Non-Uniform Degradation Behavior Across Device Width in RF Power - - PowerPoint PPT Presentation

Non-Uniform Degradation Behavior Across Device Width in RF Power GaAs PHEMTs

• A. A. Villanueva1, J. A. del Alamo1,
• T. Hisaka2, K. Hayashi2

and M. Somerville3

1Massachusetts Institute of Technology 2Mitsubishi Electric 3Olin College of Engineering

Sponsor: Mitsubishi Electric

Motivation

• Electrical degradation is serious

concern in RF power GaAs PHEMTs

– Under stressing: RD and Imax → Pout

• Degradation mechanisms identified [1]-

[3], but no studies of uniformity

• This study: investigate degradation

across device width

[1] del Alamo et al (IEDM 2004) [2] Meneghesso et al (1996) [3] Hisaka et al (GaAs IC 2003) S G D

W

Outline

• Introduction
• Experimental
• PHEMT Degradation

– Light Emission

• TLM Degradation

– Light Emission – Materials Analysis

• Conclusions

Introduction

• hmics
• Experimental RF power PHEMTs
• Lg = 0.25 μm, Wg = 50 μm

ft ~ 40-50 GHz, BVDG,off ~ 12-15 V

Source Gate Drain

channel supply cap etch-stop supply buffer

With step-stress:

• RD
• Imax (from VT )

InGaAs AlGaAs AlGaAs n- GaAs n+ GaAs GaAs

Stressing: ID = 400 mA/mm, step VDGo+VT. In air @ 300 K.

0.90 0.95 1.00 1.05 1.10 100 200 300 400 500 600 700 time [min] RD/RD(0), Imax/Imax(0)

6.0 6.2 6.4 6.6 6.8 7.0 7.2 7.4

VDGo+VT [V]

RD

Imax VDGo+VT

Light Emission & Degradation

• RD degradation due to

surface corrosion [3], high E involved

• High E impact

ionization (II) recombination light emission

• Light-emission picture:

spatial view of II, E

G D S

+

• +
• hν

hν +

• [3] Hisaka et al (GaAs IC 2003)

100 200 300 400 500 600 700 800 900 6.4 6.6 6.8 7.0 7.2 7.4 7.6 7.8 8.0 8.2 time [min] VDS [V]

Light-emission: Experimental

• Astronomical-grade

CCD sensor

• Stressing: constant

VGS & constant VDS

– VDS stepped

• Photographs taken:

– at frequent intervals – at fixed (low) value of VDS

camera DUT probes microscope

VGS =0.3 V

200 400 600 800 2 4 6 8 x 10

5

time [min] Ihv / ID [a.u.] 200 400 600 800 6.6 7.0 7.4 7.8 8.2 VDS [V]

• VDS → Ihν
• For constant VDS,

Ihν constant, but eventually

RD → VDGo

VDS

Light-Emission vs. Stressing (1)

Ihν/ID

VGS =0.3 V

200 400 600 800 2 4 6 8 x 10

5

time [min] Ihv / ID [a.u.] 200 400 600 800 6.6 7.0 7.4 7.8 8.2 VDS [V]

• VDS → Ihν
• For constant VDS,

Ihν constant, but eventually

RD → VDGo

VDS

Light-Emission vs. Stressing (1)

Ihν/ID

VGS =0.3 V

Light-Emission vs. Stressing (2)

45 µ m

t = 0 min 218 min 428 min 638 min 849 min

30 µm

S D

VDS = 6.6 V

45 µm W=50 µm

VGS = 0.3 V

• Initially, light concentrates in center ~30 µm of width
• With stressing: (1) light spreads out along width

(2) weakens in intensity

200 400 600 800 1 2 3 4 5 6 time [min] Ihv / ID [a.u.]

10 20 30 40 50 60 2 4 6 8 10 12 14 width (um) Ihv / ID [a.u.]

Light Emission vs. Width

t = 0 min t =849 min t = 428 min

VDS = 6.6 V light from source side VGS = 0.3 V

• 1st half: light spreads out → Ihv
• 2nd half: intensity decreases → Ihv

Ihν / ID [a.u.]

VDS = 6.6 V VGS = 0.3 V

width [μm] ihv / ID [a.u.]

Light Emission During Stressing

• During stressing, at high bias:
• early stages: degradation peaks in center
• advanced stages: degradation peaks at edges

10 20 30 40 50 60 20 40 60 80 100 120 140 160 width [um] Ihv / ID [a.u.] t = 0 min, VDS = 6.6 V t = 218 min, VDS = 7.0 V t = 428 min, VDS = 7.4 V t = 639 min, VDS = 7.8 V

device width

width [μm]

200 400 600 800 2 4 6 8 x 10

5

time [min] Ihv / ID [a.u.] 200 400 600 800 6.6 7.0 7.4 7.8 8.2 VDS [V]

ihv / ID [a.u.]

50 100 150 200 1 2 3 x 10

5

time [min] Ihv / ID [a.u.] 50 100 150 200 5.00 5.25 5.50 5.75 6.00 VD [V]

VD

Light-Emission of TLMs (1)

Ihν / ID

TLM: same structure as PHEMT, but no gate

• VD → Ihv
• Constant VD → Ihv

R → II

+

• +
• hν

hν +

Light Emission of TLMs (2)

• Light initially

concentrated in center

• With stressing:

– Light spreads out over width of TLM – Ihν (for constant voltage) (Similar to PHEMT light emission behavior)

t = 0 min 67 min 100 min 133 min 167 min 201 min 90 µm W=100 µm 55 µm VD = 5.0 V 5.0 V 5.25 V 5.50 V 5.75 V 6.0 V

20 40 60 80 100 0.5 1.0 1.5 2.0 2.5 3.0 width (μm) Ihv / ID [a.u.]

Light Emission vs. Width (TLMs)

• During stressing, at high bias, light “peaks” at edges

→ similar behavior in PHEMTs

t = 0 min, VD = 5 V t = 201 min, VD = 6 V t = 100 min, VD = 5.25 V

width [μm] ihv / ID [a.u.] device width

Origin of Non-Uniform II

3 possible causes for non-uniform II across device width:

• Non-uniform ID
• Non-uniform T
• Non-uniform E-field

Non-Uniform Drain Current ?

• Non-uniform ID

– but II ∝ ID, so requires edges be “shut off”

G S D ID

Non-Uniform Temperature?

• Non-uniform T

– but edges should be cooler → more II

G S D

Non-Uniform Electric Field?

• Non-uniform E-field

– II ∝ exp(-1/E), small ΔE → large ΔII – from non-uniform recess

G S D

Recess Non-Uniformity (TLMs)

x y

narrower wider

L=2.4 µm

Examined top view of entire recess area recess is shorter in the center

W = 60 μm

10 20 30 40 50 60 0.5 0.6 0.7 0.8 0.9 1.0 x [um] length [um ]

Recess vs. Width (TLMs)

• Nominal recess: 0.7 µm
• Actual recess varies:

– Center: ~0.6-0.7 µm – Edges: ~0.8-0.9 µm

In center: electric field → II → degradation Same phenomenon likely happening in PHEMTs

AFM W = 60 µm

width [μm]

Conclusions

• Non-uniform recess geometry non-uniform E
• Areas of higher E areas more susceptible to

degradation

• To improve long-term device reliability: must

identify & minimize non-uniformities in device geometry