The story so far: • Landauer formula + scattering matrix approach: general way of treating (noninteracting, small bias) two-terminal conductance of quantum coherent system attached to classical reservoirs at absolute zero, independent of details of the quantum system. • Subtle issues about conductance: ballistic system has finite conductance. • Energy relaxation processes typically modeled as taking place in leads or contacts, resulting in very nonthermal / nonequilibrium electronic distributions in “active” region of device.
On the plate today: • Zeroth order effect of interactions • Multiterminal generalization of Landauer formula: the Buttiker formula. • Reciprocity relations • Finite temperature and larger biases • Combining scattering matrices
“Resistivity dipole” - Coulomb interactions When computing chemical potential changes, we showed abrupt changes (a) at interfaces between contacts and leads; and (b) across a scatterer of transmittance T. While µ may change abruptly, we know electrostatic potential cannot, because of screening. Quick accounting for averaged electron-electron interactions: Poisson equation and screening length.
contact 1
contact 2
T lead 1
lead 2
µ 1-T -k states
+k states T x
“Resistivity dipole” - Coulomb interactions Discontinuity in chemical potential leads to smeared discontinuity in electrostatic potential. Charge builds up microscopically like a dipole around the scatterer. Whole system is solved self-consistently. In systems with poor screening, effects of interfaces can be very big!
µ φ screening length electron density -
electric field
+
Buttiker formula (1988) Treats multiple probe measurements such that all probes are on equal footing:
2e I p = ∑ (Tq ← p µ p − Tp ←q µ q ) h q Contributions from scattering with to/from terminals q.
Net current out of terminal p Rewriting
I p = ∑ (GqpV p −G pqVq )
2e 2 G pq ≡ Tp←q h
q
Sum rule (guarantees I = 0 when all V are same):
∑G
qp
q
= ∑ G pq q
I p = ∑ G pq (V p − Vq ) equivalent to Kirchhoff’s law q
Buttiker formula Using this formula, potential of terminal n is determined by potentials of other terminals weighted by transmission functions:
Vn
∑G V = ∑G q≠n
nq q
q≠n
nq
Note that, in general, Gqp ≠ G pq though Gqp (+ B ) = G pq (− B )
“reciprocity” -- not easy to show in general.
Buttiker formula: 4-terminal example ⎛ I1 ⎞ ⎡G12 + G13 + G14 ⎜ ⎟ ⎢ − G21 ⎜ I2 ⎟ ⎢ ⎜I ⎟ = ⎢ − G31 ⎜ 3⎟ ⎢ ⎜I ⎟ − G41 ⎝ 4⎠ ⎣
− G12
− G13
G21 + G23 + G24 − G32
− G23
− G42
− G14
G31 + G32 + G34 − G43
⎤⎛ V1 ⎞ ⎥⎜V ⎟ − G24 ⎥⎜ 2 ⎟ ⎥⎜ V3 ⎟ − G34 ⎥⎜⎜ ⎟⎟ G41 + G42 + G43 ⎦⎝V4 ⎠
Can set V4 = 0 without loss of generality…. ⎛ I1 ⎞ ⎡G12 + G13 + G14 ⎜ ⎟ ⎢ − G21 ⎜ I2 ⎟ = ⎢ ⎜I ⎟ ⎢ − G31 ⎝ 3⎠ ⎣
− G12 G21 + G23 + G24 − G32
− G13 − G23
⎤⎛ V1 ⎞ ⎥⎜V ⎟ ⎥⎜ 2 ⎟ G31 + G32 + G34 ⎥⎦⎜⎝ V3 ⎟⎠