Download Controllability of Mesoscopic Surface Roughness of Sputtered Al

Materials Transactions, Vol. 46, No. 12 (2005) pp. 3037 to 3043
#2005 The Japan Institute of Metals
Controllability of Mesoscopic Surface Roughness
of Sputtered Al and Al–N Films
Takashi Ishiguro*1 and Kazumi Miyamura*2
Department of Electrical Engineering, Nagaoka University of Technology, Nagaoka 940-2188, Japan
There is a concept that the surface roughness is the structure-type of the optical gradient function because the material volume fraction
gradually changes along the surface normal from the vacuum (or air) to the material. In this sense, the controlling the surface undulations
corresponds to the controlling the gradient-profile.
Sputtered Al film, AlN film, and incompletely-nitrided Al film (Al–N film) has been investigated from the view point of forming the
mesoscopic scaled surface roughness because it can be applied to, e.g., the solar absorption coating, the solar cell and/or the diffusive coloring.
The surface structures have been characterized and the growth condition dependence, such as the degree of nitriding which is controlled by
changing mixing ratio of N2 and Ar sputtering gases, the film thickness (df ), and the substrate temperature (Tsub ), are systematically investigated.
The following controllable factors about surface structure are confirmed. The characteristic of the form of protuberances on the surface depends
on the degree of nitriding. The height and the width of the protuberances can be controlled by df and/or Tsub .
(Received August 5, 2005; Accepted October 25, 2005; Published December 15, 2005)
Keywords: surface roughness, sputtered aluminum film, AlN film, aluminum–nitrogen film, gradient function, thin film
1.
Introduction
The optical absorption property and/or the coloring on the
material surface depend not only on the intrinsic optical
property of the material itself but also on its surface structure.
The scale of the surface structure concerned in this paper is
the same as the optical wavelength or less than it, i.e.,
mesoscopic surface structure. In the case of the solar energy
thermal conversion coating, e.g., the surface roughness
induced by film growth at higher substrate temperature1)
and/or the dense array of metal whisker structure2) were
applied to improve the efficiency. On the other hand, the solar
cell performance is also related to the surface roughness.3)
Special surface textures, e.g., the inverted hemisphere
structure or the inverted pyramid structure were introduced
to improve the efficiency of the solar cell.4,5)
Recently we discovered that the rf sputtered incompletelynitrided Al film (Al–N film) with submicron scaled surface
roughness showed high solar absorptance.6) It has been
combined with Al film with low emittance, then a solar
selective absorber coating with surface roughness was
successfully fabricated.7) On the other hand, the surface
roughness is also applied to realize the diffusive structural
colour. The Al film with the proper surface roughness showed
high white diffuse reflection whose colour was changed by
the coating of the transparent AlN film.8) In these examples,
the important technology is to find the experimental
condition to realize the proper surface roughness. However,
the characterization of these mesoscopic surface roughness
and the forming conditions have not been investigated
systematically. The aim of this paper is to find the relation
between the mesoscopic surface structure formed on the Al
and Al–N films and the film growth conditions such as the
degree of nitriding, the film thickness (df ), and the substrate
*1Corresponding
author, E-mail: [email protected]
address: Electronics Components Division, The Furukawa
Electric Co., Ltd., Hiratsuka, Japan
*2Present
temperature (Tsub ). Of course, several investigations at
respective formation conditions of Al and/or Al alloy films
have been performed.9–13) However, the perspective view
point to control the forming of the mesoscopic submicron
scaled surface roughness, which is useful to be gradient
function, has been lacked.
2.
Experiment
Films were deposited using an rf sputtering apparatus,
which was evacuated using a turbo-molecular pump. The
base pressure was 4 105 Pa. A water-cooled Al target
(99.999%, 100 mm diameter) was sputtered at 200 W. The
substrate was Corning # 7059 glass. The substrate temperature was changed from 303 to 413 K. A pure Ar or a mixture
of Ar (99.9995%) and N2 (99.9995%) was used as the
sputtering gas. The total working gas pressure (Ptot ) was
1.33 Pa, which is the sum of the partial pressure of N2 (PN2 )
and the partial pressure of Ar.
The composition of the film was controlled by changing
the concentration of PN2 (cN2 ¼ PN2 =Ptot 100) from 0%N2
to 13%N2 . Then the substance of the deposited film was
changed from Al to AlN via incompletely-nitrided Al (Al–N)
with increasing cN2 . It has been already confirmed by the
X-ray diffraction in our previous study that the Al–N films
are composed of Al and AlN.6)
Surface morphology was observed using an atomic force
microscope (AFM) (Nanopics, NPX100AFM head, Seiko
Instruments Inc.). The silicon AFM-cantilever which has a
tip with the radius of curvature of 20 nm was used. The
scanning-surface was performed in the atmosphere at room
temperature by using the damping-mode.
Film thickness was measured using a surface-texturemeasuring instrument (Surfcom 200B, Tokyo Seimitu Ltd.).
The deposition rate was evaluated from the smooth film with
a thickness of less than 100 nm, because the reproducibility
of the measured film thickness with the deposition time could
be kept up to only approximately 100 nm for the specimen of
3038
T. Ishiguro and K. Miyamura
Fig. 1 N2 gas concentration (cN2 ) dependence of AFM plan view images and corresponding FT-LHD-profiles of 300-nm-thick specimen
deposited at Tsub ¼ 373 K. cN2 : (a) 0%N2 , (b) 2%N2 , (c) 6%N2 , (d) 10%N2 , and (e) 13%N2 .
cN2 ¼ 0{8%N2 . One reason for the non-reproducibility of the
measured film thickness is thought to be the enhanced surface
roughness during film growth, as described later. For
convenience, we identify films from their estimated thicknesses based on the deposition rates in this paper hereafter.
Maximum film thickness, prepared in this study was 700 nm.
If need be, each film was also labeled on the basis of its
substance, concentration of PN2 (cN2 ), deposition thickness
(df ), and substrate temperature (Tsub ), e.g., Al–N(cN2 , df ,
Tsub ) represents an Al–N film with a thickness of df deposited
at Tsub using a gas mixture with cN2 .
3.
Results and Discussion
As an characterization item, Fourier transformation of the
local-height-distribution (FT-LHD) of the surface has been
defined as follows: First of all, the two dimensional
distribution of height (512 512 pixels) within an AFM’s
scanning field (10 mm 10 mm) is transformed into the two
dimensional distribution of Fourier amplitude. Corresponding intensity distribution is calculated by amplitude squared,
and is averaged around the circle with equal radius from the
origin in reciprocal space. Then the one dimensional quantity
appeared to be a profile of the section area of the averaged
intensity distribution is defined. Let us call this FT-LHDprofile, which has a meaning correspond to the Fourier
transformation of the one dimensional correlation function of
height distribution function.
3.1 Change of surface morphology by nitriding
In order to characterize the nitriding effect on the surface
morphology, the films are formed in a cN2 range between
0%N2 and 13%N2 at df ¼ 300 nm and Tsub ¼ 373 K.
Figure 1 shows the cN2 dependence of the AFM plan view
images and the corresponding respective FT-LHD-profiles.
In these images, the change of darkness from darkness toward
brightness corresponds to the elevating height. The surface of
pure Al film [Fig. 1(a)] looks like connecting protuberances
and/or range of protuberances with width about 1 mm. The
corresponding FT-LHD-profile shows a peak at around
1 mm1 . Figures 1(b) and (c) show the incompletely-nitrided
Al surfaces. They are composed of isolated protuberances
with variety of width and height. The corresponding FTLHD-profiles do not have any characteristic peak like
Fig. 2 Change of distribution of relative frequency of the height along the
surface normal of the 300-nm-thick specimen deposited at Tsub ¼ 373 K as
a function of N2 gas concentration (cN2 ).
Fig. 1(a) but rather monotonously reduce with increasing
1=d from the origin. The surface of much more nitrided film
becomes flat with sparse protuberances [Fig. 1(d)], and then
becomes smooth surface [Fig. 1(e)].
The cN2 dependence of the distribution of relative
frequency of the height (RFH) is shown in Figs. 2(a) and
(b) in which the origin of the height axis corresponds to the df
of 300 nm. It is obvious that the distributions of RFH of the
films with cN2 less than or equal to 8%N2 show relatively
broad distribution. However, there is a characteristic difference between pure Al film and Al–N films, i.e., the
distribution of RFH of the pure Al film has a maximum
position at the positive side of height, on the other hand, the
Al–N films have their maximum positions on negative side
without exception. This difference is because of the difference between the connecting protuberances and the isolated
protuberances with variety of size in the morphology as
mentioned in Figs. 1(a) and (b)–(c).
Figure 3 shows the height dependence of material volume
fraction ( f ) of 300-nm-thick specimens deposited at different
cN2 . The gradual change in f is realized in the Al–N films due
to the surface roughness, which is composed of protuber-
Controllability of Mesoscopic Surface Roughness of Sputtered Al and Al–N Films
Fig. 3 N2 gas concentration (cN2 ) dependence of height dependence of
volume fraction ( f ) of the 300-nm-thick specimen deposited at
Tsub ¼ 373 K.
ances with the variety of height and width. This is thought to
be a gradient-index optics, which is one of the reasons of the
low specular reflectance, low diffuse reflectance, and high
absorptance in visible and near infra-red regions6) as shown
in Fig. 4.
3.2
Film thickness dependence of surface morphology of
Al and Al–N films
Al(0%N2 , df ¼ 46{500 nm, 373 K) films and Al–N
(6%N2 , df ¼ 55{700 nm, 373 K) films are deposited. Re-
3039
spective df dependences of the AFM plan view images and
the FT-LHD-profiles are shown in Figs. 5 and 6. The surface
of 46-nm-thick Al film [Fig. 5(a)] is composed of fine
protuberances with the size from about one hundred-nm to a
few hundreds-nm widths. There is a tendency that these
protuberances connect each other to form the range of
protuberances. With increasing film thickness [images in
Figs. 5(b)–(e)], the size of the range of protuberances
increases up to the micrometer order. This fact can also be
seen in the change of FT-LHD-profiles as the shift of the peak
indicated by arrow. As contrasted to the way of the growth of
protuberance in the Al film, the Al–N film growth shows
different way. The surface of 55-nm-thick Al–N film
[Fig. 6(a)] is also composed of protuberances as small as
the 46-nm-thick Al film [Fig. 5(a)]. However, the respective
protuberances in the 55-nm-thick Al–N film seem to be
isolated. With increasing df [Figs. 6(b)–(e)], the size of the
protuberance increases. However, there is not a tendency that
the neighboring protuberances connect each other. Furthermore, the difference in the peak height of protuberance is
seemed to be enhanced. This fact reflects to the change of FTLHD-profiles (Fig. 6) with rather monatomic change compared with the profiles in Fig. 5.
Figures 7(a) and (b) show the comparison of df dependence of RFH between Al (0%N2 , df ¼ 100{500 nm, 373 K)
Fig. 4 N2 gas concentration (cN2 ) dependence of optical properties of the 300-nm-thick specimen deposited at Tsub ¼ 373 K: (a) specular
reflectance, (b) diffuse reflectance, and (c) absorptance.
3040
T. Ishiguro and K. Miyamura
Fig. 5 Film thickness (df ) dependence of AFM plan view images and corresponding FT-LHD-profiles of pure Al film deposited at
Tsub ¼ 373 K. df : (a) 46 nm, (b) 100 nm, (c) 200 nm, (d) 300 nm, and (e) 500 nm.
Fig. 6 Film thickness (df ) dependence of AFM plan view images and corresponding FT-LHD-profiles of Al–N film deposited at
cN2 ¼ 6%N2 , Tsub ¼ 373 K. df : (a) 55 nm, (b) 200 nm, (c) 300 nm, (d) 500 nm, and (e) 700 nm.
Fig. 7 Film thickness (df ) dependence of distribution of relative frequency
of the height along the surface normal of the specimens (a): Al (0%N2 ) and
(b): Al–N (6%N2 ) deposited at Tsub ¼ 373 K.
films and Al–N (6%N2 , df ¼ 200{700 nm, 373 K) films. The
origin of height axis corresponds to the respective df ’s. In
both cases, RFH becomes broadened with increasing df . The
width of the distribution of the Al–N film is larger than
corresponding width of the Al film. The peaks of distribution
of the Al films are located at around the origin and/or
positive side of the height axis. On the other hand, the peaks
of distribution of the Al–N films are located on negative side
of the height axis. This difference corresponds to the concave
surface or the convex surface.11) Then the difference of the
distribution gives us a hint of controlling the gradient-profile.
The local peak heights are also extracted from AFM data,
and then the distributions of the relative frequency of local
peak height (RF-LPH) are obtained as shown in Figs. 8(a)
and (b). In the case of the Al film [Fig. 8(a)], the width of the
distribution of RF-LPH is narrower than that of the Al–N
films, and the peak positions of the distribution of RF-LPH
are clearly located on positive side of the height axis. These
facts are reflection of the Al film surface morphology such as
the protuberance range. The relatively broad distributions of
RF-LPH in Al–N case [Fig. 8(b)] are due to the surface
morphology with mixture of large and small protuberances. It
is important that each characteristic of the surface morphology should remain during film growth.
Figure 9 shows the df dependence of the highest and
lowest local peak heights of surface protuberance, and the
Controllability of Mesoscopic Surface Roughness of Sputtered Al and Al–N Films
Fig. 8 Film thickness (df ) dependence of distribution of relative frequency
of the peak height along the surface normal of the specimens (a): Al
(0%N2 ) and (b): Al–N (6%N2 ) deposited at Tsub ¼ 373 K.
3041
Fig. 9 Film thickness (df ) dependence of highest and lowest local peak
heights and lowest height of specimens: (a): Al (0%N2 , 373 K) and
(b): Al–N (6%N2 , 373 K).
Fig. 10 Substrate temperature (Tsub ) dependence of AFM plan view images and corresponding FT-LHD-profiles of 300-nm-thick pure Al
film deposited at Ptot ¼ 1:33 Pa. Tsub : (a) 303 K, (b) 333 K, (c) 353 K, (d) 373 K, and (e) 413 K.
lowest height of the films. In the case of Al–N films
[Fig. 9(b)], all of these quantities increase linearly with df . In
the case of Al films [Fig. 9(a)], however, the change of
corresponding quantities deviates from linear curve at the
beginning of film deposition. Among them, the lowest height
starts to uprise from about df ¼ 200 nm, i.e., a kind of
incubation thickness. This means that the substance of Al has
a tendency of aggregation, which is thought to be one of the
reasons for the distinct surface morphology. It is concluded
that the scale-up of the surface morphology is possible by
increasing the film thickness.
3.3
Substrate temperature dependence of surface morphology
Al (0%N2 , 300 nm, Tsub ¼ 303{413 K) films are deposited.
Tsub dependence of the AFM plan view images and the FTLHD-profiles are shown in Fig. 10. The size of the
protuberance seems to increase with increasing Tsub . How-
ever, there is a tendency that some protuberances gather
together to form cluster [Figs. 10(a)–(c)] or range of
protuberances [Figs. 10(d)–(e)]. The amplitude of FTLHD-profiles of the Al film increases with increasing Tsub .
However, the maximum position of the FT-LHD-profile of
the Al film with different Tsub is located at almost similar
position of 0.5 mm1 as indicated by arrow. Namely the
typical correlation length does not change with Tsub , although
the protuberances grow with Tsub . This is contrasted with the
case of the Al film growth at an equal Tsub as shown in Fig. 5,
where the peak position of the FT-LHD-profile became small
with increasing df .
Figure 11 shows the Tsub dependence of the distribution of
the RFH of Al (0%N2 , 300 nm, Tsub ¼ 303{413 K) films.
With increasing Tsub , the width of the distribution broadens
and the maximum position moves toward the positive side of
the height axis. Similar tendency has been seen more clearly
in the Tsub dependence of the RF-LPH as shown in Fig. 12.
3042
T. Ishiguro and K. Miyamura
Fig. 11 Substrate temperature (Tsub ) dependence of distribution of relative
frequency of the height along the surface normal of Al (300 nm).
Fig. 13 Substrate temperature (Tsub ) dependence of (a): highest local peak
height, lowest local peak height, and lowest height, (b): number density of
local peak of 300-nm-thick Al film.
Fig. 12 Substrate temperature (Tsub ) dependence of distribution of relative
frequency of peak height of Al (300 nm).
These facts correspond to the formation of the range of
protuberances.
As mentioned above, the degree of the surface roughness
increases with Tsub . However, there is a certain limit.
Figure 13(a) shows the Tsub dependence of the highest and
lowest local peak heights and the lowest height of the Al
(0%N2 , 300 nm) films, i.e., the bottom of the surface valley.
Figure 13(b) also shows the correspond change of the
number density of the local peak. All of these quantities
change monotonically up to around Tsub ¼ 400 K, then
saturate at the higher Tsub . It is concluded that the size of
the surface profile can be controlled by Tsub , however, there is
a certain upper limited Tsub above that the surface profile does
not change remarkably.
3.4
Other factors of surface morphology and controllability of them
Figure 14 shows the df dependence of the number density
of local peak (NLP ) of Al and Al–N films deposited at
different Tsub ’s of 303 and 373 K. Corresponding to Fig. 14,
the dependence of the normalized surface area (SNR ) is shown
in Fig. 15. The SNR ’s value of 1.0 means a perfect smooth
surface. All of NLP ’s decrease with increasing df and are
almost saturated in more than 300 nm. Either specimen, the
NLP decreases with increasing Tsub . The NLP for the Al
Fig. 14 Film thickness (df ) dependence of number density of local peak of
specimens of (a): Al (0%N2 , 303 K), (b): Al (0%N2 , 373 K), (c): Al–N
(6%N2 , 303 K) and (d): Al–N (6%N2 , 373 K).
Fig. 15 Film thickness (df ) dependence of normalized surface area of
specimens: (a): Al (0%N2 , 303 K), (b): Al (0%N2 , 373 K), (c): Al–N
(6%N2 , 303 K), (d): Al–N (6%N2 , 373 K).
(303 K) shows the largest value among them and its SNR is
almost one. These facts mean that the surface roughness of
the Al film is not enhanced at rather low Tsub of 303 K. On the
other hand, when Tsub increases in only 373 K, the NLP of the
Al film decreases drastically, the SNR increases and then
saturates at beginning stage of the film growth, i.e., at around
Controllability of Mesoscopic Surface Roughness of Sputtered Al and Al–N Films
3043
form and the different size, i.e., from smooth to the proper
rough surface with the order of the film thickness, can be
realized and/or controlled in the Al and N binary system.
4.
Conclusion
In this study, the qualitative difference in the form of
surface protuberances between the Al film and the Al–N films
has been found. There is a tendency that the size of the
protuberances increases with df and/or Tsub . The order of the
height of the surface roughness is same as the order of df .
There are, however, proper saturated sizes. It can be
concluded that the Al and N binary rf sputtered film system
is one of the promising systems to provide the mesoscopic
surface morphology, which is useful to realize the variety of
the gradient-profile.
Fig. 16 N2 gas partial pressure concentration (cN2 ) and substrate temperature (Tsub ) dependence of FWHM of relative frequency of height of 300nm-thick specimen.
100–20 nm, although the size of protuberance is enlarged as
show in Fig. 5. On the contrary, Tsub dependence of NLP of
the Al–N film is not so strong in comparison with Al film.
The SNR of Al–N films gradually increase with df . Especially
the growth at higher Tsub is profitable to enlarge the SNR .
Finally in order to clarify how we control the surface
roughness, cN2 and Tsub dependence of the FWHM of RFH
for the 300-nm-thick specimen is illustrated in Fig. 16. The
fully nitrided AlN films (cN2 > 12%N2 ) show relatively
smooth surface, which is not affected by Tsub strongly. The
FWHM of RFH of pure Al increases with Tsub monotonically.
In the intermediate region, i.e., incompletely-nitrided Al–N
film, the FWHM of RFH is enhanced at around 6%N2 . The
over all feature of the curved surface of the FWHM of RFH
seems to be a curved surface that a ridge at cN2 ¼ 6%N2
along the Tsub axis covers on a slope connected between the
Tsub axis at cN2 ¼ 14%N2 and the monotonically increasing
curve at cN2 ¼ 0%N2 . If we remember the qualitative
difference in form of surface roughness between Al and
Al–N films, i.e., concave or convex surface, it can be
mentioned that the surface morphology with the variety of
Acknowledgment
One of the authors (T.I.) was supported by a Grant-in-Aid
for Science Research from the Ministry of Education,
Culture, Sports, Science and Technology of Japan.
REFERENCES
1) G. A. Nyberg and R. A. Buhrman: Appl. Phys. Lett. 40 (1982) 129–131.
2) J. J. Cuomo, J. F. Ziegler and J. M. Woodall: Appl. Phys. Lett. 26
(1975) 557–559.
3) H. Taniguchi, H. Sannomiya, K. Kajiwara, K. Nomoto, Y. Yamamoto,
K. Hiyoshi, H. Kumada, M. Murakami and T. Tomita: Solar Energy
Materials and Solar Cells 49 (1997) 101–106.
4) J. Zhao, A. Wang, P. Campbell and M. A. Green: IEEE Trans. Electron
Devices 46 (1999) 1978–1983.
5) M. A.Green, J. Zhao, A. Wang and S. R. Wenham: IEEE Trans.
Electron Devices 46 (1999) 1940–1947.
6) T. Ishiguro, M. Nishimura and T. Yamazaki: Jpn. J. Appl. Phys. 41
(2002) 292–300.
7) M. Nishimura and T.Ishiguro: Jpn. J. Appl. Phys. 43 (2004) 757–761.
8) M. Nishimura and T. Ishiguro: Mater. Trans. 44 (2003) 2417–2419.
9) P. S. McLeod and L. D. Hartsough: J. Vac. Sci. Technol. 14 (1977)
263–265.
10) R. S. Nowicki: J. Vac. Sci. Technol. 17 (1980) 384–387.
11) M. Furukawa, Y. Yamamoto, H. Ikakura, N. Tanaka, M. Hashimoto,
A. Sano and S. Shingubara: Appl. Opt. 35 (1996) 701–707.
12) K. Kamoshida: J. Vac. Sci. Technol. B 18 (2000) 2565–2568.
13) C. Eisenmenger-Sittner: J. Appl. Phys. 89 (2001) 6085–6091.