Measur asurem emen ents of RF noi ts of RF noise i e in InG - - PowerPoint PPT Presentation
Measur asurem emen ents of RF noi ts of RF noise i e in InG InGaAs/InAlA As/InAlAs r s recessed diodes: cessed diodes: Sign Signatures of shot-noise suppr es of shot-noise suppression ssion O. Garca-Prez, J. Mateos , S. Prez, T.
1
2
Uncorrelated carriers Correlated carriers Shot-noise SI=qI full shot noise = 1 suppressed shot noise 1 enhanced shot noise 1 Origin of correlations: Pauli exclusion principle (degenerate semiconductors) Long-range Coulomb interaction (strong space-charge effects) Fano Factor = SI qI The measurement of shot noise and the value of its corresponding Fano Factor can provide valuable insight about the transport dynamics inside semiconductor devices Shot-noise SI=qI is related to the discrete character of the electronic charge and is usually observed in electronic devices when carrier transport is ballistic or is limited by an energy barrier (Schottky diodes, tunnel diodes…)
Javier Mateos – Universidad de Salamanca
x / L 0.0 0.2 0.4 0.6 0.8 1.0 Potential (KBT/q)
5 10 15
qU/KBT 40 20 10 4 80
nc nc
thermal equilibrium thermal equilibrium
x=L N x=0
Active region Contact 1 Contact 2
U
Poissonian injection Poissonian injection
Ballistic diode Shot noise suppression due to long range Coulomb correlations
high important space-charge effects
Dc
L L
Full shot noise is recovered when the barrier disappears
Presence of a barrier that disappears when voltage is increased
3
Javier Mateos – Universidad de Salamanca
4
x / L 0.0 0.2 0.4 0.6 0.8 1.0 Potential (KBT/q)
5 10 15
qU/KBT 40 20 10 4 80
nc nc
thermal equilibrium thermal equilibrium
x=L N x=0
Active region Contact 1 Contact 2
U
Poissonian injection Poissonian injection
Position(um)
0.0 0.1 0.2 0.3 0.4 0.5 0.6
Potential (V)
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
Ballistic diode
Presence of a barrier that disappears when voltage is increased
Recessed planar diodes=Slot diodes (ungated HEMTs)
Shot noise suppression expected Quasi-ballistic transport in the channel
Potential barrier imposed by the surface charges at the recess Javier Mateos – Universidad de Salamanca
Source Drain Recess LR LS LD cap
InGaAs InGaAs InAlAs InAlAs
Channel
5
Parameter Reference design [nm] Testedrange [nm] Ls 200 200-800 Lr 200 200-800 Ld 550 300-1000
Slot diodes with different geometries have been fabricated and characterized
2-8 nm
Diodes without recess
+
Javier Mateos – Universidad de Salamanca
6
Cap layer
Source Drain
Ls Lr Ld
Applied voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Current density (x10
3 A/m)
0.0 0.2 0.4 0.6 0.8 1.0 1.2 LR=160 nm (Exp.) LR=400 nm (Exp.) LR=800 nm (Exp.) LR=160 nm (MC) LR=400 nm (MC) LR=800 nm (MC) Applied voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Current density (x10
3 A/m)
0.0 0.2 0.4 0.6 0.8 1.0 1.2 LR=160 nm (Exp.) LR=400 nm (Exp.) LR=800 nm (Exp.) LR=160 nm (MC) LR=400 nm (MC) LR=800 nm (MC)
Batch C7, type 6 - Cap 2nm
Applied voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Current density (x10
3 A/m)
0.0 0.2 0.4 0.6 0.8 1.0 1.2 LR=160 nm (Exp.) LR=400 nm (Exp.) LR=800 nm (Exp.) LR=160 nm (MC) LR=400 nm (MC) LR=800 nm (MC)
Batch C7, type 2 - Cap 4nm Batch C7, type 5 - Cap 8nm
Ls = 200nm Ld = 550nm 160nm<Lr<800nm
Contact Resistance (0.35 Ω.mm)
+ MC simulations reproduce the experimental DC curves and their dependence on Ls, Lr and Ld by adjusting the values of the surface charges and including the ohmic contact resistances
Javier Mateos – Universidad de Salamanca
7
MC simulations reproduce the experimental DC curves and their dependence on Ls, Lr and Ld by adjusting the values of the surface charges and including the ohmic contact resistances
Cap layer
Source Drain
Ls Lr Ld
Contact Resistance (0.35 Ω.mm)
+ Ls = 200nm Lr = 160nm 300nm<Ld<1000nm
Applied voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Current density (x103 A/m) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 LD=300 nm (Exp.) LD=550 nm (Exp.) LD=1000 nm (Exp.) LD=300 nm (MC) LD=550 nm (MC) LD=1000 nm (MC) Applied voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Current density (x103 A/m) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 LD=300 nm (Exp.) LD=550 nm (Exp.) LD=1000 nm (Exp.) LD=300 nm (MC) LD=550 nm (MC) LD=1000 nm (MC) Applied voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Current density (x103 A/m) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 LD=300 nm (Exp.) LD=550 nm (Exp.) LD=1000 nm (Exp.) LD=300 nm (MC) LD=550 nm (MC) LD=1000 nm (MC)
Batch C7, type 6 - Cap 2nm Batch C7, type 2 - Cap 4nm Batch C7, type 5 - Cap 8nm
Javier Mateos – Universidad de Salamanca
8
Reference plane Device PNA-X N5244 VDC S11 + Power density
Measurements performed on wafer with a VNA Agilent PNA-X N5244 (with dedicated receivers for high sensitivity noise power measurements) in the range between 20 and 30 GHz.
Javier Mateos – Universidad de Salamanca
9
Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Current (mA) 5 10 15 20
LD=300nm LD=550nm LD=1000nm No recess (1300nm)
Current (mA) 2 4 6 8 10 12 14 16 SI(x10
2/Hz)
0.0 0.5 1.0 1.5 2.0 2.5 3.0
LD=300nm LD=550nm LD=1000nm No recess (1300nm) 2qI
Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Fano Factor 0.1 1 10
LD=300nm LD=550nm LD=1000nm No recess (1300nm)
Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Current (mA) 5 10 15 20
LR=160nm LR=400nm LR=800nm No recess (1300nm)
Current (mA) 2 4 6 8 10 12 14 16 SI (x10
2/Hz)
0.0 0.5 1.0 1.5 2.0 2.5 3.0
LR=160nm LR=400nm LR=800nm No recess (1300nm) 2qI
Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Fano Factor 0.1 1 10
LR=160nm LR=400nm LR=800nm No recess (1300nm)
Ls = 200 nm, Lr = 160 nm, Ld = 300, 500, 1000 nm Ls = 200 nm, Ld = 550 nm, Lr = 160, 400, 800 nm x Non-recessed diode with L = 1.3 m
Full Shot Noise Full Shot Noise
Javier Mateos – Universidad de Salamanca
10
Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Current (mA) 5 10 15 20
LD=300nm LD=550nm LD=1000nm No recess (1300nm)
Current (mA) 2 4 6 8 10 12 14 16 SI(x10
2/Hz)
0.0 0.5 1.0 1.5 2.0 2.5 3.0
LD=300nm LD=550nm LD=1000nm No recess (1300nm) 2qI
Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Fano Factor 0.1 1 10
LD=300nm LD=550nm LD=1000nm No recess (1300nm)
Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Current (mA) 5 10 15 20
LR=160nm LR=400nm LR=800nm No recess (1300nm)
Current (mA) 2 4 6 8 10 12 14 16 SI (x10
2/Hz)
0.0 0.5 1.0 1.5 2.0 2.5 3.0
LR=160nm LR=400nm LR=800nm No recess (1300nm) 2qI
Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Fano Factor 0.1 1 10
LR=160nm LR=400nm LR=800nm No recess (1300nm)
Ls = 200 nm, Lr = 160 nm, Ld = 300, 500, 1000 nm Ls = 200 nm, Ld = 550 nm, Lr = 160, 400, 800 nm x Non-recessed diode with L = 1.3 m Noise level much lower than full shot noise level = shot noise suppression Increase of noise at high bias, when current saturation starts (not visible without recess)
Full Shot Noise Full Shot Noise
Javier Mateos – Universidad de Salamanca
11
Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Current (mA) 5 10 15 20
LD=300nm LD=550nm LD=1000nm No recess (1300nm)
Current (mA) 2 4 6 8 10 12 14 16 SI(x10
2/Hz)
0.0 0.5 1.0 1.5 2.0 2.5 3.0
LD=300nm LD=550nm LD=1000nm No recess (1300nm) 2qI
Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Fano Factor 0.1 1 10
LD=300nm LD=550nm LD=1000nm No recess (1300nm)
Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Current (mA) 5 10 15 20
LR=160nm LR=400nm LR=800nm No recess (1300nm)
Current (mA) 2 4 6 8 10 12 14 16 SI (x10
2/Hz)
0.0 0.5 1.0 1.5 2.0 2.5 3.0
LR=160nm LR=400nm LR=800nm No recess (1300nm) 2qI
Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Fano Factor 0.1 1 10
LR=160nm LR=400nm LR=800nm No recess (1300nm)
Ls = 200 nm, Lr = 160 nm, Ld = 300, 500, 1000 nm Ls = 200 nm, Ld = 550 nm, Lr = 160, 400, 800 nm x Non-recessed diode with L = 1.3 m Stronger shot noise suppression for longer Ld (and not depending on Lr)
Full Shot Noise Full Shot Noise
Javier Mateos – Universidad de Salamanca
Ohmic region Thermal noise Always at equilibrium Ss= 4KBT Rc+Rs Ohmic region Thermal noise Bias dependent resistance Sd= 4KBT Rd(V) Barrier limited transport Shot noise Sr=F·2qI Rc+Rs Sr Ss Sd Rr Rd(V)
12
Javier Mateos – Universidad de Salamanca
Position(um)
0.0 0.1 0.2 0.3 0.4 0.5 0.6
Potential (V)
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
Rc+Rs = 30 Ss= 4KBT Rc+Rs with T = 300 K Rr = ? Si=F·2qI Rd = ? Sd= 4KBT Rd with T = ?
Rc+Rs= 30
R V V Si
Si=4kT0/Rc+Rs
Rc+Rs Sr Ss Sd Rr Rd(V) Total resistance is measured RT(V)=Rc+Rs+Rr+Rd is known The values of Rr and Rc can be estimated from Monte Carlo simulations
Rr decreases and Rc strongly increases when current saturates (due to intervalley scattering) Complex non-linear behavior Noise?
13
Javier Mateos – Universidad de Salamanca
Rc+Rs Sr Ss Sd Rr Rd(V)
Rr
R V V Si
Si=g2qI Rd
R V V Si
Si=4kT0/Rd(0) Noise in Rd is initially supposed to be constant (equilibrium value of Rd)
Rc+Rs = 30 Ss= 4KBT Rc+Rs with T = 300 K Rr = ? Si=F·2qI Rd = ? Sd= 4KBT Rd with T = ?
Rc+Rs= 30
R V V Si
Si=4kT0/Rc+Rs Simplified noise model
14
Full/suppressed shot noise in the active region Simplified noise model Javier Mateos – Universidad de Salamanca
Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Fano Factor 0.1 1 10
LR=160nm LR=400nm LR=800nm Model Exp.
Current (mA) 2 4 6 8 10 12 SI (10
2/Hz)
0.0 0.5 1.0 1.5 2.0 Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Fano Factor 0.1 1 10
LD=300nm LD=550nm LD=1000nm Model Exp.
Current (mA) 2 4 6 8 10 12 SI(10
2/Hz)
0.0 0.5 1.0 1.5 2.0
15
LD=300nm LD=550nm LD=1000nm Model Exp.
Current (mA) 2 4 6 8 10 12 SI(10
2/Hz)
0.0 0.5 1.0 1.5 2.0 Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Fano Factor 0.1 1 10
LR=160nm LR=400nm LR=800nm Model Exp.
Current (mA) 2 4 6 8 10 12 SI (10
2/Hz)
0.0 0.5 1.0 1.5 2.0 Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Fano Factor 0.1 1 10
Ls = 200 nm, Lr = 160 nm, Ld = 300, 500, 1000 nm Ls = 200 nm, Ld = 550 nm, Lr = 160, 400, 800 nm
Full Shot Noise Full Shot Noise
Full shot noise in the active region F=1
The simplified noise model provide good qualitative agreement with the experimental results (that could be improved by increasing the noise temperature of both contact and drain resistances)
Javier Mateos – Universidad de Salamanca
Voltage (V) 0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 Fano Factor 0,1 1 Voltage (V) 0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 Fano Factor 0,1 1 Fano Factor 0,1 1 Voltage (V) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Fano Factor 0.1 1 Voltage (V) 0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 Fano Factor 0,1 1 Fano Factor 0,1 1
LD=300nm LD=550nm LD=1000nm Model Exp.
Current (mA) 2 4 6 8 10 12 SI(10
2/Hz)
0.0 0.5 1.0 1.5 2.0
LR=160nm LR=400nm LR=800nm Model Exp.
Current (mA) 2 4 6 8 10 12 SI (10
2/Hz)
0.0 0.5 1.0 1.5 2.0
LR=160nm LR=400nm LR=800nm Model Exp.
Current (mA) 2 4 6 8 10 12 SI (10
2/Hz)
0.0 0.5 1.0 1.5 2.0
LD=300nm LD=550nm LD=1000nm Model Exp.
Current (mA) 2 4 6 8 10 12 SI(10
2/Hz)
0.0 0.5 1.0 1.5 2.0
16
Ls = 200 nm, Lr = 160 nm, Ld = 300, 500, 1000 nm Ls = 200 nm, Ld = 550 nm, Lr = 160, 400, 800 nm
Full Shot Noise Full Shot Noise
Completely suppressed shot noise in the active region F=0 Full shot noise in the active region F=1
Noise in the active region (and therefore the signature of full/suppressed shot noise) is only visible on the total noise at intermediate voltages Noise at high bias depend just
due to its high resistance
Javier Mateos – Universidad de Salamanca
InGaAs/InAlAs diodes with different dimensions show potential signs of shot noise suppression in the structures due to the presence of a potential barrier
regions of the devices shows that contact, source and drain resistances strongly affect the value of the total noise
could just be visible on the total noise at intermediate bias (before the onset of intervalley scattering) where the resistance of the drain region is still low
fabricated in order to obtain conclusive results
17
Javier Mateos – Universidad de Salamanca