June 11-12 2005 Electrical Characterization of MOS Devices with Advanced Gate Stacks Eric M. Vogel Leader, CMOS and Novel Devices Group and Director, NIST AML Nanofab Semiconductor Electronics Division Gaithersburg, MD 20899 Advanced Dielectrics Ultrathin and high-κ gate dielectrics present numerous challenges to MOS device characterization. Slide No. 2 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Scope • This 2-part lecture will focus on the understanding necessary to characterize MOS capacitors and gated diodes on bulk silicon having advanced gate dielectrics using the most common lower frequency (<~ 1 MHz) techniques: 1) Capacitance, 2) Conductance, 3) Charge-Pumping, 4) Tunneling. • For FET Parameter Extraction see G. Ghibaudo lecture • For LF Noise see C. Clayes lecture • For SOI see S. Cristoloveanu lecture and A. Zaslavsky lecture • For Reliability, Defect generation, and Device Instability see J. Stathis lecture • For High Frequency Device Characterization see G. Dambrine lecture Slide No. 3 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” OUTLINE • Capacitance-Voltage • Conductance • Tunneling • Charge-Pumping Slide No. 4 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” CAPACITANCE-VOLTAGE • Measurement Issues • Theory (Without Interface States) • Parameter Extraction • Including Interface States Slide No. 5 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” C-V Measurement Issues • Conditions • Errors • Equivalent Circuits Slide No. 6 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Measurement Conditions • The ac voltage should be as small as possible to ensure small signal approximation while still allowing accurate measurement. • Assuming that the device capacitance is properly extracted from the measured capacitance, either parallel or series mode may be used since the one can be derived from the other (Cs=Cp*(1+D2), D=1/ωRpCp). • Quasi-static C-V measurements generally can not be performed (using standard q-s meters) due to the large leakage currents. Slide No. 7 Cm Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Gm Measurement Errors Relative Measurement Accuracy for HP4284A • The relative measurement Calculations from HP4284A Precision LCR Meter Operation Manual p. 9-7 to 9-15 accuracy of an LCR meter This is only valid for: medium/long integration 100Hz <= F <= 1MHz depends on the frequency 1 m cable and the nominal capacitance Vdc < 20 V 30mV <= Vac <= 150mV <or> Vac = 10mV,20mV,25mV and conductance of the device Cp-Gp measurement mode short and open correction performed under test. • We have developed a spreadsheet which calculates the relative measurement accuracy of the HP4284A. Slide No. 8 Input C (F) G (S) F (Hz) Vac (Vrms) Output C_relacc (+/-%) G_relacc (+/-%) D_relacc (+/-%) 1.30E-10 1.00E-02 1.00E+03 5.00E-02 1288.996212 D too large (D>0.1)* 157833.5641 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 1.05296E-05 (FYI: Cal (FYI: Cal (FYI: Cal Measurement Errors • The relative capacitance measurement error of an LCR meter increases with: → decreasing frequency → increasing conductance → decreasing capacitance Capacitance Error (+/- %) 105 -11 104 103 102 101 100 10-1 10-2 102 103 104 Frequency (Hz) Slide No. 9 -3 C = 10 F, G = 10 S -6 C = 10-11 F, G = 10 S -10 -6 C = 10 F, G = 10 S Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 105 106 Capacitor Equivalent Circuits Gc Cc Rs Lo Slide No. 10 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Equivalent Circuits Transistor Equivalent Circuit† Capacitor LCR Measurement Equivalent Circuit Cm Gm Gc Cc Rs †K. Lo Slide No. 11 Ahmed, E. Ibok, G. Yeap, Q. Xiang, B. Ogle, J. J. Wortman, and J. R. Hauser, IEEE Trans. Elec. Dev., vol. 46, pp. 1650-1655, 1999. Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Examples of Measured Behavior 600 F = 1 MHz Capacitance (pF) Capacitance (pF) 140 F = 10 kHz 500 400 300 200 100 2 10 Hz 3 10 Hz 4 10 Hz 105 Hz 106 Hz 120 100 80 60 ISSG 2.1 nm -5 2 Area = 5x10 cm 40 20 0 -3 -2 -1 0 Vg (V) 1 2 0 -3 -2 -1 0 1 2 3 Vg (V) • For thicker dielectrics, a simple • For thinner or leakier reduction in capacitance at dielectrics, capacitance rollhigh frequencies due to series over, negative capacitance, and resistance is many times increasing capacitance with observed. bias is sometimes observed. Eric M. Vogel June 11-12, 2005 Slide No. 12 “Electrical Characterization of MOS Devices” Modeling “Measured” Capacitance • Series resistance alone cannot explain an increase in or negative measured capacitance at high frequency. 1000 Capacitance (pF) 800 Rs=20Ω, L=10µH 600 Rs=20Ω, L=45µH Rs=103Ω, L=0µH 400 F=1MHz 200 0 • Inductance is necessary to observe an increase in or negative capacitance at high frequency. Gm = Correct Capacitance Rs=20Ω, L=0µH -200 -3 -2 -1 1 Vg (V) Gc (Gc Rs + 1) − ω 2Cc2 Rs (G R c 2 2 s − ω C c Lo + 1) + ω (C c Rs + Gc Lo ) 2 ( 2 ) Cc 1 − ω 2Cc Lo − Gc2 Lo Cm = (C R GR −ω C L 11-12, + 1) + ω Eric (M. Vogel June 2005 2 2 c Slide No. 13 0 s c o 2 c 2 ) + G L s c o “Electrical Characterization of MOS Devices” Correcting Measured Capacitance • Previous work provided methodologies to correct measured capacitance for leakage and series resistance1,2. • Recent work has provided the methodology that includes a series inductance3. • It is unknown whether the series inductance is due to measurement (e.g. cabling)4 or a physical phenomenon5. 1K. J.Yang and C. Hu, IEEE Trans. Elec. Dev., vol. 46, pp. 1500-1501, 1999. M. Vogel, W. K. Henson, C. A. Richter, and J. S. Suehle, IEEE Trans. Elec. Dev., vol. 47, p. 601, 2000. 3H.-T. Lue, C.-Y. Liu, and T.-Y. Tseng, IEEE Elec. Dev. Lett., vol. 23, pp. 553-555, 2002. 4A. Nara, N. Yasuda, H. Satake, and A. Toriumi, IEEE Trans. Semi. Manuf., vol. 15, pp. 209-213, 2002. 5M. Matsumura, and Y. Hirose, Jap. J. Appl. Phys., vol. 39, pp. L123-L125, 2000. 2E. Slide No. 14 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” C-V Theory • Potential and Charge Balance • Built-in Voltage • Total Semiconductor Charge • Regions of Operation • Capacitance • Quantum Mechanical Effects Slide No. 15 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Equations for C-V Potential Balance: Vg = Vsub − V poly + Vox + Vbi Charge Balance: Qsub (Vsub ) + Q poly (V poly ) + Qox = 0 • To calculate C-V, we need to solve the above equations. • Vpoly is measured from the oxide-poly interface to the bulk of the poly. Vpoly = 0 for metal gate electrodes. • Qox is charge in the oxide that is fixed with bias. • We will first neglect defects in the dielectric that change occupancy with applied bias (interface states). • Vbi is the built-in potential between the gate and substrate. Slide No. 16 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Potential Balance Vg = Vsub − V poly + Vox + Vbi Vbi = (E vac Potential balance − E f , poly ) @ poly bulk q Q poly (V poly ) = CoxVox − (E vac − E f ,sub ) @ sub bulk q Using Gauss’ Law Qsub Vg = Vsub − V poly + V fb − Cox Qox V fb = Vbi − Cox Slide No. 17 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” More later Charge Density in a Semiconductor ρ = q ( p − n + N d+ − N a− ) Classical Charge Density (cm-3) in a Semiconductor ⎛ 2Nv ⎞ ⎛ − η − E gap ⎞ 2 N c ⎛η ⎞ ⎟ ⎜ ⎟⎟ − F1 2 ⎜⎜ F1 2 ⎜⎜ ⎟⎟ + φt π ⎟ ⎜ π ⎝ ⎠ ⎝ φt ⎠ ⎟ ρ = q⎜ Nd Na ⎟ ⎜ − ⎜ 1 + 2 exp⎛ η − ( Ed − Ec ) ⎞ 1 + 4 exp⎛ ( Ea − Ec ) − η ⎞ ⎟ ⎟ ⎜ ⎜ ⎟⎟ ⎜ φt φt ⎝ ⎝ ⎠ ⎠⎠ ⎝ F1/ 2 (η ) = Slide No. 18 2 π ∫ ∞ 0 E dE 1 + exp( E − η ) η ≡ E f − Ec Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” φt ≡ kT q Calculating the Built-in Voltage Vbi = (E vac − E f , poly ) @ poly bulk − (E vac − E f ,sub ) @ sub bulk q q ⎛ 2Nv ⎞ ⎛ − η − E gap ⎞ 2 N c ⎛η ⎞ ⎟ ⎜ ⎟⎟ − F1 2 ⎜⎜ F1 2 ⎜⎜ ⎟⎟ + φt π ⎟ η ≡ E f − Ec ⎜ π ⎝ ⎠ ⎝ φt ⎠ ⎟ ρ = q⎜ Nd Na ⎟ ⎜ − ⎜ 1 + 2 exp⎛ η − ( Ed − Ec ) ⎞ 1 + 4 exp⎛ ( Ea − Ec ) − η ⎞ ⎟ ⎟ ⎜ ⎜ ⎟⎟ kT ⎜ φ φ ⎝ ⎝ ⎠ ⎠ ⎠ φt ≡ t t ⎝ q E 2 ∞ F1/ 2 (η ) = dE ∫ 0 π 1 + exp( E − η ) • To calculate Vbi set ρ = 0 and find Ef in the bulk of the semiconductor. Slide No. 19 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Calculating Total Semiconductor Charge Qsub (Vsub ) + Q poly (V poly ) + Qox = 0 ⎛ 2Nv ⎞ ⎛ − η − E gap ⎞ 2 N c ⎛η ⎞ ⎟ ⎜ ⎟⎟ − F1 2 ⎜⎜ F1 2 ⎜⎜ ⎟⎟ + φt π ⎟ ⎜ π ⎝ ⎠ ⎝ φt ⎠ ⎟ ρ = q⎜ Nd Na ⎟ ⎜ − ⎜ 1 + 2 exp⎛ η − ( Ed − Ec ) ⎞ 1 + 4 exp⎛ ( Ea − Ec ) − η ⎞ ⎟ ⎟ ⎜ ⎜ ⎟⎟ ⎜ φt φt ⎝ ⎝ ⎠ ⎠⎠ ⎝ • To calculate Qsub(Vsub) or Qpoly(Vpoly), one must integrate ρ from the oxide/semi interface to the bulk of the semiconductor. Slide No. 20 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Calculating Total Semiconductor Charge Qsub = −Ε surf ε η bulk = (E f − Ec ) bulk = φb F3 / 2 (η ) = Slide No. 21 2 π ∫ ∞ 0 Ε 2surf ⎡ 4Nv ⎛ ⎛ − φb − E gap ⎞ ⎛ − φb − Vsub − E gap ⎞ ⎞ ⎤ ⎜ ⎜ ⎟ ⎟⎟ ⎟⎟ + ⎥ − F F ⎢ ⎟ 3 2 ⎜⎜ ⎜ 3 2⎜ π φ φ 3 ⎝ ⎠ ⎝ ⎠⎠ ⎥ t t ⎝ ⎢ ⎢ 4N ⎛ ⎥ ⎛ φb ⎞ ⎛ φb + Vsub ⎞ ⎞ c ⎜⎜ F3 2 ⎜⎜ ⎟⎟ − F3 2 ⎜⎜ ⎢ ⎥ ⎟⎟ ⎟⎟ + ⎢3 π ⎝ ⎥ ⎝ φt ⎠ ⎠ ⎝ φt ⎠ ⎢ ⎥ ⎛ ⎞ ⎛ ⎞ ⎛ ⎞ − − ( ) E E φ d c ⎢ ⎜ ⎥ ⎜ 1 + 2 exp⎜ b ⎟ ⎟⎟ ⎥ 2qφt ⎢ ⎜ Vsub φt ⎟ ⎜ ⎟ ⎝ ⎠ = + ln + ⎥ N ⎜ ⎟⎟ ε ⎢ d ⎜ φt ⎛ ⎞ + − − V E E φ ( ) sub d c ⎢ ⎜ ⎥ ⎟ ⎟ ⎟⎟ ⎜ 1 + 2 exp⎜ b ⎜ ⎢ ⎝ ⎥ φt ⎝ ⎠ ⎠⎠ ⎝ ⎢ ⎥ ⎛ ⎛ ( Ea − E c ) − φb ⎞ ⎞ ⎞ ⎥ ⎢ ⎛ ⎜ ⎜ + 1 4 exp ⎜ ⎟ ⎟⎟ ⎥ ⎢ φt V ⎝ ⎠ ⎟⎟ ⎥ ⎢ N a ⎜ − sub + ln⎜ ⎜ ⎛ ( Ea − E c ) − φb − Vsub ⎞ ⎟ ⎟ ⎥ ⎢ ⎜ φt ⎟ ⎟ ⎟⎟ ⎥ ⎜ 1 + 4 exp⎜ ⎢ ⎜⎜ φt ⎝ ⎠ ⎠⎠ ⎦ ⎝ dE ⎣ ⎝ E3 2 1 + exp( E − η ) Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Solving Potential and Charge Balance Potential Balance: Charge Balance: Qsub Vg = Vsub − V poly + V fb − Cox Qsub (Vsub ) + Q poly (V poly ) + Qox = 0 • Given: Vg, Nsub, Npoly, Cox, Qox • The above 2 equations can be solved for 2 unknowns: Vsub, Vpoly Slide No. 22 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Regions of Operation 1.4 3e-6 Inversion 1.2 2e-6 1.0 Depletion 0.6 Qsub Vsub 1e-6 Ef-Ec=-nkT/q 0.8 0.4 0.2 Ef-Ec=-Egap+nkT/q 0.0 -0.2 -2 -1 0 Vg 1 2 3 Inversion -1e-6 Accumulation -3e-6 -0.4 -3 0 -2e-6 Accumulation -4 Depletion 4 -4e-6 -4 -3 -2 -1 0 1 2 Vg • Accumulation occurs when the Ef is near the valence band. • Inversion occurs when the Ef is near the conduction band. Slide No. 23 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 3 4 Common Approximations 3e-6 • Qinv = Cox (Vg-Vt) Qacc 2e-6 Depletion 1e-6 Qsub • Qacc = Cox (Vg-Vfb) 0 Inversion -1e-6 Accumulation -2e-6 -3e-6 Qinv -4e-6 -4 -3 -2 -1 0 1 Vg Slide No. 24 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 2 3 4 Quasi-static vs. Deep Depletion 3 4 Quasi-static Deep Depletion 1 2 Vox Vsub 3 Quasi-static Deep Depletion 2 1 0 -1 0 -2 -1 -4 -3 -2 -1 0 Vg 1 2 3 4 -3 -4 -3 -2 -1 0 Vg 1 2 3 • The previous analysis assumes minority carrier generation can keep up with the dc bias (quasi-static). • However, deep depletion is typically seen for MOS capacitors with ultra-thin oxides. Slide No. 25 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 4 Calculating Capacitance Total Device Capacitance: Substrate/Poly Capacitance: Slide No. 26 ( Ctot = (Csub ) + C Csub ( poly ) = −1 −1 poly dQsub ( poly ) dVsub ( poly ) Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” +C ) −1 −1 ox 1.2 1.2 1.0 1.0 Nsub = 10 0.6 cm -3 LF HF DD 0.4 0.2 0.8 2 17 C (µF/cm ) 0.8 2 C (µF/cm ) Typical C-V Characteristics 14 0.6 LF HF DD 0.4 0.2 0.0 Nsub = 6x10 -3 cm 0.0 -4 -3 -2 -1 0 Vg (V) 1 2 3 4 -4 -3 -2 -1 0 1 2 3 Vg (V) • High frequency means that carriers can respond to the dc signal but not the ac signal • Low Frequency means that carriers can respond to the dc and ac signals. • Deep-depletion means Eric thatM.the carriers cannot respond to Vogel June 11-12, 2005 the dc or ac signals. “Electrical Characterization of MOS Devices” Slideeither No. 27 4 Quantum Mechanical Effects • The previous analysis was based on a classical description of charge in the semiconductor. • The large band bending in the semiconductor causes the formation of a potential well. • This results in a quantization of the density of states and a shifting of the carrier centroid away from the semiconductor/insulator interface. • The most fundamental and computationally expensive approach to handle these quantum-mechanical effects is to solve the Schroedinger and Poisson equations together. Slide No. 28 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Quantum Mechanical Effects • The splitting of energy levels and the shifting of the carriers away from the interface leads to a decrease of the inversion layer or accumulation layer charge density as a functio of the surface potential as compared to classical simulation. • van Dort et al. pursued a more computationally efficient approach of modeling these effects via an increase in the effective bandgap of silicon. • Hareland et al. improved on van Dort’s work by providing a more detailed description of the bandgap widening based on a comparison to a rigorous self-consistent SchroedingerPoisson solution. 13 ⎛ ε ⎞ 23 ∆E g = 5.92 × 10 −8 ⎜ si ⎟ (Ε s ) ⎝ 4kTq ⎠ Slide No. 29 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Quantum Mechanical Effects 3.5 Classical 3.0 2.5 QM 2 C (µF/cm ) • Quantum Mechanical Effects result in a drop of the maximum capacitance and a slight shift of the threshold voltage. 2.0 1.5 Metal Gate Tox = 1.0 nm 1.0 0.5 Nsub = 2x1017cm-3 0.0 -4 -3 -2 -1 0 Vg (V) Slide No. 30 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 1 2 3 4 C-V Parameter Extraction • Methodology • Oxide Thickness • Substrate Doping • Polysilicon Doping • Flatband Voltage • Oxide Charge and Workfunction Slide No. 31 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Parameter Extraction using Modeling 1.6 • NCSU CVC program fits experimental C-V data using a model that has the following parameters: Vfb, Tox, Nsub, Npoly 1.4 2 C (µF/cm ) 1.2 • NCSU CVC does not include interface states. 1.0 0.6 Vfb = -0.987 V Tox = 2.01 nm 0.4 Nsub = 2.97x10 0.8 Measured Modeled 0.2 cm 20 cm Npoly = 1.61x10 -3 0.0 -4 -3 -2 -1 0 Vg (V) Slide No. 32 17 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 1 2 3 -3 Impact of Simulation Code Slide No. 33 1.50 1.25 1.00 2 C (µF/cm ) • Simulators show a difference of up to 20% in the calculated accumulation capacitance. • This discrepancy leads to large inaccuracies in the values of dielectric thickness extracted from C-V. • Possible reasons include: the use of approximations for quantum effects vs. Schrödinger equation, wave function boundary conditions, and type of carrier statistics. 0.75 0.50 UTQuant [30] NIST Schred [32] NCSU [25] NEMO [29] Berkeley [31] Tox = 2.0 nm 18 -3 Nsub = 10 cm 0.25 Npoly = 1020 cm-3 n-channel, n-poly gate 0.00 -3 -2 -1 0 Vg (V) Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 1 2 3 Oxide Thickness Definitions • The Equivalent Oxide Thickness (EOT) is obtained from the gate dielectric capacitance alone. • EOT must be determined from C-V measurements using a fitting or extraction algorithm which includes QM effects, polysilicon depletion, etc. • The Capacitance Equivalent Thickness (CET) is determined by simply taking the ∈SiO2xArea/Cmeas where Cmeas is the measured capacitance in inversion or accumulation at some defined voltage. Slide No. 34 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Oxide Thickness Definitions • In order to achieve a small EOT, the interfacial thickness must be controlled. 2.0 1.5 EOT (nm) • EOT is the thickness of SiO2 which would produce the same capacitance as that obtained from a high-K dielectric. 1.0 T(SiO2) = 1 nm, k(High-k) = 20 T(SiO2) = 1 nm, k(High-k) = 40 T(SiO2) = 0.5 nm, k(High-k) = 20 T(SiO2) = 0.5 nm, k(High-k) = 40 0.5 0.0 0 1 2 3 4 5 Physical Thickness of High-κ Dielectric (nm) Slide No. 35 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Thickness Extraction +C ) −1 −1 ox • The maximum capacitance in accumulation is close to Cox. • The minimum capacitance is due to the substrate capacitance. 14 2 −1 poly C (µF/cm ) ( Ctot = (Csub ) + C −1 12 Ctot 10 Csub Cox 8 6 4 2 0 -4 -3 -2 -1 0 Vg (V) Slide No. 36 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 1 2 3 4 Thickness Extraction 1.75 1.70 1.65 CET (nm) • The capacitance in accumulation (represented here by CET) is not strongly impacted by the substrate or polysilicon doping. 1.60 1.55 1.50 • The CET does depend strongly on whether the gate is metal or polysilicon. Slide No. 37 19 -3 20 -3 Npoly = 10 cm EOT = 1.0 nm N-channel Device Vg = -2 V Npoly = 10 cm Metal 1.45 1.40 1015 1016 1017 1018 -3 Nsub (cm ) Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 1019 Substrate Doping Extraction 1.2 1.0 0.8 2 C (µF/cm ) • The substrate doping strongly impacts the minimum capacitance but does not strongly impact the maximum capacitance in accumulation. 0.6 0.4 17 Nsub = 10 0.2 • The minimum capacitance can be used to determine substrate doping. Slide No. 38 -3 cm 14 Nsub = 6x10 0.0 -4 -3 -2 -1 0 Vg (V) Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 1 2 3 cm 4 -3 Polysilicon Doping Extraction 3.0 • The capacitance in inversion can be used to determine polysilicon doping. 2.5 2 C (µF/cm ) • The depletion of polysilicon results in a large drop of the capacitance. Metal Poly=1020 cm-3 2.0 1.5 1.0 Poly=5x1019 cm-3 Tox = 1.0 nm 0.5 N = 2x1017cm-3 sub 0.0 -4 -3 -2 -1 0 Vg (V) Slide No. 39 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 1 2 3 4 Flatband Voltage Extraction • There are numerous sources of possible error with this technique. Slide No. 40 18 10 10-1 Cfb/Cox • If the oxide capacitance and substrate doping is known, the flatband capacitance (and hence Vfb) can be found. 100 17 10 16 10 15 10 10-2 Na 10-3 10-1 -3 14 0 =1 cm 100 Tox (nm) Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 101 Workfunction and Oxide Charge Extraction EOT ⎤ 1 ⎡ V fb = φms − xρ ( x )dx ⎥ ⎢ ∫ ε ox ⎣ 0 ⎦ EOT 1 EOT ⎤ 1 ⎡ V fb = φms − xρ ( x )dx + ∫ xρ f 2bulk ( x )dx ⎥ ⎢ ∫ ε ox ⎣ 0 ⎦ EOT 1 For a stacked dielectric Slide No. 41 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Workfunction and Oxide Charge Extraction V fb = φms − − − 1 ε ox 1 ⎡1 2⎤ EOT ρ ⎥⎦ ε ox ⎢⎣ 2 f 1bulk [Q f 1int f R. Jha, et al., IEEE EDL 25, 420 (2004) + Q f 2 int f − (ρ f 1bulk − ρ f 2bulk )EOT2 ]EOT 1 ⎡1 ⎤ ( ) EOT Q ρ ρ − − f 1bulk f 2 bulk 2 f 1int f ⎥ EOT2 ⎢ ε ox ⎣ 2 ⎦ For a SiO2/high-k stack where EOT1(2) is the EOT of the highk (SiO2), Qf1(2)int are the charges at the high-k/SiO2 (SiO2/Si) interface, ρf1(2)bulk are charges uniformly distributed within the high-k (SiO2). Slide No. 42 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Workfunction and Oxide Charge Extraction R. Jha, et al., IEEE EDL 25, 420 (2004) If interface charges dominate bulk charges: V fb = φms − V fb = φms − 1 ε ox 1 ε ox [Q [Q f 1int f ]EOT + [Q 1 f 1int f 1 ε ox + Q f 2 int f ]EOT + f 2 int f 1 ε ox [Q EOT ] f 1int f EOT2 ] The intercepts and slopes of Vfb vs. EOT with varying EOT1 and EOT2 can provide Φms, Qf1intf, and Qf2intf. Slide No. 43 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Effective Workfunction • The vacuum work function of a metal does not necessarily equal the effective work function of the metal due to charge transfer, defects, and dipoles. Slide No. 44 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” C-V Including Interface States • Theory • Interface State Capacitance • Interface State Density “Extraction” Slide No. 45 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Including Interface States • Qit is charge in the oxide that changes occupancy with bias. • Dita(d) is the density of acceptor(donor)-like interface states Qsub (Vsub ) + Q poly (V poly ) + Qit (Vsub ) + Qox = 0 Qsub Qit (Vsub ) Vg = Vsub − V poly + V fb − − Cox Cox ⎡− qDita ( Et − Ei )Fsa (( Et − Ei ) − (E f − Ei )) ⎤ Qit (Vsub ) = ∫ ⎢ ⎥dEt + qDitd ( Et − Ei )Fsd (( Et − Ei ) − (E f − Ei ))⎦ Ev ⎣ −1 −1 ⎛ ⎛ ⎛ − (Et − E f ) ⎞ ⎞ ⎛ (Et − E f )⎞ ⎞ ⎟⎟ ⎟⎟ Fsd (Et − E f ) = ⎜⎜1 + 2 exp⎜⎜ ⎟⎟ ⎟⎟ Fsa (Et − E f ) = ⎜⎜1 + 0.25 exp⎜⎜ Ec ⎝ Slide No. 46 ⎝ φt ⎠⎠ ⎝ ⎝ Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” φt ⎠⎠ Interface State Capacitance • At very low frequencies (quasi-static), interface states respond to the ac signal over the entire bias range resulting in a capacitance. ( Ctot = (Cit + Csub ) + C Slide No. 47 −1 −1 poly +C ) −1 −1 ox 0.35 2 Capacitance (µF/cm ) • At very high frequencies (infinite), interface states do not respond to the ac signal. 0.30 0.25 0.20 Tox = 10 nm 0.15 Dit = 1012 cm-2eV-1 Ideal HF HF with Dit 0.10 QS with Dit 0.05 -3 -2 -1 0 Vg (V) Cit ,QS ∆Qita + ∆Qitd = ∆Vsub Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 1 2 3 Interface State Capacitance tan (ωτ )P(V )dV ∫ ωτ = (c p ) Cit = qDit −1 p s s p −∞ τp σp ≡ −1 p cp vth s = capture cross section Capacitance (µF/cm2) ∞ 2.5 Tox = 1.0 nm 2.0 Dit = 1012 cm-2eV-1 FET Cap (102 Hz) Cap (103 Hz) Cap (104 Hz) Cap (105 Hz) Cap (106 Hz) 1.5 1.0 0.5 0.0 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 Vg (V) • At intermediate frequencies, SRH theory must be used including the effect of surface potential variation across the interfacial plane. •P(Vs) is the probability thatEric the band bending is V , ps is the free M. Vogel June 11-12, 2005 s carrier “Electrical Characterization of MOS Devices” Slide No. 48 concentration. Dit Extraction EOT = 0.62 nm Nsub = 4x1017 cm-3 3 Exp. Data (105 Hz) Simulation: 105 Hz (Dit profile 1) 2 σs = 4 (kT/q) σp = 10-14 cm2 1 Simulation: Quasi-static (Dit profile 2) Simulation: No Dit 0 -2.0 -1.5 -1.0 -0.5 0.0 Gate Voltage (V) 10 8 profile 1 profile 2 -1 Dit (x10 cm eV ) 6 12 -2 • Proper modeling requires including the interface state capacitance as a function of frequency. 2 • Some have attempted extracting Dit from the “hump” observed in C-V using a quasi-static approach. Capacitance (µF/cm ) 4 4 2 0 Slide No. 49 Eric M. Vogel June-1.0 11-12, -0.5 2005 0.0 E - E (eV) “Electrical Characterization of MOS Devices” t i 0.5 1.0 Dit Extraction on Thick Oxides 2) Terman: A HF CV curve is measured and compared to a theoretical ideal (no Dit) CV curve to obtain the amount of voltage stretch-out. Slide No. 50 0.35 2 Capacitance (µF/cm ) 1) Low-High Frequency: A quasistatic (QS) and a high-frequency (HF) CV curve is measured and interface state capacitance is determined. 0.30 0.25 0.20 Tox = 10 nm 0.15 12 0.10 -2 Dit = 10 cm eV Ideal HF HF with Dit -1 QS with Dit 0.05 -3 -2 -1 0 Vg (V) Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 1 2 3 Dit Extraction on Thin Dielectrics 2) Terman: With decreasing EOT (increasing dielectric capacitance), the voltage shift (Qit/Cox) due to charging of traps (stretch-out) becomes smaller. Slide No. 51 ∆V (V) 1) Low-High Frequency: Quasistatic measurements cannot be performed on advanced dielectrics due to leakage current 100 10 Nit = 1012 cm-2 -1 11 -2 Nit = 10 cm 10-2 10 -2 Nit = 10 cm 10-3 10-4 0 2 4 6 EOT (nm) Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 8 10 12 Dit Extraction • However the number of parameters that can provide a reasonable fit is large. 4 17 2 Capacitance (µF/cm ) • Dit can be extracted by properly modeling the frequency dependence of the interface state capacitance. 3 5 F = 10 Hz 2 1 σs = 4 (kT/q), σp = 10-14 cm2 σs = 1 (kT/q), σp = 10-14 cm2 σs = 4 (kT/q), σp = 10 -17 0 -2.0 -1.5 2 cm -1.0 -0.5 Gate Voltage (V) Slide No. 52 -3 Nsub = 4x10 cm EOT = 0.62 nm Dit profile 1 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 0.0 CONDUCTANCE • Theory • Parameter Extraction Slide No. 53 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Conductance Theory rc ( x, t ) = c p p( x, t )nT f ( x, t ) (cm -3 s −1 ) re ( x, t ) = e p nT [1 − f ( x, t )] (cm -3 s −1 ) i p ( x, t ) = q[rc ( x, t ) − re ( x, t )] (A cm3 ) Applying small signal approximation q2 Gp = nT f 0 c p p0 ( x ) (mhos cm 3 ) kT Slide No. 54 Rate of hole capture Rate of hole emission Hole recombination current due to semi. bulk traps Hole conductance due to semi. bulk traps Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Conductance Theory [ G p = Cit (2ωτ p ) ln 1 + (ωτ p ) −1 2 ] 1 τp = exp( Vsub cp Na Hole conductance due to distribution of surface states ) σ p is the hole capture cross - section v is the thermal velocity cp = σ p v Including band bending fluctuations across the interfacial plane with variance of banding bending, σs qD (2πσ = it ω 2ωτ p Gp Slide No. 55 ) 2 −1 2 ∞ s ( ) ⎛ η2 ⎞ 2 ( ) ⎟ ⎜ exp exp ( − η ) ln 1 + ωτ exp 2η dη − p ∫−∞ ⎜⎝ 2σ s2 ⎟⎠ Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Conductance Theory Cm Measured Capacitance Gm Measured Conductance Cc Capacitance corrected for Rs Gc Conductance corrected for Rs Gt DC Tunneling Conductance Gac Rs Gc corrected for Tunneling Series Resistance Dit ω Cx Gp Slide No. 56 Interface State Density Angular Frequency Oxide and Poly Capacitance Interface Trap Conductance Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Conductance Parameter Extraction 1. Determine Rs (see previous section). 2. Determine Cox (see previous section). 3. Determine Gt = dIg(Vg)/dVg 4. Measure Cm(F) and Gm(F) for the gate biases (trap energies) of interest. 5. Correct Cm and Gm for Rs using (see previous section if inductance is an issue) 2 Cm CmCc Rs − Gm ω Cc = 2 (1 − Gm Rs ) + ω 2Cm2 Rs2 Gc = Gm Rs − 1 6. Correct Gc for tunneling using Gac = Gc − Gt Slide No. 57 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” Conductance Parameter Extraction ωC Gac = 2 ω Gc + ω 2 (C x − Cc )2 Gp 2 x 14 -9 2 Gp/ω (10 S/rad/cm ) 7. Determine interface trap conductance as function of frequency for each gate bias using 16 Vg= -0.60 V Symbols: Experiment Lines: Model 12 10 tox = 2.0 nm -0.55 V -0.65 V 8 Rs = 20 Ω -0.70 V 6 4 2 0 102 103 104 Frequency (Hz) Slide No. 58 -0.80 V Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 105 106 Conductance Parameter Extraction 16 14 -9 2 Gp/ω (10 S/rad/cm ) 7. Determine the peak frequency and G p associate ω fp conductance for each gate bias. 8. Determine the conductance at 5fp or fp/5 for each gate bias. Gp Gp ω Slide No. 59 5 fp ω Vg= -0.60 V Symbols: Experiment Lines: Model 12 10 tox = 2.0 nm -0.55 V -0.65 V 8 Rs = 20 Ω -0.70 V 6 -0.80 V 4 2 0 102 103 104 Frequency (Hz) fp 5 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 105 106 Conductance Parameter Extraction Gp ω Gp 5 fp ω fp 5 p 1.0 [<Gp>/ω]/[<Gpp>/ω]f 9. Determine σs using the following plot and the previously determined 0.9 0.8 High (5fp) 0.7 0.6 Low (fp/5) 0.5 0.4 0 1 2 3 σs (in units of kT/q) Slide No. 60 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 4 5 Conductance Parameter Extraction 0.4 fd(σs) 10. Determine fd using the following plot. 0.3 0.2 0.1 0.0 0 1 2 3 σs (in units of kT/q) Slide No. 61 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 4 5 Conductance Parameter Extraction 2.7 2.6 2.5 11. Determine ξp using the following plot. ξp 2.4 2.3 2.2 2.1 2.0 1.9 0 1 2 3 σs (in units of kT/q) Slide No. 62 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 4 5 Conductance Parameter Extraction 7 12. Calculate the interface state density and trap time constant using Slide No. 63 -1 -2 4 3 Dit (10 ⎛ Gp ⎞ −1 Dit = ⎜ ⎟ [ f D (σ s )q ] ⎝ ω ⎠ fp ζp 1 τp = = exp( Vsub ω p σ p vN a 5 10 cm eV ) 6 ~2.0 nm RTO 2.0 nm, 20 Ω 2.2 nm, 20 Ω 1.8 nm, 20 Ω 2.0 nm, 10 Ω 2.0 nm, 18 Ω 2.0 nm, 30 Ω 2 1 ) 0 -0.35 -0.30 -0.25 Et-Ei (eV) Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” -0.20 -0.15 Conductance of Ultra-thin Oxides 1e-7 8e-8 Gp/ω (S/rad/cm2) • The very large tunneling currents associated with ultrathin oxides begins to affect the conductance Symbols: Experiment Lines: Model tox = 1.4 nm Rs = 120 Ω 6e-8 Vg = -0.70 4e-8 2e-8 Vg = -0.65 Vg = -0.75 0 1e+3 1e+4 1e+5 Frequency (Hz) Slide No. 64 Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 1e+6 Sensitivity of Conductance with Rs • The expected measured conductance was determined using modeling for various values of device properties. A smaller change in expected Gm with Dit indicates less sensitivity. Slide No. 65 2 • Gm (fp(Gp/ω)) (S/cm ) 100 10-1 10-2 Gt = 0 10-3 Rs, τp 10-4 0 Ω, 10-4 s 100 Ω, 10-4 s -6 0 Ω, 10 s -6 100 Ω, 10 s -6 999 Ω, 10 s 10-5 10-6 109 1010 1011 -2 1012 -1 Dit (cm eV ) Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 1013 Sensitivity of Conductance with Gt • A smaller change in expected Gm with Dit indicates less sensitivity. Slide No. 66 100 2 The expected measured conductance was determined using modeling for various values of device properties. Gm (fp(Gp/ω)) (S/cm ) • 10-1 Rs = 0 10-2 10-3 τ p , Gt 10-4 10-4 s, 0 S/cm2 10-4 s, 10-3 S/cm2 10-6 s, 0 S/cm2 10-6 s, 10-3 S/cm2 10-5 10-6 109 1010 1011 Dit (cm-2eV-1) Eric M. Vogel June 11-12, 2005 “Electrical Characterization of MOS Devices” 1012 1013 Interface State Time Constant 1013 Dit (cm-2eV-1) 1012 1011 10-3 ZrO2-SiN 10-4 ZrO2-SiOx 10-5 SiO2 Solid Symbols: D it 1010 O pen Symbols: τ 109 -0.30 -0.25 -0.20 -0.15 E - Ei (eV) Slid