1Semiconductor Electronics Division, NIST

1
Degradation and Breakdown of Ultra-thin Silicon
Dioxide by Electron and Hole Injection Eric M.
Vogel1 M. Edelstein1, J. Suehle1, D. Heh2, and
J. Bernstein2
1Semiconductor Electronics Division, NIST 2Center
for Reliability Engineering, University of
Maryland
2
E vs. 1/E Lifetime Extrapolation - Neither are
Valid
3
Motivation 1 Determining physical model for
oxide breakdown (Vg lt 7.5 V)
Energetic Carrier Models
Electric Field Models
(Thermochemical-E)
Anode Hole Injection
Trap Creation/ Hydrogen Release
Motivation 2 Comprehensive and self-consistent
understanding of MOSFET degradation and breakdown
that includes hot-carrier and uniform tunneling
stress conditions
4
Charge-to-Breakdown (Qbd) for CVS

• For CVS, the Qbd is independent of substrate
  bias.
• For ultra-thin oxides (lt 3.0 nm), Qbd versus
  Vg is approximately independent of thickness.

5
Time-to-Breakdown (tbd) versus Vg for CVS

• At a given Vg, the tbd increases with
  increasing thickness due to the smaller current
  flowing in a thicker oxide at a given gate
  voltage.

6
Time-to-Breakdown (tbd) versus Eox for CVS

• At a given Eox, the tbd decreases with
  increasing thickness because a larger gate
  voltage is required for a thinner oxide to obtain
  a given electric field.

7
Experimental for SHE/CVS

• N-channel MOSFETs
• lt100gt silicon, Na 2x1017 cm-3
• tox 2.0 nm to 3.4 nm
• n polysilicon
• Constant Voltage Stress (CVS),
• Vg gt 0, Vs Vd Vb 0
• Substrate Hot Electron (SHE) Stress,
• Vg gt 0, Vs Vd 0, Vb lt 0, Vinj lt Vb
• Stress Induced Leakage Current (SILC)
• and sinusoidal Charge Pumping (CP)
• used to monitor electrically active defects.

SHE set-up
8
SHE Gate Current Density Characteristics

• For low gate voltages, the SHE current
  dominates the gate current and is dependent on
  the injector bias.
• For higher gate voltages, the normal tunneling
  current dominates the gate current
  characteristics.

9
SHE Band-diagram

• The gate current is the sum of the SHE current
  and the tunnel current.
• The SHE current has an energy distribution at
  the interface determined mainly by the substrate
  bias, and a density that can be controlled using
  the injector bias.
• The tunneling carriers have energy in the
  silicon that corresponds approximately to the
  bottom of the conduction band and a density that
  is determined by the gate bias.

10
tbd versus Vg for CVS and SHE

• The tbd versus Vg characteristics are inversely
  proportional to the gate current density.

11
Representative Dit Build-Up

• For thin oxides (lt 3.0 nm) the number of
  interface states at breakdown measured using
  charge-pumping is approximately independent of
  thickness and stress condition (CVS vs. SHE).

12
Summary SHE and CVS Reliability

• The results confirm that energetic electrons are
  responsible for degradation and breakdown of
  ultra-thin silicon dioxide.

Energetic Carrier Models
Anode Hole Injection
Trap Creation/ Hydrogen Release
13
Experimental for SHH/CVS

• P-channel MOSFETs
• lt100gt silicon, Nd 2x1017 cm-3
• tox 2.0 nm and 3.0 nm
• p polysilicon
• Constant Voltage Stress (CVS),
• Vg ltgt 0, Vs Vd Vb 0
• Substrate Hot Hole (SHH) Stress,
• Vg lt 0, Vs Vd 0, Vb gt 0, Vinj gt Vb
• Stress Induced Leakage Current (SILC)
• and sinusoidal Charge Pumping (CP)
• used to monitor electrically active defects.

SHH set-up
14
Substrate Hot Hole (SHH) Injection vs. Anode
Hole Injection (AHI)
Proposed AHI on a n-channel MOSFET (Vg gt 0,
Vb0)
SHHI on a p-channel MOSFET (Vglt0, Vbgt0, VinjgtVb)
15
Gate Voltage Dependence of CP Defects Produced
by SHH Stress

• For SHH stress, the number
• of defects produced per hole
• injected is independent of gate
• voltage (oxide field).
• The Nbd for SHH stress is
• much greater than the Nbd for
• CVS.
• SILC and CP show similar
• results.

Breakdown occurred after the final defect
measurement shown
16
Temperature Dependence of Defect Generation by
SHHs

• Defect generation by SHHs is
• decreased at higher temperatures.
• Defect generation by CVS is
• increased at higher temperatures.

17
Temperature Dependence of Breakdown by SHHs

• Breakdown by SHHs is observed
• to have a very weak dependence
• on temperature as compared to
• breakdown by CVS.
• This is because defect generation
• by the trapping of holes has a very
• weak temperature dependence.

18
Effect of Trapped Holes on CVS Breakdown

• These results show that prior
• injection of holes does not result
• in a reduction of subsequent CVS
• Qbd.
• This again illustrates the
• ineffectiveness of defects
• generated by holes to cause
• breakdown.
• This suggests that the recently
• theorized1 hole-catalyzed
• thermochemical electric field
• model is incorrect.

1J. W. McPherson et al., J. Appl. Phys. 88, 5351
(2000).
19
Calculation of Ratio of Holes to Electrons
1) Measure the tunnel gate current Itun
(Vb0,Vg) 2) For high Vinj, measure the SHH gate
current at the Vb of interest Ishh (Vbgt0, Vg,
Vinjhigh) 3) For high Vinj, measure the hot
hole current impinging the interface Ids (Vbgt0,
Vg, Vinjhigh) 4) For high Vinj, calculate the
transmission coefficient for hot holes T ?
Ishh (Vbgt0, Vg, Vinjhigh) - Itun (Vb0,Vg)/Ids
(Vbgt0, Vg) Assume that T is independent of
injector bias and time. 6) Calculate the ratio
of holes to electrons for any injector bias R
? T x Ids (Vbgt0, Vg, Vinj)/ Itun (Vb0,Vg)
20
Bulk Defect Generation Measured Using SILC

• The rate of bulk defect
• generation as a function of hole
• charge injected is observed to
• be similar even though the ratio
• of hole current to tunneling
• current is different by two orders
• magnitude.

21
Effect of Hole to Electron Ratio on Subsequent
Breakdown

• SHHs were injected to 2 C/cm2
• with two different hole to electron
• ratios.
• The number of defects at
• 2 C/cm2 was on the order of the
• CVS Nbd.
• Subsequent CVS breakdown
• measurements were performed.
• The breakdown susceptibility
• was not effected by the hole to
• electron ratio.

22
Summary

• The energetic electron fluence entering the
  anode is responsible for
• the degradation and breakdown of ultra-thin
  silicon dioxide under
• constant voltage tunneling stress.
• Hole trapping alone can not explain the
  breakdown of ultra-thin
• silicon dioxide under tunneling stress
  conditions.
• The interaction of tunneling electrons with
  defects created by holes
• is not a viable mechanism for explaining the
  catastrophic breakdown
• of ultra-thin silicon dioxide.
• The results imply that hydrogen release from the
  anode may be
• responsible for the degradation and breakdown of
  ultra-thin silicon
• dioxide.