Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
il~717~43
The present invention relates to a photodetector
whose maximum sensitivity is between 0.8 and 2 ~m, comprising
a P type substrate of Hg1 x Cd Te, and an N type zone for-
ming a P-N junction with the substrate.
In photodetectors, it may be desirable that the
response ti~e be as short as possible, particularly when
such detectors are used for detecting the light signals
transmitted by optical fibres, the digital rate in this
type of link being extremely high.
It is known that the attenuation in optical fibre
transmi~3ion is minimum for two values of the wave-length
of the transmitted signal, namely 1.3 jum and 1.6 ~um.
Applicants'French Patent Application No. 80 16788,
published under No. 2 488 o48 on February 5, 1982, already
relates to photodetectors which, by the use of a P type
substrate of Hg1 Cd Te, have a very short response time
and are sensitive to the above wave-lengths, the molar
fraction x being selected within a range of from 0.4 to
0.9, as a function of the wave-length to be received.
However, such photodectors do not necessarily fur-
nish, on the one hand, a considerable output current and,
on the other hand, a very high signal-*o-noise ratio. In
other words, as the light intensity passing in the optical
fibres is weak, the current gain of the dectectors, i.e. the
ratio, at a given wave-length, between the light intensity
received and the electric current created, may not be
sufficient.
A process which partially overcomes this drawback
is known, whereby the photodector is inversely biased, under
a high voltage, in order to place its working in the break-
down region.
.,,~.~
~1'7'79~3
In fact, this high bias is possible since the~unction
useæ a very weakly doped ~ubstrate and therefore ha~ a weak
capacitance, so that the time constant, or the response time,
o~ the detector i8 weak especially at high inverse potentials.
However, one drawback of this prooes~ is that the wor-
king point, due to thls high lnverse bias, is near the un-
controlled breakdown voltage, ln other words near the break-
down voltage of the ~unction.
This inverse bia~ potential, of the ordor of a few tens
or hundred~ of volts, is furnished by a more or les~ complex
or expensive BUpp~y neces~itating a ~ource of energy.
It is an object of the present invention to overcome
these drawbacks and to propose a photodetector having a high
gain, and consequently being able to operate under a weak
inverse bias potential.
To thi~ end, the present invention relates to a photo-
detector having its maximum sensitivity bet~een 0.8 and 2 ym,
comprising a P type sub~trate of Hg1 x CdX ~e, x being selec-
t~d within a rangc extending from 0~4 to 0.9, and a doped N-
type zone formed on the ~ubstrate, the resulting Junctionbeing of the P-N type, wheroin x is selected a~ a funotion
of the desired current gain.
It is known that ~uch photodetectors present the pheno-
menon of the ~pectral respon~e, and ~ore particularly the cut-
oif wavo-l~ngth, being dopendant on s.
~ hus, as the value o~ x i8 already imposed a priori by
the choice of the desired wave-length, it would appear lnof-
~cctual to soek to discover whether anothor phenomenon, al80
depending on the value o~ x, could not be the origin of a
greater electric output curront.
-- 2
1~7'~943
Applicants nevertheless sought to di~cover whether the
current gain, in other words the figure of merit of the junc-
tion, likewise advantageously depended on ~. Applicants not
only demon~trated thi~ phenomenon, namely that the figure of
merit, of which the value involves that of the output current,
effectively depends on the molar fraction x, but they also
obæerved that the value of x, propibious to a greater current,
was very close to the value optimali~ing the detection of a
oertain wave-length.
More particularly, Applicants have shown that the figure
o~ merit, and therefore the output current, reached its maxi-
mum for a value of x close to but different from ths one which
corresponds~to the optimal dete¢tion, with a short response
time, of a radiation o* wave-length of t.3 pm.
The invention will be more readily understood on reading
the iollowing description with reference to the accompanying
drawings, in which s
- Fig~. 1 to 7 illu~trate the dif~erent pha~es of the
procsss for manufaoturing the photodetector according to the
invention;
- Fig. 8 shows the ~pectral re~ponse ourves of the detec-
tor of the invention, for several value~ of the inverse bias;
- Fig, 9 showsJ on the one hand, the curve of the cut-off
wave-le~gth and, on the other hand, the vPlue of the figure
o~ merit, function~ of the molar fraction of the substrate of
the detector according to the invention, and
- Fig. 10 show~ the curve of the current gain of the de-
tector according to the invention as a function of the inverse
bias voltage, for the chosen value of ~.
~0 Referring now to the drawings, the detector shown therein
_ ~ _
11~7'~9~3
is well adapted for a wave-length close to 1.3 ~m. It compri-
ses a substrate constituted by an alloy of mercury, cadmium
and tellurium, Hg1 x CdX Te in which the molar fraction x is
between 0.4 and 0.9 as indicated in French Application No
80 16788, published under No. 2 488 040.
To mam1facture the detector of the invention, one
starts with P type substrate 1 (cf. Fig. 1), formed by a
crystal of Hg1 CdX Te of very high purity, therefore having
a very low concentration of P carriers, of the order of
1015/cm3. Such a crystal may be obtained, for example, by the
method described in an article by R. Triboulet, entitled
"Cd Te and Cd Te:Hg alloys crystal growth using stoichio-
metric and off-stoichiometric zone passing techniques",
which appeared in the Revue de Physique Appliquée of
February 1977.
On the substrate 1 is deposited a layer 2 ~f Cd
Te, for example by cathode sputerring. A masking layer 3,
preferably o ZnS, SiO or SiO2, is then deposited on layer 2
(Fig. 2). At least one opening 10 is then made in the layers 2
and 3, so as to lay bare part of the surface la of the
substrate 1 (Fig. 3). A planar P-N junction i9 then made by
a firYt diffu~ion in the substrate 1 of an element such as
cadmium or mercury or of an impurity such as aluminium,
indium or boron (arrow 5, Fig. 4).
A zone 6 of N type is thus obtained, with a doping
of about 10 atomsJcm3, wnich forms a junction 7 with the
substrate P.
However, with a view to reducing the curvature of
this junction 7 at its ends, and therefore to reducing the
valus of the electric field, and consequently the risk of
breakdown in these regions, one of the same elements
may previously be
11~;J';J943
di~fused, unde~ th~ ~ame oonditions, in a¢cordance with the
so-cal:Led keeper ring technique (Fig. 5), in order finally
to obt~n 8 zone 8 o~ ~ type with a larger curv~ture.
Another procese which may be used to form the N 20ne
is the ionic implantation of atoms having, after appropriate
~nnealing, an electrical N type activity~ such as atoms of
indium In, aluminium Al or boron B.
~ layer 4 of the ~ame nature as layer 3 i~ then deposited
over the whole suriace, then an openlng 9 i9 made in layer 4
(Fig. 6) by chemical etching and this opening i8 filled with
a conducting metal such a~ Al or In in order to make a con-
tact 10 ~Fig. 7).
The above-de~cribed detector ~ u~ed ~or detecting radia-
tion in the near infrared.
The choice of the value of the molar fraction ~ depends
on the following con~iderations.
To define the structure o~ the electronic bands, the
conduction band, ~orbidden band and valence band model ma~ be
used.
The arrival o~ a photon of the radiation received create~,
ii its energy i9 ~ufficient, a hole-electron palr which parti-
cipateo ln a current proportional to the lnoident ~lu~.
It wlll be readlly understood that, when the wave-length
of the radiation exceeds a oertain threshold, the energy o~
the photons, proportional to ~ no longer su~ficient to
create hole-electron pair~0
This i~ why, for a given value o~ ~, i.e. for a glven
value ~E o~ the width o~ the forbidden band, there is a cut
off wave-length ~c~ ~uch that, when the radiation has a
wave-length ;~ > ~ c~ no current is oreated in the ~unotion,
llt7 ~9,~3
this explaining the appearance of the curve~ of Fi~. 8, which
represent the response ~ of the detector, i.e. the ratio bet-
ween the current created and the light inten~ity received,
as a fl~ction of the wave-length ~ of the radiation received,
~or three values of the blas potential V of the ~nction, and
a predetermined value of x. It i8 noted that, in the e~ample
of Fig. 8, the current gain is already Gonsiderable for a
weak inverse bias.
The curve 100 of Fig. 9 represents the cut-o~ wave-
tO length Ao as a function o~ ~, in lum.
~ hus, in order optimally to receive a radiation o~ wave-
len~th 1.3 ~m, the cut-oii wave-lsngth must be greater than
1.~ pm, but preferably falrly near ae indicated by the spec-
tral response curves (Fig. 8)~
Thus, to detect a radiation o~ a wave-length o~ 1.3 pm,
which corresponds to a mlnimum attenuation in the optical fi-
bree, the optimum value o~ ~ would be near 0.7, considering
the curve 100 of Fig. 9. However, the above-mentioned draw-
back, namely the weak output current of the detector, would
no~ nece~sarily be overcome.
. Another band, a so-called epin-orblt band, exi~ts ln
the etructure o~ the electronic bands o~ the ~emlconduotor
envi~aged.
It has been found that the energy di~erenee ~B' between
the ~pin-orbit band and the valence band was capable, depen-
ding o~ its value, to play a role concerning the generation~-
recombi~ations of hole~electrons created by the photonsO More
partioularl~, for ~ wide range o~ values of ~ ' i8 15 to
10 times greater than ~R and, in thi~ ca~e, the spin-orbit
band ha~ only a slight influence on the generations-recombi-
- 6 -
il~7'79~3
nations o~ the hole-electrone. However, for certain values
of ~, between about 0.6 and 0.8; the value o~ ~' is ~airly
close to that of ~, both having for order o~ magnitude
0.9 eV. It is readlly appreciated that the properties o~
creation and of recombination o~ the hole-electron pairs
will then be substa~tially di~ferent from those oorresponding
to the model which does not take into account the epln-orbit
band.
The phenomenon is as ~ollow~. When the spin-orbit band
has little or non influenoe on the photoelectrio phenomenon,
a current of electron~ and a current o~ holes, which are
substantially equal, i.e. ionisation coe~ficients and ~ of
~ubstantially equal values, are created by the radiation
received.
However, when ~ iæ in a range extending ~rom about 0.6
to 0.8, the influencer1o~ the spin-orbit band i8 real and very
dif~erent ionisation coefficients a and ~ are obtained. In
~act, the ratio l~ proportional to the term
P = (1 _ ~ ) 2 , called ~igure oi merit, illustrated by
~B
curve 101 of Fig. 9, which repre~ents the value of the figure
oi merit F as a ~u~ction o~ the oompo~ition o~ the sub~trate,
i.e. of ~. It i~ observed that thi~ ~alue pa~ses throu~h a
ma~imum ~or a value o~ ~ subatantlally equal to 0.67 and that
it ie greater than 10 when ~ is included between 0.55 and 0.85.
The value~ o~ ~ and ~ are then ~ery different, the ourrent
gain as a ~unction o~ the bias potential i8 better controlled
and the signal-to-noise ratio ia higher.
It i~ therefore necessary to choo~e, on the one hand, a
value of ~ such that the worth ~actor and consequently the
output current are high, and, on the other hand, a value of
11'7'7943
optimali~Qing the reception of a radiation o~ wave-length 1.3 pm,
optimal value ~or optical fibre transmis~lon.
The two most satisfactory value~o~ ~, one for the cur-
rent, the other for the wave-length, are different.
The present invention therefore relates to a detector
capable both o~ detecting a radiation of wave-length o~ 1.3 ~m
and o~ furnishing a high output current, i.e. havlng a high
ourrent gain, without having to apply a considerable inverse
bia6 potential. Thi~ charaoteri~tic i~ also to be found on
the curve o~ Fig. 10, representing the gain G defined herein~
above, function of the inverse bias potential V, for the
chosen value o~ ~. It is ob~erved that thi~ gain begins to
increase for a voltage of about -4 or -5 volts, to reach a
value of 30, which is more than ~ati~actory, for a value o~
V of the order of -10 volt~, con~idered as weak.
For x = 0.67, which therefore corresponds to a maximum
gain, the cut-o~f wave-length ~ c i~ equal to t.43 ~m, whlch
length is per~ectly suitable for receiving a radiation of
wave-length of 1.3 ~.
The characteri~tice o~ the photo-detectors are advanta-
geou~ly a~ ~ollow~s
- Active area s 1 x 10 4 cm2
- Inverse bias potential V : 10 volts
- Current gain ~ 2 30
- Saturation current : < 1nA
- Dark current (V = -10V) : ~ 10~A
- Total capacitance (V = -10V) s ~ lpF
- Junction capacitance (V = -10V~ s < 0.2 pP
- Resi~tance-area product (V=Ov) : 6 ~ 104 Q ~cm2
- Response in current ( ~ = 1.3 ~m) : > 0.5 A/w
- Operational temperature : 300E.
