close

Вход

Забыли?

вход по аккаунту

?

el%3A19800358

код для вставкиСкачать
OH absorption loss at 1-39 fim was 10 dB/km, from which the
residual OH content was estimated to be about 015 ppm.
To investigate the origins of OH ion contamination, OH ion
distribution profiles in the rod preforms were measured
optically.7 Fig. 3 shows the measured OH ion distribution
profile in the preform corresponding to the fibre of Fig. 2. It is
noticeable in Fig. 3 that a fair number of OH ions exist at the
boundary between the deposited cladding region and the low
OH content silica tube region. This contamination is undoubtedly caused by OH ion penetration during the elongation and jacketing processes with an oxyhydrogen burner. The
OH absorption loss calculated as an overlap integral of the
observed OH profile and the power distribution of the fundamental H E n mode is about 7 dB/km for the fibre of Fig. 2.
Musashino Electrical Communication Laboratory for supplying a laser diode operating at 1-52 fim.
S. TOMARU
M. KAWACHI
T. EDAHIRO
Ibaraki Electrical Communication Laboratory
Nippon Telegraph & Telephone Public Corporation
Tokai, Ibaraki 319-11, Japan
References
1
3
50
5
, ' \
/
\
1
i
»
/
\
\
\
\
y
6
K)
fabrication
SUDO, s., KAWACHI, M., EDAHIRO, T., and CHIDA, K.: '212 km graded-
index v.a.d. fibre with low-loss and wide bandwidth', Electron.
Lett., 1980,16, pp. 152-154
7
08
IZAWA, T., SUDO, s., and HANAWA, F.: 'Continuous
process for high-silica fiber preforms', Trans. IECE Japan, 1979,
E62, pp. 779-785
cutoff
05
OKADA, M., KAWACHI, M., and KAWANA, A.: 'Improved chemical
vapour deposition method for long-length optical fibre', ibid., 1978,
14, pp. 89-90
it
5
0-1
MIYA, T., TERUNUMA, v., HOSAKA, T., and MIYASHITA, T.: 'Ultimate
low-loss single-mode fibre at 1-55 /im\ ibid., 1979,15, pp. 106-108
4
£
KAWACHI, M., KAWANA, A., and MIYASHITA, T.: 'Low-loss single-
mode fibre at the material-dispersion-free wavelength of 1-27 fitn',
Electron. Lett., 1977, 13, pp. 442-443
100
E
=f
FRENCH, w. G., and TASKER, G. w.: 'Fabrication of graded and
single-mode fibers with silica core'. Topical meeting on optical fibre
transmission, Williamsburg, T.D. Tu Z-2-1, 1975
2
CD
TJ
in"
1/1
13th May 1980
12
\U
12
U
wavelength, pm
KAWACHI, M., HORIGUCHI, M., KAWANA, A., a n d MIYASHITA, T.:
'OH-ion distribution in preforms of high-silica optical fiber', Japan.
J. Appl. Phys., 1978, 17, pp. 1975-1981
16
Gzl£)
Fig. 2 Loss spectrum of 21 km single-mode fibre
0013-5194/80/130511 -02$l .50/0
deposited cladding
.low-OH-content
silica tube
/,
300
silica tube
E
°:200
LASER ANNEALING OF LOW DOSE
Se-IMPLANTED GaAs STUDIED BY
D.L.T.S.
100
Indexing terms: Charge carriers, Doping
-10
-5
0
distance,mm
10
|178/3|
Fig. 3 OH ion distribution profile in preform corresponding tofibreof
Fig. 2
This value is almost comparable to the observed OH absorption
loss of 10 dB/km in Fig. 2. The slight deviation might be due
to the residual OH content uniformly distributed in both core
and the cladding regions.
The length of single-mode fibres made by the v.a.d. methoc
is, at present, limited by the size of low OH content silica tube
used as the first jacketing tube. The solution to this problem
will be sought in increasing the deposited cladding thickness at
the stage of porous preform fabrication. The deposition
technique for obtaining a deposited cladding/core diameter
ratio larger than 5 is soon taken in hand. This will make it
possible to produce longer and lower-loss single-mode fibres
by v.a.d. without use of any special low OH content silica
jacketing tube.
In conclusion, a 21 km single-mode fibre with loss values of
0-6 dB/km at 1-2 and 1-55 /mi, and 0-7 dB/km at 1-3 /mi, was
made by v.a.d. It has been shown that the v.a.d. method has
much potential in the production of long low-loss single-mode
fibres.
Acknowledgments: The authors would like to express their
thanks to N. Niizeki, H. Takata and N. Inagaki for their helpful suggestions and encouragements. They are also indebted to
S. Takahashi, M. Nakahara, K. Okamoto and S. Sudo for their
helpful discussions and technical help, and G. Iwane at the
512
Deep level transient spectroscopy (d.l.t.s.) has been applied to
the study of deep levels in GaAs following post-implantation
annealing using a Q-switched ruby laser. High concentrations
(> 10 15 cm" 3 ) of deep trapping levels are observed in the
laser melt region using a forward-bias voltage pulse.
Introduction: Recent publications12 have reported on the notable absence of deep levels in GaAs after laser annealing, as
measured using transient capacitance spectroscopy. In Reference 1 the measurements werfr limited to depths greater than
5000 A, this lower limit being determined by the zero-bias
depletion capacitance associated with the Schottky-barrier
diodes. Recent work3 however suggests that the depth of melting during laser irradiation was less than the zero-bias
depletion width. The melt depth is estimated to be of the order
of several thousand angstroms for a Q-switched ruby laser
pulse of energy density 0-5 Jem" 2 .
In this experiment deep levels are observed within the melt
region by using a forward-bias voltage pulse, the effect of
which is to reduce the depletion width to approximately 1000
A. This pulse is used to investigate the effect of laser annealing
on the deep levels in the melt region of both unimplanted and
selenium implanted n-type GaAs.
Experiment: Low doses of Se+ ions (1012 cm" 2 ) were implanted at 300 keV, at room temperature, into n-type vapourphase epitaxial (v.p.e.) GaAs. The samples were subsequently
laser annealed, without an encapsulant, using a single 25 ns
pulse from a Q-switched ruby laser of energy density 0-5
Jcm~2. For comparison, unimplanted material was also annealed by the same method. In each case 1 mm diameter aluminium dots were evaporated onto the active surfaces to form
ELECTRONICS LETTERS
19th June 1980
Vol. 16 No. 13
Schottky-barrier contacts, and tin dots were alloyed to the n+
substrates for ohmic contacts.
D.L.T.S. is based on the measurement of capacitance transients associated with an abrupt change in the voltage bias
applied to a Schottky diode. The transients are attributed to
deep levels present in the reverse-bias depletion region. Individual levels are characterised by varying the sample temperature, typically in the range - 150°C to + 150°C, to bring each
level in turn into the measurement window of the system. The
signal processing is performed here using a correlator based on
the work of Miller et a/.4
•0-83 eV
-100
-50
0
*50
temperature, *C
+100
greater than the zero-bias depletion width, the concentration
of deep levels as a function of depth is measured by varying the
reverse bias while keeping the pulse height (0-5 V) constant.
Low temperature (77 K) point-by-point C/V measurements
are used to determine the free carrier concentration profiles.
Results: The effect of using a zero and forward-bias pulse is
demonstrated in Fig. 1 for unimplanted GaAs, before and after
laser annealing. As previously reported, the A-centre (083 eV)
has decreased from a density of 1014 cm" 3 to less than 1013
cm" 3 as a result of the annealing process. In addition, high
concentrations of traps are now observed within the melt
region using the forward-bias pulse. The concentration of traps
is calculated from the height of the d.l.t.s. peak to be greater
than the background carrier concentration of 2 x 1015 cm" 3 .
In contrast, outside the region of melting, at depths greater
than 1 nm, the total trap concentration is less than 3 x 1013
cm"3, as shown in Fig. 2.
D.L.T.S. spectra and concentration profiles for the Se+ implanted samples are shown in Figs. 3 and 4, respectively. Again,
high concentrations of traps over an energy band of approximately 0-2 eV-0-7 eV are observed in forward bias. The
decrease in the free carrier concentration towards the surface
in Fig. 4 implies that the Se+ has failed to become electrically
active. In addition, the zero-bias depletion width is greater
than would be expected from a highly doped w-type layer.
Fig. 1 D.L.T.S. spectra for unimplanted GaAs before and after laser
annealing. Baselines (dotted) vertically displaced for clarity. Reference
time constant = 30 ms
(i) As-grown v.p.e. GaAs (zero and forward-bias pulse)
(ii) a Forward-bias pulse
b Zero-bias pulse
c Zero-bias pulse ( x 3 )
In these d.l.t.s. measurements the samples are reverse-biased
to 4 V. The filling and emptying of traps is performed using
voltage pulses where the diode is either zero-biased or forwardbiased for the 2 ms pulse duration. In the latter case the capacitance meter (Boonton type 72B) is gated out for the duration of
the pulse to avoid the problem of measurement associated with
the high shunt conductance of the diode. The depletion depth
under forward bias is estimated by extrapolating the reversebias C/V curve in the forward voltage direction. The uncertainty in the estimate of depth for the forward voltage
measurement precludes any detailed study of the trap profiles
in the forward-bias depletion region. However, for depths
-100
-50
0
+50
temperature, "C
*100
Fig. 3 D.L.TS. spectra for Se* implanted GaAs after laser annealing.
Reference time constant = 30 ms
a Forward-bias pulse
b Zero-bias pulse
15
10
16
10
approximate
melt depth
approximate
• melt depth
ion
*—p rang
range
o
£ 15
K)
300 keV
13
K)
05
10
15
20
depth, pm
2-5
0-5
Fig. 2 Concentration profiles for unimplanted GaAs before and after
laser annealing
a
b
c
d
Free carrier profile before laser annealing
Free carrier profile after laser annealing
0-83 eV trap profile before laser annealing
0-79 eV trap profile after laser annealing
ELECTRONICS LETTERS
19th June 1980
1-0
V5
20
depth, pm
25
Fig. 4 Concentration profiles for Se+ implanted GaAs after laser annealing. (Unimplanted background concentration approximately 5 x 1015
cm-*)
a Free carrier profile
b Total trap profile
Vol. 16
No. 13
513
Conclusions: The high concentration of traps (> 1015 cm 3) in
the region of melting probably accounts at least in part for the
inability of the laser anneil treatment to activate electrically
the low dose Se+ implants. This problem has been discussed in
other publications,5 where only doses greater than 1014 cm" 2
give rise to significant electrical activity (of the order 10-40%)
and where the mobilities are always lower than expected. The
introduction of high concentrations of traps in the region of
melting may account for these low mobilities.
Finally, the comparison of traps in both the implanted and
the unimplanted samples shows that it is the laser annealing
process itself that is responsible for the introduction of the
majority of the traps.
magnitude response passing through a prescribed cutoff
frequency.
Acknowledgments: The authors would like to thank Philips
Research Laboratories for the use of their ultrasonic bonding
facility. We acknowledge the financial support of the UK
Science Research Council. Finally, we wish to thank M. H.
Badawi and Prof. K. G. Stephens for their useful comments.
where IN is the filter order and K is the degree offlatnessin
the stopband. Substitution of q = jCl gives
N. G. EMERSON
B. J. SEALY
Department of Electronic & Electrical Engineering
University of Surrey, Guildford, Surrey, England
To obtain a transitional filter between two adjacent values of
K, say Ki and K2 with K2 = Kv - 1, we add an extra term in
the numerator of eqn. 2 as
Design procedure
Transitional filters between two adjacent values of K: The
elegant design of Miller2 is comparatively simple and readily
applicable to our proposal. From exprs. 4 and 6 of Reference 2
the maximally f.i.r. filter function in the g-plane is rewritten
here again as
7th May 1980
(2)
X(7)°- +c "-
References
1
EMERSON, N. c , and SEALY, B. J.: 'Effects oflaser irradiation of GaAs
observed by d.l.t.s.\ Electron. Lett., 1979, 15, pp. 553-554
2
YUBA, Y., GAMO, K., MURAKAMI, K., and NAMBA, s.: 'Laser-irradiation
If CN-K2 is now allowed to vary between 0 < CN-Kl < {N-K2)>
a class of transitional filters whose magnitude responses lie
between those with the flatness parameters Kx and K2 is obtained. Fig. 1 shows the case of transitional responses with
N = 10, Kt = 6 and K2 = 5.
effects on unencapsulated GaAs studied by capacitance spectroscopy', Appl. Phys. Lett., 1979, 35
3
BADAWI, M. H., SEALY, B. J., and CLEGG, J. B.: 'Redistribution of
chromium in semi-insulating GaAs:Cr during laser annealing' (to
be published)
4
(3)
MILLER, G. L., RAMIREZ, J. v., and ROBINSON, D. A. H.: 'A correlation
method for semiconductor transient signal measurements', J. Appl.
Phys., 1975, 46
5
BADAWI, M. H., SEALY, B. J., STEPHENS, K. G., a n d AKINTUNDE, J. A.:
'Improvement in electrical properties of laser annealed ionimplanted GaAs', Japan. J. Appl. Phys., 1980 (to be published)
0013-5194/80/130512-03S1.50/0
0-4-
0-2 -
MAXIMALLY FLAT F.I.R. FILTER WITH
PRESCRIBED CUTOFF FREQUENCY
0 2 TT
Indexing terms: Filters, Digital control
Maximally flat f.i.r. digital filter design provides the advantage of giving a closed-form solution, but there still remains a
problem of designing such a filter whose magnitude response
passes through a prescribed cutoff frequency point. It is
described here how to generate a class of transitional maximally flat f.i.r. digital filters to overcome such a difficulty.
0-6 Tt
0 8 TT
Fig. 1 Transitional maximally flat fix. digital filters with N = 10,
Kv=6andK2
=5
Introduction: Maximally flat f.i.r. digital filter design1"4 offers
an advantageous feature over other optimal designs,5 e.g. the
minimax design, in providing the weighting coefficients in
closed form. There still, however, remains a problem of choosing the coefficients of this filter type with its magnitude response passing through an arbitrary specified cutoff frequency.
So far, for a given filter of order 2N, the 'flatness' parameter K
must be chosen in accordance with expr. 9 of Reference 1 to
give the filter cutoff point near to the desired value. But as is
seen from Fig. 2 of Reference 1, the magnitude responses never
exactly pass through the desired cutoff frequency. The purpose
of this paper is twofold. First, it is shown that for any two
adjacent values of K, a class of transitional maximally flat f.i.r.
filters can be generated. It is then shown further that for any
given filter order 2N, one can design such an f.i.r.filterwith its
514
0 4 TT
Cs values
252
210
-ooo-oo-
—° —
— o—
168
126
84
42
0
F.I.R. filter with arbitrary specified cutoff point: For a specified
cutoff point in the z-plane, the corresponding cutoff point in
the g-plane is determined via
i
OD
(4)
= tan 2
where QOA and ClOD are the analogue and digital cutoff
frequencies, respectively. The sampling frequency is normalised to unity. Next, the two adjacent values of K (i.e. Ki
and K2) between which the magnitude response with the
specified cutoff point falls into are determined. This can be
ELECTRONICS LETTERS
19th June 1980
Vol. 16 No. 13
Документ
Категория
Без категории
Просмотров
0
Размер файла
375 Кб
Теги
3a19800358
1/--страниц
Пожаловаться на содержимое документа