Charge Density in MOS
Charge Density in MOS
Charge Density in MOS
6.1 INTRODUCTION
INTRODUCTION
Charges and their location for thermally oxidized silicon.
(1) (2) (3) (4) Interface trapped charge (Qit, Nit, Dit) Fixed oxide charge (Qf, Nf) Oxide trapped charge (Qot, Not) Mobile oxide charge (Qm, Nm)
Capacitance-Voltage Curves
Qs=Qp+Qb+Qn+Qit
Capacitances of an MOS capacitor for various bias conditions as discussed in the text.
7
In order for the inversion charge to be able to respond, Jscr = qniW/g Jd = CdVg/dt
W in m, tox in nm, g in s
8
is the dimensionless semiconductor surface electric field. Us=s/kT, UF=qF/kT = 1 is the intrinsic Debye length
9
10
11
Low-frequency (lf), high-frequency (hf), and deep-depletion (dd) normalized SiO2-Si capacitance-voltage curves of an MOS-C; (a) p-substrate NA= 1017 cm-3, (b) n-substrate ND = 1017 cm-3, tox= 10nm, T=300K.
12
(a)
(b)
Effect of sweep direction on the hf MOS-C capacitance on an p-substrate, (a) entire C-VG curve, (b) enlarged portion of (a) showing the dc sweep direction; f=1 MHz.
13
Flatband Voltage
normalized CFB
14
CFB/COX versus NA as a function of tox for the SiO2 -Si system at T=300K.
15
Schematic illustration of an MOS-C with finite gate doping density, showing gate depletion for positive gate voltage.
16
Low-frequency and high-frequency capacitance-voltage curves for various n+ polysilicon gate doping densities. The lowest Chf curve is for ND (gate) =1018 cm-3. Substrate NA =1016 cm-3, tox =10nm.
17
Capacitance Measurement
(a)
(b)
Block diagram of circuits to measure the current and charge of an MOS capacitor.
19
Ideal (line) and experimental (point) MOS-C curves. NA =51016 cm-3, tox=20nm, T=300K, CFB/Cox=0.77.
21
Fixed Charge
22
Potential band diagram of (a) n+ polysilicon-p substrate, and (b) p+ polysilicon-n substrate at flatband.
24
25
Work function difference as a function of doping density for polysilicon-SiO2 MOS devices.
27
28
Mobile Charge
Drift time for Na, Li, K, and Cu for an oxide electric field of 106 V/cm and tox =100 nm.
29
30
CIf and Chf measured at T=250OC. The mobile charge density is determined from the area between the two curves.
32
33
Semiconductor band diagram illustrating the effect of interface traps; (a) V=0, (b) V>0, (c) V<0. Electron-occupied interface traps are indicated by the small horizontal heavy lines and unoccupied traps by the light lines
34
(a) (c)
Theoretical ideal (Dit=0) and Dit 0 (a) hf , (b) If and (c) experimental lf C-V curves. (b)
35
CS=Cb+Cn
36
High- and low-frequency C-VG curves showing the offset C/Cox due to interface traps.
37
Interface trapped charge density from the hf curve and the offset C/Cox.
38
Conductance Method
Cox Rit G Cit Cp Gp C m Gm CS Cox Cox Rit Cit rS (a) (b) (c) (d) Gt
(a) MOS-C with interface trap time constant it=RitCit , (b) simplified circuit of (a), (c) measured circuit, (d) including series rs resistance and tunnel conductance Gt.
39
40
Gp/ versus for a single level, a continuum and experimental data. For all curves: Dit =1.9109 cm-2 eV-1, it=710-5s.
41
Interface trapped charge density versus energy from the quasi-static and conductance methods. (a) (111) n-Si, (b) (100) n-Si.
42
High-Frequency Methods
Terman method:
43
Circuit diagram and energy bands for charge pumping measurement. The figures are explained in the text.
44
(c)
(d)
(e)
(f)
45
46
Dit=7109cm-2eV-1
(a) Icp as a function of tstep showing e at the point where Icp begins to saturate. (b) insulator trap density versus insulator depth from the insulator/Si interface for Al2O3 and SiO2.
50
51
Interface trap density as a function of energy through the band gap for various measurement techniques. Reprinted with permission
52
Charge pumping current versus base voltage for two voltage pulse heights before and after gate leakage current correction. tox=1.8nm, f=1kHz.
53
54
55
56
57
58
DC-IV
(a) DC-IV measurement set-up. (b) surface space charge region for different gate bias.
59
With the surface in strong inversion or accumulation, the recombination rate is low. The rate is highest with the surface in depletion.
If charge is injected into oxide, leading to a VT shift, ID will also shift. IDVTQot
60
61
62
C-V Measurements
Assume:
1. The interface trap capacitance is negligible in accumulation at 100kHz to 1MHz. 2. The differential interface trap charge density between flatband and accumulation is negligible. 3. The oxide charge density is negligible. 4. Quantum effects are negligible.
63
C-V Measurements
64
1/C versus 1/(VG-VFB) for two oxide thicknesses. Data adapted from Ref. 101.
65
D=Gp/Cp.
(a) MOSC equivalent circuit with tunnel conductance and series resistance. (b) parallel and (c) series equivalent circuit .
66
Measurement error dependence on device area and oxide thickness. The two-frequency measured capacitance is in error less than 4% in the shaded region. At higher frequencies the D=1.1 border shifts to thinner oxides. Adapted from ref. 109
67
I-V Measurements
2 J FN AE ox exp( B / E ox )
J dir
68