news & views of myosin, such as myosin V and the closely related myosin XI, are plus-end directed because they lack the insert found in myosin VI (Fig. 1b)5. Like myosins, class 14 kinesins (minus-end-directed microtubulebased motors) also use a lever arm to amplify structural changes occurring in the motor domain to drive motility 6. Bryant and colleagues have previously engineered myosin VI to move backwards by changing the length of the lever arm7. To do this, they inserted a domain between the motor domain and the lever arm that could switch between a rigid and a flexible conformation in response to calcium concentration7. This proof-of-principle experiment showed that motor directionality could be controlled by an external stimulus. However, varying calcium concentrations in a localized manner, or over short timescales, is difficult. Optical signals would instead provide the ideal level of control8. Bryant and colleagues now achieve optical control of myosin and kinesin motors by adding a protein domain called LOV (light, oxygen or voltage) that changes conformation in response to blue light 9. The LOV2 domain, in particular, undergoes a conformational change in the blue light that causes the undocking of an alpha helix from the photosensitive domain10 (Fig. 1c). Based on this principle, the researchers designed a series of chimeras in which the LOV2 domain is inserted between the motor and the lever arm of myosin VI, myosin XI or kinesin-14. All of their designs are based on the principle that the LOV2 domain adopts
a rigid hairpin conformation in the dark and an open and more flexible conformation in the light. This switch allows them to control the position of the lever arm relative to the motor cores. For myosin VI, which is normally a minus-end-directed motor (Fig. 1a), the researchers first engineered it to be plus-end directed using an approach they previously developed7 (Fig. 1c, top panel). This myosin VI also contains the LOV2 domain and can switch its directionality from plusto minus-end-directed in response to light (Fig. 1c, bottom panel). Furthermore, a variation of this myosin VI construct was engineered to switch, when irradiated with blue light, from faster to slower plus-enddirected motility. Likewise, myosin XI, normally a fast, plus-end-directed motor (Fig. 1b) could be converted, in response to light, from a slow to fast plus-end-directed motor (Fig. 1d). As a final proof of the versatility of their approach, Bryant and colleagues used a similar strategy to generate kinesin-14 motors that changed direction when illuminated. To determine the speed and directionality of different engineered motor chimeras, Bryant and colleagues performed gliding filament assays, in which myosin or kinesin motors labelled with a fluorescent tag are tethered to glass by the non-motor end of the protein. Fluorescent actin filaments or microtubules, with one end of the filament marked to determine polarity, are then observed as the motors push them along the glass. These light-induced speed changes
occur on the order of seconds and can be repeated more than 50 times. These studies show that motor activity can be controlled reversibly with high spatial and fast temporal resolution. Although the motors used in these studies are much slower than the natural motors they were derived from, the experiments are an exciting starting point to further design myosins and kinesins. For example, the motors used by Bryant and colleagues are monomeric, but both myosin VI and XI can travel for long distances without dissociating from their tracks when dimeric5,11. One of the most exciting potential applications of this technology will be to express lighttunable motors in cells, allowing optical control of motor behaviour in living cells and organisms. ❐ Samara L. Reck-Peterson is in the Department of Cell Biology, Harvard Medical School, Boston, Massachusetts 02115, USA. e-mail: [email protected] References 1. 2. 3. 4. 5. 6.
Kull, F. J. & Endow, S. A. J. Cell Sci. 126, 9–19 (2013). Nakamura, M. et al. Nature Nanotech. 9, 693–697 (2014). Vale, R. D. & Milligan, R. A. Science 288, 88–95 (2000). Menetrey, J. et al. Nature 435, 779–785 (2005). Tominaga, M. et al. EMBO J. 22, 1263–1272 (2003). Endres, N. F., Yoshioka, C., Milligan, R. A. & Vale, R. D. Nature 439, 875–878 (2006). 7. Chen, L., Nakamura, M., Schindler, T. D., Parker, D. & Bryant, Z. Nature Nanotech. 7, 252–256 (2012). 8. Kim, B. & Lin, M. Z. Biochem. Soc. Trans. 41, 1183–1188 (2013). 9. Herrou, J. & Crosson, S. Nature Rev. Microbiol. 9, 713–723 (2011). 10. Harper, S. M., Neil, L. C. & Gardner, K. H. Science 301, 1541–1544 (2003). 11. Rock, R. S. et al. Proc. Natl Acad. Sci. USA 98, 13655–13659 (2001).
SPINTRONICS
Electrons act constructively
Interference effects in semiconductor quantum structures provide an elegant way to electrically map the strength and direction of spin–orbit fields.
Tomas Jungwirth and Jörg Wunderlich
E
lectrons in solid-state systems can behave like waves, and, in both metals and semiconductors, quantumrelativistic effects can influence the motion of the charge carriers. One of these effects is the spin–orbit interaction, which describes the coupling of the orbital and spin degrees of freedom of the electrons, and manifests itself as an effective momentum-dependent magnetic field seen by the carriers. The interference of relativistic, spin–orbit-coupled electrons has in the past been considered primarily 662
in terms of destructive interference1. Furthermore, previous measurements of the strength of the spin–orbit coupling by phase coherent transport could only return approximate values. Writing in Nature Nanotechnology, Junsaku Nitta and colleagues at Tohoku University and the University of Regensburg now report observing constructive interference effects in relativistic electrons in semiconductor quantum structures, and use these effects to precisely map the relativistic field, including its momentum-dependent direction2.
When electrons in a conductor scatter off intentionally introduced dopants, alloy disorder or unintentional impurities the constructive interference of backscattered electron waves ψ+ and ψ– (corresponding to time-reversed paths) leads to a phenomenon known as weak localization (Fig. 1a). The backscattering probability given by |ψ+ + ψ–|2 has the incoherent contribution |ψ+|2 + |ψ–|2 and an interference contribution ψ*+ψ– + ψ*–ψ+. In the absence of a magnetic field, which breaks time-reversal symmetry, the phase shift acquired by the two
NATURE NANOTECHNOLOGY | VOL 9 | SEPTEMBER 2014 | www.nature.com/naturenanotechnology
© 2014 Macmillan Publishers Limited. All rights reserved
news & views counter-propagating waves is the same, the interference is constructive, and the probability of backscattering is doubled by the interference contribution. This increases the localization of the carriers and, therefore, the resistivity of the sample. In early studies of weak localization, physicists were already intrigued by the role of spin and spin–orbit coupling on coherent transport 1. The spin–orbit interaction can be viewed as a momentum-dependent Zeeman coupling of an effective magnetic field Beff to the electron spin. During each scattering event, Beff changes its direction, causing the spin to precess. The constructive interference (and weak localization) persists only for weak spin–orbit coupling; in that case, the spin–orbit coupling time, τso ~1/Beff, is much larger than the spin dephasing time, τi, due to inelastic scattering. Naively, for a strong spin–orbit coupling for which τso τB. With decreasing Bp, the constructive interference contribution strengthens, resulting in an increase of the measured resistance. When, however, τB >> τso at small Bp, weak anti-localization takes over and the resistance drops. The resulting maximum in the resistance at τB ≈ τso, marking the crossover from weak localization to weak anti-localization, has traditionally provided the means for inferring the approximate strength of the spin–orbit coupling 1. Nitta and colleagues carried out an experiment in which the interference contribution to electron transport can provide not just an approximate scale of the
a
b
2D
2D
ky kx Beff WAL
WL Ψ+ Ψ−
c
d
1D
Bin + Beff
1D
ky Bin WL
Deph.
Bin + Beff
Beff
Figure 1 | Interference transport effects in 2D and 1D spin–orbit-coupled systems. a, Constructive interference of backscattered electron waves (ψ+ and ψ-) passing along time-reversed 2D paths leads to weak localization (WL) in the absence of spin–orbit coupling. Red and blue arrows represent scattering events of the two waves. The inset is a zoom-in of the interference occurring at the scattering site. b, Left: Spins randomly precess when scattered in the presence of the momentum-dependent spin–orbit field Beff. The spin–orbit coupling leads to a prevailing destructive interference of backscattered electron waves passing along a time-reversed 2D path and to the weak anti-localization (WAL). Right: The electron spin (light blue arrow) precesses around the effective field Beff. The direction of Beff in momentum space kx–ky is also shown. c, The variation of the direction of Beff is quenched in 1D wires (the direction of Beff in momentum space depends only on ky; right) restoring the constructive interference and weak localization despite the presence of spin–orbit coupling. d, Right: Adding a Zeeman coupling of an in-plane magnetic field Bin breaks time-reversal and introduces a change in the direction of the net spin-precession axis Beff + Bin when the carrier scatters. Left: The resulting dephasing (Deph.) of backscattered waves suppresses weak localization.
spin–orbit coupling, but also an accurate measurement of its momentum-dependent direction in different semiconductor quantum structures. Unexpectedly, weak anti-localization plays no role in these experiments. Indeed, in structures with strong spin–orbit coupling, the researchers could rely solely on weak localization, and on tuning its contribution to transport by controlling the dephasing, following a concept previously proposed by the group3. The researchers studied transport in a two-dimensional (2D) electron gas in an InGaAs-based heterostructure by patterning microwires with widths smaller than the characteristic spin-precession length in the spin–orbit field. This quasi-1D character of transport implies that the carrier spins experience only the spin–orbit-field component given by the momentum component along the wire. Backscattering in this geometry changes the sign of the spin–orbit field, but does not rotate its spin-precession axis. The randomization of
NATURE NANOTECHNOLOGY | VOL 9 | SEPTEMBER 2014 | www.nature.com/naturenanotechnology
© 2014 Macmillan Publishers Limited. All rights reserved
spin rotations is suppressed in the quasi-1D wire and the interference contribution to transport has the weak localization form, despite the strong spin–orbit coupling (Fig. 1c). The researchers applied an in-plane magnetic field Bin, which couples to the carrier spin. Because this momentumindependent Zeeman coupling breaks time-reversal symmetry, the spin-precession axis given by Beff + Bin rotates when the carrier is scattered in the quasi-1D channel, and the resulting dephasing suppresses weak localization (Fig. 1d). The dephasing is maximized when Bin is orthogonal to Beff and minimized when the two fields are parallel. The researchers studied a semiconductor heterostructure in which the momentumdependent spin–orbit field has two types of spin–orbit coupling with different symmetry, called Rashba and Dresselhaus. The direction of the total Rashba–Dresselhaus field is determined by the ratio of their respective strengths, α and β. These can be tuned 663
news & views independently, for example by applying a voltage on a gate electrode. The method used by Nitta and colleagues gives precise information on the direction of the total spin– orbit field and, therefore, also of the α/β ratio. As well as introducing a new intriguing concept in the physics of interference transport phenomena, the method has practical implications for the research of spintronic devices. Spin-transistor structures, in which electrical control of on and off states is achieved using the spin of the electrons flowing in the transistor channel4–6, rely on coherent precession of
spins of carriers injected into the channel. When α/β = 1, the spin–orbit-field axis becomes independent of momentum, and confining the electrons to the narrow quasi-1D channel is not required to preserve spin coherence7. The method developed by Nitta and colleagues allows this optimal spin–orbit-coupling field to be found and to be used to manipulate spins without destroying the spin information. ❐ Tomas Jungwirth and Jörg Wunderlich are at the Institute of Physics ASCR, v.v.i., Cukrovarnicka 10, 162 53 Praha 6, Czech Republic. Tomas Jungwirth
is also at the School of Physics and Astronomy, University of Nottingham, Nottingham NG7 2RD, UK. Jörg Wunderlich is also at the Hitachi Cambridge Laboratory, Cambridge CB3 0HE, UK. e-mail: [email protected] References
1. Bergmann, G. Solid State Commun. 42, 815–817 (1982). 2. Sasaki, A. et al. Nature Nanotech. 9, 703–709 (2014). 3. Scheid, M., Kohda, M., Kunihashi, Y., Richter, K. & Nitta, J. Phys. Rev. Lett. 101, 266401 (2008). 4. Datta, S. & Das, B. Appl. Phys. Lett. 56, 665–667 (1990). 5. Koo, H. C. et al. Science 325, 1515–1518 (2009). 6. Wunderlich, J. et al. Science 330, 1801–1804 (2010). 7. Bernevig, B. A., Orenstein, J. & Zhang, S.‑C. Phys. Rev. Lett. 97, 236601 (2006).
TWO-DIMENSIONAL MATERIALS
Atomically thin p–n junctions
Van der Waals heterostructures consisting of a single MoS2 monolayer and a single WSe2 monolayer can be used to form p–n junctions.
Su-Fei Shi and Feng Wang
A
p–n junction is formed at the boundary between a hole-doped (p-type) and electron-doped (n-type) semiconductor, and creates an electrical field due to electron transfer from the n-type region to the p-type region. This built-in electrical field leads to diode behaviour, which is characterized by asymmetric charge transport under forward and backward biases. Such semiconductor diodes are ubiquitous in modern electronics and optoelectronics, ranging from rectifying diodes, frequency
mixing diodes and tunnelling diodes to photodiodes, light-emitting diodes and laser diodes. A key parameter in semiconductor diodes is the thickness of the p–n junction, which is typically 10 nm to 1 μm depending on the diode type. Atomically thin layers of two-dimensional (2D) semiconductors, such as transition metal dichalcogenides, offer opportunities for engineering novel semiconductor p–n junctions. Writing in Nature Nanotechnology, Philip Kim and co-workers at Columbia University,
b
a
MoS2
WSe2 Ec
+
+
+
−
−
−
Ec Ev Ev
Figure 1 | Atomically thin p–n junctions in van der Waals heterostructures. a, Schematic of the theoretical band alignment of WSe2 and MoS2 (Ec, bottom of conduction band; Ev, top of valence band). A type-II junction is formed in the stacked MoS2–WSe2 heterostructure. b, Schematic of the p–n junction formed by a monolayer of WSe2 and a monolayer of MoS2. The blue, yellow, purple and green spheres represent molybdenum (Mo), sulphur (S), tungsten (W) and selenium (Se) atoms, respectively. Electrons are transferred from MoS2 to WSe2, generating the illustrated charge distribution (red, positive charge; blue, negative charge) and a built-in electrical field in the heterostructure. 664
Korea University, Yonsei University and the University of Florida now report the thinnest possible p–n junction, formed in a van der Waals stacked heterostructure composed of one WSe2 monolayer and one MoS2 monolayer 1. Monolayers of MoS2 and WSe2 are direct-bandgap semiconductors with unique optical and electrical optical properties2,3. Recently, it has been demonstrated that lateral p–n junctions can be created electrostatically in a monolayer of WSe2 (refs 4–6). While these lateral p–n junctions exhibit interesting phenomena such as electroluminescence, their width is defined by the charge depletion region, as in conventional semiconductor diodes. Advances in 2D layered materials have also allowed vertically stacked heterostructures of graphene, boron nitride and transition metal dichalcogenides to be fabricated7. Kim and colleagues have used such methods to create layers of the different transition metal dichalcogenides. These layers of MoS2 and WSe2 have significant band offset and form a type-II heterojunction when stacked together (Fig. 1a). MoS2 layers are naturally n-doped, whereas WSe2 layers are p-doped; the MoS2–WSe2 van der Waals heterostructure creates vertical p–n junctions with built-in electrical fields that are confined within a thickness of merely two unit cells (Fig. 1b). The atomically thin MoS2–WSe2 p–n junctions show highly asymmetric current–voltage characteristics, similar
NATURE NANOTECHNOLOGY | VOL 9 | SEPTEMBER 2014 | www.nature.com/naturenanotechnology
© 2014 Macmillan Publishers Limited. All rights reserved
a rigid hairpin conformation in the dark and an open and more flexible conformation in the light. This switch allows them to control the position of the lever arm relative to the motor cores. For myosin VI, which is normally a minus-end-directed motor (Fig. 1a), the researchers first engineered it to be plus-end directed using an approach they previously developed7 (Fig. 1c, top panel). This myosin VI also contains the LOV2 domain and can switch its directionality from plusto minus-end-directed in response to light (Fig. 1c, bottom panel). Furthermore, a variation of this myosin VI construct was engineered to switch, when irradiated with blue light, from faster to slower plus-enddirected motility. Likewise, myosin XI, normally a fast, plus-end-directed motor (Fig. 1b) could be converted, in response to light, from a slow to fast plus-end-directed motor (Fig. 1d). As a final proof of the versatility of their approach, Bryant and colleagues used a similar strategy to generate kinesin-14 motors that changed direction when illuminated. To determine the speed and directionality of different engineered motor chimeras, Bryant and colleagues performed gliding filament assays, in which myosin or kinesin motors labelled with a fluorescent tag are tethered to glass by the non-motor end of the protein. Fluorescent actin filaments or microtubules, with one end of the filament marked to determine polarity, are then observed as the motors push them along the glass. These light-induced speed changes
occur on the order of seconds and can be repeated more than 50 times. These studies show that motor activity can be controlled reversibly with high spatial and fast temporal resolution. Although the motors used in these studies are much slower than the natural motors they were derived from, the experiments are an exciting starting point to further design myosins and kinesins. For example, the motors used by Bryant and colleagues are monomeric, but both myosin VI and XI can travel for long distances without dissociating from their tracks when dimeric5,11. One of the most exciting potential applications of this technology will be to express lighttunable motors in cells, allowing optical control of motor behaviour in living cells and organisms. ❐ Samara L. Reck-Peterson is in the Department of Cell Biology, Harvard Medical School, Boston, Massachusetts 02115, USA. e-mail: [email protected] References 1. 2. 3. 4. 5. 6.
Kull, F. J. & Endow, S. A. J. Cell Sci. 126, 9–19 (2013). Nakamura, M. et al. Nature Nanotech. 9, 693–697 (2014). Vale, R. D. & Milligan, R. A. Science 288, 88–95 (2000). Menetrey, J. et al. Nature 435, 779–785 (2005). Tominaga, M. et al. EMBO J. 22, 1263–1272 (2003). Endres, N. F., Yoshioka, C., Milligan, R. A. & Vale, R. D. Nature 439, 875–878 (2006). 7. Chen, L., Nakamura, M., Schindler, T. D., Parker, D. & Bryant, Z. Nature Nanotech. 7, 252–256 (2012). 8. Kim, B. & Lin, M. Z. Biochem. Soc. Trans. 41, 1183–1188 (2013). 9. Herrou, J. & Crosson, S. Nature Rev. Microbiol. 9, 713–723 (2011). 10. Harper, S. M., Neil, L. C. & Gardner, K. H. Science 301, 1541–1544 (2003). 11. Rock, R. S. et al. Proc. Natl Acad. Sci. USA 98, 13655–13659 (2001).
SPINTRONICS
Electrons act constructively
Interference effects in semiconductor quantum structures provide an elegant way to electrically map the strength and direction of spin–orbit fields.
Tomas Jungwirth and Jörg Wunderlich
E
lectrons in solid-state systems can behave like waves, and, in both metals and semiconductors, quantumrelativistic effects can influence the motion of the charge carriers. One of these effects is the spin–orbit interaction, which describes the coupling of the orbital and spin degrees of freedom of the electrons, and manifests itself as an effective momentum-dependent magnetic field seen by the carriers. The interference of relativistic, spin–orbit-coupled electrons has in the past been considered primarily 662
in terms of destructive interference1. Furthermore, previous measurements of the strength of the spin–orbit coupling by phase coherent transport could only return approximate values. Writing in Nature Nanotechnology, Junsaku Nitta and colleagues at Tohoku University and the University of Regensburg now report observing constructive interference effects in relativistic electrons in semiconductor quantum structures, and use these effects to precisely map the relativistic field, including its momentum-dependent direction2.
When electrons in a conductor scatter off intentionally introduced dopants, alloy disorder or unintentional impurities the constructive interference of backscattered electron waves ψ+ and ψ– (corresponding to time-reversed paths) leads to a phenomenon known as weak localization (Fig. 1a). The backscattering probability given by |ψ+ + ψ–|2 has the incoherent contribution |ψ+|2 + |ψ–|2 and an interference contribution ψ*+ψ– + ψ*–ψ+. In the absence of a magnetic field, which breaks time-reversal symmetry, the phase shift acquired by the two
NATURE NANOTECHNOLOGY | VOL 9 | SEPTEMBER 2014 | www.nature.com/naturenanotechnology
© 2014 Macmillan Publishers Limited. All rights reserved
news & views counter-propagating waves is the same, the interference is constructive, and the probability of backscattering is doubled by the interference contribution. This increases the localization of the carriers and, therefore, the resistivity of the sample. In early studies of weak localization, physicists were already intrigued by the role of spin and spin–orbit coupling on coherent transport 1. The spin–orbit interaction can be viewed as a momentum-dependent Zeeman coupling of an effective magnetic field Beff to the electron spin. During each scattering event, Beff changes its direction, causing the spin to precess. The constructive interference (and weak localization) persists only for weak spin–orbit coupling; in that case, the spin–orbit coupling time, τso ~1/Beff, is much larger than the spin dephasing time, τi, due to inelastic scattering. Naively, for a strong spin–orbit coupling for which τso τB. With decreasing Bp, the constructive interference contribution strengthens, resulting in an increase of the measured resistance. When, however, τB >> τso at small Bp, weak anti-localization takes over and the resistance drops. The resulting maximum in the resistance at τB ≈ τso, marking the crossover from weak localization to weak anti-localization, has traditionally provided the means for inferring the approximate strength of the spin–orbit coupling 1. Nitta and colleagues carried out an experiment in which the interference contribution to electron transport can provide not just an approximate scale of the
a
b
2D
2D
ky kx Beff WAL
WL Ψ+ Ψ−
c
d
1D
Bin + Beff
1D
ky Bin WL
Deph.
Bin + Beff
Beff
Figure 1 | Interference transport effects in 2D and 1D spin–orbit-coupled systems. a, Constructive interference of backscattered electron waves (ψ+ and ψ-) passing along time-reversed 2D paths leads to weak localization (WL) in the absence of spin–orbit coupling. Red and blue arrows represent scattering events of the two waves. The inset is a zoom-in of the interference occurring at the scattering site. b, Left: Spins randomly precess when scattered in the presence of the momentum-dependent spin–orbit field Beff. The spin–orbit coupling leads to a prevailing destructive interference of backscattered electron waves passing along a time-reversed 2D path and to the weak anti-localization (WAL). Right: The electron spin (light blue arrow) precesses around the effective field Beff. The direction of Beff in momentum space kx–ky is also shown. c, The variation of the direction of Beff is quenched in 1D wires (the direction of Beff in momentum space depends only on ky; right) restoring the constructive interference and weak localization despite the presence of spin–orbit coupling. d, Right: Adding a Zeeman coupling of an in-plane magnetic field Bin breaks time-reversal and introduces a change in the direction of the net spin-precession axis Beff + Bin when the carrier scatters. Left: The resulting dephasing (Deph.) of backscattered waves suppresses weak localization.
spin–orbit coupling, but also an accurate measurement of its momentum-dependent direction in different semiconductor quantum structures. Unexpectedly, weak anti-localization plays no role in these experiments. Indeed, in structures with strong spin–orbit coupling, the researchers could rely solely on weak localization, and on tuning its contribution to transport by controlling the dephasing, following a concept previously proposed by the group3. The researchers studied transport in a two-dimensional (2D) electron gas in an InGaAs-based heterostructure by patterning microwires with widths smaller than the characteristic spin-precession length in the spin–orbit field. This quasi-1D character of transport implies that the carrier spins experience only the spin–orbit-field component given by the momentum component along the wire. Backscattering in this geometry changes the sign of the spin–orbit field, but does not rotate its spin-precession axis. The randomization of
NATURE NANOTECHNOLOGY | VOL 9 | SEPTEMBER 2014 | www.nature.com/naturenanotechnology
© 2014 Macmillan Publishers Limited. All rights reserved
spin rotations is suppressed in the quasi-1D wire and the interference contribution to transport has the weak localization form, despite the strong spin–orbit coupling (Fig. 1c). The researchers applied an in-plane magnetic field Bin, which couples to the carrier spin. Because this momentumindependent Zeeman coupling breaks time-reversal symmetry, the spin-precession axis given by Beff + Bin rotates when the carrier is scattered in the quasi-1D channel, and the resulting dephasing suppresses weak localization (Fig. 1d). The dephasing is maximized when Bin is orthogonal to Beff and minimized when the two fields are parallel. The researchers studied a semiconductor heterostructure in which the momentumdependent spin–orbit field has two types of spin–orbit coupling with different symmetry, called Rashba and Dresselhaus. The direction of the total Rashba–Dresselhaus field is determined by the ratio of their respective strengths, α and β. These can be tuned 663
news & views independently, for example by applying a voltage on a gate electrode. The method used by Nitta and colleagues gives precise information on the direction of the total spin– orbit field and, therefore, also of the α/β ratio. As well as introducing a new intriguing concept in the physics of interference transport phenomena, the method has practical implications for the research of spintronic devices. Spin-transistor structures, in which electrical control of on and off states is achieved using the spin of the electrons flowing in the transistor channel4–6, rely on coherent precession of
spins of carriers injected into the channel. When α/β = 1, the spin–orbit-field axis becomes independent of momentum, and confining the electrons to the narrow quasi-1D channel is not required to preserve spin coherence7. The method developed by Nitta and colleagues allows this optimal spin–orbit-coupling field to be found and to be used to manipulate spins without destroying the spin information. ❐ Tomas Jungwirth and Jörg Wunderlich are at the Institute of Physics ASCR, v.v.i., Cukrovarnicka 10, 162 53 Praha 6, Czech Republic. Tomas Jungwirth
is also at the School of Physics and Astronomy, University of Nottingham, Nottingham NG7 2RD, UK. Jörg Wunderlich is also at the Hitachi Cambridge Laboratory, Cambridge CB3 0HE, UK. e-mail: [email protected] References
1. Bergmann, G. Solid State Commun. 42, 815–817 (1982). 2. Sasaki, A. et al. Nature Nanotech. 9, 703–709 (2014). 3. Scheid, M., Kohda, M., Kunihashi, Y., Richter, K. & Nitta, J. Phys. Rev. Lett. 101, 266401 (2008). 4. Datta, S. & Das, B. Appl. Phys. Lett. 56, 665–667 (1990). 5. Koo, H. C. et al. Science 325, 1515–1518 (2009). 6. Wunderlich, J. et al. Science 330, 1801–1804 (2010). 7. Bernevig, B. A., Orenstein, J. & Zhang, S.‑C. Phys. Rev. Lett. 97, 236601 (2006).
TWO-DIMENSIONAL MATERIALS
Atomically thin p–n junctions
Van der Waals heterostructures consisting of a single MoS2 monolayer and a single WSe2 monolayer can be used to form p–n junctions.
Su-Fei Shi and Feng Wang
A
p–n junction is formed at the boundary between a hole-doped (p-type) and electron-doped (n-type) semiconductor, and creates an electrical field due to electron transfer from the n-type region to the p-type region. This built-in electrical field leads to diode behaviour, which is characterized by asymmetric charge transport under forward and backward biases. Such semiconductor diodes are ubiquitous in modern electronics and optoelectronics, ranging from rectifying diodes, frequency
mixing diodes and tunnelling diodes to photodiodes, light-emitting diodes and laser diodes. A key parameter in semiconductor diodes is the thickness of the p–n junction, which is typically 10 nm to 1 μm depending on the diode type. Atomically thin layers of two-dimensional (2D) semiconductors, such as transition metal dichalcogenides, offer opportunities for engineering novel semiconductor p–n junctions. Writing in Nature Nanotechnology, Philip Kim and co-workers at Columbia University,
b
a
MoS2
WSe2 Ec
+
+
+
−
−
−
Ec Ev Ev
Figure 1 | Atomically thin p–n junctions in van der Waals heterostructures. a, Schematic of the theoretical band alignment of WSe2 and MoS2 (Ec, bottom of conduction band; Ev, top of valence band). A type-II junction is formed in the stacked MoS2–WSe2 heterostructure. b, Schematic of the p–n junction formed by a monolayer of WSe2 and a monolayer of MoS2. The blue, yellow, purple and green spheres represent molybdenum (Mo), sulphur (S), tungsten (W) and selenium (Se) atoms, respectively. Electrons are transferred from MoS2 to WSe2, generating the illustrated charge distribution (red, positive charge; blue, negative charge) and a built-in electrical field in the heterostructure. 664
Korea University, Yonsei University and the University of Florida now report the thinnest possible p–n junction, formed in a van der Waals stacked heterostructure composed of one WSe2 monolayer and one MoS2 monolayer 1. Monolayers of MoS2 and WSe2 are direct-bandgap semiconductors with unique optical and electrical optical properties2,3. Recently, it has been demonstrated that lateral p–n junctions can be created electrostatically in a monolayer of WSe2 (refs 4–6). While these lateral p–n junctions exhibit interesting phenomena such as electroluminescence, their width is defined by the charge depletion region, as in conventional semiconductor diodes. Advances in 2D layered materials have also allowed vertically stacked heterostructures of graphene, boron nitride and transition metal dichalcogenides to be fabricated7. Kim and colleagues have used such methods to create layers of the different transition metal dichalcogenides. These layers of MoS2 and WSe2 have significant band offset and form a type-II heterojunction when stacked together (Fig. 1a). MoS2 layers are naturally n-doped, whereas WSe2 layers are p-doped; the MoS2–WSe2 van der Waals heterostructure creates vertical p–n junctions with built-in electrical fields that are confined within a thickness of merely two unit cells (Fig. 1b). The atomically thin MoS2–WSe2 p–n junctions show highly asymmetric current–voltage characteristics, similar
NATURE NANOTECHNOLOGY | VOL 9 | SEPTEMBER 2014 | www.nature.com/naturenanotechnology
© 2014 Macmillan Publishers Limited. All rights reserved
