Memristor element with a magnetic domain wall in a magnetic free layer moved by spin orbit torque, synapse element and neuromorphic processor including the same
Disclosed are a memristor element, a synapse element and a neuromorphic processor including the same. The memristor element includes a free layer including a domain wall; and a fixed layer including a material of which a magnetization direction is fixed, wherein a position of the domain wall in the free layer is changeable based on a spin orbit torque (SOT) generated by a current introduced from an outside, and wherein a resistance value, measured through both ends of the fixed layer, is based on the position of the domain wall and on a Hall voltage.
1. A memristor element comprising:
a free layer including a domain wall; and
a fixed layer including a material of which a magnetization direction is fixed,
wherein a position of the domain wall in the free layer is changeable based on a spin orbit torque (SOT) generated by a current injected to the fixed layer from an outside through a first position and output through a second position such that the current traverses through the fixed layer, and
wherein a resistance value of the fixed layer between the first and second positions is based on the position of the domain wall and on a Hall voltage,
wherein the Hall voltage is based on the position of the domain wall,
wherein the Hall voltage has at least a lower value, a higher value, and an intermediate value,
wherein the memristor element is configured to store non-binary data based the resistance value, and
wherein a direction of the domain wall position change is parallel to a direction of the injected current that causes the domain wall position to change.
2. The memristor element of claim 1 , further comprising:
a first electrode configured to accept the current introduced from the outside; and
a second electrode configured to provide a voltage pulse to the outside.
3. The memristor element of claim 1 , wherein the free layer includes a synthetic antiferromagnet (SAF) material.
4. The memristor element of claim 1 , wherein the fixed layer includes at least one of a metal iridate or an oxidized layer of a ferromagnet (FM)/heavy metal bilayer.
5. The memristor element of claim 1 , further comprising:
a spin orbit torque writing line with a direction of spin parallel to a magnetization direction of the free layer.
6. The memristor element of claim 1 , further comprising:
a spin orbit torque writing line with a direction of spin perpendicular to a magnetization direction of the free layer.
7. A synapse element comprising:
at least one transistor; and
at least one memristor element electrically connected to the at least one transistor and configured to store a weight for an artificial neural network operation, and
wherein the at least one memristor element includes
a free layer including a domain wall; and
a fixed layer including a material of which a magnetization direction is fixed, and
wherein a position of the domain wall is changeable based on a spin orbit torque (SOT) generated by a current injected to the fixed layer from an outside through a first position and output through a second position such that the current traverses through the fixed layer,
wherein a resistance value of the fixed layer between the first and second positions is based on the position of the domain wall and on a Hall voltage,
wherein the Hall voltage is based on the position of the domain wall,
wherein the Hall voltage has at least a lower value, a higher value, and an intermediate value,
wherein the memristor element is configured to store non-binary data based the resistance value, and
wherein a direction of the domain wall position change is parallel to a direction of the injected current that causes the domain wall position to change.
8. The synapse element of claim 7 , wherein the at least one memristor element further includes:
a first electrode configured to accept the current introduced from the outside; and
a second electrode configured to provide a voltage pulse to the outside.
9. The synapse element of claim 7 , wherein
a first terminal of a first transistor, among the at least one transistor, is connected to a word line,
a second terminal of the first transistor is connected to a source line,
a third terminal of the first transistor is connected to one end of a first memristor element, among the at least one memristor element, and
another end of the first memristor element is connected to a bit line.
10. The synapse element of claim 7 , wherein
one end of a first memristor element, among the at least one memristor element, is connected to a word line,
wherein another end of the first memristor element is connected to a first node,
wherein one end of a second memristor element, among the at least one memristor element, is connected to the word line,
wherein another end of the second memristor element is connected to a second node,
wherein a first terminal of a first transistor, among the at least one transistor, is connected to the first node,
wherein a second terminal of the first transistor is connected to the second node,
wherein a third terminal of the first transistor is connected to a bit line,
wherein a first terminal of a second transistor, among the at least one transistor, is connected to the second node,
wherein a second terminal of the second transistor is connected to the first node, and
wherein a third terminal of the second transistor is connected to the bit line.
11. The synapse element of claim 7 , wherein the free layer includes a synthetic antiferromagnet (SAF) material.
12. The synapse element of claim 7 , wherein the fixed layer includes at least one of a metal iridate or an oxidized layer of a ferromagnet (FM)/heavy metal bilayer.
13. A neuromorphic processor comprising:
a synapse element array including a plurality of synapse elements,
wherein each of the plurality of synapse elements includes a first memristor element including
a free layer including a domain wall, and
a fixed layer including a material of which a magnetization direction is fixed,
wherein a position of the domain wall in the free layer is changeable based on a spin orbit torque (SOT) generated by a current injected to the fixed layer from an outside through a first position and output through a second position such that the current traverses through the fixed layer,
wherein a resistance value of the fixed layer between the first and second positions is based on the position of the domain wall and on a Hall voltage,
wherein the Hall voltage is based on the position of the domain wall,
wherein the Hall voltage has at least a lower value, a higher value, and an intermediate value,
wherein the memristor element is configured to store non-binary data based the resistance value, and
wherein a direction of the domain wall position change is parallel to a direction of the injected current that causes the domain wall position to change.
14. The neuromorphic processor of claim 13 , wherein the first memristor element further includes:
a first electrode configured to accept the current introduced from the outside; and
a second electrode configured to provide a voltage pulse to the outside.
15. The neuromorphic processor of claim 13 , wherein
each of the plurality of synapse elements further includes a first transistor,
a first terminal of the first transistor is connected to a word line,
a second terminal of the first transistor is connected to a source line,
a third terminal of the first transistor is connected to one end of the first memristor element, and
another end of the first memristor element is connected to a bit line.
16. The neuromorphic processor of claim 15 , wherein
the neuromorphic processor is configured to supply the current supplied from the outside to the plurality of synapse elements such that the position of the domain wall is moveable based on a signal supplied through the word line, and
wherein the resistance value is changed based on the position of the moveable domain wall.
17. The neuromorphic processor of claim 15 , wherein the neuromorphic processor is configured to derive a result of an artificial neural network operation based on a current output from the bit line.
18. The neuromorphic processor of claim 13 , wherein
each of the plurality of synapse elements further includes a first transistor, a second transistor, and a second memristor element,
one end of the first memristor element is connected to a word line,
another end of the first memristor element is connected to a first node,
one end of the second memristor element is connected to the word line,
another end of the second memristor element is connected to a second node,
a first terminal of the first transistor is connected to the first node,
a second terminal of the first transistor is connected to the second node,
a third terminal of the first transistor is connected to a bit line,
a first terminal of the second transistor is connected to the second node,
a second terminal of the second transistor is connected to the first node, and
a third terminal of the second transistor is connected to the bit line.
19. The neuromorphic processor of claim 13 , wherein the free layer includes a synthetic antiferromagnet (SAF) material.
20. The neuromorphic processor of claim 13 , wherein the fixed layer includes at least one of a metal iridate or an oxidized layer of a ferromagnet (FM)/heavy metal bilayer.