U.S. patent application number 11/451298 was filed with the patent office on 2006-12-14 for damage evaluation method of compound semiconductor member, production method of compound semiconductor member, gallium nitride compound semiconductor member, and gallium nitride compound semiconductor membrane.
This patent application is currently assigned to SUMITOMO ELECTRIC INDUSTRIES, LTD.. Invention is credited to Akihiro Hachigo, Keiji Ishibashi, Takayuki Nishiura.
Application Number | 20060281201 11/451298 |
Document ID | / |
Family ID | 36930381 |
Filed Date | 2006-12-14 |
United States Patent
Application |
20060281201 |
Kind Code |
A1 |
Hachigo; Akihiro ; et
al. |
December 14, 2006 |
Damage evaluation method of compound semiconductor member,
production method of compound semiconductor member, gallium nitride
compound semiconductor member, and gallium nitride compound
semiconductor membrane
Abstract
A method of evaluating damage of a compound semiconductor
member, comprising: a step of performing spectroscopic ellipsometry
measurement on a surface of the compound semiconductor member; and
a step of evaluating damage on the surface of the compound
semiconductor member, using a spectrum in a wavelength band
containing a wavelength corresponding to a bandgap of the compound
semiconductor member, in a spectrum of an optical constant obtained
by the spectroscopic ellipsometry measurement.
Inventors: |
Hachigo; Akihiro;
(Itami-shi, JP) ; Nishiura; Takayuki; (Itami-shi,
JP) ; Ishibashi; Keiji; (Itami-shi, JP) |
Correspondence
Address: |
MCDERMOTT WILL & EMERY LLP
600 13TH STREET, N.W.
WASHINGTON
DC
20005-3096
US
|
Assignee: |
SUMITOMO ELECTRIC INDUSTRIES,
LTD.
|
Family ID: |
36930381 |
Appl. No.: |
11/451298 |
Filed: |
June 13, 2006 |
Current U.S.
Class: |
438/16 |
Current CPC
Class: |
G01N 21/211
20130101 |
Class at
Publication: |
438/016 |
International
Class: |
H01L 21/66 20060101
H01L021/66; G01R 31/26 20060101 G01R031/26 |
Foreign Application Data
Date |
Code |
Application Number |
Jun 13, 2005 |
JP |
P2005-172567 |
Claims
1. A method of evaluating damage of a compound semiconductor
member, comprising: a step of performing spectroscopic ellipsometry
measurement on a surface of the compound semiconductor member; and
a step of evaluating damage on the surface of the compound
semiconductor member, using a spectrum in a wavelength band
containing a wavelength corresponding to a bandgap of the compound
semiconductor member, in a spectrum of an optical constant obtained
by the spectroscopic ellipsometry measurement.
2. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the optical constant is an
imaginary part of a complex index of refraction, and wherein the
step of evaluating the damage uses a maximum absolute value of a
slope of the spectrum in the wavelength band.
3. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the optical constant is an
imaginary part of a complex index of refraction, and wherein the
step of evaluating the damage uses an absolute value of an extremum
of a first derivative of the spectrum in the wavelength band.
4. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the optical constant is an
imaginary part of a complex index of refraction, and wherein the
step of evaluating the damage uses a wavelength at which an
absolute value of a slope of the spectrum in the wavelength band is
maximum.
5. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the optical constant is an
imaginary part of a complex index of refraction, and wherein the
step of evaluating the damage uses a maximum of the spectrum in the
wavelength band.
6. A method of evaluating damage of a compound semiconductor
member, comprising: a step of performing spectroscopic ellipsometry
measurement on a surface of the compound semiconductor member; and
a step of evaluating damage on the surface of the compound
semiconductor member, using a peak in a separate wavelength band
located on a longer wavelength side than a wavelength band
containing a wavelength corresponding to a bandgap of the compound
semiconductor member, in a spectrum of an imaginary part of a
complex index of refraction obtained by the spectroscopic
ellipsometry measurement.
7. A method of evaluating damage of a compound semiconductor
member, comprising: a step of performing spectroscopic ellipsometry
measurement on a surface of a damage layer in the compound
semiconductor member having a compound semiconductor region and the
damage layer provided on the compound semiconductor region; and a
step of evaluating damage on the surface of the damage layer in the
compound semiconductor member, using a peak appearing in a spectrum
of an imaginary part of a complex index of refraction with at least
one reflection of light between the compound semiconductor region
and the damage layer, in the spectrum of the imaginary part of the
complex index of refraction obtained by the spectroscopic
ellipsometry measurement.
8. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the optical constant is an
imaginary part of a complex dielectric constant, and wherein the
step of evaluating the damage uses a maximum absolute value of a
slope of the spectrum in the wavelength band.
9. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the optical constant is an
imaginary part of a complex dielectric constant, and wherein the
step of evaluating the damage uses an absolute value of an extremum
of a first derivative of the spectrum in the wavelength band.
10. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the optical constant is an
imaginary part of a complex dielectric constant, and wherein the
step of evaluating the damage uses a wavelength at which an
absolute value of a slope of the spectrum in the wavelength band is
maximum.
11. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the optical constant is an
imaginary part of a complex dielectric constant, and wherein the
step of evaluating the damage uses a maximum of the spectrum in the
wavelength band.
12. A method of evaluating damage of a compound semiconductor
member, comprising: a step of performing spectroscopic ellipsometry
measurement on a surface of the compound semiconductor member; and
a step of evaluating damage on the surface of the compound
semiconductor member, using a peak in a separate wavelength band
located on a longer wavelength side than a wavelength band
containing a wavelength corresponding to a bandgap of the compound
semiconductor member, in a spectrum of an imaginary part of a
complex dielectric constant obtained by the spectroscopic
ellipsometry measurement.
13. A method of evaluating damage of a compound semiconductor
member, comprising: a step of performing spectroscopic ellipsometry
measurement on a surface of a damage layer in the compound
semiconductor member having a compound semiconductor region and the
damage layer provided on the compound semiconductor region; and a
step of evaluating damage on the surface of the damage layer in the
compound semiconductor member, using a peak appearing in a spectrum
of an imaginary part of a complex dielectric constant with at least
one reflection of light between the compound semiconductor region
and the damage layer, in the spectrum of the imaginary part of the
complex dielectric constant of refraction obtained by the
spectroscopic ellipsometry measurement.
14. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the optical constant is a real
part of a complex index of refraction, and wherein the step of
evaluating the damage uses a maximum absolute value of a slope in a
portion located on a shorter wavelength side than a wavelength
corresponding to a maximum in the spectrum in the wavelength
band.
15. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the optical constant is a real
part of a complex index of refraction, and wherein the step of
evaluating the damage uses a maximum absolute value of a slope in a
portion located on a longer wavelength side than a wavelength
corresponding to a maximum in the spectrum in the wavelength
band.
16. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the optical constant is a real
part of a complex index of refraction, and wherein the step of
evaluating the damage uses a wavelength at which an absolute value
of a slope in a portion located on a shorter wavelength side than a
wavelength corresponding to a maximum in the spectrum in the
wavelength band is maximum.
17. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the optical constant is a real
part of a complex index of refraction, and wherein the step of
evaluating the damage uses a maximum of the spectrum in the
wavelength band.
18. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the optical constant is a real
part of a complex dielectric constant, and wherein the step of
evaluating the damage uses a maximum absolute value of a slope in a
portion located on a shorter wavelength side than a wavelength
corresponding to a maximum in the spectrum in the wavelength
band.
19. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the optical constant is a real
part of a complex dielectric constant, and wherein the step of
evaluating the damage uses a maximum absolute value of a slope in a
portion located on a longer wavelength side than a wavelength
corresponding to a maximum in the spectrum in the wavelength
band.
20. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the optical constant is a real
part of a complex dielectric constant, and wherein the step of
evaluating the damage uses a wavelength at which an absolute value
of a slope in a portion located on a shorter wavelength side than a
wavelength corresponding to a maximum in the spectrum in the
wavelength band is maximum.
21. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the optical constant is a real
part of a complex dielectric constant, and wherein the step of
evaluating the damage uses a maximum of the spectrum in the
wavelength band.
22. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the compound semiconductor
member is a compound semiconductor substrate.
23. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the compound semiconductor
member is a compound semiconductor membrane provided on a
substrate.
24. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the compound semiconductor
member is comprised of a monocrystalline material or
polycrystalline material.
25. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the bandgap is not less than
1.6.times.10.sup.-19 J.
26. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the compound semiconductor
member is comprised of a nitride compound semiconductor containing
at least one of B, Al, and Ga.
27. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the compound semiconductor
member is comprised of an oxide compound semiconductor containing
at least one of Be and Zn.
28. The damage evaluation method of the compound semiconductor
member according to claim 1, wherein the compound semiconductor
member is comprised of a ZnSe compound semiconductor.
29-46. (canceled)
47. A gallium nitride compound semiconductor member wherein an
absolute value of a difference between an imaginary part of a
complex index of refraction at 360 nm and an imaginary part of a
complex index of refraction at 370 nm obtained by spectroscopic
ellipsometry measurement on a surface is not less than 0.045.
48. A gallium nitride compound semiconductor member wherein an
absolute value of an imaginary part of a complex index of
refraction at 370 nm obtained by spectroscopic ellipsometry
measurement on a surface of the gallium nitride compound
semiconductor member is not more than 0.18.
49. A gallium nitride compound semiconductor member wherein in a
spectrum of an imaginary part of a complex index of refraction
obtained by spectroscopic ellipsometry measurement on a surface of
the gallium nitride compound semiconductor member, a wavelength at
which an absolute value of a slope of a spectrum in a wavelength
band of 300 to 400 nm is maximum, is not less than 350 nm.
50. A gallium nitride compound semiconductor member wherein an
absolute value of a difference between an imaginary part of a
complex dielectric constant at 360 nm and an imaginary part of a
complex dielectric constant at 370 nm obtained by spectroscopic
ellipsometry measurement on a surface of the gallium nitride
compound semiconductor member is not less than 0.24.
51. A gallium nitride compound semiconductor member wherein an
absolute value of an imaginary part of a complex dielectric
constant at 370 nm obtained by spectroscopic ellipsometry
measurement on a surface of the gallium nitride compound
semiconductor member is not more than 0.9.
52. A gallium nitride compound semiconductor member wherein in a
spectrum of an imaginary part of a complex dielectric constant
obtained by spectroscopic ellipsometry measurement on a surface of
the gallium nitride compound semiconductor member, a wavelength at
which an absolute value of a slope of a spectrum in a wavelength
band of 300 to 400 nm is maximum, is not less than 350 nm.
53. A gallium nitride compound semiconductor member wherein an
absolute value of a difference between a real part of a complex
index of refraction at 365 nm and a real part of a complex index of
refraction at 375 nm obtained by spectroscopic ellipsometry
measurement on a surface of the gallium nitride compound
semiconductor member is not less than 0.035.
54. A gallium nitride compound semiconductor member wherein in a
spectrum of a real part of a complex index of refraction obtained
by spectroscopic ellipsometry measurement on a surface of the
gallium nitride compound semiconductor member, a maximum of a
spectrum in a wavelength band of 300 to 400 nm is not less than
2.7.
55. A gallium nitride compound semiconductor member wherein an
absolute value of a difference between a real part of a complex
dielectric constant at 365 nm and a real part of a complex
dielectric constant at 375 nm obtained by spectroscopic
ellipsometry measurement on a surface of the gallium nitride
compound semiconductor member is not less than 0.13.
56. A gallium nitride compound semiconductor member wherein in a
spectrum of a real part of a complex dielectric constant obtained
by spectroscopic ellipsometry measurement on a surface of the
gallium nitride compound semiconductor member, a maximum of a
spectrum in a wavelength band of 300 to 400 nm is not less than
7.2.
57. A gallium nitride compound semiconductor member wherein a
thickness of a layer containing at least one of an oxide film and
an uneven layer formed on a surface of the gallium nitride compound
semiconductor member is not more than 6 nm.
58. The gallium nitride compound semiconductor member according to
claim 47, said gallium nitride compound semiconductor member being
a gallium nitride compound semiconductor substrate.
59. The gallium nitride compound semiconductor member according to
claim 47, said gallium nitride compound semiconductor member being
a gallium nitride compound semiconductor membrane provided on a
substrate.
60. A gallium nitride compound semiconductor membrane formed on a
gallium nitride compound semiconductor member wherein an absolute
value of a difference between an imaginary part of a complex index
of refraction at 360 nm and an imaginary part of a complex index of
refraction at 370 nm obtained by spectroscopic ellipsometry
measurement on a surface of the gallium nitride compound
semiconductor member is not less than 0.045.
61. A gallium nitride compound semiconductor membrane formed on a
gallium nitride compound semiconductor member wherein an absolute
value of an imaginary part of a complex index of refraction at 370
nm obtained by spectroscopic ellipsometry measurement on a surface
of the gallium nitride compound semiconductor member is not more
than 0.18.
62. A gallium nitride compound semiconductor membrane formed on a
gallium nitride compound semiconductor member wherein in a spectrum
of an imaginary part of a complex index of refraction obtained by
spectroscopic ellipsometry measurement on a surface of the gallium
nitride compound semiconductor member, a wavelength at which an
absolute value of a slope of a spectrum in a wavelength band of 300
to 400 nm is maximum, is not less than 350 nm.
63. A gallium nitride compound semiconductor membrane formed on a
gallium nitride compound semiconductor member wherein an absolute
value of a difference between an imaginary part of a complex
dielectric constant at 360 nm and an imaginary part of a complex
dielectric constant at 370 nm obtained by spectroscopic
ellipsometry measurement on a surface of the gallium nitride
compound semiconductor member is not less than 0.24.
64. A gallium nitride compound semiconductor membrane formed on a
gallium nitride compound semiconductor member wherein an absolute
value of an imaginary part of a complex dielectric constant at 370
nm obtained by spectroscopic ellipsometry measurement on a surface
of the gallium nitride compound semiconductor member is not more
than 0.9.
65. A gallium nitride compound semiconductor membrane formed on a
gallium nitride compound semiconductor member wherein in a spectrum
of an imaginary part of a complex dielectric constant obtained by
spectroscopic ellipsometry measurement on a surface of the gallium
nitride compound semiconductor member, a wavelength at which an
absolute value of a slope of a spectrum in a wavelength band of 300
to 400 nm is maximum, is not less than 350 nm.
66. A gallium nitride compound semiconductor membrane formed on a
gallium nitride compound semiconductor member wherein an absolute
value of a difference between a real part of a complex index of
refraction at 365 nm and a real part of a complex index of
refraction at 375 nm obtained by spectroscopic ellipsometry
measurement on a surface of the gallium nitride compound
semiconductor member is not less than 0.035.
67. A gallium nitride compound semiconductor membrane formed on a
gallium nitride compound semiconductor member wherein in a spectrum
of a real part of a complex index of refraction obtained by
spectroscopic ellipsometry measurement on a surface of the gallium
nitride compound semiconductor member, a maximum of a spectrum in a
wavelength band of 300 to 400 nm is not less than 2.7.
68. A gallium nitride compound semiconductor membrane formed on a
gallium nitride compound semiconductor member wherein an absolute
value of a difference between a real part of a complex dielectric
constant at 365 nm and a real part of a complex dielectric constant
at 375 nm obtained by spectroscopic ellipsometry measurement on a
surface of the gallium nitride compound semiconductor member is not
less than 0.13.
69. A gallium nitride compound semiconductor membrane formed on a
gallium nitride compound semiconductor member wherein in a spectrum
of a real part of a complex dielectric constant obtained by
spectroscopic ellipsometry measurement on a surface of the gallium
nitride compound semiconductor member, a maximum of a spectrum in a
wavelength band of 300 to 400 nm is not less than 7.2.
70. A gallium nitride compound semiconductor membrane formed on a
gallium nitride compound semiconductor member wherein a thickness
of a layer containing at least one of an oxide film and an uneven
layer formed on a surface of the gallium nitride compound
semiconductor member is not more than 6 nm.
Description
BACKGROUND OF THE INVENTION
[0001] 1. Field of the Invention
[0002] The present invention relates to a damage evaluation method
of a compound semiconductor member, a production method of a
compound semiconductor member, a gallium nitride compound
semiconductor member, and a gallium nitride compound semiconductor
membrane.
[0003] 2. Related Background Art
[0004] Compound semiconductors have various merits in comparison
with Si. For example, the compound semiconductors permit control of
the bandgap through adjustment of compositions. Furthermore, the
compound semiconductors have such optical properties as direct
transition and wide bandgap, and are thus suitably applied to
optical devices such as LEDs or LDs. Since the compound
semiconductors have high carrier mobility, they are also suitably
applied to high-speed devices.
[0005] In producing such compound semiconductor devices as the
optical devices or high-speed devices, a substrate used is a
compound semiconductor substrate, or a laminated substrate in which
a compound semiconductor membrane is formed on an amorphous
substrate such as a glass substrate. For example, a compound
semiconductor membrane or electrodes are formed on a surface of the
compound semiconductor substrate or laminated substrate. The device
characteristics of the compound semiconductor devices are
significantly affected by an interface between the compound
semiconductor substrate or laminated substrate and the compound
semiconductor membrane or by interfaces between the compound
semiconductor substrate or laminated substrate and the electrodes.
Therefore, it is important to evaluate the interfaces in the
compound semiconductor devices.
[0006] In production of the compound semiconductor devices, damage
occurs on the foregoing interfaces in several production processes.
For example, since surface roughness of the compound semiconductor
substrate or laminated substrate affects the device
characteristics, the surface of the compound semiconductor
substrate or laminated substrate is subjected to polishing or
etching. This process produces scratches or distortion on the
surface to cause damage on the surface. For example, dry etching or
wet etching or the like is used in forming a thin film or fine
pattern of nanometer size on the surface of the compound
semiconductor substrate or laminated substrate. At this time,
damage is caused on the surface of the compound semiconductor
substrate or laminated substrate or on the surface of the thin film
or fine pattern.
[0007] When a compound semiconductor device is produced, for
example, by growing an epitaxial film on the surface of the
compound semiconductor substrate or compound semiconductor membrane
with the surface including the damage as described above, the
device characteristics are degraded by virtue of the damage
existing at the interface between the compound semiconductor
substrate or compound semiconductor membrane and the epitaxial
film.
[0008] Methods for evaluating the damage on the surface of the
compound semiconductor substrate or compound semiconductor membrane
include methods using X-ray diffraction, scanning electron
microscope (SEM), cathodoluminescence, or the like as usually
adopted methods.
[0009] On the other hand, Japanese Patent Application Laid-Open No.
11-87448 discloses a method of evaluating damage on a substrate by
ellipsometry. This method is to evaluate a depth of a damage layer
or a level of damage from a change rate of intensity of reflected
light from the substrate.
[0010] Japanese Patent Application Laid-Open No. 2005-33187
discloses a method of subjecting a wafer to an etching treatment
and evaluating a surface structure of the wafer after the etching
treatment by ellipsometry. This method is to project polarized
light onto the wafer and to evaluate the surface structure of the
wafer, using polarized light reflected from the wafer.
Specifically, the surface structure of the wafer is estimated from
a phase difference .DELTA. and an amplitude ratio .psi. between the
polarized light projected onto the wafer and the polarized light
reflected from the wafer.
SUMMARY OF THE INVENTION
[0011] However, the methods disclosed in the foregoing Patent
Applications are not satisfactory yet in accuracy of damage
evaluation on a surface of a compound semiconductor member, and
there are desires for further improvement in the accuracy.
[0012] An object of the present invention is therefore to provide a
damage evaluation method of a compound semiconductor member and a
production method of a compound semiconductor member capable of
evaluating a level of damage on a surface with high accuracy and to
provide a gallium nitride compound semiconductor member and a
gallium nitride compound semiconductor membrane with a low level of
damage.
[0013] In order to solve the above problem, a damage evaluation
method of a compound semiconductor member according to the present
invention is (1) a method of evaluating damage of a compound
semiconductor member, comprising: a step of performing
spectroscopic ellipsometry measurement on a surface of the compound
semiconductor member; and a step of evaluating damage on the
surface of the compound semiconductor member, using a spectrum in a
wavelength band containing a wavelength corresponding to a bandgap
of the compound semiconductor member, in a spectrum of an optical
constant obtained by the spectroscopic ellipsometry measurement.
Here the term "spectrum of an optical constant" refers to data of
an optical constant at wavelengths, for example.
[0014] Preferably, (2) the optical constant is an imaginary part of
a complex index of refraction, and the step of evaluating the
damage uses a maximum absolute value of a slope of the spectrum in
the wavelength band.
[0015] Preferably, (3) the optical constant is an imaginary part of
a complex index of refraction, and the step of evaluating the
damage uses an absolute value of an extremum of a first derivative
of the spectrum in the wavelength band.
[0016] Preferably, (4) the optical constant is an imaginary part of
a complex index of refraction, and the step of evaluating the
damage uses a wavelength at which an absolute value of a slope of
the spectrum in the wavelength band is maximum.
[0017] Preferably, (5) the optical constant is an imaginary part of
a complex index of refraction, and the step of evaluating the
damage uses a maximum of the spectrum in the wavelength band.
[0018] Another damage evaluation method of a compound semiconductor
member according to the present invention is (6) a method of
evaluating damage of a compound semiconductor member, comprising: a
step of performing spectroscopic ellipsometry measurement on a
surface of the compound semiconductor member; and a step of
evaluating damage on the surface of the compound semiconductor
member, using a peak in a separate wavelength band located on a
longer wavelength side than a wavelength band containing a
wavelength corresponding to a bandgap of the compound semiconductor
member, in a spectrum of an imaginary part of a complex index of
refraction obtained by the spectroscopic ellipsometry
measurement.
[0019] Still another damage evaluation method of a compound
semiconductor member according to the present invention is (7) a
method of evaluating damage of a compound semiconductor member,
comprising: a step of performing spectroscopic ellipsometry
measurement on a surface of a damage layer in the compound
semiconductor member having a compound semiconductor region and the
damage layer provided on the compound semiconductor region; and a
step of evaluating damage on the surface of the damage layer in the
compound semiconductor member, using a peak appearing in a spectrum
of an imaginary part of a complex index of refraction with at least
one reflection of light between the compound semiconductor region
and the damage layer, in the spectrum of the imaginary part of the
complex index of refraction obtained by the spectroscopic
ellipsometry measurement.
[0020] Preferably, (8) the optical constant is an imaginary part of
a complex dielectric constant, and the step of evaluating the
damage uses a maximum absolute value of a slope of the spectrum in
the wavelength band.
[0021] Preferably, (9) the optical constant is an imaginary part of
a complex dielectric constant, and the step of evaluating the
damage uses an absolute value of an extremum of a first derivative
of the spectrum in the wavelength band.
[0022] Preferably, (10) the optical constant is an imaginary part
of a complex dielectric constant, and the step of evaluating the
damage uses a wavelength at which an absolute value of a slope of
the spectrum in the wavelength band is maximum.
[0023] Preferably, (11) the optical constant is an imaginary part
of a complex dielectric constant, and the step of evaluating the
damage uses a maximum of the spectrum in the wavelength band.
[0024] Another damage evaluation method of a compound semiconductor
member according to the present invention is (12) a method of
evaluating damage of a compound semiconductor member, comprising: a
step of performing spectroscopic ellipsometry measurement on a
surface of the compound semiconductor member; and a step of
evaluating damage on the surface of the compound semiconductor
member, using a peak in a separate wavelength band located on a
longer wavelength side than a wavelength band containing a
wavelength corresponding to a bandgap of the compound semiconductor
member, in a spectrum of an imaginary part of a complex dielectric
constant obtained by the spectroscopic ellipsometry
measurement.
[0025] Still another damage evaluation method of a compound
semiconductor member according to the present invention is (13) a
method of evaluating damage of a compound semiconductor member,
comprising: a step of performing spectroscopic ellipsometry
measurement on a surface of a damage layer in the compound
semiconductor member having a compound semiconductor region and the
damage layer provided on the compound semiconductor region; and a
step of evaluating damage on the surface of the damage layer in the
compound semiconductor member, using a peak appearing in a spectrum
of an imaginary part of a complex dielectric constant with at least
one reflection of light between the compound semiconductor region
and the damage layer, in the spectrum of the imaginary part of the
complex dielectric constant of refraction obtained by the
spectroscopic ellipsometry measurement.
[0026] Preferably, (14) the optical constant is a real part of a
complex index of refraction, and the step of evaluating the damage
uses a maximum absolute value of a slope in a portion located on a
shorter wavelength side than a wavelength corresponding to a
maximum in the spectrum in the wavelength band.
[0027] Preferably, (15) the optical constant is a real part of a
complex index of refraction, and the step of evaluating the damage
uses a maximum absolute value of a slope in a portion located on a
longer wavelength side than a wavelength corresponding to a maximum
in the spectrum in the wavelength band.
[0028] Preferably, (16) the optical constant is a real part of a
complex index of refraction, and the step of evaluating the damage
uses a wavelength at which an absolute value of a slope in a
portion located on a shorter wavelength side than a wavelength
corresponding to a maximum in the spectrum in the wavelength band
is maximum.
[0029] Preferably, (17) the optical constant is a real part of a
complex index of refraction, and the step of evaluating the damage
uses a maximum of the spectrum in the wavelength band.
[0030] Preferably, (18) the optical constant is a real part of a
complex dielectric constant, and the step of evaluating the damage
uses a maximum absolute value of a slope in a portion located on a
shorter wavelength side than a wavelength corresponding to a
maximum in the spectrum in the wavelength band.
[0031] Preferably, (19) the optical constant is a real part of a
complex dielectric constant, and the step of evaluating the damage
uses a maximum absolute value of a slope in a portion located on a
longer wavelength side than a wavelength corresponding to a maximum
in the spectrum in the wavelength band.
[0032] Preferably, (20) the optical constant is a real part of a
complex dielectric constant, and the step of evaluating the damage
uses a wavelength at which an absolute value of a slope in a
portion located on a shorter wavelength side than a wavelength
corresponding to a maximum in the spectrum in the wavelength band
is maximum.
[0033] Preferably, (21) the optical constant is a real part of a
complex dielectric constant, and the step of evaluating the damage
uses a maximum of the spectrum in the wavelength band.
[0034] Preferably, (22) the compound semiconductor member is a
compound semiconductor substrate. Preferably, (23) the compound
semiconductor member is a compound semiconductor membrane provided
on a substrate. Preferably, (24) the compound semiconductor member
is comprised of a monocrystalline material or polycrystalline
material. Preferably, (25) the bandgap is not less than
1.6.times.10.sup.-19 J.
[0035] Preferably, (26) the compound semiconductor member is
comprised of a nitride compound semiconductor containing at least
one of B, Al, and Ga. Preferably, (27) the compound semiconductor
member is comprised of an oxide compound semiconductor containing
at least one of Be and Zn. Preferably, (28) the compound
semiconductor member is comprised of a ZnSe compound
semiconductor.
[0036] A production method of a compound semiconductor member
according to the present invention is (29) a method of producing a
compound semiconductor member, comprising: a step of performing
spectroscopic ellipsometry measurement on a surface of the compound
semiconductor member; and a step of determining that the compound
semiconductor member is nondefective when a maximum absolute value
of a slope of a spectrum in a wavelength band containing a
wavelength corresponding to a bandgap of the compound semiconductor
member, in a spectrum of an imaginary part of a complex index of
refraction obtained by the spectroscopic ellipsometry measurement
is not less than a predetermined threshold.
[0037] Another production method of a compound semiconductor member
according to the present invention is (30) a method of producing a
compound semiconductor member, comprising: a step of performing
spectroscopic ellipsometry measurement on a surface of the compound
semiconductor member; and a step of determining that the compound
semiconductor member is nondefective when an absolute value of an
extremum of a first derivative of a spectrum in a wavelength band
containing a wavelength corresponding to a bandgap of the compound
semiconductor member, in a spectrum of an imaginary part of a
complex index of refraction obtained by the spectroscopic
ellipsometry measurement is not less than a predetermined
threshold.
[0038] Another production method of a compound semiconductor member
according to the present invention is (31) a method of producing a
compound semiconductor member, comprising: a step of performing
spectroscopic ellipsometry measurement on a surface of the compound
semiconductor member; and a step of determining that the compound
semiconductor member is nondefective when a wavelength at which an
absolute value of a slope of a spectrum in a wavelength band
containing a wavelength corresponding to a bandgap of the compound
semiconductor member, in a spectrum of an imaginary part of a
complex index of refraction obtained by the spectroscopic
ellipsometry measurement is maximum, is not less than a
predetermined threshold.
[0039] Another production method of a compound semiconductor member
according to the present invention is (32) a method of producing a
compound semiconductor member, comprising: a step of performing
spectroscopic ellipsometry measurement on a surface of the compound
semiconductor member; and a step of determining that the compound
semiconductor member is nondefective when a maximum absolute value
of a slope of a spectrum in a wavelength band containing a
wavelength corresponding to a bandgap of the compound semiconductor
member, in a spectrum of an imaginary part of a complex dielectric
constant obtained by the spectroscopic ellipsometry measurement is
not less than a predetermined threshold.
[0040] Still another production method of a compound semiconductor
member according to the present invention is (33) a method of
producing a compound semiconductor member, comprising: a step of
performing spectroscopic ellipsometry measurement on a surface of
the compound semiconductor member; and a step of determining that
the compound semiconductor member is nondefective when an absolute
value of an extremum of a first derivative of a spectrum in a
wavelength band containing a wavelength corresponding to a bandgap
of the compound semiconductor member, in a spectrum of an imaginary
part of a complex dielectric constant obtained by the spectroscopic
ellipsometry measurement is not less than a predetermined
threshold.
[0041] Another production method of a compound semiconductor member
according to the present invention is (34) a method of producing a
compound semiconductor member, comprising: a step of performing
spectroscopic ellipsometry measurement on a surface of the compound
semiconductor member; and a step of determining that the compound
semiconductor member is nondefective when a wavelength at which an
absolute value of a slope of a spectrum in a wavelength band
containing a wavelength corresponding to a bandgap of the compound
semiconductor member, in a spectrum of an imaginary part of a
complex dielectric constant obtained by the spectroscopic
ellipsometry measurement is maximum, is not less than a
predetermined threshold.
[0042] Another production method of a compound semiconductor member
according to the present invention is (35) a method of producing a
compound semiconductor member, comprising: a step of performing
spectroscopic ellipsometry measurement on a surface of the compound
semiconductor member; and a step of determining that the compound
semiconductor member is nondefective when a maximum absolute value
of a slope in a portion located on a shorter wavelength side than a
wavelength corresponding to a maximum of a spectrum in a wavelength
band containing a wavelength corresponding to a bandgap of the
compound semiconductor member, in a spectrum of a real part of a
complex index of refraction obtained by the spectroscopic
ellipsometry measurement is not less than a predetermined
threshold.
[0043] Another production method of a compound semiconductor member
according to the present invention is (36) a method of producing a
compound semiconductor member, comprising: a step of performing
spectroscopic ellipsometry measurement on a surface of the compound
semiconductor member; and a step of determining that the compound
semiconductor member is nondefective when a maximum absolute value
of a slope in a portion located on a longer wavelength side than a
wavelength corresponding to a maximum of a spectrum in a wavelength
band containing a wavelength corresponding to a bandgap of the
compound semiconductor member, in a spectrum of a real part of a
complex index of refraction obtained by the spectroscopic
ellipsometry measurement is not less than a predetermined
threshold.
[0044] Another production method of a compound semiconductor member
according to the present invention is (37) a method of producing a
compound semiconductor member, comprising: a step of performing
spectroscopic ellipsometry measurement on a surface of the compound
semiconductor member; and a step of determining that the compound
semiconductor member is nondefective when a wavelength at which an
absolute value of a slope in a portion located on a shorter
wavelength side than a wavelength corresponding to a maximum in a
spectrum in a wavelength band containing a wavelength corresponding
to a bandgap of the compound semiconductor member, in a spectrum of
a real part of a complex index of refraction obtained by the
spectroscopic ellipsometry measurement is maximum, is not less than
a predetermined threshold.
[0045] Another production method of a compound semiconductor member
according to the present invention is (38) a method of producing a
compound semiconductor member, comprising: a step of performing
spectroscopic ellipsometry measurement on a surface of the compound
semiconductor member; and a step of determining that the compound
semiconductor member is nondefective when a maximum of a spectrum
in a wavelength band containing a wavelength corresponding to a
bandgap of the compound semiconductor member, in a spectrum of a
real part of a complex index of refraction obtained by the
spectroscopic ellipsometry measurement is not less than a
predetermined threshold.
[0046] Another production method of a compound semiconductor member
according to the present invention is (39) a method of producing a
compound semiconductor member, comprising: a step of performing
spectroscopic ellipsometry measurement on a surface of the compound
semiconductor member; and a step of determining that the compound
semiconductor member is nondefective when a maximum absolute value
of a slope in a portion located on a shorter wavelength side than a
wavelength corresponding to a maximum of a spectrum in a wavelength
band containing a wavelength corresponding to a bandgap of the
compound semiconductor member, in a spectrum of a real part of a
complex dielectric constant obtained by the spectroscopic
ellipsometry measurement is not less than a predetermined
threshold.
[0047] Another production method of a compound semiconductor member
according to the present invention is (40) a method of producing a
compound semiconductor member, comprising: a step of performing
spectroscopic ellipsometry measurement on a surface of the compound
semiconductor member; and a step of determining that the compound
semiconductor member is nondefective when a maximum absolute value
of a slope in a portion located on a longer wavelength side than a
wavelength corresponding to a maximum of a spectrum in a wavelength
band containing a wavelength corresponding to a bandgap of the
compound semiconductor member, in a spectrum of a real part of a
complex dielectric constant obtained by the spectroscopic
ellipsometry measurement is not less than a predetermined
threshold.
[0048] Another production method of a compound semiconductor member
according to the present invention is (41) a method of producing a
compound semiconductor member, comprising: a step of performing
spectroscopic ellipsometry measurement on a surface of the compound
semiconductor member; and a step of determining that the compound
semiconductor member is nondefective when a wavelength at which an
absolute value of a slope in a portion located on a shorter
wavelength side than a wavelength corresponding to a maximum of a
spectrum in a wavelength band containing a wavelength corresponding
to a bandgap of the compound semiconductor member, in a spectrum of
a real part of a complex dielectric constant obtained by the
spectroscopic ellipsometry measurement is maximum, is not less than
a predetermined threshold.
[0049] Another production method of a compound semiconductor member
according to the present invention is (42) a method of producing a
compound semiconductor member, comprising: a step of performing
spectroscopic ellipsometry measurement on a surface of the compound
semiconductor member; and a step of determining that the compound
semiconductor member is nondefective when a maximum of a spectrum
in a wavelength band containing a wavelength corresponding to a
bandgap of the compound semiconductor member, in a spectrum of a
real part of a complex dielectric constant obtained by the
spectroscopic ellipsometry measurement is not less than a
predetermined threshold.
[0050] Preferably, (43) the compound semiconductor member is a
compound semiconductor substrate. Preferably, (44) the compound
semiconductor member is a compound semiconductor membrane provided
on a substrate.
[0051] Preferably, the production method of the compound
semiconductor member further comprises (45) a step of forming a
thin film on the surface of the compound semiconductor member,
after the step of determining that the compound semiconductor
member is nondefective.
[0052] Preferably, the production method of the compound
semiconductor member further comprises (46) a step of forming an
electrode on the surface of the compound semiconductor member,
after the step of determining that the compound semiconductor
member is nondefective.
[0053] A gallium nitride compound semiconductor member according to
the present invention is (47) a gallium nitride compound
semiconductor member wherein an absolute value of a difference
between an imaginary part of a complex index of refraction at 360
nm and an imaginary part of a complex index of refraction at 370 nm
obtained by spectroscopic ellipsometry measurement on a surface is
not less than 0.045.
[0054] Another gallium nitride compound semiconductor member
according to the present invention is (48) a gallium nitride
compound semiconductor member wherein an absolute value of an
imaginary part of a complex index of refraction at 370 nm obtained
by spectroscopic ellipsometry measurement on a surface of the
gallium nitride compound semiconductor member is not more than
0.18.
[0055] Still another gallium nitride compound semiconductor member
according to the present invention is (49) a gallium nitride
compound semiconductor member wherein in a spectrum of an imaginary
part of a complex index of refraction obtained by spectroscopic
ellipsometry measurement on a surface of the gallium nitride
compound semiconductor member, a wavelength at which an absolute
value of a slope of a spectrum in a wavelength band of 300 to 400
nm is maximum, is not less than 350 nm.
[0056] Still another gallium nitride compound semiconductor member
according to the present invention is (50) a gallium nitride
compound semiconductor member wherein an absolute value of a
difference between an imaginary part of a complex dielectric
constant at 360 nm and an imaginary part of a complex dielectric
constant at 370 nm obtained by spectroscopic ellipsometry
measurement on a surface of the gallium nitride compound
semiconductor member is not less than 0.24.
[0057] Still another gallium nitride compound semiconductor member
according to the present invention is (51) a gallium nitride
compound semiconductor member wherein an absolute value of an
imaginary part of a complex dielectric constant at 370 nm obtained
by spectroscopic ellipsometry measurement on a surface of the
gallium nitride compound semiconductor member is not more than
0.9.
[0058] Still another gallium nitride compound semiconductor member
according to the present invention is (52) a gallium nitride
compound semiconductor member wherein in a spectrum of an imaginary
part of a complex dielectric constant obtained by spectroscopic
ellipsometry measurement on a surface of the gallium nitride
compound semiconductor member, a wavelength at which an absolute
value of a slope of a spectrum in a wavelength band of 300 to 400
nm is maximum, is not less than 350 nm.
[0059] Still another gallium nitride compound semiconductor member
according to the present invention is (53) a gallium nitride
compound semiconductor member wherein an absolute value of a
difference between a real part of a complex index of refraction at
365 nm and a real part of a complex index of refraction at 375 nm
obtained by spectroscopic ellipsometry measurement on a surface of
the gallium nitride compound semiconductor member is not less than
0.035.
[0060] Still another gallium nitride compound semiconductor member
according to the present invention is (54) a gallium nitride
compound semiconductor member wherein in a spectrum of a real part
of a complex index of refraction obtained by spectroscopic
ellipsometry measurement on a surface of the gallium nitride
compound semiconductor member, a maximum of a spectrum in a
wavelength band of 300 to 400 nm is not less than 2.7.
[0061] Still another gallium nitride compound semiconductor member
according to the present invention is (55) a gallium nitride
compound semiconductor member wherein an absolute value of a
difference between a real part of a complex dielectric constant at
365 nm and a real part of a complex dielectric constant at 375 nm
obtained by spectroscopic ellipsometry measurement on a surface of
the gallium nitride compound semiconductor member is not less than
0.13.
[0062] Still another gallium nitride compound semiconductor member
according to the present invention is (56) a gallium nitride
compound semiconductor member wherein in a spectrum of a real part
of a complex dielectric constant obtained by spectroscopic
ellipsometry measurement on a surface of the gallium nitride
compound semiconductor member, a maximum of a spectrum in a
wavelength band of 300 to 400 nm is not less than 7.2.
[0063] Still another gallium nitride compound semiconductor member
according to the present invention is (57) a gallium nitride
compound semiconductor member wherein a thickness of a layer
containing at least one of an oxide film and an uneven layer formed
on a surface of the gallium nitride compound semiconductor member
is not more than 6 nm.
[0064] Preferably, (58) the gallium nitride compound semiconductor
member is a gallium nitride compound semiconductor substrate.
Preferably, (59) the gallium nitride compound semiconductor member
is a gallium nitride compound semiconductor membrane provided on a
substrate.
[0065] A gallium nitride compound semiconductor membrane according
to the present invention is (60) a gallium nitride compound
semiconductor membrane formed on a gallium nitride compound
semiconductor member wherein an absolute value of a difference
between an imaginary part of a complex index of refraction at 360
nm and an imaginary part of a complex index of refraction at 370 nm
obtained by spectroscopic ellipsometry measurement on a surface of
the gallium nitride compound semiconductor member is not less than
0.045.
[0066] Another gallium nitride compound semiconductor membrane
according to the present invention is (61) a gallium nitride
compound semiconductor membrane formed on a gallium nitride
compound semiconductor member wherein an absolute value of an
imaginary part of a complex index of refraction at 370 nm obtained
by spectroscopic ellipsometry measurement on a surface of the
gallium nitride compound semiconductor member is not more than
0.18.
[0067] Still another gallium nitride compound semiconductor
membrane according to the present invention is (62) a gallium
nitride compound semiconductor membrane formed on a gallium nitride
compound semiconductor member wherein in a spectrum of an imaginary
part of a complex index of refraction obtained by spectroscopic
ellipsometry measurement on a surface of the gallium nitride
compound semiconductor member, a wavelength at which an absolute
value of a slope of a spectrum in a wavelength band of 300 to 400
nm is maximum, is not less than 350 nm.
[0068] Still another gallium nitride compound semiconductor
membrane according to the present invention is (63) a gallium
nitride compound semiconductor membrane formed on a gallium nitride
compound semiconductor member wherein an absolute value of a
difference between an imaginary part of a complex dielectric
constant at 360 nm and an imaginary part of a complex dielectric
constant at 370 nm obtained by spectroscopic ellipsometry
measurement on a surface of the gallium nitride compound
semiconductor member is not less than 0.24.
[0069] Still another gallium nitride compound semiconductor
membrane according to the present invention is (64) a gallium
nitride compound semiconductor membrane formed on a gallium nitride
compound semiconductor member wherein an absolute value of an
imaginary part of a complex dielectric constant at 370 nm obtained
by spectroscopic ellipsometry measurement on a surface of the
gallium nitride compound semiconductor member is not more than
0.9.
[0070] Still another gallium nitride compound semiconductor
membrane according to the present invention is (65) a gallium
nitride compound semiconductor membrane formed on a gallium nitride
compound semiconductor member wherein in a spectrum of an imaginary
part of a complex dielectric constant obtained by spectroscopic
ellipsometry measurement on a surface of the gallium nitride
compound semiconductor member, a wavelength at which an absolute
value of a slope of a spectrum in a wavelength band of 300 to 400
nm is maximum, is not less than 350 nm.
[0071] Still another gallium nitride compound semiconductor
membrane according to the present invention is (66) a gallium
nitride compound semiconductor membrane formed on a gallium nitride
compound semiconductor member wherein an absolute value of a
difference between a real part of a complex index of refraction at
365 nm and a real part of a complex index of refraction at 375 nm
obtained by spectroscopic ellipsometry measurement on a surface of
the gallium nitride compound semiconductor member is not less than
0.035.
[0072] Still another gallium nitride compound semiconductor
membrane according to the present invention is (67) a gallium
nitride compound semiconductor membrane formed on a gallium nitride
compound semiconductor member wherein in a spectrum of a real part
of a complex index of refraction obtained by spectroscopic
ellipsometry measurement on a surface of the gallium nitride
compound semiconductor member, a maximum of a spectrum in a
wavelength band of 300 to 400 nm is not less than 2.7.
[0073] Still another gallium nitride compound semiconductor
membrane according to the present invention is (68) a gallium
nitride compound semiconductor membrane formed on a gallium nitride
compound semiconductor member wherein an absolute value of a
difference between a real part of a complex dielectric constant at
365 nm and a real part of a complex dielectric constant at 375 nm
obtained by spectroscopic ellipsometry measurement on a surface of
the gallium nitride compound semiconductor member is not less than
0.13.
[0074] Still another gallium nitride compound semiconductor
membrane according to the present invention is (69) a gallium
nitride compound semiconductor membrane formed on a gallium nitride
compound semiconductor member wherein in a spectrum of a real part
of a complex dielectric constant obtained by spectroscopic
ellipsometry measurement on a surface of the gallium nitride
compound semiconductor member, a maximum of a spectrum in a
wavelength band of 300 to 400 nm is not less than 7.2.
[0075] Still another gallium nitride compound semiconductor
membrane according to the present invention is (70) a gallium
nitride compound semiconductor membrane formed on a gallium nitride
compound semiconductor member wherein a thickness of a layer
containing at least one of an oxide film and an uneven layer formed
on a surface of the gallium nitride compound semiconductor member
is not more than 6 nm.
BRIEF DESCRIPTION OF THE DRAWINGS
[0076] FIG. 1 is a flowchart showing steps in a damage evaluation
method of a compound semiconductor member and in a production
method of a compound semiconductor member according to an
embodiment.
[0077] FIG. 2 is an illustration schematically showing a
spectroscopic ellipsometry measurement step.
[0078] FIG. 3 is a sectional view schematically showing a compound
semiconductor membrane provided on a substrate.
[0079] FIG. 4A is a graph schematically showing a spectrum SP1 of
the imaginary part K of the complex index of refraction obtained by
spectroscopic ellipsometry measurement, and FIG. 4B is a graph
schematically showing a spectrum SP2 of the imaginary part
.epsilon..sub.2 of the complex dielectric constant obtained by
spectroscopic ellipsometry measurement.
[0080] FIG. 5A is a graph schematically showing a spectrum SP3 of
the real part N of the complex index of refraction obtained by
spectroscopic ellipsometry measurement, and FIG. 5B is a graph
schematically showing a spectrum SP4 of the real part
.epsilon..sub.1 of the complex dielectric constant obtained by
spectroscopic ellipsometry measurement.
[0081] FIG. 6A is a sectional view schematically showing a compound
semiconductor substrate in a thin film forming step, and FIG. 6B is
a sectional view schematically showing a compound semiconductor
membrane in the thin film forming step.
[0082] FIG. 7A is a sectional view schematically showing a compound
semiconductor substrate in an electrode forming step, and FIG. 7B
is a sectional view schematically showing a compound semiconductor
membrane in the electrode forming step.
[0083] FIG. 8 shows PL intensities of a peak near 365 nm in
respective emission spectra obtained by photoluminescence
measurement on surfaces of monocrystalline GaN substrates of
Experiment Example 1 to Experiment Example 5.
[0084] FIG. 9 is a graph showing part of spectra of the imaginary
part K of the complex index of refraction obtained by spectroscopic
ellipsometry measurement.
[0085] FIG. 10 is a graph showing part of spectra of the imaginary
part K of the complex index of refraction obtained by spectroscopic
ellipsometry measurement.
[0086] FIG. 11 is a graph showing first derivatives of spectra K1
to K5 shown in FIG. 9.
[0087] FIG. 12 is a graph showing part of spectra of the imaginary
part .epsilon..sub.2 of the complex dielectric constant obtained by
spectroscopic ellipsometry measurement.
[0088] FIG. 13 is a graph showing portions located on the longer
wavelength side of spectra M1 to M5 shown in FIG. 12.
[0089] FIG. 14 is a graph showing first derivatives of spectra M1
to M5 shown in FIG. 12.
[0090] FIG. 15 is a graph showing part of spectra of the real part
N of the complex index of refraction obtained by spectroscopic
ellipsometry measurement.
[0091] FIG. 16 is a graph showing first derivatives of spectra P1
to P5 shown in FIG. 15.
[0092] FIG. 17 is a graph showing part of spectra of the real part
.epsilon..sub.1 of the complex dielectric constant obtained by
spectroscopic ellipsometry measurement.
[0093] FIG. 18 is a graph showing first derivatives of spectra Q1
to Q5 shown in FIG. 17.
[0094] FIG. 19 is a graph showing a relation between PL intensities
shown in FIG. 8, and maximum absolute values of slopes of spectra
of the imaginary part K of the complex index of refraction and the
real part N of the complex index of refraction.
[0095] FIG. 20 is a graph showing a relation between PL intensities
shown in FIG. 8, and wavelengths to achieve maximum absolute values
of slopes of spectra of the imaginary part K of the complex index
of refraction and the real part N of the complex index of
refraction.
[0096] FIG. 21 is a graph showing part of spectra of the imaginary
part K of the complex index of refraction obtained by spectroscopic
ellipsometry measurement.
[0097] FIG. 22 is a graph showing part of spectra of the real part
N of the complex index of refraction obtained by spectroscopic
ellipsometry measurement.
[0098] FIG. 23 is a graph showing part of spectra of the imaginary
part K of the complex index of refraction obtained by spectroscopic
ellipsometry measurement.
[0099] FIG. 24 is a graph showing first derivatives of spectra P8
to P10 shown in FIG. 22.
[0100] FIG. 25 is a graph showing a relation between
cathodoluminescence intensities and maximum absolute values of
slopes.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0101] Embodiments of the present invention will be described below
in detail with reference to the accompanying drawings. In the
description of the drawings identical or equivalent elements will
be denoted by the same reference symbols, without redundant
description.
[0102] FIG. 1 is a flowchart showing steps in a damage evaluation
method of a compound semiconductor member and in a production
method of a compound semiconductor member according to an
embodiment. The damage evaluation method of the compound
semiconductor member according to the embodiment includes a
spectroscopic ellipsometry measurement step S1 and a damage
evaluation step S2. The production method of the compound
semiconductor member according to the embodiment includes the
spectroscopic ellipsometry measurement step S1 and the damage
evaluation step S2 and, preferably, further includes a thin film
forming step S3 and an electrode forming step S4.
[0103] (Spectroscopic Ellipsometry Measurement Step)
[0104] FIG. 2 is an illustration schematically showing the
spectroscopic ellipsometry measurement step. The spectroscopic
ellipsometry measurement step S1 is to perform spectroscopic
ellipsometry measurement on a surface 10a of a compound
semiconductor substrate 10 (compound semiconductor member). The
spectroscopic ellipsometry measurement is carried out preferably
with a spectroscopic ellipsometer 16.
[0105] The spectroscopic ellipsometer 16 has a stage 17 for
supporting the compound semiconductor substrate 10, and a light
source 12 for emitting light LT1 toward the surface 10a of the
compound semiconductor substrate 10. The wavelength of light LT1
may be fixed or variable. The light LT1 emitted from the light
source 12 is guided through a filter 18 to be converted into
linearly polarized light. Therefore, the linearly polarized light
LT1 is incident to the surface 10a of the compound semiconductor
substrate 10. Light LT2 reflected on the surface 10a is incident to
a light detection unit 14. A computer C is connected through an
optical system (not shown) for measuring the amplitude ratio .psi.
and phase difference .DELTA. of light LT2 as described later, to
the light detection unit 14. The computer C saves the data of
amplitude ratio .psi. and phase difference .DELTA.. The
spectroscopic ellipsometer 16 is preferably provided with an angle
adjusting mechanism (not shown) capable of manually or
automatically adjusting the angle of incidence of the light LT1 and
the angle of reflection of the light LT2 relative to the surface
10a.
[0106] The spectroscopic ellipsometry measurement is carried out
well by use of the spectroscopic ellipsometer 16. First, initial
adjustment of the spectroscopic ellipsometer 16 is carried out
according to need. Specifically, for example, the compound
semiconductor substrate 10 is mounted on the stage 17, and
thereafter, while reference light is projected from the direction
perpendicular to the surface 10a, onto the compound semiconductor
substrate 10, the inclination of stage 17 is adjusted so as to
match the reference light with its reflected light. Subsequently,
the light LT1 is projected from a direction shifted by a
predetermined angle from the direction perpendicular to the surface
10a, onto the compound semiconductor substrate 10, and the light
source 12 and light detection unit 14 are set at a predetermined
angle of incidence by adjusting the distance between the light
source 12 and the compound semiconductor substrate 10, the angle of
incidence of the light LT1, the distance between the light
detection unit 14 and the compound semiconductor substrate 10, etc.
so as to maximize the intensity of the light LT2 incident to the
light detection unit 14.
[0107] The light LT2 consists of a component orthogonal to the
surface 10a (hereinafter referred to as "orthogonal component Y")
and a component parallel to the surface 10a (hereinafter referred
to as "parallel component X"). The light detection unit 14 is able
to detect the orthogonal component Y and parallel component X.
Where the reflection coefficient of the orthogonal component Y is
Rp and the reflection coefficient of the parallel component X is
Rs, Eq (1) below holds. With this Eq (1) the amplitude ratio .psi.
and phase difference .DELTA. are obtained as measured data. In Eq
(1), i indicates an imaginary number. Rp/Rs=tan(.psi.)exp(i.DELTA.)
(1)
[0108] The data of amplitude ratio .psi. and phase difference
.DELTA. is stored in a storage unit (not shown) such as a memory
and a hard disk in the computer C connected to the light detection
unit 14. If the wavelength of light LT1 is variable, data of
amplitude ratio .psi. and phase difference .DELTA. at each
wavelength is obtained as measured data. When the wavelength of
light LT1 is variable, the obtained data includes the number of
measurement wavelengths times those in the case where the
wavelength of light LT1 is fixed. Therefore, a single measurement
enables measurement of optical properties of the compound
semiconductor substrate 10, thicknesses of respective layers
constituting the compound semiconductor substrate 10, and so on.
The spectroscopic ellipsometry measurement also allows measurement
of surface roughness of not more than 1 nm.
[0109] (Damage Evaluation Step)
[0110] The damage evaluation step S2 is to perform an analysis
described below, using the measured data, to obtain spectra of
optical constants (the real part .epsilon..sub.1 of the complex
dielectric constant, the imaginary part .epsilon..sub.2 of the
complex dielectric constant, the real part N of the complex index
of refraction, and the imaginary part K of the complex index of
refraction) of the compound semiconductor substrate 10.
[0111] For obtaining the spectra of the optical constants, a model
structure of the compound semiconductor substrate 10 is first
estimated. The model structure of the compound semiconductor
substrate 10 has, for example, a compound semiconductor region 13,
and a damage layer 11 provided on the compound semiconductor region
13. A layer 15 containing at least one of an oxide film and an
uneven layer due to surface roughness, for example, may be formed
on the damage layer 11. The layer 15 may also be an oxide film with
an uneven surface. The layer 15 is preferably, for example, a layer
in which an oxide and air are mixed. In the layer 15, for example,
the oxide and air are mixed 50% by volume each, but the volume
ratio of the oxide and air may be modified.
[0112] Next, an optical simulation is carried out with input of
such parameters as thicknesses, refractive indices, extinction
coefficients, etc. of the compound semiconductor region 13, damage
layer 11, layer 15, etc. in the model structure, and fitting is
performed between the result of the optical simulation and the
measured data of the spectroscopic ellipsometry measurement.
Furthermore, with feedback of the result of the fitting, another
optical simulation is carried out again with reentry of the above
parameters. This sequential operation is repeatedly carried out to
determine the best model structure. The optical simulation and
fitting are carried out well by use of optimum analysis software in
the computer C.
[0113] Next, using the model structure determined as described
above, the complex dielectric constant .epsilon. and the complex
index of refraction n are calculated from the measured data (data
of amplitude ratio .psi. and phase difference .DELTA. for each
wavelength). Since there are a variety of known dielectric
functions, it is preferable first to calculate the complex
dielectric constant .epsilon. and thereafter to calculate the
complex index of refraction n. For example, the complex index of
refraction n is calculated from the complex dielectric constant
.epsilon., using the dielectric function represented by Eq (2)
below. n= {square root over (.epsilon.)} (2)
[0114] The complex dielectric constant .epsilon. and the complex
index of refraction n are represented by Eqs (3), (4) below. In Eq
(3), .epsilon..sub.1 indicates the real part of the complex
dielectric constant and .epsilon..sub.2 the imaginary part of the
complex dielectric constant. In Eq (4) N indicates the real part of
the complex index of refraction and K the imaginary part of the
complex index of refraction.
.epsilon.=.epsilon..sub.1+i.epsilon..sub.2 (3) n=N+iK (4)
[0115] The spectra of the optical constants (the real part
.epsilon..sub.1 of the complex dielectric constant, the imaginary
part .epsilon..sub.2 of the complex dielectric constant, the real
part N of the complex index of refraction, and the imaginary part K
of the complex index of refraction) are obtained using the above
Eqs (3), (4). The damage evaluation step S2 is to evaluate damage
on the surface 10a of the compound semiconductor substrate 10,
using a spectrum in a wavelength band containing a wavelength
corresponding to the bandgap of the compound semiconductor
substrate 10, in the spectra of the optical constants obtained by
the spectroscopic ellipsometry measurement. Examples of such damage
include damage, scratches, distortion, etc. due to polishing,
etching, or the like.
[0116] In the damage evaluation method, execution of the foregoing
spectroscopic ellipsometry measurement on the surface 10a of the
compound semiconductor substrate 10 generates the exciton being a
state of an electron and a hole coupled by Coulomb force. The
exciton significantly affects, particularly, the spectra of the
optical constants in the wavelength band containing the wavelength
corresponding to the bandgap. Therefore, selective use of a part of
the spectra of the optical constants can decrease the noise
component, in comparison with the damage evaluation using the whole
spectra of the optical constants obtained by the spectroscopic
ellipsometry measurement. Therefore, it is feasible to achieve
nondestructive and highly accurate evaluation on the level of
damage on the surface 10a of the compound semiconductor substrate
10. Furthermore, the above damage evaluation method also enables
simultaneous measurement of the thickness and optical properties of
the damage layer 11, the thickness of the layer 15, the thickness
of the oxide film, surface roughness, and so on.
[0117] It is noted that the tendency of influence of damage on the
spectra of the optical constants is the same even if the model
structure or the dielectric function used is different from the
aforementioned model structure or dielectric function.
[0118] When the compound semiconductor substrate 10 is made, for
example, of a monocrystalline material or polycrystalline material,
the monocrystalline material or polycrystalline material turns into
the amorphous form in a damaged region and thus it becomes easier
to discriminate the damaged region from the other region without
damage. For this reason, the damage becomes easier to detect and an
improvement is made in the accuracy of damage evaluation.
[0119] The bandgap of the compound semiconductor substrate 10 is
preferably not less than 1.6.times.10.sup.-19 J (1 eV). In this
case, the exciton is more likely to be generated by the
spectroscopic ellipsometry measurement, so that the level of damage
on the surface 10a of the compound semiconductor substrate 10 can
be evaluated with higher accuracy. Particularly, where the compound
semiconductor substrate 10 is made of a wide-gap semiconductor such
as GaN, AlN, BN, ZnSe, or ZnO, the influence of exciton becomes
stronger. On the other hand, even in cases where the compound
semiconductor substrate 10 is made of a compound semiconductor with
a small bandgap, the effect by exciton can be actualized by cooling
approximately to the temperature of liquid nitrogen.
[0120] The compound semiconductor substrate 10 is preferably made
of a nitride compound semiconductor containing at least one of B,
Al, and Ga. Also, the compound semiconductor substrate 10 is
preferably made of an oxide compound semiconductor containing at
least one of Be and Zn. Furthermore, the compound semiconductor
substrate 10 is preferably made of a ZnSe compound semiconductor.
In each case, the bandgap of the compound semiconductor substrate
10 is increased, and thus the effect of exciton becomes easier to
exhibit.
[0121] More specifically, the compound semiconductor substrate 10
is made, for example, of a III-V compound semiconductor such as
GaAs or InP, a nitride compound semiconductor such as BN, GaN, AlN,
or InN, a II-VI compound semiconductor such as ZnO or ZnS, an oxide
compound semiconductor such as Be.sub.xO.sub.y, ZnO,
Ga.sub.2O.sub.3, or Al.sub.2O.sub.3, a ZnSe compound semiconductor
such as ZnSe, a ternary compound semiconductor such as GaAlN or
InGaN, or a quaternary or higher compound semiconductor. These
compound semiconductors may be doped with an impurity.
[0122] For example, in a case where the compound semiconductor
substrate 10 is made of a gallium nitride compound semiconductor,
the gallium nitride compound semiconductor suitably applicable is
of the wurtzite structure or the zinc blende (cubic crystal)
structure. In the case of the wurtzite structure, the surface 10a
may be any one of the (0001) face called the C-plane, the (10-10)
face called the M-plane, the (11-20) face called the A-plane, the
(01-12) face called the R-plane, and the (10-11) face called the
S-plane. The C-plane can be either a Ga plane consisting of Ga or
an N plane consisting of N. Since the Ga plane is normally more
resistant to etching, the surface 10a is preferably the Ga plane,
but the surface 10a may be the N plane.
[0123] On the occasion of performing the spectroscopic ellipsometry
measurement, a compound semiconductor membrane 20 (compound
semiconductor member) shown in FIG. 3 may also be used instead of
the compound semiconductor substrate 10.
[0124] FIG. 3 is a sectional view schematically showing a compound
semiconductor membrane provided on a substrate. The substrate 22
shown in FIG. 3 is, for example, an amorphous substrate such as a
glass substrate, or a monocrystalline substrate such as a sapphire
substrate or Si substrate. From the viewpoint of eliminating
influence of light reflected from the back surface of the substrate
22, the back surface of the substrate 22 is preferably provided
with surface unevenness. However, if the analysis software is able
to utilize a program that can take account of the influence of the
light reflected from the back surface, the back surface may be a
specular surface. The constituent material of the compound
semiconductor membrane 20 may be the same as the compound
semiconductor substrate 10. In this case, the spectroscopic
ellipsometry measurement is carried out by projecting the light LT1
onto a surface 20a of the compound semiconductor membrane 20.
[0125] The model structure of the compound semiconductor membrane
20 has, for example, a compound semiconductor region 23, and a
damage layer 21 provided on the compound semiconductor region 23. A
layer 25 made of a constituent material similar to the layer 15 may
be formed on the damage layer 21.
[0126] Method 1 to Method 16 of damage evaluation using the
spectrum of the real part .epsilon..sub.1 of the complex dielectric
constant, the spectrum of the imaginary part .epsilon..sub.2 of the
complex dielectric constant, the spectrum of the real part N of the
complex index of refraction, or the spectrum of the imaginary part
K of the complex index of refraction will be described below in
detail with reference to FIGS. 4A, 4B, 5A and 5B.
[0127] FIG. 4A is a graph schematically showing the spectrum SP1 of
the imaginary part K of the complex index of refraction obtained by
spectroscopic ellipsometry measurement. The vertical axis of the
graph represents the imaginary part K of the complex index of
refraction, and the horizontal axis the wavelength .lamda.. Energy
or wave number may be taken on the horizontal axis. With reference
to FIG. 4A, Method 1 to Method 4 will be described below in detail
as methods of performing the damage evaluation using a spectrum SA1
in a wavelength band A from wavelength .lamda..sub.a to wavelength
.lamda..sub.b, in the spectrum SP1 of the imaginary part K of the
complex index of refraction. The wavelength band A contains a
wavelength .lamda..sub.Eg corresponding to the bandgap Eg of the
compound semiconductor substrate 10. The wavelength .lamda..sub.a
is preferably set, for example, to (0.9.times..lamda..sub.Eg) and
the wavelength .lamda..sub.b is preferably set, for example, to
(1.1.times..lamda..sub.Eg).
[0128] <Method 1>
[0129] Method 1 is to evaluate damage on the surface 10a of the
compound semiconductor substrate 10, using a maximum absolute value
G1.sub.max of the slope of the spectrum SA1 in the wavelength band
A. The maximum G1.sub.max is the same value as an absolute value of
an extremum of the first derivative of the spectrum SA1. The
maximum G1.sub.max tends to decrease with increasing level of
damage. Therefore, the use of the maximum G1.sub.max in the damage
evaluation enables the level of damage to be quantified.
[0130] Method 1 is suitably applicable to production of the
compound semiconductor substrate 10. The compound semiconductor
substrate is determined to be nondefective when the maximum
G1.sub.max is not less than a predetermined threshold, whereby the
compound semiconductor substrate 10 with a low level of damage can
be produced at a high yield.
[0131] <Method 2>
[0132] Method 2 is to evaluate damage on the surface 10a of the
compound semiconductor substrate 10, using a wavelength
.lamda.1.sub.max at which the absolute value of the slope of the
spectrum SA1 in the wavelength band A is maximum G1.sub.max. The
wavelength .lamda.1.sub.max tends to decrease with increasing level
of damage. Therefore, the use of the wavelength .lamda.1.sub.max in
the damage evaluation enables the level of damage to be
quantified.
[0133] Method 2 is suitably applicable to production of the
compound semiconductor substrate 10. The compound semiconductor
substrate is determined to be nondefective when the wavelength
.lamda.1.sub.max is not less than a predetermined threshold,
whereby the compound semiconductor substrate 10 with a low level of
damage can be produced at a high yield.
[0134] <Method 3>
[0135] Method 3 is to evaluate damage on the surface 10a of the
compound semiconductor substrate 10, using a maximum P1.sub.max of
the spectrum SA1 in the wavelength band A. The maximum P1.sub.max
tends to decrease with increasing level of damage. Therefore, the
use of the maximum P1.sub.max in the damage evaluation enables the
level of damage to be quantified.
[0136] Method 3 is suitably applicable to production of the
compound semiconductor substrate 10. The compound semiconductor
substrate is determined to be nondefective when the maximum
P1.sub.max is not less than a predetermined threshold, whereby the
compound semiconductor substrate 10 with a low level of damage can
be produced at a high yield.
[0137] <Method 4>
[0138] Method 4 uses a peak SB1 in another wavelength band B
located on the longer wavelength side than the wavelength band A.
The wavelength band B is a wavelength band from wavelength
.lamda..sub.c larger than the wavelength .lamda..sub.b to
wavelength .lamda..sub.d. The peak SB1 appears in the spectrum SP1
of the imaginary part K of the complex index of refraction when
light is reflected at least once (e.g., with multiple reflection)
between the compound semiconductor region 13 and the damage layer
11. The peak SB1 is observed when the level of damage is high. The
use of this peak SB1 enables the level of damage on the surface 10a
to be evaluated with high accuracy.
[0139] The compound semiconductor substrate 10 is preferably one of
substrate A1 to substrate A3 described below. In each case, a
gallium nitride compound semiconductor substrate is obtained with a
low level of damage on its surface. Where the compound
semiconductor substrate 10 is made, for example, of a gallium
nitride compound semiconductor, the wavelength .lamda..sub.Eg is
about 365 nm. [0140] (Substrate Al) Gallium nitride compound
semiconductor substrate in which the absolute value of the
difference between the imaginary part K of the complex index of
refraction at 360 nm and the imaginary part K of the complex index
of refraction at 370 nm is not less than 0.045. [0141] (Substrate
A2) Gallium nitride compound semiconductor substrate in which the
absolute value of the imaginary part K of the complex index of
refraction at 370 nm is not more than 0.18. [0142] (Substrate A3)
Gallium nitride compound semiconductor substrate in which the
wavelength .lamda.1.sub.max at which the absolute value of the
slope of the spectrum SA1 in the wavelength band A is maximum is
not less than 350 nm, where the wavelength .lamda..sub.a is 300 nm
and the wavelength .lamda..sub.b 400 nm.
[0143] FIG. 4B is a graph schematically showing the spectrum SP2 of
the imaginary part .epsilon..sub.2 of the complex dielectric
constant obtained by spectroscopic ellipsometry measurement. The
vertical axis of the graph represents the imaginary part
.epsilon..sub.2 of the complex dielectric constant, and the
horizontal axis the wavelength .lamda.. Energy or wave number may
be taken on the horizontal axis. With reference to FIG. 4B, Method
5 to Method 8 will be described in detail as methods of performing
damage evaluation using a spectrum SA2 in the wavelength band A in
the spectrum SP2 of the imaginary part .epsilon..sub.2 of the
complex dielectric constant.
[0144] <Method 5>
[0145] Method 5 is to evaluate damage on the surface 10a of the
compound semiconductor substrate 10, using a maximum absolute value
G2.sub.max of the slope of the spectrum SA2 in the wavelength band
A. The maximum G2.sub.max is the same value as an absolute value of
an extremum of the first derivative of the spectrum SA2. The
maximum G2.sub.max tends to decrease with increasing level of
damage. Therefore, the use of the maximum G2.sub.max in the damage
evaluation enables the level of damage to be quantified.
[0146] Method 5 is suitably applicable to production of the
compound semiconductor substrate 10. The compound semiconductor
substrate is determined to be nondefective when the maximum
G2.sub.max is not less than a predetermined threshold, whereby the
compound semiconductor substrate 10 with a low level of damage can
be produced at a high yield.
[0147] <Method 6>
[0148] Method 6 is to evaluate damage on the surface 10a of the
compound semiconductor substrate 10, using a wavelength
.lamda.2.sub.max at which the absolute value of the slope of the
spectrum SA2 in the wavelength band A is maximum G2.sub.max. The
wavelength .lamda.2.sub.max tends to decrease with increasing level
of damage. Therefore, the use of the wavelength .lamda.2.sub.max in
the damage evaluation enables the level of damage to be
quantified.
[0149] Method 6 is suitably applicable to production of the
compound semiconductor substrate 10. The compound semiconductor
substrate is determined to be nondefective when the wavelength
.lamda.2.sub.max is not less than a predetermined threshold,
whereby the compound semiconductor substrate 10 with a low level of
damage can be produced at a high yield.
[0150] <Method 7>
[0151] Method 7 is to evaluate damage on the surface 10a of the
compound semiconductor substrate 10, using a maximum P2.sub.max of
the spectrum SA2 in the wavelength band A. The maximum P2.sub.max
tends to decrease with increasing level of damage. Therefore, the
use of the maximum P2.sub.max in the damage evaluation enables the
level of damage to be quantified.
[0152] Method 7 is suitably applicable to production of the
compound semiconductor substrate 10. The compound semiconductor
substrate is determined to be nondefective when the maximum
P2.sub.max is not less than a predetermined threshold, whereby the
compound semiconductor substrate 10 with a low level of damage can
be produced at a high yield.
[0153] <Method 8>
[0154] Method 8 uses a peak SB2 in another wavelength band B
located on the longer wavelength side than the wavelength band A.
The peak SB2 appears in the spectrum SP2 of the imaginary part
.epsilon..sub.2 of the complex dielectric constant when light is
reflected at least once (e.g., with multiple reflection) between
the compound semiconductor region 13 and the damage layer 11. The
peak SB2 is observed when the level of damage is high. The use of
this peak SB2 enables the level of damage on the surface 10a to be
evaluated with high accuracy.
[0155] The compound semiconductor substrate 10 is preferably one of
substrate A4 to substrate A6 described below. In each case, a
gallium nitride compound semiconductor substrate is obtained with a
low level of damage on its surface. [0156] (Substrate A4) Gallium
nitride compound semiconductor substrate in which the absolute
value of the difference between the imaginary part .epsilon..sub.2
of the complex dielectric constant at 360 nm and the imaginary part
.epsilon..sub.2 of the complex dielectric constant at 370 nm is not
less than 0.24. [0157] (Substrate A5) Gallium nitride compound
semiconductor substrate in which the absolute value of the
imaginary part .epsilon..sub.2 of the complex dielectric constant
at 370 nm is not more than 0.9. [0158] (Substrate A6) Gallium
nitride compound semiconductor substrate in which the wavelength
.lamda.2.sub.max at which the absolute value of the slope of the
spectrum SA2 in the wavelength band A is maximum is not less than
350 nm, where the wavelength .lamda..sub.a is 300 nm and the
wavelength .lamda..sub.b 400 nm.
[0159] FIG. 5A is a graph schematically showing a spectrum SP3 of
the real part N of the complex index of refraction obtained by
spectroscopic ellipsometry measurement. The vertical axis of the
graph represents the real part N of the complex index of
refraction, and the horizontal axis the wavelength .lamda.. Energy
or wave number may be taken on the horizontal axis. With reference
to FIG. 5A, Method 9 to Method 12 will be described in detail as
methods of performing damage evaluation using a spectrum SA3 in the
wavelength band A in the spectrum SP3 of the real part N of the
complex index of refraction.
[0160] <Method 9>
[0161] Method 9 is to evaluate damage on the surface 10a of the
compound semiconductor substrate 10, using a maximum absolute value
G31.sub.max of the slope in a portion (.lamda..sub.a to
.lamda.3.sub.p) located on the shorter wavelength side than the
wavelength .lamda.3.sub.p corresponding to a maximum P3.sub.max in
the spectrum SA3 in the wavelength band A. The maximum G31.sub.max
tends to decrease with increasing level of damage. Therefore, the
use of the maximum G31.sub.max in the damage evaluation enables the
level of damage to be quantified.
[0162] Method 9 is suitably applicable to production of the
compound semiconductor substrate 10. The compound semiconductor
substrate is determined to be nondefective when the maximum
G31.sub.max is not less than a predetermined threshold, whereby the
compound semiconductor substrate 10 with a low level of damage can
be produced at a high yield.
[0163] <Method 10>
[0164] Method 10 is to evaluate damage on the surface 10a of the
compound semiconductor substrate 10, using a maximum absolute value
G32.sub.max of the slope in a portion (.lamda.3.sub.p to
.lamda..sub.b) located on the longer wavelength side than the
wavelength .lamda.3.sub.p corresponding to the maximum P3.sub.max
in the spectrum SA3 in the wavelength band A. The maximum
G32.sub.max tends to decrease with increasing level of damage.
Therefore, the use of the maximum G32.sub.max in the damage
evaluation enables the level of damage to be quantified.
[0165] Method 10 is suitably applicable to production of the
compound semiconductor substrate 10. The compound semiconductor
substrate is determined to be nondefective when the maximum
G32.sub.max is not less than a predetermined threshold, whereby the
compound semiconductor substrate 10 with a low level of damage can
be produced at a high yield.
[0166] <Method 11>
[0167] Method 11 is to evaluate damage on the surface 10a of the
compound semiconductor substrate 10, using a wavelength
.lamda.3.sub.max at which the absolute value of the slope in the
portion (.lamda..sub.a to .lamda.3.sub.p) located on the shorter
wavelength side than the wavelength .lamda.3.sub.p corresponding to
the maximum P3.sub.max in the spectrum SA3 in the wavelength band
A-is maximum G31.sub.max. The wavelength .lamda.3.sub.max tends to
decrease with increasing level of damage. Therefore, the use of the
wavelength .lamda.3.sub.max in the damage evaluation enables the
level of damage to be quantified.
[0168] Method 11 is suitably applicable to production of the
compound semiconductor substrate 10. The compound semiconductor
substrate is determined to be nondefective when the wavelength
.lamda.3.sub.max is not less than a predetermined threshold,
whereby the compound semiconductor substrate 10 with a low level of
damage can be produced at a high yield.
[0169] <Method 12>
[0170] Method 12 is to evaluate damage on the surface 10a of the
compound semiconductor substrate 10, using the maximum P3.sub.max
of the spectrum SA3 in the wavelength band A. The maximum
P3.sub.max tends to decrease with increasing level of damage.
Therefore, the use of the maximum P3.sub.max in the damage
evaluation enables the level of damage to be quantified.
[0171] Method 12 is suitably applicable to production of the
compound semiconductor substrate 10. The compound semiconductor
substrate is determined to be nondefective when the maximum
P3.sub.max is not less than a predetermined threshold, whereby the
compound semiconductor substrate 10 with a low level of damage can
be produced at a high yield.
[0172] The compound semiconductor substrate 10 is preferably
substrate A7 or substrate A8 described below. In either case, a
gallium nitride compound semiconductor substrate is obtained with a
low level of damage on its surface. [0173] (Substrate A7) Gallium
nitride compound semiconductor substrate in which the absolute
value of the difference between the real part N of the complex
index of refraction at 365 nm and the real part N of the complex
index of refraction at 375 nm is not less than 0.035. [0174]
(Substrate A8) Gallium nitride compound semiconductor substrate in
which the maximum P3.sub.max of the spectrum SA3 in the wavelength
band A is not less than 2.7, where the wavelength .lamda..sub.a is
300 nm and the wavelength .lamda..sub.b 400 nm.
[0175] FIG. 5B is a graph schematically showing a spectrum SP4 of
the real part .epsilon..sub.1 of the complex dielectric constant
obtained by spectroscopic ellipsometry measurement. The vertical
axis of the graph represents the real part .epsilon..sub.1 of the
complex dielectric constant, and the horizontal axis the wavelength
.lamda.. Energy or wave number may be taken on the horizontal axis.
With reference to FIG. 5B, Method 13 to Method 16 will be described
in detail as methods of performing damage evaluation using a
spectrum SA4 in the wavelength band A in the spectrum SP4 of the
real part .epsilon..sub.1 of the complex dielectric constant.
[0176] <Method 13>
[0177] Method 13 is to evaluate damage on the surface 10a of the
compound semiconductor substrate 10, using a maximum absolute value
G41.sub.max of the slope in a portion (.lamda..sub.a to
.lamda.4.sub.p) located on the shorter wavelength side than a
wavelength .lamda.4.sub.p corresponding to a maximum P4.sub.max in
the spectrum SA4 in the wavelength band A. The maximum G41.sub.max
tends to decrease with increasing level of damage. Therefore, the
use of the maximum G41.sub.max in the damage evaluation enables the
level of damage to be quantified.
[0178] Method 13 is suitably applicable to production of the
compound semiconductor substrate 10. The compound semiconductor
substrate is determined to be nondefective when the maximum
G41.sub.max is not less than a predetermined threshold, whereby the
compound semiconductor substrate 10 with a low level of damage can
be produced at a high yield.
[0179] <Method 14>
[0180] Method 14 is to evaluate damage on the surface 10a of the
compound semiconductor substrate 10, using a maximum absolute value
G42.sub.max of the slope in a portion (.lamda.4.sub.p to
.lamda..sub.b) located on the longer wavelength side than the
wavelength .lamda.4.sub.p corresponding to the maximum P4.sub.max
in the spectrum SA4 in the wavelength band A. The maximum
G42.sub.max tends to decrease with increasing level of damage.
Therefore, the use of the maximum G42.sub.max in the damage
evaluation enables the level of damage to be quantified.
[0181] Method 14 is suitably applicable to production of the
compound semiconductor substrate 10. The compound semiconductor
substrate is determined to be nondefective when the maximum
G42.sub.max is not less than a predetermined threshold, whereby the
compound semiconductor substrate 10 with a low level of damage can
be produced at a high yield.
[0182] <Method 15>
[0183] Method 15 is to evaluate damage on the surface 10a of the
compound semiconductor substrate 10, using a wavelength
.lamda.4.sub.max at which the absolute value of the slope in the
portion (.lamda..sub.a to .lamda.4.sub.p) located on the shorter
wavelength side than the wavelength .lamda.4.sub.p corresponding to
the maximum P4.sub.max in the spectrum SA4 in the wavelength band A
is maximum G41.sub.max. The wavelength .lamda.4.sub.max tends to
decrease with increasing level of damage. Therefore, the use of the
wavelength .lamda.4.sub.max in the damage evaluation enables the
level of damage to be quantified.
[0184] Method 15 is suitably applicable to production of the
compound semiconductor substrate 10. The compound semiconductor
substrate is determined to be nondefective when the wavelength
.lamda.4.sub.max is not less than a predetermined threshold,
whereby the compound semiconductor substrate 10 with a low level of
damage can be produced at a high yield.
[0185] <Method 16>
[0186] Method 16 is to evaluate damage on the surface 10a of the
compound semiconductor substrate 10, using the maximum P4.sub.max
of the spectrum SA4 in the wavelength band A. The maximum
P4.sub.max tends to decrease with increasing level of damage.
Therefore, the use of the maximum P4.sub.max in the damage
evaluation enables the level of damage to be quantified.
[0187] Method 16 is suitably applicable to production of the
compound semiconductor substrate 10. The compound semiconductor
substrate is determined to be nondefective when the maximum
P4.sub.max is not less than a predetermined threshold, whereby the
compound semiconductor substrate 10 with a low level of damage can
be produced at a high yield.
[0188] The compound semiconductor substrate 10 is preferably
substrate A9 or substrate A10 described below. In either case, a
gallium nitride compound semiconductor substrate is obtained with a
low level of damage on its surface. [0189] (Substrate A9) Gallium
nitride compound semiconductor substrate in which the absolute
value of the difference between the real part .epsilon..sub.1 of
the complex dielectric constant at 365 nm and the real part
.epsilon..sub.1 of the complex dielectric constant at 375 nm is not
less than 0.13. [0190] (Substrate A10) Gallium nitride compound
semiconductor substrate in which the maximum P4.sub.max of the
spectrum SA4 in the wavelength band A is not less than 7.2, where
the wavelength a is 300 nm and the wavelength .lamda..sub.b 400
nm.
[0191] The compound semiconductor substrate 10 is preferably
substrate A11. In this case, a gallium nitride compound
semiconductor substrate is obtained with a low level of damage on
its surface. [0192] (Substrate A11) Gallium nitride compound
semiconductor substrate the thickness of the layer 15 formed on the
surface 10a of which is not more than 6 nm.
[0193] Damage may also be evaluated for the compound semiconductor
membrane 20 shown in FIG. 3, instead of the compound semiconductor
substrate 10. In this case, use of Method 1 to Method 16 permits
the level of damage on the surface 20a of the compound
semiconductor membrane 20 provided on the substrate 22 to be
evaluated with high accuracy and permits the level of damage to be
quantified. Since the influence of damage on the compound
semiconductor membrane 20 becomes relatively large, the damage
becomes easier to detect even if the level of damage is low.
[0194] Furthermore, use of Method 1 to Method 16 permits the
compound semiconductor membrane 20 with a low level of damage to be
produced at a high yield.
[0195] The compound semiconductor membrane 20 is preferably one of
membrane B1 to membrane B11 described below. In each case, a
gallium nitride compound semiconductor membrane is obtained with a
low level of damage on its surface. Where the compound
semiconductor membrane 20 is made, for example, of a gallium
nitride compound semiconductor, the wavelength .lamda..sub.Eg is
about 365 nm. [0196] (Membrane B1) Gallium nitride compound
semiconductor membrane in which the absolute value of the
difference between the imaginary part K of the complex index of
refraction at 360 nm and the imaginary part K of the complex index
of refraction at 370 nm is not less than 0.045. [0197] (Membrane
B2) Gallium nitride compound semiconductor membrane in which the
absolute value of the imaginary part K of the complex index of
refraction at 370 nm is not more than 0.18. [0198] (Membrane B3)
Gallium nitride compound semiconductor membrane in which the
wavelength .lamda.1.sub.max at which the absolute value of the
slope of the spectrum SA1 in the wavelength band A is maximum is
not less than 350 nm, where the wavelength .lamda..sub.a is 300 nm
and the wavelength .lamda..sub.b 400 nm. [0199] (Membrane B4)
Gallium nitride compound semiconductor membrane in which the
absolute value of the difference between the imaginary part
.epsilon..sub.2 of the complex dielectric constant at 360 nm and
the imaginary part .epsilon..sub.2 of the complex dielectric
constant at 370 nm is not less than 0.24. [0200] (Membrane B5)
Gallium nitride compound semiconductor membrane in which the
absolute value of the imaginary part .epsilon..sub.2 of the complex
dielectric constant at 370 nm is not more than 0.9. [0201]
(Membrane B6) Gallium nitride compound semiconductor membrane in
which the wavelength .lamda.2.sub.max at which the absolute value
of the slope of the spectrum SA2 in the wavelength band A is
maximum is not less than 350 nm, where the wavelength .lamda..sub.a
is 300 nm and the wavelength .lamda..sub.b 400 nm. [0202] (Membrane
B7) Gallium nitride compound semiconductor membrane in which the
absolute value of the difference between the real part N of the
complex index of refraction at 365 nm and the real part N of the
complex index of refraction at 375 nm is not less than 0.035.
[0203] (Membrane B8) Gallium nitride compound semiconductor
membrane in which the maximum P3.sub.max of the spectrum SA3 in the
wavelength band A is not less than 2.7, where the wavelength
.lamda..sub.a is 300 nm and the wavelength .lamda..sub.b 400 nm.
[0204] (Membrane B9) Gallium nitride compound semiconductor
membrane in which the absolute value of the difference between the
real part .epsilon..sub.1 of the complex dielectric constant at 365
nm and the real part .epsilon..sub.1 of the complex dielectric
constant at 375 nm is not less than 0.13. [0205] (Membrane B10)
Gallium nitride compound semiconductor membrane in which the
maximum P4.sub.max of the spectrum SA4 in the wavelength band A is
not less than 7.2, where the wavelength .lamda..sub.a is 300 nm and
the wavelength .lamda..sub.b 400 nm. [0206] (Membrane B11) Gallium
nitride compound semiconductor membrane the thickness of the layer
25 formed on the surface 20a of which is not more than 6 nm.
[0207] (Thin Film Forming Step)
[0208] FIG. 6A is a sectional view schematically showing a compound
semiconductor substrate in the thin film forming step. FIG. 6B is a
sectional view schematically showing a compound semiconductor
membrane in the thin film forming step. The thin film forming step
S3 is preferably carried out after the damage evaluation step
S2.
[0209] The thin film forming step S3 is to form a thin film 30 on
the surface 10a of the compound semiconductor substrate 10, as
shown in FIG. 6A. The thin film 30 is formed, for example, by an
epitaxial growth method. The thin film 30 is, for example, a
compound semiconductor film, an oxide film, a ZnO film, an
amorphous film, or the like. As the thin film 30 is formed on the
surface 10a of the compound semiconductor substrate 10 with a low
level of damage, an improvement is made in crystallinity and in
surface roughness of the thin film 30. A layer 15 may be interposed
between the surface 10a of the compound semiconductor substrate 10
and the thin film 30.
[0210] The thin film 30 is preferably a gallium nitride compound
semiconductor membrane formed on any one of the aforementioned
substrate A1 to substrate A11. Since this gallium nitride compound
semiconductor membrane is formed on one of substrate A1 to
substrate A11 with a low level of damage on the surface, an
improvement is made in crystallinity and in surface roughness.
[0211] In the thin film forming step S3, as shown in FIG. 6B, a
thin film 32 may be formed on a surface 20a of compound
semiconductor membrane 20. The thin film 32 is formed, for example,
by an epitaxial growth method. The thin film 32 can be a film
similar to the thin film 30. As the thin film 32 is formed on the
surface 20a of the compound semiconductor membrane 20 with a low
level of damage, an improvement is made in crystallinity and
surface roughness of the thin film 32. A layer 25 may be interposed
between the surface 20a of the compound semiconductor membrane 20
and the thin film 32.
[0212] The thin film 32 is preferably a gallium nitride compound
semiconductor membrane formed on any one of the aforementioned
membrane B1 to membrane B11. Since this gallium nitride compound
semiconductor membrane is formed on one of membrane B1 to membrane
B11 with a low level of damage on the surface, an improvement is
made in crystallinity and in surface roughness.
[0213] (Electrode Forming Step)
[0214] FIG. 7A is a sectional view schematically showing a compound
semiconductor substrate in the electrode forming step. FIG. 7B is a
sectional view schematically showing a compound semiconductor
membrane in the electrode forming step. The electrode forming step
S4 is preferably carried out after the damage evaluation step S2
and more preferably carried out after the thin film forming step
S3.
[0215] The electrode forming step S4 is to form an electrode 40,
for example, of a metal film or the like on the thin film 30, as
shown in FIG. 7A. In this case, the thin film 30 has excellent
crystallinity and reduced surface roughness, and occurrence of
damage can be suppressed at the interface between the thin film 30
and the electrode 40.
[0216] The electrode 40 may also be formed directly on the surface
10a of the compound semiconductor substrate 10. In that case, when
the compound semiconductor substrate 10 with a low level of damage
is used, occurrence of damage can be suppressed at the interface
between the compound semiconductor substrate 10 and the electrode
40.
[0217] The electrode forming step S4 may also be to form an
electrode 42 on the thin film 32, as shown in FIG. 7B. In this
case, the thin film 32 has excellent crystallinity and reduced
surface roughness, and occurrence of damage can be suppressed at
the interface between the thin film 32 and the electrode 42.
[0218] The electrode 40 may also be formed directly on the surface
20a of the compound semiconductor membrane 20. In that case, when
the compound semiconductor membrane 20 with a low level of damage
is used, occurrence of damage can be suppressed at the interface
between the compound semiconductor membrane 20 and the electrode
42.
[0219] A compound semiconductor device can be produced through the
steps described above.
[0220] The preferred embodiments of the present invention were
described above in detail, but it is noted that the present
invention is not limited to the above embodiments.
[0221] Subsequently, Experiment Examples associated with the above
embodiments will be described.
EXPERIMENTAL EXAMPLE 1
[0222] First, a monocrystalline GaN ingot was sliced to prepare a
monocrystalline GaN substrate with the diameter of 2 inches. The
surface of the monocrystalline GaN substrate prepared was polished
and thereafter the surface was dry-etched by reactive ion etching
(RIE). The conditions for dry etching were as follows. [0223]
Etching gas: Ar gas [0224] Supplied power: 200 W [0225] Pressure in
chamber: 1.3 Pa (10 mTorr) [0226] Etching time: 10 minutes
[0227] Thereafter, in order to eliminate damage from the surface,
the monocrystalline GaN substrate was immersed in a 5% NH.sub.4OH
solution at 40.degree. C. for 15 minutes to effect wet etching. The
monocrystalline GaN substrate of Experiment Example 1 was obtained
as described above.
EXPERIMENTAL EXAMPLE 2
[0228] First, a monocrystalline GaN ingot was sliced to prepare a
monocrystalline GaN substrate with the diameter of 2 inches. The
surface of the monocrystalline GaN substrate prepared was roughly
polished and thereafter the surface was further polished by means
of diamond abrasive grains with the grain size of 0.5 .mu.m.
Thereafter, the surface was cleaned with isopropyl alcohol. The
monocrystalline GaN substrate of Experiment Example 2 was obtained
as described above.
EXPERIMENTAL EXAMPLE 3
[0229] A monocrystalline GaN substrate of Experiment Example 3 was
obtained in the same manner as in Experiment Example 2 except that
diamond abrasive grains with the grain size of 0.1 .mu.m were used
instead of the diamond abrasive grains with the grain size of 0.5
.mu.m.
EXPERIMENTAL EXAMPLE 4
[0230] A monocrystalline GaN substrate of Experiment Example 4 was
obtained by effecting dry etching in Experiment Example 1 on a
monocrystalline GaN substrate obtained in the same manner as in
Experiment Example 3.
EXPERIMENTAL EXAMPLE 5
[0231] A monocrystalline GaN substrate of Experiment Example 5 was
obtained by effecting wet etching with a diluted H.sub.3PO.sub.4
solution on a monocrystalline GaN substrate obtained in the same
manner as in Experiment Example 3.
[0232] (Photoluminescence Measurement and Fluorescence Microscope
Measurement)
[0233] Photoluminescence measurement was carried out for each of
the surfaces of the monocrystalline GaN substrates of Experiment
Example 1 to Experiment Example 5. In the photoluminescence
measurement a He--Cd laser capable of emitting a laser beam with
the wavelength of 325 nm was used as a light source. The laser beam
was made incident normally to the surface of each monocrystalline
GaN substrate to obtain an emission spectrum. In each emission
spectrum a peak was observed near 365 nm.
[0234] The photoluminescence measurement was carried out at
wavelength intervals of 0.5 nm and values near the peak were
interpolated by a normal distribution. The background was adjusted
by linear approximation of wing portions of the peak.
[0235] Fluorescence microscope measurement was carried out for each
of the surfaces of the monocrystalline GaN substrates of Experiment
Example 1 to Experiment Example 5. The fluorescence microscope
measurement was carried out using an optical system capable of
transmitting light of wavelengths of not less than 345 nm.
[0236] It was found from the results of the photoluminescence
measurement and the fluorescence microscope measurement that there
was a correlation between photoluminescence intensities (PL
intensities) in the photoluminescence measurement and fluorescence
intensities in the fluorescence microscope measurement.
[0237] FIG. 8 shows the PL intensities of the peak near 365 nm in
the respective emission spectra obtained by the photoluminescence
measurement on the surfaces of the monocrystalline GaN substrates
of Experiment Example 1 to Experiment Example 5. The values of PL
intensities are relative values with respect to 1 for the PL
intensity of Experiment Example 2. FIG. 8 shows that the PL
intensity decreases and the level of damage on the surface
increases, in the order of Experiment Example 1, Experiment Example
4, Experiment Example 5, Experiment Example 3, and Experiment
Example 2.
[0238] (Spectroscopic Ellipsometry Measurement)
[0239] The spectroscopic ellipsometry measurement was carried out
for the surfaces of the monocrystalline GaN substrates of
Experiment Example 1 to Experiment Example 5. The spectroscopic
ellipsometer 16 used was the spectroscopic ellipsometer available
from SOPRA Co. The spectroscopic ellipsometry measurement was
conducted at angles of incidence of the light LT1, 65.degree.,
70.degree., and 75.degree..
[0240] (Damage Evaluation)
[0241] The model structure of the monocrystalline GaN substrate was
a monocrystalline GaN substrate with a damage layer as a surface
layer. Furthermore, it was assumed that there was a mixed layer of
an oxide film and an uneven layer on the damage layer. The mixed
layer was assumed to be a mixture of oxide and air 50% each by use
of effective medium approximation. The optical simulation and
fitting were conducted using this model structure. These resulted
in obtaining the spectrum of the real part .epsilon..sub.1 of the
complex dielectric constant, the spectrum of the imaginary part
.epsilon..sub.2 of the complex dielectric constant, the spectrum of
the real part N of the complex index of refraction, and the
spectrum of the imaginary part K of the complex index of
refraction.
[0242] FIG. 9 is a graph showing part of spectra of the imaginary
part K of the complex index of refraction obtained by the
spectroscopic ellipsometry measurement. Spectra K1 to K5 in the
graph indicate spectra in the wavelength range of 280 to 400 nm of
the imaginary part K of the complex index of refraction of the
monocrystalline GaN substrates of Experiment Example 1 to
Experiment Example 5, respectively. It was found from the graph
that the spectra K1 to K5 had a spread on the longer wavelength
side and on the shorter wavelength side than 365 nm. It was also
found that values of the imaginary part K of the complex index of
refraction on the longer wavelength side and on the shorter
wavelength side than 365 nm tended to increase with increasing
level of damage.
[0243] In the spectra K1 to K5, absolute values of the difference
between the imaginary part K of the complex index of refraction at
360 nm and the imaginary part K of the complex index of refraction
at 370 nm were 0.215, 0.044, 0.138, 0.200, and 0.188, respectively.
In the spectra K1 to K5, absolute values of the imaginary part K of
the complex index of refraction at 370 nm were 0.012, 0.023, 0.031,
0.118, and 0.187, respectively.
[0244] FIG. 10 is a graph showing part of spectra of the imaginary
part K of the complex index of refraction obtained by the
spectroscopic ellipsometry measurement. The spectra K11 to K15 in
the graph indicate spectra in the wavelength range of 700 to 1600
nm of the imaginary part K of the complex index of refraction of
the monocrystalline GaN substrates of Experiment Example 1 to
Experiment Example 5, respectively. As shown in the graph, the
spectrum K12 only had a peak, and the spectra K11, K13 to K15 had
the value of zero. It was thus found that the spectroscopic
ellipsometry measurement of the monocrystalline GaN substrate with
a high level of damage provided nonzero values of the imaginary
part K of the complex index of refraction in the wavelength range
of 700 to 1600 nm.
[0245] FIG. 11 is a graph showing the first derivatives of the
spectra K1 to K5 shown in FIG. 9. Specifically, the imaginary part
K of the complex index of refraction was differentiated with
respect to wavelength. Spectra L1 to L5 indicating the first
derivatives in the graph represent the first derivatives of the
spectra K1 to K5 (slopes of spectra K1 to K5), respectively. As
shown in the graph, each of the spectra L1 to L5 has an extremum.
It was found that with increasing level of damage, the absolute
value of the extremum (the maximum absolute value of the slope of
spectra K1 to K5) tended to decrease and the wavelength at the
extremum tended to decrease.
[0246] In the spectra L2, L3, the wavelengths at which the absolute
value of the slope of the spectrum in the wavelength band of 300 to
400 nm was maximum, were 347 nm and 361 nm, respectively.
[0247] FIG. 12 is a graph showing part of spectra of the imaginary
part .epsilon..sub.2 of the complex dielectric constant obtained by
spectroscopic ellipsometry measurement. Spectra M1 to M5 in the
graph represent spectra in the wavelength band of 280 to 400 nm of
the imaginary part .epsilon..sub.2 of the complex dielectric
constant of the monocrystalline GaN substrates of Experiment
Example 1 to Experiment Example 5, respectively. It was found from
the graph that the spectra M1 to M5 had a spread on the longer
wavelength side and on the shorter wavelength side than 365 nm. It
was also found that with increasing level of damage, values of the
imaginary part .epsilon..sub.2 of the complex dielectric constant
on the longer wavelength side and on the shorter wavelength side
than 365 nm tended to increase.
[0248] In the spectra M1, M4, M5, M3, and M2, absolute values of
the difference between the imaginary part .epsilon..sub.2 of the
complex dielectric constant at 360 nm and the imaginary part
.epsilon..sub.2 of the complex dielectric constant at 370 nm were
1.2, 1.195, 1.048, 0.759, and 0.235, respectively. In the spectra
M1, M4, M5, M3, and M2, absolute values of the imaginary part
.epsilon..sub.2 of the complex dielectric constant at 370 nm were
0.13, 0.16, 0.63, 0.73, and 1.00, respectively.
[0249] FIG. 13 is a graph showing part of spectra of the imaginary
part .epsilon..sub.2 of the complex dielectric constant obtained by
spectroscopic ellipsometry measurement. Spectra M11 to M15 in the
graph represent spectra in the wavelength band of 700 to 1700 nm of
the imaginary part .epsilon..sub.2 of the complex dielectric
constant of the monocrystalline GaN substrates of Experiment
Example 1 to Experiment Example 5, respectively. As shown in the
graph, the spectrum M12 only had a peak, and the spectra M11, M13
to M15 had the value of zero. It was thus found that the
spectroscopic ellipsometry measurement of the monocrystalline GaN
substrate with a high level of damage provided nonzero values of
the imaginary part .epsilon..sub.2 of the complex dielectric
constant in the wavelength band of 700 to 1700 nm.
[0250] FIG. 14 is a graph showing the first derivatives of the
spectra M1 to M5 shown in FIG. 12. Specifically, the imaginary part
.epsilon..sub.2 of the complex dielectric constant was
differentiated with respect to wavelength. The spectra N1 to N5
showing the first derivatives in the graph indicate the first
derivatives of the spectra M1 to M5 (slopes of spectra M1 to M5),
respectively. As shown in the graph, each of the spectra N1 to N5
has an extremum. It was also found that with increasing level of
damage, the absolute value of the extremum (the maximum absolute
value of the slope of spectra M1 to M5) tended to decrease and the
wavelength at the extremum tended to decrease.
[0251] In the spectra N2, N3, the wavelengths at which the absolute
value of the slope of the spectrum in the wavelength band of 300 to
400 nm was maximum, were 347 nm and 363 nm, respectively.
[0252] FIG. 15 is a graph showing part of spectra of the real part
N of the complex index of refraction obtained by spectroscopic
ellipsometry measurement. Spectra P1 to P5 in the graph show
spectra in the wavelength band of 280 to 400 nm of the real part N
of the complex index of refraction of the monocrystalline GaN
substrates of Experiment Example 1 to Experiment Example 5,
respectively. It was found from the graph that with increasing
level of damage, the maximum of spectra P1 to P5 tended to
decrease, the wavelength at the maximum tended to decrease, and the
half width of the peak tended to increase.
[0253] In the spectra P1, P4, P5, P3, and P2, absolute values of
the difference between the real part N of the complex index of
refraction at 365 nm and the real part N of the complex index of
refraction at 375 nm were 0.15, 0.134, 0.126, 0.08, and 0.029,
respectively. In the spectra P2 and P3, maxima of the spectra in
the wavelength band of 300 to 400 nm were 2.69 and 2.733,
respectively.
[0254] FIG. 16 is a graph showing the first derivatives of the
spectra P1 to P5 shown in FIG. 15. Specifically, the real part N of
the complex index of refraction was differentiated with respect to
wavelength. Spectra U1 to U5 showing the first derivatives in the
graph represent the first derivatives of the spectra P1 to P5
(slopes of spectra P1 to P5), respectively. As shown in the graph,
the spectra U1 to U5 have a maximum in a portion located on the
shorter wavelength side than the wavelength where the value of the
first derivative is zero. It was found from this fact that in the
portion located on the shorter wavelength side than the wavelength
where the value of the first derivative was zero, the absolute
value of the extremum (the maximum absolute value of the slope of
spectra P1 to P5) tended to decrease and the wavelength at the
extremum tended to decrease, with increasing level of damage.
[0255] The spectra U1 to U5 have a minimum in a portion located on
the longer wavelength side than the wavelength where the value of
the first derivative is zero. It was found from this fact that in
the portion located on the longer wavelength side than the
wavelength where the value of the first derivative was zero, the
absolute value of the extremum (the maximum absolute value of the
slope of spectra P1 to P5) decreased with increasing level of
damage.
[0256] FIG. 17 is a graph showing part of spectra of the real part
.epsilon..sub.1 of the complex dielectric constant obtained by
spectroscopic ellipsometry measurement. Spectra Q1 to Q5 in the
graph represent spectra in the wavelength band of 280 to 400 nm of
the real part .epsilon..sub.1 of the complex dielectric constant of
the monocrystalline GaN substrates of Experiment Example 1 to
Experiment Example 5, respectively. It was found from the graph
that with increasing level of damage, the maximum of spectra Q1 to
Q5 tended to decrease, the wavelength at the maximum tended to
decrease, and the half width of the peak tended to increase.
[0257] In the spectra Q2, Q3, Q5, Q4, and Q1, absolute values of
the difference between the real part 6l of the complex dielectric
constant at 365 nm and the real part .epsilon..sub.1 of the complex
dielectric constant at 375 nm were 0.125, 0.326, 0.416, 0.589, and
0.69, respectively. In the spectra Q2 and Q3, maxima of spectra in
the wavelength band of 300 to 400 nm were 7.15 and 7.43,
respectively.
[0258] FIG. 18 is a graph showing the first derivatives of the
spectra Q1 to Q5 shown in FIG. 17. Specifically, the real part
.epsilon..sub.1 of the complex dielectric constant was
differentiated with respect to wavelength. Spectra V1 to V5 showing
the first derivatives in the graph represent the first derivatives
of the spectra Q1 to Q5 (slopes of spectra Q1 to Q5), respectively.
As shown in the graph, each of the spectra V1 to V5 has a maximum
in a portion located on the shorter wavelength side than the
wavelength where the value of the first derivative is zero. It was
found from this fact that in the portion located on the shorter
wavelength side than the wavelength where the value of the first
derivative was zero, the absolute value of the extremum (the
maximum absolute value of the slope of spectra Q1 to Q5) tended to
decrease and the wavelength at the extremum tended to decrease,
with increasing level of damage.
[0259] Each of the spectra V1 to V5 has a minimum in a portion
located on the longer wavelength side than the wavelength where the
value of the first derivative is zero. It was found from this fact
that in the portion located on the longer wavelength side than the
wavelength where the value of the first derivative was zero, the
absolute value of the extremum (the maximum absolute value of the
slope of spectra Q1 to Q5) decreased with increasing level of
damage.
[0260] FIG. 19 is a graph showing a relation between the PL
intensities shown in FIG. 8, and maximum absolute values of the
slopes of the spectra of the imaginary part K of the complex index
of refraction and the real part N of the complex index of
refraction. Plots D1 to D5 in the graph represent maximum absolute
values G1.sub.max of the slopes of the spectra K1 to K5 shown in
FIG. 9 (absolute values of extrema of the spectra L1 to L5 shown in
FIG. 11), respectively. Plots E1 to E5 represent maximum absolute
values G3.sub.max of the slopes of the spectra P1 to P5 shown in
FIG. 15, respectively. It was found from the graph that with
increasing level of damage, the values of plots D1 to D5, E1 to E5
tended to decrease. It was thus found that with increasing level of
damage, the maximum absolute values G3.sub.max of the slopes of
spectra P1 to P5 shown in FIG. 15 decreased.
[0261] FIG. 20 is a graph showing a relation between the PL
intensities shown in FIG. 8, and wavelengths where the absolute
values of the slopes of the spectra of the imaginary part K of the
complex index of refraction and the real part N of the complex
index of refraction are maximum. Plots F1 to F5 in the graph
represent wavelengths at the maximum absolute value of the slopes
of the spectra K1 to K5 shown in FIG. 9 (wavelengths of extrema of
spectra L1 to L5 shown in FIG. 11), respectively. In addition,
plots G1 to G5 represent wavelengths at the maximum absolute value
of the slope (first derivative) in the portion located on the
shorter wavelength side than the wavelength corresponding to the
maximum in the spectra P1 to P5 shown in FIG. 15, respectively. It
was found from the graph that with increasing level of damage, the
values of plots F1 to F5, G1 to G5 tended to decrease.
[0262] Next, Table 1 presents the thickness of the mixed layer in
the model structure of the monocrystalline GaN substrate. It was
found from Table 1 that the thickness of the mixed layer increased
with increasing level of damage. TABLE-US-00001 TABLE 1 Thickness
of Mixed Layer[nm] Experiment Example 2 7.2 Experiment Example 3
3.8 Experiment Example 5 1.6 Experiment Example 4 1.1 Experiment
Example 1 0.7
EXPERIMENTAL EXAMPLE 6
[0263] A monocrystalline GaN substrate obtained in the same manner
as in Experiment Example 1 was cleaned with HCl solution and 50%
NH.sub.4OH to obtain a monocrystalline GaN substrate of Experiment
Example 6.
EXPERIMENTAL EXAMPLE 7
[0264] A monocrystalline GaN substrate obtained in the same manner
as in Experiment Example 6 was cleaned with a solution as a 1:1
mixture of 50% NH.sub.4OH and hydrogen peroxide to obtain a
monocrystalline GaN substrate of Experiment Example 7. The
spectroscopic ellipsometry measurement was carried out in the same
manner as in Experiment Example 1 to Experiment Example 5, for each
of surfaces of the monocrystalline GaN substrates of Experiment
Examples 6, 7.
[0265] FIG. 21 is a graph showing part of spectra of the imaginary
part K of the complex index of refraction obtained by the
spectroscopic ellipsometry measurement. Spectra K6, K7 in the graph
represent spectra in the wavelength band of 250 to 550 nm of the
imaginary part K of the complex index of refraction of the
monocrystalline GaN substrates of Experiment Examples 6, 7,
respectively. A maximum was observed near 365 nm in the graph. A
conceivable reason for it is that repetitions of cleaning decreased
the level of damage to facilitate generation of exciton.
EXPERIMENTAL EXAMPLE 8
[0266] A surface of a monocrystalline InP substrate with the
diameter of 3 inches having the (110) face as a principal surface
was polished to be mirror-finished, thereby obtaining a
monocrystalline InP substrate of Experiment Example 8.
EXPERIMENTAL EXAMPLE 9
[0267] A monocrystalline InP substrate of Experiment Example 9 was
obtained in the same manner as in Experiment Example 8, except that
dry etching by the reactive ion etching method was carried out,
instead of the mirror finishing.
EXPERIMENTAL EXAMPLE 10
[0268] A monocrystalline InP substrate obtained in the same manner
as in Experiment Example 9 was subjected to wet etching, thereby
obtaining a monocrystalline InP substrate of Experiment Example
10.
[0269] (Cathodoluminescence Intensity Measurement)
[0270] Cathodoluminescence intensity measurement was carried out
for the monocrystalline InP substrates of Experiment Example 8,
Experiment Example 9, and Experiment Example 10. The measurement
results are presented in Table 2. The cathode luminescence
intensities are relative values, based on the cathodoluminescence
intensity of the monocrystalline InP substrate of Experiment
Example 8 being defined as 1. The cathodoluminescence intensities
are values obtained by integrating an intensity distribution near
the wavelength of 900 nm within a field of view at the
magnification of 10000. TABLE-US-00002 TABLE 2 Cathodoluminescence
Intensity (Relative Intensity) Experiment Example 8 1.00 Experiment
Example 9 1.18 Experiment Example 10 1.26
[0271] It was found from Table 2 that the level of damage tended to
decrease in the order of Experiment Example 8, Experiment Example
9, and Experiment Example 10.
[0272] (Spectroscopic Ellipsometry Measurement)
[0273] The spectroscopic ellipsometry measurement was carried out
for surfaces of the monocrystalline InP substrates of Experiment
Example 8, Experiment Example 9, and Experiment Example 10. The
spectroscopic ellipsometer 16 used was the spectroscopic
ellipsometer available from SOPRA Co. The angle of incidence of the
light LT1 was set at 75.degree. to achieve high measurement
sensitivity. The measurement wavelength range was 300 to 1200
nm.
[0274] (Damage Evaluation)
[0275] The model structure of the monocrystalline InP substrate was
a monocrystalline InP substrate having a damage layer as a surface
layer. Furthermore, a mixed layer of an oxide film and an uneven
layer was assumed on the damage layer. The mixed layer was assumed
to be a layer as a mixture of oxide and air 50% each by effective
medium approximation. Optical simulation and fitting were conducted
using this model structure. They resulted in obtaining the spectrum
of the real part .epsilon..sub.1 of the complex dielectric
constant, the spectrum of the imaginary part .epsilon..sub.2 of the
complex dielectric constant, the spectrum of the real part N of the
complex index of refraction, and the spectrum of the imaginary part
K of the complex index of refraction.
[0276] FIG. 22 is a graph showing part of spectra of the real part
N of the complex index of refraction obtained by the spectroscopic
ellipsometry measurement. Spectra P8 to P10 in the graph represent
spectra in the wavelength band of 800 to 1100 nm of the real part N
of the complex index of refraction of the monocrystalline InP
substrates of Experiment Example 8 to Experiment Example 10,
respectively.
[0277] FIG. 23 is a graph showing part of spectra of the imaginary
part K of the complex index of refraction obtained by the
spectroscopic ellipsometry measurement. Spectra K8 to K10 in the
graph represent spectra in the wavelength band of 800 to 1100 nm of
the imaginary part K of the complex index of refraction of the
monocrystalline InP substrates of Experiment Example 8 to
Experiment Example 10, respectively.
[0278] FIG. 24 is a graph showing the first derivatives of the
spectra P8 to P10 shown in FIG. 22. Specifically, the imaginary
part K of the complex index of refraction was differentiated with
respect to wavelength. Spectra L8 to L10 showing the first
derivatives in the graph represent the first derivatives of the
spectra K8 to K10, respectively. As shown in the graph, each of the
spectra L8 to L10 has an extremum. It was found that with
increasing level of damage, the absolute value of the extremum (the
maximum absolute value of the slope of spectra P8 to P10) tended to
decrease and the wavelength at the extremum tended to decrease.
[0279] FIG. 25 is a graph showing a relation between
cathodoluminescence intensities and maximum absolute values of
slopes. Plots H8 to H10 in the graph represent the maximum absolute
values G1.sub.max of the slopes of the spectra K8 to K10 shown in
FIG. 23 (absolute values of the extrema of the spectra L8 to L10
shown in FIG. 24), respectively. It was found from the graph that
there was a correlation between the cathodoluminescence intensities
and the maximum absolute values of the slopes and that the maximum
absolute value of the slope tended to increase with decreasing
level of damage.
EXPERIMENTAL EXAMPLE 11
[0280] A monocrystalline GaN substrate of Experiment Example 11
without damage was prepared in the same manner as in Experiment
Example 1, except that the substrate used was a monocrystalline GaN
substrate 20 mm square.
EXPERIMENTAL EXAMPLE 12
[0281] A monocrystalline GaN ingot was sliced to prepare a
monocrystalline GaN substrate 20 mm square. The surface of the
monocrystalline GaN substrate prepared was roughly polished and
thereafter the surface was further polished with diamond abrasive
grains having the grain size of 0.1 .mu.m, thereby obtaining the
monocrystalline GaN substrate of Experiment Example 12.
EXPERIMENTAL EXAMPLE 13
[0282] A monocrystalline GaN substrate of Experiment Example 13 was
prepared in the same manner as in Experiment Example 12, except
that diamond abrasive grains with the grain size of 0.5 .mu.m were
used instead of the diamond abrasive grains with the grain size of
0.1 .mu.m.
[0283] (Spectroscopic Ellipsometry Measurement)
[0284] Spectroscopic ellipsometry measurement was carried out for
the surfaces of the monocrystalline GaN substrates of Experiment
Example 11 to Experiment Example 13. The maximum absolute value
G1.sub.max of the slope of the spectrum of the imaginary part K of
the complex index of refraction was calculated for each of the
monocrystalline GaN substrates of Experiment Example 11 to
Experiment Example 13. The maximum G1.sub.max of Experiment Example
12 was 2.6 times the maximum G1.sub.max of Experiment Example 13.
The maximum G1.sub.max of Experiment Example 11 was 2.9 times the
maximum G1.sub.max of Experiment Example 13. This confirmed that
the level of damage on the surface increased in the order of
Experiment Example 11, Experiment Example 12, and Experiment
Example 13.
[0285] Next, a thin film of GaN was formed in the film thickness of
1 .mu.m by HVPE on the surfaces of the monocrystalline GaN
substrates of Experiment Example 11 to Experiment Example 13. The
conditions for forming a thin film of GaN were as follows. GaCl gas
was obtained by reaction of Ga metal with HCl gas at 880.degree. C.
[0286] Temperature of monocrystalline GaN substrate: 1000.degree.
C. [0287] Reaction gases: NH.sub.3 gas and GaCl gas [0288] Pressure
of NH.sub.3 gas: 10 kPa [0289] Pressure of GaCl gas: 0.6 Pa
[0290] After formation of the GaN thin film, the surface roughness
(Ra: arithmetic mean roughness) was measured for the GaN thin film
by AFM. A percentage of lattice strain to the bulk was measured by
X-ray diffraction. The measurement results of these are presented
in Table 3. It was found from Table 3 that the monocrystalline GaN
substrates of Experiment Example 11 and Experiment Example 12 had
satisfactory performance as substrates to be used for the compound
semiconductor devices. TABLE-US-00003 TABLE 3 Arithmetic Mean
Percentage of Lattice Roughness[nm] Strain[%] Experiment Example 11
0.82 0.01 Experiment Example 12 0.95 0.04 Experiment Example 13
1.43 0.15
[0291] The present invention provides the damage evaluation methods
of the compound semiconductor member and the production methods of
the compound semiconductor member capable of highly accurately
evaluating the level of damage on the surface and provides the
gallium nitride compound semiconductor members and gallium nitride
compound semiconductor membranes with a low level of damage.
* * * * *