronics memory arrays like STT-MRAM avoid wear-out issues plaguing charge-storage based flash memory while delivering similar densities and DRAM-like random access speeds. Further integration of spin transfer torque and exchange biasing effects in magnetic tunnel junctions and related spin logic gates could enable fast, high-capacity non-volatile memory (Kent & Worledge 2015). Such developments would fulfil long-sought universal memory aims.
Low-Power Mobile Computing
Novel logic device concepts harnessing spin orientation switching mechanisms rather than charge flow could substantially cut power consumption compared to state-of-the-art CMOS processors and related circuits. New low energy spintronics-based instant-on computing architectures may emerge for battery-constrained mobile platforms (Manipatruni et al, 2018).
The inherent dynamics of interconnected spin systems closely resemble biological neural networks. Therefore, spin-based devices theoretically provide extremely efficient artificial synapses and neuron-like functionality.
This empowers brain-inspired computing stressing complex parallelism and efficiency rather than raw sequential digital performance (Torrejon et al. 2017). Ongoing efforts seek comprehensive beyond-CMOS architectures fusing spin logic, memory, and processing analogous to biological brains.
In summary, while scaling spin-based devices to definitively surpass state-of-the-art CMOS poses immense challenges, spintronics offers strong potential for specialized computing technologies. Non-volatile embedded/main memory, ultralow energy mobile processors, and brain-inspired neuromorphic architectures represent promising near-term niches if materials, integration, and architectural obstacles can be conquered. Looking beyond the silicon era, the intrinsic physics of electron spin could transform information processing by augmenting and even replacing charge flow mechanisms. Therefore, spintronics research appears well-positioned to usher electronics into uncharted territories beyond classical and quantum computing frontiers.