Note: Descriptions are shown in the official language in which they were submitted.
background of the Invention
This invention relates to semiconductor devices
for controlling light such as optical modulators and
nonlinear optical devices.
Conventional semiconductor optical modulators have
traditionally made use of the Franæ~Keldysh effect to
modulate an incident light beam. According to the Franz-
Keldysh effect, the band structure of the semiconductor
material is shifted by application of an electric field
however, the Franz-Keldysh effect is characterized by small
shifts in optical properties such as absorption and index
of refraction thereby limiting the modulation to be
correspondingly small In order to achieve the deep
modulation required for a practical device it is necessary
to use either a high electric field or a device naming
long optical path length or both. For example, a
modulator described by Honda in V. S. Patent No. 3,791,717
issued February 12, 1g74 uses a high electric field l105
to 106 volts per centimeter) in a semiconductor having a
crystal of long optical path length (10 microns.
An optical modulator which uses a heterojunction
semiconductor device in which optical absorption and no-
emission are controlled by an electric field is described
by Clang et at in I. S. Patent 4,208,667. Clang et at uses
a heterojunction super lattice having two different
material arrangements alternatively to form a semiconductor
heterojunction in which the bottom of the conduction band
of a first layer is lower than the conduction band in the
second layer, and also the top of the valence band is lower
than the corresponding band in the second layer. This
device and super lattice allows electrons and holes to be
excited by photo absorption wherein the electrons collect in
one layer and holes in the adjacent layer. Charge carriers
recombine by making a transition into the adjacent layer
-- 2
with the subsequent emission of light. The rate of recomb
bination is controlled by an electric field applied to the
super lattice. It is a property of the Clang et. at. device
that the emitted light is of a different frequency from the
incident light and that the light is emitted in all direct
lions. It should also be noted that the incident and
emitted light beams are not collinear through the devices.
As a result, the Clang et at device is impractical as an
optical modulator.
Nonlinear optical devices have been made using
heterojunction semiconductor materials. These devices are
characterized by an operating point determined by optical
cavity gain or finesse. These nonlinear optical devices
exhibit particular aspects such as bistability, amplifica-
lion, photonic modulation or the like. A problem with this
type of nonlinear optical device is that the operating
point is selected by the choice of materials and other
design parameters during fabrication of the device. There
fore, the operating point cannot be conveniently controlled
at the time the device is in use.
Summary of toe Invention
In accordance with an aspect of the invention
there is provided a semiconductor apparatus comprising: a
multiple layer heterostructure having first and second
material layers having first and second band gaps, respect
lively, and a semiconductor layer having a third band gap
and being positioned between said material layers, the
bottom of the conduction band of said semiconductor layer
being below the bottom of the conduction bands of said
material layers, and the top of the valence band of said
semiconductor layer being above the tops of the valence
bands of said material layers, the thickness of said semi-
conductor layer being sufficient for carrier confinement
effects within said semiconductor layer to influence the
optical properties of said multiple layer heterostructure;
and means for applying an electric field to said multiple
' 'I, 6`
So
- pa -
layer heterostructure in order to vary an optical absorb-
lion coefficient and an index of refraction of said
multiple layer heterostructure in response to said electric
field wherein said means for applying an electric field
to said multiple layer heterostructure comprises: a p doped
semiconductor layer and; a n doped semiconductor layer and
said semiconductor layer having a third band gap is located
between said p doped semiconductor layer and said n doped
semiconductor layer.
In accordance with the present invention, the
foregoing problems are solved by employing a semiconductor
apparatus comprising a multiple layer heterostructure and
means for applying an electric field to the multiple layer
heterostructure in order to vary optical absorption Coffey-
clients and an index refraction of the multiple layer
heterostructure in response to the electric field. The
multiple layer heterostructure includes first and second
material layers having first and second band gaps, respect
lively, and a semiconductor layer having a third band gap
and being positioned between the wider band gap layers. The
bottom of the conduction band of the first semiconductor
layer is below the bottom of the conduction bands of the
first and second material layers. The top of the valence
band of the first semiconductor layer is above the top of
the valence band of the first and second
Jo
I
material layers The thickness of the first semiconductor
layer is 1000 Angstroms or less. The semiconductor
apparatus is adapted for use as an optical absorption
modulator, or an optical phase modulator or an optical
polarization modulator. By interposing the apparatus
between mirrors, the resulting apparatus is an electrically
tuned Fairy Pert cavity. The apparatus is also adapted
for other nonlinear or bistable applications in which
modification of the optical properties of a semiconductor
by application of an electric field is useful. In one
embodiment the electric field is conveniently applied by
fabricating the multiple layer heterostructure as the
intrinsic region of a PIN semiconductor structure.
iffy Description of the Drawing
FIG. 1 is a cross-sectional drawing of a MOW with
an electric potential applied perpendicular to the layer
planes;
FIG. 2 is a eross-sectional drawing of a MOW with
an electric field applied substantially parallel to the
layer planes;
FAKE. 3 is representative optical absorption for
different applied electric fields for a MOW;
FIG. 4 is a graph showing the shift in excitor
peak with applied voltage;
FIG. 5 is a graph showing variation in optical
absorption with applied voltage for a MOW;
FIG. 6 is a graph showing operation of an
absorption modulator;
FOG. 7 is a cross section of a MOW structure;
FIG. 3 is an energy band diagram for a
semiconductor,
FIG 9 is an energy band diagram for a
heterojunction;
FIG. 10 is an energy band diagram for a MOW
structure;
FIG. 11 is a graph showing saturation of optical
absorption versus intensity of illumination for a MOW
I
I,,
structure;
FIG. 12 is a graph showing the relationship
between optical absorption and index of refraction as
predicted by the Kramers-Rronig relationship as the optical
absorption saturates under a high intensity;
FIG 13 is a cross-sectional drawing showing an
alternate arrangement of electrodes for applying an
electric field to a MOW structure;
FIG. 14 is a top view showing an alternate
arrangement of electrodes for applying an electric field to
a MOW structure;
FUGUE is a perspective drawing showing an
alternate arrangement of electrodes for applying an
electric field to a MOW structure;
FIG. 16 is a cross-sectional drawing showing an
alternate method of attaching contacts to a MOW structure;
FIG. 17 is a cross-sectional Figure showing a PIN
semiconductor structure for applying an electric field to a
MOW;
FIG. 18 is an alternate arrangement of a PIN
structure for applying an electric field to a MOW
structure;
FIG. 19 is an alternate PIN structure for applying
an electric field to a MOW;
FIG. 20 is a graph giving optical transmission
versus photon energy for various voltages applied to a MOW
using a PIN structure;
FIG. 21 is a cross section showing a controlled
Fabry-Perot cavity;
FIG. 22 is a graph giving the optical transmission
of a controlled Fabry-Perot cavity versus applied voltage;
FIG. 23 is a cross sectional view showing a
controlled Fabry-Perot cavity;
FIG. 24 is a cross sectional view showing a
polarization modulator and
FIG. 25 shows an array of electric field
controlled MOW devices.
I
-- 5 --
Description of the Preferred Embodiment
In FIG. 1 there is shown an exemplary light
modulator 100. A source snot shown) produces unmodulated
light 110. The source may be, for example, a laser. Also,
for example, the source may be a light emitting diode.
Other light sources such as light transmitted by an optical
fiber may conveniently be employed to produce unmodulated
input light 110. Unmodulated input light 110 passes
through multiple quantum well (MOW) structure 120, after
which it is output modulated light 130. A MOW structure
120 includes both a structure with one quantum well or
structures with many quantum wells. First electrical
contact 140 and second electrical contact 150 are connected
by electrical conductors 152, 154 to source 156 of electric
potential. An electric field is applied to MOW structure
120 by the potential provided by source 156 through
electrical contacts 140 and 150. Source 156 may, for
example, be a direct current source, such a a battery.
Or, for example, source 156 may be a klystron oscillator
operating at a multiple gigahertz frequency. For example,
source 156 may provide electric potential in the 100
gigahertz frequency range. Application of an electric
field to MOW structure 120 cause the optical transmission
of MOW structure 120 to vary with the applied potential
The optical transmission of MOW 120 may either increase or
decrease with applied electric field, depending upon the
light frequency at which the transmission is considered, as
will be more fully explained hereinafter. In the
alternative embodiment shown in FIG. 1, electrical contacts
140, 150 are attached to transparent or partially
transparent electrodes 160, 162 which, or example, can be
semiconductor layers doped to be conductive, or, for
example, can be thin metallic layers. Contacts 140, 150
are attached so that an electrical potential applied
between them is conducted by electrodes 160~ 162 and
therefore produces an electric field which is substantially
perpendicular to layer planes 164 of MOW 120. For
I
-- 6 --
example, electrodes made of Agues are satisfactory for a
MOW 120 made of Agues and Gays layers. For example,
electrodes 160, 162 may be made of Al/ Or, A, A or alloys
of these metals or combinations of layers of these metals,
and with thicknesses between 0.05 micron and 0.1 micron.
Also for example, indium tin oxide electrodes may be
used.
In an embodiment which uses partially transparent
electrodes, it is convenient to choose their thickness to
achieve anti reflection properties. Such a choice is given
by the equation
T = on (up - I
where T is the electrode thickness, n is the index of
refraction of the electrodes, is the wavelength of the
light being modulated, and p is an integer, p = 1, 2,
3,....etc. For example for electrodes made of indium tin
oxide which has an index of refraction n of approximately
1.8, and the use of GaAs/A1GaAs MOW structure operating at
= 0.85 micron, the thickness is given as:
P T
micron
1 0.12
2 0.36
The use of the thicker, 0.36 micron layer is more
convenient as the thicker layer provides lower contact
resistance.
In FIG. 2 the MOW modulator 170 used in an
exemplary embodiment was fabricated by molecular beam
epitaxy on a Gays substrate (not shown and which was
removed during fabrication. A 1.45 em thick
Alto guy assay cap layer 172 is attached to 65
periods of alternate AYE Gays and AYE
Aye Gwen assay layers to form the 1.26 em thick
MOW 174, and capped by a further 1.45 em thick
Aye guy assay layer 176. The cap layers 172,
176 are transparent in the MOW band gap region. All the
I
-- 7 --
materials were unhoped with residual carrier concentration
less than ~1015cm 3. A 3 x 3mm2 sample was
leaved along the [110] and ~110] directions and glued by
the epitaxial cap layer 172~ with a transparent epoxy (not
shown) to a sapphire support 180. The entire Gays
substrate (not shown) was selectively etched off
Electrical contacts 182, 184 owe 100~ Or 186
followed by AYE A 188 were evaporated on the sample to
give an electrode spacing 190 d = 300~m. A ow tunable
oxazine 750 dye laser (not shown) was used as a light.
source to measure the absorption spectra of MOW structure
170. The light beam 192 was focused to a Siam diameter
spot 194 at the center of the interlectrode gap, with a
polarization parallel to the electric field. The
transmission, as corrected for surface reflection, was
measured as a function of the laser frequency for voltages
applied between electrical contacts 182 and 184, with the
voltage varying from 0 to 650V. The laser power was kept
as low as YO-YO to avoid carrier generation The current
passing through the sample between electrodes 182 and 184
was 10~ at 150V.
In FIG. 3 there are shown exemplary curves of the
absorption coefficient spectra for 0 volt, 400 volts, and
600 volts, as marked. The curves for 400 volts and 600
volts are shifted vertically upward to avoid crossing in
FIG. 3. Shifts of the heavy hole excitor shown as peak 200
A at 0 volts, to 200 B at 400 volts, and to 200 C at 600
volts are evident. Also peak 200 A, 200 B, 200 C show a
broadening as the voltage is increased. The light hole
30 pea 210 A at 0 volt, 210 B at 400 volts, and 210 C at 600
volts show both a shift and a broadening. The shifts are
difficult to measure up to V-200V because they are small
compared to the line width; above this value they show a
super linear dependence on V.
In FIG. 4 the shift of the heavy hole peak ~200 A,
2G0 B, 200 C in FIG. I is shown as curve 212. Roy shift
of the light hole peak (210 A, 210 B, 210 C in FIX. 3) is
~3~3~6~
-- 8 --
shown as curve 214~
The light hole-exciton peak was found to shift
more than the heavy hole one. The line width could be
measured only for the heavy hole-exciton; it was taken as
the half width at half-maximum on the low energy side of the
peak. The width variation is monotonic and approximately
linear with the applied field; varying from 2~8 me at
applied voltage = OVA to 4.3 me at applied voltage =
600V.
The precise value of the field applied to the MCKEE
is difficult to evaluate because of possible space charge
effects and contact resistances. Assuming that the sample
behaves like a simple resistor, at the center of the
inter electrode spacing one can take E = yV/d, where the
correction factor for the present exemplary geometry is
yo-yo, as discussed more fully by M. Cordon in the
article "Electric Field Modulation," published in Solid
State Physics 11, Academic Press, New York,
pages 165-275, 196g.
I the absorption spectra shown in FIG. 3 for 400
volts corresponds therefore to E~104V/cm, and for 600
volts therefore corresponds to Ea1.6x104V/cm. To
evaluate the winding energy of the excitors in MOW AYE Y
thick, we use the experimental and theoretical results of
R. C. Miller et. at. as disclosed more fully in their
article "Observation of the Excited Level of Excitors in
Gays Quantum Wells," published in Review, Volt
B24, p. 1134, in July 1981. Miller et. at. give En
(Huh e) = 9 me and Exile = t0.5 me. The radius
can be calculated by the relation; axe
constant, which gives axe and axle
AYE. The corresponding ionization fields,
HI 1.4 x 104V~cm and
HI = 1-9 x 104V/cm~ are quite consistent with our
measurements. Note that any simple perturbation analysis
of our results is invalid because of the high value of the
fields, relative to HI, which we utilize. The
~.~23~
_ g
observation of the larger binding energy of the light hole
excitor and its greater sensitivity to static fields should
be interpreted with care, as additional complication may
occur due to its proximity to the heavy hole inter band
continuum.
Referring to FIG. 5, the variations of the
absorption coefficient are shown as a function of the
applied potential at the heavy hole-exciton peak in curve
220, and 5 me below in curve 222. Positive or negative
changes larger than ~a~+4x103cm 1 are obtained in
the absorption at given photon energies as the field is
increased from 0 to 1.6 x 104V/cm. This result
compares most favorably to the case of bulk Gays where
fields up to (5 + 1) x 104V/cm are necessary to induce
changes of the absorption coefficient
2 x 102cm 1, as well as to the case of other III-
V compounds where changes pa 2 x 103cm 1 are
obtainer but only with fields as large as 4 x
105V/cm. Bulk Gays is discussed more fully by
Still man et. at. in the article "Electroabsorption in
Gays and its Application to Wave guide Detectors and
Modulators," published in Applied Physics Letters Vol. 28,
page 544 r in May 1976, and other group III-V compounds are
discussed more fully by Kingston in the article
"Electroabsorption in GaInAsP," published in
Physics Letters, Vol. 34, page 744, in June 1979.
It is important to understand that the speed at
which the absorption changes is not determined by the
excitor lifetime; rather it is related to how fast the
energy levels of the crystal can be shifted, which is a
very fast process. The speed of a modulator based on this
effect will most likely be limited by the speed at which the
"static" field can be applied. We showed that the present
device produces large variations of the absorption coeffic-
tent I I+ 4 x 103cm 1) for applied fields of threadier of 104V/cm~ this effect is usable in very high-
speed optical modulation schemes because of the fast
I
10 --
mechanisms involved and the small volume (< 100~m3)
necessary to achieve large change of transmission.
Referring to FIG. 6, there is shown an exemplary
operating frequency 230 for a laser which has its output
light beam modulated by the present invention. Solid curve
232 shows the absorption coefficient (a) plotted versus
photon energy for the case that no electric potential is
applied between contacts 182 and 184 as shown in FIG. 2.
Curve 234, curve 236 and curve 238 show the contribution of
the heavy hole, a light hole, and band gap absorption
respectively, to the absorption coefficient curve 232.
pun application of an electric potential between
contacts 182 and 184 shown in FIG. 2, the potential results
in an electric field within the MOW structure 174, and a
consequent shift on the excitor absorption. Curve 240
shows, for example, a shifted absorption coefficient a
which gives the optical absorption of the MOW structure 174
after application of electric potential between contacts
182 and 184. Curve 242 and curve 244 show the shifted
heavy hole and shifted light hole excitor absorption
respectively. The shift in the heavy hole excitor peak is
labeled "excitor shift" 2460 The band gap absorption is
shifted a Swahili amount through the Fran~-Keldysh effect as
is indicated by curve 250. The shift due to the Franz-
Keldysh effect is labeled and indicated by reference numeral 252.
At laser operating frequency 230 the optical
absorption coefficient has a low value shown by line 260
when zero electric potential it applied between contacts 182
and t84, and in contrast has a high value shown by line 262
upon the application of an electric potential between con
teats 182 and 184. The shift in the optical absorption
curve, labeled "excitor shift" 246, which occurs as a result
of the application of a modest potential between contacts 182,
1~4, causes a very large change in the optical absorption
coefficient at laser operating frequency 230, and therefore
~3~i6
"
provides deep modulation of the laser beam.
As an example of the amount of modulation which
can be achieved with a micron dimension modulator, the
decibels (DUB) of modulation can be calculated from the
expressions
= eve (V1)~ *
and
IVY)
where IVY) and ~(V13 are the intensity and absorption
coefficient at voltage V1, IVY) and ~(V2) are the
intensity and absorption coefficient at voltage V2, and T
is the length of the optical path through the MOW. The
example presented above showed a pa where
pa = avow (V1 )
of approximately 4 103 cm 1 for applied voltages
of less than 600 volts. Using equation I with various
thicknesses gives the modulation shown in Table 1.
TABLE 1
Thickness T Modulation DUB
Microns
17
Thus the "excitor shift" 246 provides excellent deep
modulation of a laser beam with only micron dimension
optical path length.
A KIWI structure, as designated in FIG. 1 by
reference numeral 120 and in FIG. 2 by reference numeral
174, is a semiconductor multiple layer heterostructure~ An
MOW is made of alternate layers of wide band gap material
and narrow band gap semiconductor material. FIG. 7 gives a
cross-sectional view of a typical MOW structure. ho
I
exemplary embodiment of an MOW uses AlxGa1 was as
the wide band gap material and Gays as the narrow band gap
semiconductor material.
Layers of narrow band gap Gays 268-1 to 268-N are
alternated with layers of wide band gap Al XGa1 Was
270-1 to 270-N+l. Convenient choices for the dimensions of
the structures in an MOW are, for the thicknesses of the
Gays 268-1 to 268-N layers .01 micron, for the thickness ox
the Algal was 270~2 to 270-N layers .01 micron,
and for the thickness of the Gays substrate 274
approximately 100 microns. The side dimensions of the
substrate 274 may be chosen conveniently as approximately 1
to 5 millimeters. The MOW structure then has layer planes
of Gays 268-1 to 268-N whose length and width are
approximately 1 to 5 millimeters and whose thickness is
approximately .01 microns. Also the alternate layers of
Algal was 270-2 through 270-N have the same
ratio of length and width to thickness, that is, 1 to 5
millimeters in length and width and approximately .01
microns in thickness. Thus, the MOW structure comprises
essentially plane layers of Gays 268-1 to 268-N interleaved
with plane layers of AlxGa1 was 270-1 to 270-N+1.
The alternate layers of Gays and A1xGa1_xAs may
be deposited using, for example molecular beam epitaxy
using methods as, for example, taught by Dangle et. at.
in U. S. patents Nos. 3,982,207, 4,205,329 and 4,261,771.
Epitaxial growth of heterostructures is further described
in the reference book by Casey and Punish, "Heterostr~cture
Lasers Part B; Materials and Operating Characteristics" r at
Chap. 6, pp. 71-155, and molecular beam epitaxy is
particularly discussed at pp. 132-144, academic Press, New
York, 1978.
Further details of an exemplary design for an MOW
multiple layer heterostructure are also given in Fig I
There is shown in FIG. 7 capping layers 270-1 and 270-Nt1.
The capping layers 270-1 and 270-N+1 are the first and last
wide band gap layers, and they are made thicker than the
it
- 13 -
layers 270-2 to 270-N which only separate layers of narrow
band gap material. An internal capping layer 270 1 may be
epitaxially crown on substrate 274 in order, for example,
to cover over any imperfections in the upper surface 278 of
S substrate 274. An external capping layer 270-N+1 may serve
to protect the underlying thinner layers from mechanical
injury. Further, the upper surface 280 of external capping
layer 270-N~1 may be shaped or treated to serve as a
partially reflecting mirror, or surface 280 may serve to
attach the multiple layer heterostructure to an external
device (not shown), or surface 280 may serve as the side of
an optical wave guide used to direct a beam of light to
propagate substantially parallel to the layer planes 268-1
to 268-N and 270-1 to 270-N+1. Surface 280 and external
capping layer 270-N~1, or internal capping layer 270-1, may
serve additional purposes which will be apparent to those
skilled in the art of optical devices. Capping layers 270-
1 and 270-N~1 are normally made from the wide band gap
material which forms the charge barrier layers, and so
additionally serve the function of preventing charges from
leaking out of the narrow band gap charge carrier material.
The narrow band gap layers 268-1 to 2G8-M are
charge carrier layers and each layer forms a quantum well.
The width of the quantum well is determined by the
thickness of the narrow band gap material. The barrier
heights for the quantum well are determined by the
differences between the conduction bands and between the
valence bands of the narrow band gap material 268-1 to 268-N
and the wide band yap material 270-1 to 270-N+1. The
on barrier heights at the junction between epitaxially grown
narrow band gap and wide band gap materials are shown in FIG.
9 for the GaAs/A1xGa1 was case. In FIG 9 both the
conduction band carrier and the valence band barrier are
shown.
The wide band gap material used for layers 270-1 to
?70-N+1 need not be a semiconductor. The layers must be
epitaxially grown upon the substrate 274, and one upon the
3~23~4~
other, The charge carriers produced by photon absorption
within the layers of narrow band gap material 268-1 to 268-N
then may propagate throughout the entire epitaxially grown
crystal with their motion limited only by the potential
barriers which occur a the boundaries of narrow band gap
material and wide band gap material, as is shown in FIG. 9
and FIG. 10 for the GaAs~A1xGa1 was case
Referring to FIG. 8, the band structure of Gays is
shown in a simplified diagram. Reference to the Gays band
structure as shown in FIG. 8 provides insight into excitor
absorption in an MOW structure. Energy is plotted along
the vertical axis 300. The valence band Eve 302 and
the conduction band Be 304 are shown along with the
energy gap ERG 306. An excitor level 310 is shown with
a binding energy EN 312 measured from the conduction
band 304.
A photon absorption transition 306 from the
valence band 302 to the excitor level 310 is shown
Transition 306 represents an excitor creation transition,
and such transitions are thought to be the cause of
resonant absorption peaks AYE and OWE as shown in FIG. 3.
After the excitor level 310 is formed as a result of photon
absorption, the excitor may break apart and form both a
conduction band electron 312 and a valence band hole 314,
The excitor is thought to break apart as a result of
ionization by a lattice vibration photon which supplies
the necessary energy
An inter band photon absorption transition 320 in
which a conduction band electron 312 and a valence band
hole 314 are formed as a result of photon absorption is
shown. Inter band photon transitions 320 are thought to
account for the band gap absorption as indicated by
reference numeral 238 in FIG. 6. The inter band photon
absorption transition 320 is a direct transition because of
the band structure of Gays.
Referring to FIG. 9, the band gap 330 of Gays and
the band gap 332 of A1~Ga1_xAs are shown for an
36
epitaxially grown junction 335. Such junctions occur
between the layers of epitaxially grown alternate layers of
Gays and AlxGa1 was as shown in Figs 1 2 and 7.
The valence band edge 338 of AlxGa1 was is
believed to be lower in energy Han the valence band edge
340 of Gays. The conduction band edge 342 of
A1xGa1 was is believed to be higher in energy
than the conduction band edge 3~4 of Gays. The total
difference between the two gaps, of Gays and
AlxGa1 was, is believed to be distributed as
approximately 15 percent 346 of the difference appears at a
lowered valence band edge 338 of AlxGa1 was, and
approximately 85 percent 348 of the difference appears as
an increase in the conduction band edge 342 of
AlxGa1 was, relative to Gus The excitor
binding energy EN for Gays is designated by reverence
numeral 345 in Figs 8 and 9, for the case in which the
Gays layer is substantially thicker than 1000 Angstroms
Referring to FIG. 10, the potentials seen by both
a conduction band electron and by a valence band hole
within a MOW structure are shown. The conduction band
electron energy barrier EKE 354 is shown. The valence
band hole energy barrier REV 356 is shown, Conduction
electrons within a narrow band gap layer 360 are trapped in
a potential well with sides of height EKE 354.
Correspondingly, valence band holes within a narrow band gap
layer 360 are trapped by the energy barrier ARC 356.
A conduction band electron or a valence band hole
may be produced within the narrow band gap layers 360.
Alternatively if the electron and hole are produced within
the wide band gap layers 3~2, they will experience
potentials due to he and he which will drive
them into the narrow band gap layers 360 where they will be
trapped by the potential barriers he and eke.
In an exemplary embodiment in which the narrow band gap
material is Gays and the wide band gap material is
AlxGa1~xAS~ the magnitude of both rev and
~3~6
- 16 -
EKE depends upon the mole fraction x of Al in the wide
band gap layers.
The energy Levels of electrons and holes trapped
within the narrow band gap layers 360 are shifted relative
to their locations in bulk material because of quantum
confinement effects arising from the thinness of the narrow
band gap layers 360~ A thickness of 1000 Angstroms or less
can cause appreciable shift in the allowed energy levels of
a semiconductor layer. Also the electrons and holes
interact to form excitor pairs. The excitor pairs occupy
energy levels which are shifted from the single particle
energy levels. All of these energy levels are affected by
an electric field applied to the narrow band gap layers 360.
Referring to FIG. 11, the effective optical thickness of
the MOW structure mounted as shown in FIG. 2 is shown
plotted versus the intensity of incident light beam l92~
The incident light beam 192 is adjusted to coincide with
peak AYE in the optical absorption curve shown in FIG. 3.
The intensity of light beam 192 was varied. The effective
optical thickness of the sample measures the total
attenuation of the light beam as it traverses the sample.
Curve 370 shows the effective optical thickness of a bulk
sample of Gays. Curve 372 shows the effective optical
thickness of the MOW device. The effective average
I intensity of light beam 192 is plotted along the lower
margin 374 of FIG. 11 and is shown to vary from 0 to
approximately S0,000 Watts/cm2. The total incident
power in beam 119 is plotted along the upper margin 376 of
FIG. 11 and is shown to vary from .01 to approximately 50
milliwatts. A comparison of the effective optical
thickness of bulk Gays and a GaAs/AlxGa1 was MOW
device at an incident light power of 0.1 milliwatts is
shown by lines 380, 382, 384. The effective optical
thickness of the bulk baas shown in curve 370 is shown to
decrease from a value of approximately 2 to a value of
approximately 1.9 at an incident power of approximately
0.1 milliwatts, an approximate change in effective optical
- 17 -
thickness of (2.0-1.9)/2 = 5%. In contrast, the effective
optical thickness of the MCKEE device is seen to vary from
approximately .75 to approximately .63 as the incident
power varies from zero to 0.1 milliwatts, or a percentage
change of approximately (.75-.63~/.75 = 16%. The decrease
in effective optical thickness with increasing beam
intensity is attributed to saturation of the optical
absorption of the material, and is commonly referred to as
a nonlinear absorption.
Referring to FIG. 12, an example of the variation
of optical absorption 390 and index of refraction 392 with
photon energy is shown as the two are related by the
Kramers-Kronig relationship using a Lorentzian absorption
line shape. The curves of optical absorption 390 and index
of refraction 392 illustrate generally the variation of
these quantities for excitor absorption over a photon
energy range in the vicinity of the band gap. Roy curves
shown in FIG. 12 illustrate the relationship between
optical adsorption as shown in FIG. 3 for an MOW structure
and the corresponding index of refraction, as that
relationship is given by the Kramers-Kronig relationship
using a Lorentzian absorption line shape.
Curve 390-A represents a large optical resonant
absorption for a low incident light intensity and a
corresponding index of refraction is shown in curve 392-A.
A smaller resonant absorption is represented by curve 390~B
for a hither incident light intensity and the
correspondingly smaller index of refraction is represented
by curve 392-~. A further smaller resonant absorption is
represented by curve 390-C for a still higher incident
light intensity and the correspondingly smaller index of
refraction is represented by curve 390-C.
For a single excitor resonance the Kramers-Kronig
model illustrated in FIG. 12 shows that for photon energy
below the resonant energy 395 the index of refraction
decreases with increasing light intensity, while for photon
energies above the resonant energy 395 the index of
- 18 -
refraction increases with increasing incident light
intensity.
For a multiple quantum well, the variation of
index of refraction with light intensity depends upon the
S interaction of at least one and possibly several excitor
resonances with the processes leading to the background
index of refraction. These interactions involve quantum
interference effects which further complicate the detailed
variation of the index of refraction with both light
intensity and photon energy. For example, the
s/AlxGal_xAs MOW, whose measured optical
absorption coefficient is shown in FIG. 3, is dominated by
two resolvable excitor absorption peaks superimposed upon
an inter band transition background. The interaction of
those absorption processes complicates the variation of
index of refraction beyond the simple predictions of the
one peak model using the Kramers-Kronig relationship.
Louvre, the model makes a useful connection between the
measured optical absorption and the index of refraction for
the practice of embodiments of the invention which depend
upon a variation of the index of refraction with incident
light intensity, or applied electric field.
FIG. 13 shows an alternative attachment of
electrical contacts to a MOW structure 403. Contacts 400,
401 provide an electric field substantially parallel with
the planes 405 of MOW 403.
JIG. 14 shows a top view of MOW 410. Electrical
contacts 412 and 414 may, for example, be deposited upon
the upper surface (shown in top view but not indicated by
30 reference numeral) of MOW 410. Contacts 412, 414 provide a
substantially uniform electric field between end 416 and
end 418 which penetrates MOW 410 and provides an electric
field substantially parallel to the layer planes of MOW
410.
Referring Jo FIG. 15, there is shown an
alternative arrangement of electrical contacts 420, 422 to
MOW structure 424. Contacts 420, 422 may be made, for
I
-- 19 --
example, by ion implantation of an MCKEE Selective ion
implantation produces conductive regions in a semiconductor
which result in electrical contacts 420, 422. Also,
diffusion may be used to make doped contacts
Referring to FIG. 16, there is shown an
alternative exemplary method of making electrical contacts
to the layer planes of an MOW. As a first step, an MOW 430
is grown epitaxially upon a substrate 432~ As a second
slept a mesa 434 of MOW material is made by selectively
1G etching away MOW material As a third step, contacts 436,
438 are regrown on substrate 432 so as to electrically
contact the edges 440, 442 of layer planes 444 of the mesa
434 of MOW material. A potential applied between contacts
436, 438 produces an electric field within mesa 434 which
is substantially parallel with the MOW layer planes 444.
The narrow spacing, 446, provides a small volume
modulator. For example, the active region could be 1
micron square, and the thickness could, for example be 20
microns to provide a modulation of 35 DUB as given in TABLE
1, for a total active volume of 20 cubic microns. Such a
small volume provides a small capacitance and operates
satisfactorily in the multiple gigahertz frequency range.
Materials other than the GaAs/AlGaAs system are
useful in practicing the invention. A MCKEE may be made in
which the narrow band gap material is Alga was
and the wide band gap material is AlxGa1 was. The
band gap of the material is larger as the fraction of Al in
the material is larger, and so the wider band gap mole
fraction x must be larger than the narrow band gap mole
fraction y.
The invention uses a super lattice in which the
conduction band of the narrow band gap layer is below the
conduction band of the wide band gap layer, and the valence
band of the narrow band gap layer is above the valence band
of the wider band gap layer, and this type of super lattice
band structure is named a Type 1 super lattice. The Type 1
super lattice is distinguished from a Type 2 super lattice in
I
- 20 -
which both the conduction band and valence band of one of
the materials are below the corresponding conduction and
valence bands of the other material. A Type 2 super lattice
device is disclosed by Clang et. at. in US. Patent No.
4,208,667 issued June 17, 1980 and entitled "Controlled
Absorption in Heterojunction Structures".
Materials which are relieved to exhibit the Type 1
super lattice band structure and are suitable for making an
MOW structure of the present invention include InGaAs,
InGaAlAs, InGaAsP, and also HgCdTe. These materials are
useful for application in the 1.5 micron to 1.3 micron
wavelength range.
Lattice-matched growth of these materials and
their band structure is discussed more fully in the books
by Casey and Punish entitled "Heterostructure Lasers, Part
A", and "Heterostructure Lasers, Part B", Academic Press,
New York, 1978. Also HgCdTe is believed to exhibit the
type 1 super lattice band structure and is therefore
suitable for making an MOW of the present invention.
In a structure using InGaAs, the narrow band gap
layers use InGaAs and the wide band gap layers use In.
These materials are lattice-matched and so are suitable for
epitaxial growth as a multiple quantum well device
In a structure using InGaAlAs as the narrow
band gap material, the wider band gap material may be In or
may be other compositions of InGaAl~s which are chosen to
have a wider band gap. Such a choice is possible because
the band gap of the material can be varied by varying the
composition, while maintain lattice-matched crystal
growth.
In a structure using InGaAsP as the narrow band gap
material, the wider band gap material may be selected from
In or may be another composition of InGaAsP which is
lattice-matched for crystal growth.
In a structure using HgCdTe as the narrow band gap
material, the wider band gap material may be made using
HgCdTe of a different composition or Cute and therefore
~;~3~9L6$~
- 21 -
wider band gap.
Additional materials which are believed to exhibit
type 1 super lattice band structure include Gas, AlGaSb, So
and Go.
Turning now to FIX&. 17, there is shown an MOW used
as the intrinsic, or I, layer of a PIN semiconductor diode
structure. An electric field may be advantageously applied
to the MOW by reverse biasing the PIN diode. The PIN diode
structure is particularly suitable for applying the
electric field perpendicular to the layer planes of the
MOW. Advantages gained from the use of the PIN structure
to apply an electric field to the MOW are that the diode is
operated in reverse bias, and this condition gives a high
resistance to current slow through the MOW. Also the
device may be made small in lateral area in order to
minimize the capacitance.
FIG. 18 shows an alternate PIN diode structure for
conveniently propagating the light in the direction
parallel to the MOW layers. The MOW may have only one
narrow band gap layer sandwiched between wide band gap layers
to form only one quantum well. Or the MOW may have several
narrow band gap layers in order to provide a stronger
interaction between the light and the optical properties of
the quantum wells, such as absorption, index of refraction
bireringence, and other polarization properties. Lateral
confinement of the light can be achieved by etching a ridge
or by other means
Turning to FIG. 19, there is shown an exemplary
embodiment of a PIN diode structure used to apply an
electric field perpendicular to the layer plane of an MOW.
The sample was grown on a So doped [100] Gays substrate.
The unhoped optically active layer contained 50 Gays wells
each 95 Angstroms thick and was surrounded by unhoped
buffer layers and doped contact layers. This structure
creates a PIN diode which was operated in reverse bias
mode. buffer and first contact layers were also made of
a super lattice of alternating layers of Galas and Gays.
Introducing the thin layers of Gays into the nominally
unhoped buffer regions was found to reduce the background
doping level by more than an order of magnitude. This
reduction of background doping level ruckuses the field
S in homogeneity across the active region and reduces the
drive voltage of the device. The device was defined
laterally by an etched mesa 600 microns in diameter. A
small hole was etched through the opaque substrate by a
selective chemical etch. The capacitance of the device was
20 pi The device was made fairly large or ease of
fabrication, but a smaller device will advantageously have
smaller capacitance.
The lower portion of FIG. to shows the internal
electric field in the various layers at two different
applied voltages as calculated within the depletion
approximation with a p-type background doping level of
2 x 1015cm 3 in the intrinsic layers. The active
layer can be switched from a low field to a high field of
approximately 6 x 104V/cm by the application ox 8
volts. Because the device is operated as a reverse biased
device, its resistance is high and its capacitance is low,
which are desirable electrical properties. The heavily
doped contact layers can easily be metallized and
contacted.
Turning to FOG. 20, there is shown the optical
transmission of an exemplary embodiment of the invention as
shown in FIG. 19. The transmission at 0 volts shows the
usual excitor peaks. Between 0 volts and 8 volts applied
there is almost a factor of 2 reduction in transmission at
30 a photon energy of 1.446 eve (857nm). Greater modulation
depth is possible with a thicker sample. The dynamical
optical response was determined by bridging the PIN diode
structure with a 50 ohm resistor and driving it with a
pulse generator with a rise time of 1.8 nanosecond. The
optical output was detected with a So avalanche photo diode
of roughly 1 nanosecond rise time. The observed 10~ to 90%
rise time was 2.8 nanoseconds. The calculated ARC rise time
~39i~
- 23 -
of the device when driven by a 50 ohm load is 2.2
nanoseconds The optical photon energy was 1.454 eve (853
no). Thus modulation occurred with a rise time limited by
the capacitance of the device.
Turning now to FIG. 21, there is shown an
illustrative embodiment of an electrically tuned Fairy-
Pert cavity. The Fabry-Rerot cavity is formed by two
substantially parallel mirrors which are partially
transparent. The MOW is placed between the two mirrors and
therefore within the Fairy Pert cavity. The MOW has an
adjustable electric field applied perpendicular to the
layer planes. The optical transmission of the Fabry-Perot
cavity for light incident substantially perpendicular to
the mirrors it high when the optical path length between
the mirrors is an integral number of one-half wavelengths
of the light The index of refraction of the MOW, and
hence the optical path length within the Fabry~Perot cavity
can be changed by changing the voltage applied to the MOW.
FIG. 22 is an illustrative graph of the optical
transmission of the Fabry-Perot cavity shown in FIG. 21.
The optical transmission is plotted versus the voltage
applied to the MOW. The index of refraction of the MOW
varies as the voltage varies, and so at values of the
voltage for which the optical path length within the
~abry-Perot cavity is an integral number of one-half
wavelengths, the optical transmission is high, and at other
values of the voltage the transmission is low.
A losing gain medium may be optionally included
within the Fabry-Perot cavity along with the voltage-
controlled MCKEE The frequency at which the gain mediumlases may be selected by a choice of the optical path
length within the Fabry-Perot cavity, and so the frequency
of losing of the gain medium may be selected by adjustment
of the voltage applied to the McKee Choosing the voltage
applied to the MOW determines the index of refraction of
the MCKEE the index of refraction determines the optical
path length within the Fahry-Perot cavity, and the losing
I
- I -
frequency will be limited to those frequencies for which a
one-half integral number of wavelengths exist within the
Eabry-Perot cavity. Losing occurs at those frequencies for
which the output spectrum of the losing medium matches the
S frequencies at which a one-half integral number of
wavelengths can exist within the Fabry-Perot cavity.
Turning now to FIG. 23, a cross sectional view of
a controlled Fabry-Perot cavity is shown. Light propagates
parallel to the layers of the MOW. Mirrors in which the
plane of the reflective surface lies substantially
perpendicular to the propagation direction of the light are
shown and define the Fairy Pert cavity. Only one
exemplary quantum well is shown because the optical path
length is long when the light beam propagates parallel to
the layers, and consequently sufficient optical interaction
is achieved with only a few quantum wells.
FIG. 24 shows a cross-sectional view of a
polarization modulator. Light propagates parallel to the
layer planes of the MOW, and it guided by the guiding
I layers. The directions of propagation of the incoming
light is indicated by the direction k. The polarization of
the incoming light is described by reference to the Zeus
coordinate axis in which the y axis is shown to lie along
the direction of propagation of the incoming light, that
is, the direction k coincides with the y axis. The
polarization of the incoming light wave is defined by the
components of the electric field of the incoming light as
they are resolved along the x and z axes. En is the
component of the light wave electric field resolved along
the x axis and lies parallel to the layer planes of the
MCKEE. En is the component of the Lotte electric
field resolved along the z axis and lies perpendicular to
the layer planes of the MOW. The optical transmission of
the MOW along the direction parallel to the layer planes,
which is along the y axis, is different for the two
polarization components En and En. This difference
in transmission is a birefringence of the MOW.
'~L%~3~6
- 25 -
The birefringence of the MOW may be varied by
application of an electric field to the MOW. For example,
an electric field may be applied to the MOW in a direction
substantially perpendicular to the layer planes.
Alternatively the electric field may be applied parallel to
the layer planes. For a parallel application of the
electric field to the layers of the MOW there are two
possibilities, the first being parallel to the propagation
direction of the light, that is along the y axis. The
second possibility is perpendicular to the propagation
direction of the light, that is along the x axis. Each of
these possible applications of an electric field may be
used to cause variation in the polarization of the outgoing
light. A downstream polarization filter (not shown)
permits intensity modulation of the light as it emerges
from the filter.
The input light may be tuned to a frequency which
optimizes the modification ox absorption property of the
MOW by the applied electric field.
In FIG. 6, the laser operating frequency 230
indicates an exemplary adjustment of the laser frequency
for operation of a device as an absorption modulator. We
turn now to optimization of the MOW device as a modulator
of optical path lengths. The laser may be operated at a
I frequency AYE, which is well below the energy of the large
absorption as modified by the electric field. The
influence of the electric field on the index of refraction
extends below the absorption peak. Because the index of
refraction effect extends below the absorption peak it is
possible to tune the low frequency much as shown by AYE in
FIG. 6 in order to optimize the apparatus to modify the
optical path lengths through the MOW. The light frequency
for optimum operation of the device will depend upon the
polarization of the incident light.
The optical absorption and the index of refraction
may be saturated by increasing the intensity of the input
light. The saturation may be influenced by application of
~3~i6
- I -
an electric field to the MOW. For example, a bistable
optical device may be caused by application of an electric
field to become not bistable. Also the operating
characteristics of a nonlinear optical device may be
changed by application of an electric field.
In FIG. 25 an array of electric field controlled
MOW devices is shown. An array of electric field
controlled type of MOW devices may be grown on a single
substrate. These MOW devices may form a logical array of
optical switching elements. The logic of the element may
be modified by application of electric fields to the
individual elements. A different electric field may be
applied to each MOW device. An array of MOW devices which
are individually controlled by electric fields is a
programmed logical array.
It is to be understood that the above-described
embodiments are simply illustrative of the principles of
the invention. Various other modifications and changes may
be made by those skilled in the art which will embody the
principles of the invention and fall within the spirit and
scope thereof.
