Progress in the Development of

1
Progress in the Development of MBE-grown HEMTs
on SiC(0001)
Christiane Poblenz, Siddharth Rajan, Andrea
Corrion, Umesh K. Mishra and James
Speck Materials and ECE Departments University
of California Santa Barbara, CA
CANE Review April 12, 2005
2
Outline

• Review of current HEMT structure and progress
• -TD reduction (two-step buffer growth)
• -Uniformity (modulated growth)
• Buffer Leakage
• - Carbon Doping
• - Controlled growth of AlN on SiC
• Fe-doping for semi-insulating GaN (Andrea
  Corrion)
• Conclusions

3
Typical Growth Structure
AlGaN Surface (5 µm x 5 µm) 20 nm vertical scale
Standard Al0.30Ga0.70N Cap, 715 C
Two-Step GaN Buffer, 715 C Step 1 TD
reduction Step 2 morphology recovery TDD 1-3
x 1010 cm-3 Modulated Growth for uniformity
AlN Initiation, 740 C Demonstrated Effect of
Al/N Ratio on Leakage
CREE SiC 0.5 µm Ti Backside Metal CMP by
NovaSiC Pre-growth in-situ Ga polish
SiC Post-CMP Surface (5 µm x 5 µm)
4
Defect Reduction via High Temperature
Intermediate Growth
Previously established that intermediate growth
conditions give lowest defect density Via TD line
length minimization and subsequent TD-TD reaction
Higher temperatures can be achieved during
ammonia GaN MBE growth (900C ammonia vs
750C plasma) Future work will investigate
ammonia growth for low TDD buffers in new MBE
system
5
Temperature Sensitivity
AFM 5 µm 5 µm Non-bonded, ¼ Wafer Growth
DROPLETS
PITTING
6
Morphology Control Modulated Growth
Uniform, smooth, pit-free surfaces demonstrated
over ¼ wafer and 2 wafers (non-bonded growth)
t1
t2
FGa2
flux
FGa1
FN
time
5x5 µm AFM
50x50 µm AFM
Modulate surface coverage of Ga between (a) and
(c) Time-averaged surface coverage at (b)
Powerful in-situ monitoring RHEED QMS
7
Buffer Leakage Standard AlGaN/GaN HEMT
Problem High Leakage in UID GaN Buffers
Source-Drain I-V curves, gate region etched
Carrier Concentration 1016 cm-3 ( from C-V) Low
FMAX, Low POUT, Low PAE in HEMTs (4.2 W/mm _at_ 10
GHz, 19.5 PAE, Vds 35V)
Solution Carbon Doping via CBr4
Carrier Concentration REDUCED (1015 cm-3 from
C-V) 7.3 W/mm _at_ 10 GHz, 36 PAE, Vds 40V 12
W/mm _at_ 4 GHz, 46 PAE, Vds 70V 15.6 W/mm _at_ 4
GHz, 55 PAE, Vds 50V (with field plate)
Optimized level and thickness of carbon doping
for leakage reduction
Carbon doping is effective, but source of
carriers is unknown
8
AlN/GaN Interface Leakage Study in UID HEMTs
Al is known to significantly enhance the
incorporation of oxygen in AlGaN compared to
GaN Impact on AlN/GaN interface near SiC
substrate ?
9 nm AlGaN cap
Upper-Intermediate to Ga-Droplet GaN
(UID)
600 nm, T730ºC
100 nm, T730ºC
Al
GaN

AlN
45 nm, T740ºC
??
AlN
CMP SiC
Variable Al/N Ratio during AlN nucleation layer
9
AlN/GaN Interface Study Electrical Properties
9 nm AlGaN cap
Upper-Intermediate to Ga-Droplet GaN
600 nm, T730ºC
100 nm, T730ºC
45 nm, T740ºC Variable Al/N Ratio
AlN
CMP SiC
Al/Nlt1 buffer leakage as low as with standard
carbon doping scheme Presence of Al droplets
leads to dramatic increase in buffer
leakage Exposure of Al doplets to N flux does
not improve leakage Initial Power Performance
4.8 W/mm _at_ 4 GHz, 62 PAE, Vds30V
8.1 W/mm _at_ 4 GHz, 38
PAE, Vds50V
10
Impurity Incorporation SIMS
Si incorporation from SiC substrate contributes
to leakage in UID GaN buffers

• Reaction of Al with SiC surface during Al-rich
  NL Si soluble in liquid Al
• N-rich AlN results in buffer leakage as low as
  buffers with carbon doping

UID GaN (Ga/Ngt1, Ga droplet regime)
GaN (250 nm)
Variable Al/N ratio
AlN (250 nm)
SI 4H-SiC
11
Conclusions Future Work

• -Two-step growth method for TD reduction
• Demonstrated higher temperature growth can
  further reduce TDs
• Introduced modulated growth method for uniformity
• Carbon doping has been optimized for successful
  buffer leakage reduction
• Demonstrated effect of Al/N ratio on Si
  incorporation for AlN on SiC
• Si incorporation under Al-rich conditions is a
  likely contributor to buffer leakage

Current Future Work -Cap optimization
Introduce thin AlN interlayer (AlGaN/AlN/GaN
strucutres), Si doping, Graded AlGaN
caps -Focus on origin and elimination of gate
leakage
12
Fe-doping Outline

• Motivation
• Ga flux series dependence of Fe incorporation
  on growth regime
• Fe flux series
• Tsub series
• TEM
• CV data
• Conclusions/Future work

13
Fe-doping Motivation

• MOCVD highly-resistive GaN films obtained by
  Fe-doping
• MOCVD Fe-doped templates commercially available
• MBE carbon doping has been successful in
  reducing HEMT buffer leakage
• What about Fe-doping GaN by MBE?
• Little information on MBE of Fe-doped GaN in
  literature

14
Fe doping PA MBE

• Samples grown on MOCVD GaN templates
• Thin AlGaN layers inserted as markers
• Fe determined by SIMS
• Ga flux series
• Fe flux series
• Tsub series

a
1 um
b
Intermediate
c
d
AFM surface morphology unaffected by Fe doping
(for low Fe)
Heying et. al, J. App. Phys., 88 (1855) 2000
15
Ga flux series Ga droplet
TFe 1000 C

• No Fe incorporation in nominally Fe-doped layer
• SIMS image of Fe ion shows inhomogeneity in
  first layer

SIMS Fe ion signal (yellow)
AFM droplets highlighted
Fe concentrates in liquid Ga droplets, preventing
incorporation?
16
Ga flux series Ga droplet (2)
TFe 840 C

• Repeat, with lower Fe flux
• No incorporation into nominally Fe-doped layer
  inhomogeneous incorporation in top layer

Fe
Al
Fe appears to float on surface during growth,
preferentially concentrating in Ga droplets
17
Ga flux series Intermediate
TFe 840 C

• High intermediate

1 um
Fe
Al

• Poor SIMS depth resolution due to rough surface
• No Fe incorporation in nominally Fe-doped layer

Liquid-like Ga adlayer prevents Fe incorporation?
18
Ga flux series Intermediate (2)
TFe 840 C

• Low intermediate

Al
Fe

• Poor SIMS depth resolution due to rough surface
• Fe signal smeared out low Fe throughout
  sample

Fe doesnt incorporate controllably in
intermediate growth regime
19
Ga flux series N-rich
TFe 840 C
Typical AFM

• Fe incorporation in Fe-doped layer
• Extra peak at MOCVD/MBE interface
• Unintentional pre-growth Fe accumulation
• Occurs on all N-rich samples

Al
Fe
MOCVD/MBE interface Fe peak
Fe incorporates in N-rich growth regime
20
Fe flux series N-rich
TFe 1000 C

• Doping profile sharp turn on, slow turn off
  suggests surface segregation
• Plot logFe vs. 1/kTFe linear w/slope -3.51
• Literature value of Qsub,Fe 4.13 eV

920 C
Al
Fe
840 C
Fe concentration controllable in N-rich growth
regime
Smithells Met. Ref. Book, 6th ed.,
Butterworths, 1983
21
Tsub series N-rich
TFe 920 C
Tsub 650 C

• Little difference in Fe for N-rich samples
  grown at 650, 750 C
• Higher Tsub rougher material, poor depth
  resolution in SIMS

Al
Fe
750 C
Tsub doesnt have a significant effect on Fe
22
Recovered N-rich Fe-doped GaN
UID Ga-rich GaN
UID Ga-rich GaN
FeGaN

• Ga-rich, UID GaN interrupted growth to dry
  surface of Ga, lower Ga flux
• 100 nm of N-rich Fe-doped GaN
• Increased Ga flux - continued Ga-rich, UID growth
  to recover surface morphology

• Sharp turn-on, turn-off in Fe profile

Currently evaluating structural quality and
electrical properties of N-rich Fe-doped films
with recovered surface morphology
23
TEM
UID GaN
FeGaN
UID GaN
MOCVD GaN
500 nm
500 nm
g 1120
g 0002

• BF TEM no significant structural degradation of
  films with N-rich Fe-doped layers
• N-rich surface morphology can be recovered with
  Ga-rich overgrowth

24
Surface Morphology - High Fe
TFe1250 C
High Fe flux Ga droplet regime

• Optical Microscopy formation of dendritic
  phase on surface
• As TFe ?, Ga droplets ? , quantity of second
  phase ?
• Phase not seen for TFe lt 1075 C

1175 C
1125 C
1075 C

• AFM between second phase/droplets, material
  unaffected

20x
25
Ga-Fe Phase Diagram

• Tsub650C, liquid Ga in equilibrium with
  tetragonal FeGa3
• For high Fe fluxes, expect phase segregation in
  presence of droplets
• XRD extra peaks for highest Fe flux
  corresponding to second phase

26
CV measurements

• Three N-rich samples measured one UID and two
  Fe-doped
• All three show large increase in charge near
  MOCVD/MBE interface
• No clear trend with increasing Fe
• Hall measurements of Fe-doped films underway

27
Conclusions/Future work

• Fe incorporation strongly dependent on growth
  regime
• Little/no incoporation under Ga-rich conditions
• Controllable incorporation under N-rich
  conditions
• Tsub has little effect on Fe incorporation
  (unlike C)
• TEM no significant structural degradation due
  to Fe-doping
• Currently performing structural/electrical
  characterization of recovered N-rich Fe-doped
  films
• XRD/TEM
• Hall resistivity, carrier concentration,
  mobility
• Quadropole mass spectrometry (QMS) how does Fe
  affect Ga wetting layer?