U.S. patent application number 11/530336 was filed with the patent office on 2008-09-25 for mobile charge induced periodic poling and device.
Invention is credited to Slmon John Field, Lee Lisheng Huang.
Application Number | 20080231942 11/530336 |
Document ID | / |
Family ID | 39157717 |
Filed Date | 2008-09-25 |
United States Patent
Application |
20080231942 |
Kind Code |
A1 |
Huang; Lee Lisheng ; et
al. |
September 25, 2008 |
MOBILE CHARGE INDUCED PERIODIC POLING AND DEVICE
Abstract
Devices and methods are disclosed for realizing a high quality
bulk domain grating structure utilizing mobile charges that are
generated by means of photo-excitation in a substrate. An effect of
light exposure (UV, visible, or a combination of wavelengths) is to
generate photo-induced charges. The application of a voltage across
the substrate combined with the application of light exposure
causes a photo-induced current to flow through the substrate. The
photo-induced charges (behaving like virtual electrode inside the
material) and the photo-induced current result in both reduction of
the coercive field required for domain inversion in the material
and improve realization of the domain inversion pattern, which
previously has not been possible at room temperature.
Inventors: |
Huang; Lee Lisheng; (Palo
Alto, CA) ; Field; Slmon John; (Los Gatos,
CA) |
Correspondence
Address: |
HELLER EHRMAN LLP
4350 La Jolla Village Drive, 7th Floor
San Diego
CA
92122
US
|
Family ID: |
39157717 |
Appl. No.: |
11/530336 |
Filed: |
September 8, 2006 |
Current U.S.
Class: |
359/326 |
Current CPC
Class: |
G02F 1/3775 20130101;
G02F 1/3558 20130101 |
Class at
Publication: |
359/326 |
International
Class: |
G02F 1/35 20060101
G02F001/35 |
Claims
1. A domain grating device, comprising: a substrate with first and
second opposing surfaces, the substrate having an inverted domain
grating structure which extends through the entire substrate; and
wherein, an inverted domain duty cycle at the first surface is
greater than 50% and less than 100%, and an inverted domain duty
cycle at the second surface is less than 50% and greater than
0%.
2. The device of claim 1, wherein the substrate is a ferroelectric
substrate.
3. The device of claim 1, wherein the substrate is made of a
material selected from at least one of, MgO doped congruent lithium
niobate, stoichiometric Lithium Niobate, Stoichiometric Lithium
Tantalate, MgO:Stoichiometric Lithium Niobate, MgO:Stiochiometric
Lithium Tantalate, ZnO:Lithium Niobate, In:Lithium Niobate,
Ti:Lithium Niobate and Er:Lithium Niobate.
4. The device of claim 1, wherein a thickness of the substrate is
between 100 .mu.m and 2 mm.
5. The device of claim 1 wherein a thickness of the substrate is
between 450 .mu.m and 550 .mu.m
6. The device of claim 1 wherein a thickness of the substrate is
between 950 .mu.m and 1150 .mu.m
7. The device of claim 1, wherein a thickness of the substrate is
at least 2 mm.
8. The device of claim 1, wherein said grating structure has an
average period between 4 .mu.m and 13 .mu.m.
9. The device of claim 1, wherein said grating structure has an
average period of less than 4 .mu.m.
10. The device of claim 1, wherein said grating structure has an
average period that is greater than 15 .mu.m and the substrate has
a thickness greater than 1 mm.
11. The device of claim 1, wherein the inverted domains are tapered
in size from the first surface to the second surface.
12. The device of claim 1, wherein a domain pattern on each of the
first and second surfaces is a visible domain pattern.
13. The device of claim 12, wherein the visible domain pattern is
used to predict a performance of the frequency converter for
process yield purposes.
14. The device of claim 1, wherein the surface at which the
inverted domains have a greater duty cycle is the surface where a
pattern electrode is disposed during a poling process
15. The device of claim 1, further comprising: a repeated inverted
domain grating structure extending through the entire substrate
forming a frequency conversion device; a laser pump source; and a
means configured to provide phasematching between said laser pump
source and said frequency conversion device.
16. The device of claim 15, where said means configured to provide
phasematching between said laser pump source and said frequency
conversion device is a heating device configured to maintain a
particular temperature of said frequency conversion device.
17. The device of claim 15, where said means configured to provide
phasematching between said laser pump source and said frequency
conversion device is a means to tune the wavelength of the pump
source to match that of the frequency converter.
18. The device of claim 15, where said laser pump source comprises
a semiconductor diode laser.
19. The device of claim 15, where said laser pump source comprises
a diode pumped solid state laser.
20. The device of claim 15, where the optical beam emitted by said
pump laser source passes through the region of the domain grating
which has substantially 50% duty cycle.
21. The device of claim 15, further comprising a means for
efficient optical coupling between said laser pump source and said
frequency conversion device.
22. The device of claim 15, wherein the inverted domain device is
incorporated inside the laser cavity of said pump laser.
23. The device of claim 15, wherein the inverted domain device is
at least partially coated with a conductive coating forming a
charge dissipating closed-loop.
24. The device of claim 15, wherein said pump laser source
comprises a plurality of individual laser beams.
25. The device of claim 15, wherein the substrate is made of a
material selected from at least one of, MgO doped congruent lithium
niobate, stoichiometric Lithium Niobate, Stoichiometric Lithium
Tantalate, MgO:Stoichiometric Lithium Niobate, MgO:Stiochiometric
Lithium Tantalate, ZnO:Lithium Niobate, In:Lithium Niobate,
Ti:Lithium Niobate and Er:Lithium Niobate.
26. The device of claim 15, wherein a thickness of the substrate is
between 100 .mu.m and 2 mm.
27. The device of claim 15, wherein a period of the grating
structure is between 4 .mu.m and 7 .mu.m.
28. The device of claim 15, wherein inverted domains are tapered in
size from the first surface to the second surface.
29. The device of claim 15, further comprising: an optical coupling
means; a spatial light modulator; a projector lens element; and a
screen.
30. The device of claim 29, wherein the substrate is made of a
material selected from at least one of, MgO doped congruent lithium
niobate, stoichiometric Lithium Niobate, Stoichiometric Lithium
Tantalate, MgO:Stoichiometric Lithium Niobate, MgO:Stiochiometric
Lithium Tantalate, ZnO:Lithium Niobate, In:Lithium Niobate,
Ti:Lithium Niobate and Er:Lithium Niobate.
31. The device of claim 29, wherein a thickness of the substrate is
between 100 .mu.m and 2 mm.
32. The device of claim 29, wherein the grating structure has an
average period between 4.0 .mu.m and 13 .mu.m.
33. The device of claim 32, wherein inverted domains are tapered in
size from the first surface to the second surface.
34. A method of creating a domain grating device in a substrate
comprising: providing electrical contacts to first and second
opposing surfaces of said substrate; generating mobile charges in
said substrate; and, applying potentials to said electrical
contacts creating a patterned current flow through said substrate;
forming an inverted domain grating structure, wherein a domain duty
cycle at the first surface is greater than 50% and less than 100%,
and a domain duty cycle at the second surface is less than 50% and
greater than 0%.
35. (canceled)
36. The method of claim 34, wherein a domain duty cycle at the
first surface is greater than 50% and less than 100%, and a domain
duty cycle at the second surface is 0%
37. The method of claim 34, wherein mobile charges are generated in
the substrate in combination with the application of a patterned
electric field.
38. The method of claim 34, further comprising the steps of
annealing the domain grating device at a temperature between 500
and 650 degrees centigrade for a time of between 24 and 72 hours
providing a closed-loop electrical discharge path between the
opposing faces of the domain grating device during said
annealing.
39. The method of claim 34, further comprising: generating said
mobile charges by exposing the substrate to substantially spatially
uniform optical illumination through at least one of the first and
second surfaces to form an illuminated face.
40. The method of claim 39, wherein the optical illumination
consists of wavelengths from 250 nm to 600 nm.
41. The method of claim 39, wherein the optical illumination
consists of wavelengths greater than or equal to 400 nm
42. The method of claim 39, wherein the optical illumination is
filtered to remove wavelengths shorter than 320 nm.
43. The method of claim 39, further comprising: disposing a uniform
transparent electrode on the illuminated face.
44. The method of claim 43, further comprising: disposing a
patterned electrode on an opposite face relative to the illuminated
face.
45. The method of claim 44, further comprising: applying a high
voltage to the electrical contacts to cause a photocurrent to flow
through the substrate; and controlling a magnitude and duration of
the high voltage and illumination to create a domain inversion
structure through the thickness of the substrate.
46. The method of claim 45, further comprising: applying
illumination to the substrate simultaneously with application of a
first voltage.
47. The method of claim 46, further comprising: terminating the
application of illumination in response to a parameter selected
from at least one of, time, current flow and charge transfer.
48. The method of claim 47, further comprising: applying a second
voltage to the substrate crystal after the illumination is
terminated to cause a poling current to flow.
49. The method of claim 48, further comprising: terminating the
second voltage in response to a parameter selected from at least
one of, time, current flow and charge transfer to create a domain
inversion structure wherein a domain duty cycle at the first
surface is greater than 50% and less than 100%, and a domain duty
cycle at the second surface is less than 50% and greater than
0%.
50. The method of claim 49, further comprising: applying a time
delay between termination of the illumination and application of
the second voltage.
Description
BACKGROUND OF THE INVENTION
[0001] 1. Field of the Invention
[0002] This invention relates generally to high quality short
period bulk domain inversion structures (gratings), and more
particularly to high quality short period bulk domain inversion
structures (gratings) that are fabricated in substrate materials
such as MgO doped congruent lithium niobate using electric field
poling.
[0003] 2. Description of the Related Art
[0004] Quasi phase matching (QPM) is an efficient way to achieve
nonlinear optical interactions. The approach was first proposed by
Bloembergen et al, U.S. Pat. No. 3,384,433, using a domain
inversion grating structure to achieve QPM. Such a domain grating
structure can be usefully realized in an optically transparent
ferroelectric material, such as LiNbO3, LiTaO3, and KTP. There are
many different ways to achieve inverted domain structures in these
materials.
[0005] A periodic poled material structure can be grown directly
within the material by modifying a parameter during the growth
process, such as temperature, or a dopant concentration. Ming et.
al. ("The growth striation and ferroelectric domain structures in
Czochralski grown LiNbO3 single crystals" Journal of materials
Science, v11, p. 1663, 1982.) used variation of temperature, growth
rate and solute concentration during Czochralski growth to create a
periodic structure in Lithium Niobate. Laser heated pedestal growth
is disclosed in U.S. Pat. No. 5,171,400 by Magel et. al. from
Stanford University. This method can produce gratings with periods
as short as 6 .mu.m and 4 .mu.m, but it is difficult grow long
lengths and curvature of the domains limits the lateral dimensions
and efficiency.
[0006] Impurity doping or material removal in some ferroelectric
materials (such as lithium niobate and KTP) can result in domain
inversion. In lithium niobate, periodic domain inversion gratings
can be achieved through high temperature processes such as titanium
indiffusion, lithium outdiffusion (in air, or enhanced with surface
layers of SiO2 and MgO) or proton exchange. A mechanism for the
domain inversion was proposed by one of the present inventors,
based on space charge field of impurity gradients (Huang et al. "A
discussion on domain inversion in LiNbO3" Appl. Phys. Lett. v65. p.
1763, 1994). Byer et al, at Stanford University, U.S. Pat. No.
5,036,220, demonstrated a waveguide frequency converter wherein the
domain structure was created using titanium indiffusion in lithium
niobate.
[0007] Due to the typically shallow impurity diffusion depths, the
inverted domains are also typically shallow and generally
triangular or semicircular in depth in lithium niobate.
[0008] A high voltage may be used to generate domain inversion at
room temperature. Papuchon (U.S. Pat. No. 4,236,785) demonstrated
patterned electric field inplane poling on lithium niobate to
achieve waveguide quasi-phasematched nonlinear interactions Short
period domain inversion in Z-cut congruent lithium niobate was
first demonstrated by Yamada at Sony in 1992, U.S. Pat. No.
5,193,023 but the described process suffered from limitations in
the material thickness and high instances of destructive electrical
breakdown. Since this first report many different techniques of
applying the electric field have been demonstrated, generally
enabling electric field induced domain inversion to be achieved at
or near to room temperature, in contrast to the methods of Class 2.
Approaches include the use of patterned metal electrodes, patterned
insulators with liquid electrodes, U.S. Pat. No. 5,800,767, U.S.
Pat. No. 5,519,802, and corona discharge charging (Harada et al,
"Bulk periodically poled MgO:LiNbO3 by corona discharge method",
Appl. Phys. Lett V 69, #18, 1996, Fuji Photo Film Co Ltd). The
common feature of all of these approaches is the creation of a
localized electric field modulation (or patterned electric field)
on one face of the crystal substrate.
[0009] Bombardment with a high energy electron beam can be used to
induce bulk domain inversion in congruent lithium niobate at room
temperature as demonstrated by Yamada from Sony (Yamada et at
"Fabrication of periodically reversed domain structure for SHG in
LiNbO3 by direct electron beam lithography at room temperature"
Elect. Lett. Vol 27 p. 828, 1991), without the use of an applied
voltage. Ito et. al. also performed electron beam writing of domain
grating in lithium nioate (Ito et al. "Fabrication of periodic
domain grating in LiNbO3 by electron beam writing for application
of nonlinear optical processes" Elect Lett. Vol 27p. 1221, 1991).
The high energy electrons incident on the substrate penetrate the
surface and are trapped inside the substrate. These localized
trapped electrons in the material result in localized high electric
field that causes domain inversion. Earlier work by Keys et al
(Keys et al, "Fabrication of domain reversed gratings for SHG in
lithium niobate by electron beam bombardment". Electronics Letters,
V26, #3 p 188, 1990) used a mask to pattern the bombardment of a
high energy electron beam on congruent lithium niobate, and
combined with an elevated temperature and a small applied voltage,
this was demonstrated to provide patterned domain inversion.
[0010] In essence, all the methods described above are electric
field poling. The orientation of the internal dipole moment is
reversed under the influence of the local and global electric
field. In direct growth, and impurity diffusion approaches the
electric field is generated from a temperature gradient, or a
dopant gradient. With electron beam bombardment the electric field
is created by the trapped electrons injected into the substrate
from a high energy beam.
[0011] Early work in electric field poling for QPM applications
concentrated largely on congruent lithium niobate since this is by
far the most widely available nonlinear optical material and also
one of the most versatile, with a transparency range from about 400
nm to 5 microns in wavelength. However, as applications have come
to be developed for the visible spectrum, the large numbers of
defects in the congruent crystal structure, together with trace
impurities incorporated during the growth process, give rise to a
property called photorefractivity. The photorefractive effect is
caused by the directional drift of photo-excited charges generated
by absorption of visible and UV light within the material, which
creates a space-charge electric field. The space-charge electric
field leads, via the electro-optic effect, to a refractive index
change which distorts the optical beam passing through the crystal.
In order to be used in applications using or generating visible
light, congruent lithium niobate needs to be doped with about 5%
MgO, as shown by Bryan et. al, (Bryan et al. "Increased optical
damage resistance in Lithium Niobate" Appl. Phys. Lett. V44. p 847,
1984) to overcome the effects of structural defects and eliminate
the photorefractive effect.
[0012] However the MgO dopant in MgO:CLN brings an even bigger
challenge in realizing periodic domain structures. Many groups of
researchers around the world have been working on electric field
poling of MgO:CLN. For example, corona poling was attempted by Fuji
(R10,R19); the use of elevated temperatures was attempted by
Mitsubishi Cable (U.S. Pat. No. 6,565,648), and Matsushita
(Mizuuchi et al "Electric field poling in Mg doped LiNbO3", Jnl
Appl Phys, V96, #11, 2004, Mizuuchi et al "Efficient second
harmonic generation of 340 nm light in a 1.4 .mu.m periodically
poled bulk MgO:LiNbO3", Jpn J Appl Phys V42, p 90-91, 2003);
ultra-violet light, and laser light energy assisted poling has been
attempted by several other groups (Muller et al "Influence of
ultraviolet illumination on the poling characteristics of lithium
niobate crystals", Apl Phys Lett V83 #9 p 1824 2003, Valdivia et al
"Nano scale surface domain formation on the +Z face of lithium
niobate by pulsed ultraviolet laser illumination", Appl Phys Lett
V86 2005, Fujimura et al "Fabrication of domain inverted gratings
in MgO:LiNbO3 by applying voltage under ultraviolet irradiation
through photomask at room temperature", Elect Lett V39 #9 p 719
2003, Dierolf et al "Direct write method for domain inversion
patterns in LiNbO3", Apl Phys Lett V84 #20 p 3987 2004). However
short-period-domain-grating structures have not been achieved at
room temperature in a reliable and repeatable manner.
[0013] Part of the difficulty in poling MgO:CLN is the observation
that there is current flow through the substrate other than the
poling displacement current during the poling process. This current
flow results in preferential growth of domains which are formed
early in the poling process and disrupts the domain seeding
uniformity and therefore the uniformity of the final grating
pattern.
[0014] It is also found that the domain wall boundary in Mg doped
CLN seems to not be aligned as rigidly along the crystal axis as in
the undoped CLN material. Since the inverted domain structure does
not strictly follow the crystal structure, it is fundamentally
challenging for the inverted domain to propagate through the entire
thickness of the substrate while maintaining the lateral dimensions
of the masking pattern applied on one surface of the substrate.
[0015] Accordingly, there is a need to provide an improved domain
inverted grating device with high efficiency and high resistance to
photorefractive effects and a fabrication method able to control
the domain growth through the bulk of the crystal for short period
domain inversion gratings for applications in high power visible
light generation.
SUMMARY
[0016] Accordingly, an object of the present invention is to
provide a domain grating device, and its associated fabrication
methods, that has controlled domain growth through the bulk of the
crystal for uniform short period domain inversion gratings.
[0017] Another object of the present invention is to provide a
domain grating device fabrication method, using the generation of
mobile charges in a substrate to improve the seeding of inverted
domains and to guide the growth of the domains through the bulk of
the substrate for improved poling quality.
[0018] Another object of the present invention is to provide an
improved domain grating device that results in high efficiency bulk
domain grating devices for applications in generating high power
visible laser light. This is achieved in a domain grating device
that has a substrate with first and second opposing surfaces. The
substrate has an inverted domain grating structure that extends
through the entire substrate. An inverted domain duty cycle at the
first surface is greater than 50% and less than 100%, and an
inverted domain duty cycle at the second surface is less than 50%
and greater than 0% ensuring a region of 50% duty cycle within the
substrate.
[0019] In another embodiment of the present invention, a method is
provided for creating an improved domain grating device. A
substrate is provided with first and second opposing surfaces.
Optical illumination is used to generate mobile charges and
patterned current flows. An inverted domain grating structure is
formed that extends through the entire substrate. A domain duty
cycle at the first surface is greater than 50% and less than 100%,
and a domain duty cycle at the second surface is less than 50% an
greater than 0%.
BRIEF DESCRIPTION OF THE DRAWINGS
[0020] FIG. 1. illustrates a prior art domain grating device.
[0021] FIG. 2. illustrates one embodiment of a domain grating
device with the tapered domain grating of the present
invention.
[0022] FIG. 3. illustrates a calculation of device conversion
efficiency vs grating duty cycle.
[0023] FIG. 4. illustrates a calculation of device conversion
efficiency as a function of depth for one embodiment of the present
invention.
[0024] FIG. 5 illustrates an optical beam position for optimum
performance for one embodiment of a domain grating device of the
present invention.
[0025] FIG. 6. illustrates an electrode configuration with uniform
illumination through the back face for one embodiment of the
present invention.
[0026] FIG. 7. illustrates the effect of domain inversion mask
misalignment on poling for one embodiment of the present
invention.
[0027] FIG. 8. illustrates a typical experimental configuration for
an illuminated electric field poling process that can be used with
one embodiment of the present invention.
[0028] FIG. 9. illustrates characteristic electrical traces for a
single pulse illuminated poling process for one embodiment of the
present invention.
[0029] FIG. 10. illustrates characteristic electrical traces for a
dual pulse illuminated poling process for one embodiment of the
present invention.
[0030] FIG. 11. illustrates an insulating mask electrode
configuration with illumination through the back face for one
embodiment of the present invention.
[0031] FIG. 12. illustrates an insulating mask electrode
configuration with UV illumination through the front face for one
embodiment of the present invention.
[0032] FIG. 13. illustrates interference UV pattern poling for one
embodiment of the present invention.
[0033] FIG. 14. illustrates the combination of a UV interference
pattern with an insulating mask layer for poling in one embodiment
of the present invention.
[0034] FIG. 15. illustrates E-beam induced poling for one
embodiment of the present invention.
[0035] FIG. 16. illustrates a single pass frequency doubled visible
laser source using a domain grating device for one embodiment of
the present invention.
[0036] FIG. 17. illustrates an intra-cavity frequency doubled
visible laser source using a domain grating device for one
embodiment of the present invention.
[0037] FIG. 18. illustrates a projection laser light display system
for one embodiment of the present invention.
[0038] FIG. 19. illustrates a scanning laser light display system
for one embodiment of the present invention.
[0039] FIG. 20. illustrates a difference frequency generator using
a domain grating device for one embodiment of the present
invention.
DETAILED DESCRIPTION
[0040] In one embodiment of the present invention, an improved
domain inversion structure is provided that has optimized
efficiency, reliable fabrication and ease of characterization. This
improved domain inversion structure is the result of a new high
voltage electric field poling based fabrication process which
involves the generation of mobile charges within the substrate that
is to be poled, accompanied by the application of a patterned high
voltage electric field. This combination results in a patterned
current flow through the substrate and creates a patterned domain
inversion structure within the substrate. Various combinations of
charge generation and voltage application can be used to tailor the
size and shape of the domain inverted regions.
[0041] In one embodiment, a frequency conversion device is
provided. The improved domain inversion structure resulting from
the new fabrication process is shown schematically in FIG. 2. The
inverted domains 202 are tapered in size from one face of the
crystal to the other, and may extend through the entire thickness
of the crystal 206. Domain inverted devices in lithium niobate made
by the prior art fabrication techniques have been unable to provide
this combination of advantageous properties. A typical prior art
device is shown in FIG. 1. Here, although the domain walls extend
through the entire thickness 106 of the crystal 101, they are
substantially parallel to the Z-axis and the dimensions of the
inverted domains 105 are substantially constant through the
material thickness. Whilst, in principle, this constant domain
dimension could be seen as a good property since it allows the
fabrication of high aspect ratio domains, and provides a uniform
grating structure throughout the thickness of the crystal, it is
also exceedingly difficult to fabricate the grating so that the
domains have exactly the optimum dimensions.
[0042] In frequency conversion applications utilizing a first order
QPM grating the optimum efficiency is achieved with a 50/50 duty
cycle between the two anti-parallel domain orientations. The
importance of the domain duty cycle can be seen in FIG. 3, where
the efficiency of a phasematched interaction is plotted as a
function of the duty cycle of the grating.
[0043] When the poling process is controlled by computer based on
the amount of charge that has been transferred onto the crystal (a
measure of the amount of domain inversion that has occurred), if
there are a number of defects in the lithographically patterned
poling mask that cause a number of domains to merge together, the
charge involved in causing the merging will result in less domain
inversion in the remaining pattern, potentially resulting in a
lower than desired duty cycle overall. Since in undoped congruent
lithium niobate the inverted domain is generally uniform in
dimension throughout the bulk of the substrate, a domain grating
that is either over or under duty cycle at the surface as a result
of one of the above parameters will not provide optimum conversion
efficiency. FIG. 1 depicts the conventional grating device 101 from
prior art process described in R8. In this device, the domain
grating is over duty cycle, i.e. the inverted domain 105 is larger
than the un-inverted domain 104 in the grating region. Once the
device is fabricated, the efficiency of the device is fixed and any
difference in duty cycle from the optimum 50/50 results in a
decrease in performance compared to the optimum device.
[0044] In one embodiment of the present invention, a tapered domain
grating device is provided where an ideal first order QPM 50/50
duty cycle is realized inside the crystal for efficient nonlinear
interactions as shown schematically in FIG. 2. The inverted domains
202 are tapered in size from one face of the crystal to the other
and may extend through the entire thickness of the crystal 206.
Even if the domain inversion is not precisely controlled at the
patterned surface, the taper of the domains in the present
invention provides for a 50/50 duty cycle region within the bulk of
the material where the dimensions of the inverted 202 and
uninverted 203 domains are matched. For example, for a grating
period of 4 .mu.m, a substrate thickness of 0.5 mm, an inverted
domain 202 dimension of 3 .mu.m on the top (patterned) face and 1
.mu.m on the bottom, there will be a 120 um thick region within the
crystal where the duty cycle ranges from 47% to 53%, for which the
efficiency is 98% of the theoretical maximum, as indicated in FIG.
4. Even when a more strongly tapered domain merges on the front
face, FIG. 4 illustrates that as long as the domain is under duty
cycle on the back face (less than 50/50), there still a region
inside the material where the 50/50 duty cycle is found. Thus, an
advantage of the improved domain inversion structure of the present
invention is that even if the duty cycle of the domain inversion
structure is not precisely controlled to be 50/50 at the patterned
surface, the taper of the domains provides for a 50/50 duty cycle
region within the bulk of the material where the dimensions of the
inverted 202 and uninverted 203 domains are matched.
[0045] When the device of FIG. 2 is used as a frequency conversion
element, adjustment of the optical beam position with respect to
the depth of the crystal will result in an optimally high
efficiency nonlinear interaction as indicated in FIG. 5., wherein
the beam 502 is at the position having the optimum efficiency. This
post-fabrication efficiency optimization is not possible for
devices produced from prior art processes with vertical domain
walls. Since the exact duty cycle of the domain grating on the
masked face of the device of FIG. 2 is not critical as long as it
is greater than 50/50 and the domain taper angle is controlled such
that the duty cycle on the unmasked face is less than 50/50, the
degree of control required over the poling process is relaxed
compared to that required to fabricate an exact 50/50 duty cycle
grating with vertical walls. Thus, the tapered domain structure
greatly increases the poling process latitude in terms of exact
current and charge control in a mass production process, while
preserving the ability to provide optimum device efficiency. The
trade-off in this case is that the optimum duty cycle region of the
crystal is necessarily limited by the domain taper angle. Therefore
the uniformity of the device of FIG. 2 in the vertical direction
through the crystal substrate will generally be worse than that of
the device of FIG. 1 with vertical domain walls, but the peak
conversion efficiency will generally be higher.
[0046] Highly efficient domain gratings can be achieved utilizing
this device structure, for example an effective nonlinear
coefficient of greater than 16 pm/V has been achieved for blue
frequency doubling with a grating period of 4.45 um.
[0047] In one embodiment of the present invention, fabrication
methods are provided that involve the generation of mobile charges
within the substrate that is to be poled, accompanied by the
application of a patterned high voltage electric field. This
combination results in a patterned current flow through the
substrate and creates a patterned domain inversion structure within
the substrate. Various combinations of charge generation, voltage
application and patterning mask can be used to tailor the size and
shape of the domain inverted regions.
[0048] One basic approach of the invention is shown in FIG. 6 and
combines a voltage applied between a patterned electrode 602 on one
face of the substrate and a uniform transparent electrode 605 on
the other and a transversely spatially uniform light illumination
606 incident to the said substrate through the said transparent
electrode. The transversely spatially uniform light illumination
creates photo-induced charges within the bulk of the substrate with
a density and distribution dependent on the illumination intensity,
illumination wavelength(s) and the absorption of the crystal
substrate. The applied high voltage electric field causes a
photo-current to flow through the substrate due to the presence of
the photo-excited charges. The patterned nature of the electric
field, or in other terms, the patterned nature of the photocurrent
flow on one face of the crystal substrate, and the mobile charge
profile in depth resulting from the material absorption define the
pattern of the resulting domain inverted features. The patterned
electrode in FIG. 6 is depicted as a patterned metal layer 602, but
it may also take the form of a patterned insulating layer overlaid
with a conductor such as a metal or a liquid. In this case, the
patterned insulating layer preferably acts as a current mask to
pattern the flow of the photocurrent in response to the
illumination and the applied field. It may also provide a
modulation of the electric field strength within the crystal
substrate. The mobile charges which are generated in the bulk of
the material by the spatially uniform illumination and form the
photocurrent in response to the applied electric field are
constrained to flow through the patterned mask apertures, creating
a large current density at the apertures which provides a strong
effect in lowering the coercive field and enabling the controlled
seeding of patterned domain inversion. The present inventors have
observed that the effective coercive field of the crystal substrate
is generally decreased as the magnitude of the photocurrent is
increased, for instance by increasing the intensity of the
illumination, or by altering the spectrum of the illumination.
[0049] Here, transversely spatially uniform light illumination
means that the incident illumination is substantially uniform in
intensity and spectrum across the transverse dimensions of the
illuminated portion of the substrate surface, which is generally
equivalent to the poling area. This uniformity is highly desirable
in order to achieve a uniform spatial domain inversion pattern over
the entire poling area. It should be noted that the illumination
will not be uniform in the direction of the light propagation in
the bulk of the material since absorption will decrease the
intensity and change the spectral mix of the illumination as the
light propagates into the material. The transverse uniformity of
the domain inversion pattern should be maintained as long as the
properties of the bulk substrate are uniform over the transverse
dimensions of the illuminated region. The phrase "uniform
illumination" should be taken to refer to illumination which is
substantially uniform over the transverse dimensions of the
illuminated region. It should also be noted that the spatially
uniform light illumination does not have to be spatially uniform at
all times. That is, the time averaged intensity and/or spectral
content over a time period considerably shorter than the total
illumination time should be substantially uniform, but the
instantaneous intensity may vary across the illumination area. The
requirement for illumination uniformity is based on a requirement
for substantially uniform mobile charge generation across the
illuminated region so that the electric field poling process is
initiated at substantially the same point in time across the entire
aperture, and is able to proceed at substantially the same rate
over the illuminated area. In the presence of a material
non-uniformity, such as a dopant concentration nonuniformity, the
illumination uniformity can be tailored to result in uniform mobile
charge generation in the substrate, that is, the illumination
intensity can be varied transversely across the substrate such that
even in the presence of the dopant concentration non-uniformity,
the mobile charge generation remains substantially uniform and the
resulting domain inversion pattern is also uniform. The applied
electric field should also preferably be substantially uniformly
applied across the illuminated area in order to provide uniform
domain inversion, i.e. there should be no significant monotonic
field variation across the transverse dimension of the substrate,
although localized modulation of the electric field is desired for
optimum poling. In the case of the material non-uniformity
described above, the strength of the applied electric field could
be varied across the transverse dimensions of the poling area
instead of the illumination uniformity, in order to counteract the
material uniformity. Alternatively, both the electric field and the
illumination may vary across the poling area in such a way that
their combination produces a uniform poling pattern, with or
without the presence of a material non-uniformity.
[0050] The distributed nature of the photo-charges within the
material and the patterned photo-current flow effectively define
the domain inversion pattern throughout the bulk of the crystal
substrate, leading to very high aspect ratio domain inversion
features and very high quality short period domain inversion
gratings. The current masking windows in the patterned electrode on
one of the surfaces of the substrate, and the higher conductivity
of the seeded domain walls growing in the bulk of the substrate
enable a patterned current flow through the bulk of the crystal due
to the photo-excited charges. These "patterned" moving charges can
be envisioned as a virtual electrode in the bulk of the
substrate.
[0051] Controlling the combination of applied voltage, illumination
intensity, illumination wavelength(s), illumination time,
photocurrent/charge and poling charge, enables control to be
exercised over the size and shape of the resulting domain inverted
features. It has been shown by the present inventors that with the
appropriate illumination and voltage parameters (as described in
the following preferred embodiments) it is possible to create a
uniform short period (<4 micron) domain inversion grating in
even 1 mm thick MgO:CLN crystal substrates. The domain inverted
features can propagate through the entire thickness of the
substrate and can be observed by HF etching on both the front and
back faces of the substrate. Thus, domains with aspect ratios of
greater than 250:1 can be fabricated uniformly over large areas
limited only by the uniformity of the optical illumination,
electric field application, and the uniformity of the substrate
material itself.
[0052] Highly efficient frequency conversion devices can be
fabricated using the new fabrication process described here. For
example an effective nonlinear coefficient of greater than 16 pm/V
has been achieved for blue frequency doubling with a grating period
of 4.45 um.
[0053] In one embodiment of the present invention, the combination
of patterned electric field and uniform illumination minimizes
merging and provides improved uniformity and repeatability of
domain inversion as compared to a patterned electric field alone.
In addition, the peak high voltage that must be applied in
conjunction with the uniform illumination is significantly lower
that that required for uniform domain seeding at short periods
without illumination, which substantially eliminates the
possibility of destructive electrical breakdown of the crystal
substrate during poling. This is especially important when
considering scaling the electric field poling process to full wafer
areas, and enables a far more robust process for high yield, high
volume manufacturing that that of the application of a high voltage
alone.
[0054] It has also been determined by the present inventors that
the illuminated poling process of the present invention is more
resistive to the deleterious effects of thermally induced domain
inversion defects. These thermally induced domain inversion
defects, often called "heat defects", result from the lithographic
processing of Z-cut lithium niobate wafers. The heating and cooling
cycles during the lithography process lead to the build up of
pyroelectrically generated charges on the wafer surface, which can
lead to the spontaneous domain inversion of small defect-seeded
regions. Thus, a wafer that was uni-domain at the start of the
lithographic process may end up with a large number of small,
isolated domain inverted regions within it by the end of the
process. During the electric field poling process these domains
cause merging of the desired pattern and tend to grow to the
detriment of the desired domain pattern, leading to reduced domain
inversion pattern quality. The present illuminated poling invention
has been found to resist the effects of these heat defects and
suppress the tendency for the formation of large merged regions
around each defect site during the poling process. Thus the desired
domain inversion pattern can still be uniformly seeded and grown
with good pattern quality, even when the poled area is increased to
allow for wafer scale processing for high volume manufacturing.
[0055] It should be noted that others have attempted to use the
formation of such "heat defects" as a method of seeding the domain
inversion process in MgO:CLN to create uniform domain inversion
gratings (Nakamura et al, "Periodic poling of magnesium oxide doped
lithium niobate", Jnl. Appl, Phys, Vol 91, No 7, 2002, p 4528-34).
However, the random nature of the "heat defect" locations makes
this impractical for short period gratings. and with the
application of a high voltage pulse alone, the preferential
expansion of the "heat defects" during the poling pulse causes
significant merging and loss of quality in the domain grating.
[0056] The size and shape of the inverted domain feature can be
controlled by setting appropriate values for the illumination and
voltage applied to the substrate. In particular it has been found
that varying the illumination spectrum can be used to control the
degree of penetration of the inverted domain into and through the
crystal substrate for given set of poling parameters. With the
appropriate illumination spectrum, which contains some optical
power at wavelengths below about 320 nm, it is possible to
terminate the domains before the illuminated face (the unpatterned
face) of the crystal substrate. If the very short wavelength light
is removed from the illumination spectrum, using an absorption
filter for example, full penetration of the domains through the
substrate can be readily achieved. Even with the very short
wavelengths removed from the spectrum, it is still possible to
terminate the domains within the bulk of the substrate by
decreasing the illumination time and/or the illumination intensity
compared to the values used to produce a fully penetrated domain
inversion pattern.
[0057] In another embodiment of the present invention, domain
inversion of 0.5 mm thick .about.5% MgO doped congruent lithium
niobate is achieved using a patterned metal electrode on one face
and a transparent planar electrode on the other, combined with
UV/visible illumination through the transparent electrode. As shown
in FIG. 6, a patterned metal electrode 602 is disposed on a first
face (generally termed the "front" face) of the crystal substrate
601, preferably the +z face, using standard photolithographic
techniques as follows. Firstly the wafer surface is chemically
cleaned using BOE, acetone and IPA in sequence. An oxygen plasma
ashing process is used to ensure the removal of any remaining
hydrocarbon contamination. A metal layer, for example tantalum, is
deposited on the cleaned crystal surface using a sputter or e-beam
deposition process. Care is taken to choose the suitable metal to
use on the suitable surface of a given ferroelectric material. A
layer of photoresist, e.g. Shipley 3312 or AZ 5214 is spun onto the
metal layer and soft-baked to remove the excess solvent. The
photoresist is exposed using standard lithographic techniques
(contact or projection lithography) and developed into the pattern
of the desired electrode. The metal is than etched either using wet
chemical etching or plasma etching to form the patterned electrode
602. Alternatively the metal electrode fabrication can be performed
using a lift-off process where the photoresist is patterned on the
substrate first, and then metal is then deposited using e.g. an
e-beam deposition technique. The metal that is deposited over the
top of the resist is removed (or "lifted off") using solvent
typically with ultrasonic agitation, while the metal deposited on
the substrate surface remains. After the metal electrode is
patterned, an insulating dielectric layer 607 is preferably
deposited to cover the metal electrode. This dielectric layer
should have good electrical insulating properties especially at the
contact interface between the dielectric layer and the substrate
surface. While the patterned metal lines provide the electric field
modulation and the photo-current paths, the insulating layer
between the lines provides a current masking capability to limit
the photo-current to flow only to the metal lines. A photoresist
layer e.g. Shiply 3312 or AZ 4210, can serve as the insulating
layer after it is cross linked (hard baked) at a suitable
temperature, e.g. 140-200.degree. C. depending on the type of
resist. Alternatively an SiO.sub.2 layer deposited by sputtering,
evaporation or spin-on-glass can form the insulator. Electrical
contact holes 608 to the metal electrode through the insulating
layer are also provided by well known lithography processes.
[0058] With regard to the photolithographically patterned mask used
to define the domain inversion pattern, it has been found by the
present inventors that a long line feature is not suitable for
poling of short grating periods due to the fact that domain
structure tends to follow the hexagonal crystal structure, where
one of the sides of the hexagon is oriented along the y-axis of the
crystal in LiNbO3. When the line feature in the poling mask is
defined at an angle with respect to the y-axis, the resulting poled
domain generally either displays a jagged edge or expands outwards
from the mask feature to form an elongated hexagonal domain, as
illustrated in FIG. 7.
[0059] The domain expands sideways away from the patterned mask
feature 705 until the edges of the domain 706 are aligned with the
crystal y-axis 704. Thus, the width of the domain is no longer
defined by the width of the photolithographically defined mask, but
rather it is defined by the effective width 702 of the lithographic
feature perpendicular to the y-axis. Hence, the greater the angle
703 of the feature with respect to the y-axis, the greater the
width the poled domain becomes before the sides of the domain are
parallel to the y-axis.
[0060] For optimal poling quality of fine pitch gratings the poling
mask should be aligned so that the grating bars are exactly
parallel to the y-axis. However, there are a number of difficulties
involved in this alignment. Firstly, there is typically a tolerance
of .+-.0.25 to .+-.0.5 degrees in the angular accuracy of the
orientation flat provided by the wafer manufacturer. Secondly, the
beveling process applied to the edges of the wafer to remove edge
chips and prevent wafer breakage during processing often leads to a
slight curvature of the orientation flat, further reducing the
effective accuracy. Thirdly, some angular error will be introduced
when the poling mask is physically aligned to the orientation flat
due to the resolution of the mask aligner and the finite length of
the orientation flat at the edge of the wafer.
[0061] It can be seen from the domain expansion argument above that
for a given period of grating 712 in FIG. 7b. there is a maximum
allowable misalignment angle 710 for a particular length of feature
714. Basically, when the effective width 711 of the feature
perpendicular to the y-axis becomes equal to or greater than half
the grating period 712, there is a significantly increased
likelihood that adjacent domains will merge together and the
grating structure will be lost.
[0062] Thus, for a robust production domain inversion process we
can define a maximum line length in the poling mask, such that the
achievable angular alignment accuracy does not cause the domain
grating pattern to merge. For example, if the alignment accuracy is
.+-.0.5 degrees and the grating period is 4 .mu.m, the maximum
feature length that can be allowed is L=2 .mu.m/sin(0.5
degrees)=.about.230 .mu.m. In this case, a number of line features
are patterned on the mask separated by a small distance, to make up
the full width of the desired poling region.
[0063] In the case of the patterned metal poling mask used in this
embodiment of the invention, the mask features can effectively be
broken into bars of the desired length simply by depositing and
patterning an insulating layer on the surface of the crystal
substrate before the deposition and patterning of the metal mask
layer. The patterned insulating layer should consist of a series of
lines disposed substantially perpendicular to the desired grating
bars, and spaced apart by the desired bar length. The metal layer
may then be deposited over the top of the insulator and then
patterned to provide the grating lines. Where the patterned
insulator is interposed between the metal layer and the crystal
substrate the required voltage to achieve domain inversion will be
increased, effectively preventing domain inversion from occurring,
and hence breaking the patterned metal poling mask into a number of
bars of the desired length. In other embodiments of this invention
a patterned insulating layer is used to provide the poling mask. In
this case the poling bars are defined simply by the length of the
openings in the photomask that are transferred into the insulating
mask and no extra processing is required.
[0064] To improve the angular alignment accuracy it is possible to
provide domain inverted alignment features which more precisely
define the crystal y-axis direction. An initial poling pattern
consisting of a few narrow bars parallel to the y-axis is aligned
to the wafer orientation flat. The pattern is then poled into the
crystal and the domains allowed to expand out from the mask pattern
so that their edges are parallel to the y-axis. The poling mask may
then be removed and the crystal surface etched in HF to reveal the
poled features. (Preferably, only the area immediately surrounding
the poled features is exposed to the HF to avoid possible damage of
the surface still to be poled.) A second poling mask consisting of
the desired grating pattern is then aligned using the poled
features to define the crystal y-axis, thus achieving improved
accuracy between the grating lines and the crystal axis. This two
step process should allow longer individual lines to be poled than
are generally possible with single step alignment to the wafer
orientation flat.
[0065] During the lithography process, there are many thermal
processes such as resist baking. It is preferable to control the
thermal ramp rates of the crystal substrate/wafer during these
baking processes, and also preferable to provide some form of
discharge path for the pyrolectrically generated charges. MgO:CLN
is very prone to the generation of "heat defects", regions of
domain inversion created as a result of pyroelectric charge
accumulation on the wafer surface during heating and cooling
cycles. In general the "heat defect" domain inversion sites are
problematical for the fabrication of high quality short period
domain inversion gratings since they tend to lead to merges between
adjacent domains and defects, reducing the quality of the grating
and the efficiency of any QPM optical frequency conversion process
using the grating.
[0066] Despite the observation that the illuminated electric field
poling process of the present invention is significantly more
tolerant of, or resistant to, the deleterious effects of "heat
defects" than the prior art electric field poling process, it is
still preferable to minimize the number of defects that are formed
in order to maximize the quality of the final domain inversion
grating.
[0067] In the poling process, the electrical contact to the metal
electrode may be made by a probe contact 610. A transparent
electrode 605 on the back face is created by, for example, a
solution of lithium chloride in de-ionized water. The liquid can be
confined using an `o` ring 604 with a quartz cover plate 611, or
simple a tape cut out or silicone gel or grease barrier.
[0068] A typical experimental setup for the poling is shown in the
FIG. 8. A computer 805 controls a high voltage pulsed power supply
806 and a UV/visible light source 804. The electrical contacts 808
to the crystal 801 are connected to the high voltage power supply,
and a light sensor 807 is preferably incorporated into the circuit
to provide an optical monitor for the computer control. Preferably,
the PC controls the voltage supply, the light source shutter and
the timing sequence. Alternatively some features such as the
illumination time may be controlled by independent timers and the
optical monitor used to provide process sequencing via the PC. The
computer can also preferably monitor the current in the poling
circuit, using for example an optically isolated current monitor
809, to provide control of the charge delivered in the various
phases of the process.
[0069] The light output from the light source (in this instance
coupled via a light-pipe) 803 is arranged to provide sufficient
illumination intensity and uniformity across the electrical contact
area. Typical intensities of .about.10 W/cm.sup.2 at the output of
the light-pipe (broadband, all wavelengths from a high pressure
mercury bulb) are used. A beam shaping/expanding system 802 can be
used to increase the illumination beam diameter and/or uniformity
on the substrate. Typically about 0.5 W/cm.sup.2 of total light
intensity is incident at the surface the substrate. Higher and
lower intensities may be used with the appropriate adjustments in
illumination time and applied voltage to achieve domain
inversion.
[0070] A typical sequence of voltage and UV light poling is shown
in FIG. 9. as follows: an initial voltage of about 2500V
(.about.5000V/mm) 901 is applied to the substrate in the absence of
illumination, ramping up from zero volts in about 60 ms. At this
time no poling occurs because the voltage is significantly below
the coercive field required to achieve domain inversion, and
therefore no current flows in the poling circuit. The shutter of
the illumination source is then opened, illuminating the
unpatterned face of the wafer with both visible and ultra violet
wavelengths 902 (a simple broadband mercury lamp source is used in
this example. It is also possible to use a combination of one or
more narrowband light sources to achieve the same effect, as long
as the wavelength(s) and intensity(ies) are chosen so as to produce
a similar quantity and distribution of photo-excited charges within
the crystal substrate.) After the illumination begins a
photo-induced current 904 starts to flow through the substrate
under the influence of the high voltage applied across it. This
current generally increases gradually over a time frame of 100 ths
to 10 ths of seconds, and then tends to reach a plateau. Once the
photo-current has increased to a sufficient value, which is
determined largely by the area of exposure and the applied voltage,
domain inversion occurs within the crystal substrate despite the
applied electric-field being nominally below the coercive field.
This domain inversion appears to be seeded from the patterned face
and is thought to result at least in part from the effects of the
concentration of the photo-induced current flow in the small
features of metal electrode. It is thought that domain inversion
occurs even though the applied voltage is significantly below the
nominal coercive field of the material at least in part because the
mobile charges generated in the substrate by the illumination
decrease the effective coercive field of the illuminated
material.
[0071] The observed current flow 903 is now composed of two
components, the photocurrent 904 due to the charges generated by
the illumination, and the poling current 905, the displacement
current due to the domain inversion. Typically the photocurrent
remains substantially constant after its initial growth period,
whereas the poling current 905 typically increases to a maximum and
then decreases again as the poling is completed. Thus the poling
process can be controlled by monitoring the current flow and
terminating the applied voltage 901 when either the current 903 or
transferred charge reaches some predetermined value (which is also
dependent on the magnitude of the photo-current).
[0072] After poling, the insulating layer over the metal electrode
is stripped off and the metal electrode is etched off. The
substrate is then etched in hydro-fluoric acid (HF) to reveal the
poled pattern. It has been found that in general the inverted
domains resulting from the above described process are tapered,
with a wider line width on the patterned electrode face, and a
narrower width on the face with the transparent uniform electrode.
The width on the patterned face and depth into the substrate both
generally increase with increasing voltage, increasing illumination
intensity and increasing illumination time.
[0073] It has been observed with some combinations of illumination
spectrum, illumination dose (i.e. light intensity.times.time) and
applied voltage, that the inverted domains are terminated inside
the bulk of the crystal and do not generally reach the uniformly
illuminated face. It has been found that the inclusion of short
wavelength UV radiation around or below the band gap (.about.320 nm
in MgO:CLN) in the illumination spectrum has the effect of
terminating the domains in the bulk of the crystal.
[0074] In general it can be desirable that the domains penetrate
completely through the crystal substrate for optimum device
performance and for ease of device characterization, so it is
preferable to filter the illumination to remove the shortest
wavelengths. A dichroic or absorptive filter may be used to provide
selectivity in the wavelengths that are removed.
[0075] After etching of the top and bottom faces of the crystal to
reveal the domain inversion patterns, the quality of the domain
inversion grating device can be estimated using FIG. 4. The wafer
substrate containing the domain inversion gratings may be diced to
separate the individual gratings, which may have different periods
corresponding to different patterns on the photolithographic mask.
The end faces of each device may then be optically polished and,
preferably, coating with anti-reflection coatings, ready for use as
a quasi-phasematched frequency conversion device as shown in FIG.
5. The position of the optical beam within the crystal can be
adjusted in depth to utilize the optimum 50/50 duty cycle region of
the grating which results from the tapered domain structure.
[0076] After poling, different portions of the crystal have
opposite domain orientations. There is a resulting crystal
discontinuity at the boundary between the opposite polarity
domains. At this boundary, a refractive index pattern can be
observed using transmission illumination and crossed polarizers, or
a Nomarski microscope. This refractive index pattern may be the
result of the uncompensated charges at the boundary, causing a
refractive index change via the electro-optic effect, or from
stress at the boundary via the elasto-optic effect. This refractive
index pattern becomes less pronounced after the sample is exposed
to UV or short wavelength visible illumination, thermal annealing
or simply left at room temperature for some extended period of
time.
[0077] In order to effectively use the periodically poled (domain
inverted) frequency conversion device for the generation of visible
light, the discontinuity of the crystal at the domain wall boundary
needs to be addressed carefully. The boundary and the associated
refractive index change can act as an extra scattering source,
increasing the optical loss in the device. In addition, new
phenomena such as green induced IR absorption (GRIIRA) and Blue
induced IR absorption (BLIIRA) are associated with this boundary
structure, and the defects introduced by the domain boundary.
[0078] In one embodiment of the present invention, to alleviate the
effects of the boundary defect structure on the visible light
generation process, a high temperature annealing process is used. A
discharging closed loop is formed by placing the domain inverted
sample between two semi-conductive silicon wafers which are
electrically connected to dissipate pyroelectric charges. The
sample stack is then placed into a high temperature oven or
furnace, typically in an ambient air atmosphere, although
alternative oxidizing and reducing atmospheres of, for instance,
oxygen and argon respectively may be preferred for some
applications. The temperature of the furnace is raised slowly from
room temperature up to typically between 500 C and 600 C in about 5
hours. The samples are left at this temperature for a relatively
long period, typically around 48 hours, before being cooled down to
room temperature. Preferably the cooling is performed at a slow
rate of a few degrees centigrade per minute, preferably as low as
0.5 C/min. The electrically shorted high temperature annealing
process significantly improves the performance of the visible
frequency conversion device, especially for short wavelengths in
the blue spectrum, by reducing the boundary defect density,
uncompensated bonds and charges, and stresses at the domain
boundaries. For short period frequency conversion devices for
visible applications care must be taken not to significantly reduce
the material, i.e. to use an atmosphere containing at least some
oxygen. It is also necessary to maintain the annealing temperature
below the threshold which causes domain boundary motion and domain
merging. In MgO:CLN, this domain boundary motion is typically
observed in short period domain inversion structures at
temperatures in excess of about 650 C, indicating that annealing
temperatures are preferably below this value.
[0079] As noted above, the exposure of the domain inverted sample
to UV and visible radiation appears to reduce the magnitude of the
refractive index change at the domain wall boundary. Therefore, it
may be advantageous to illuminate the domain inverted sample with
UV and or visible light during the high temperature annealing
process. In this instance, transparent conducting material is
preferably used for the discharging loop, e.g. Indium Tin Oxide
(ITO) coated quartz, to enable simultaneous illumination and
pyroelectric charge dissipation.
[0080] In another embodiment of the present invention, the domain
inverted device is partially coated with a conductive layer.
Preferably this layer provides a conductive path linking the front
and back opposing surfaces of the domain inverted device. The
conducting layer may be deposited before or after annealing and
dicing of the domain inverted device. If the layer is deposited
before dicing, the conductive path may be completed after dicing by
for instance painting the side face of the device with conductive
silver paint which spills slightly over onto the front and back
faces. The conducting layer enables the dissipation of thermally
excited charges--pyroelectricity, and also enables the dissipation
of photocharges that drift to the edge of the crystal, where they
are no longer trapped. Thus, the conductive path over parts of at
least three faces of the domain inverted device offers the prospect
of decreasing the beam distortion and performance limiting effects
of any residual photorefractivity still present in the domain
inverted device.
[0081] In a previous embodiment of the present invention, the
domain structure is fabricated in MgO:CLN by using a simple single
applied voltage combined with illumination, as described above,
generally has a significant taper from front (patterned electrode
face) to back (uniform illumination and electrode face) surfaces.
The domain features on the uniform illumination/electrode (back)
face are generally very narrow. Thus, the optimum conversion
efficiency region, illustrated in FIG. 4 is relatively narrow due
to the strong domain taper.
[0082] In another embodiment of the present invention, the taper of
the domain is controlled in order to increase the dimension of the
domain on the back face whilst maintaining good domain quality on
the front face and increasing the size of the optimum conversion
efficiency region. This can be achieved by applying a voltage pulse
or series of pulses to the crystal after the illumination is
removed.
[0083] The substrate can be prepared as shown in FIG. 6. and as
described in the previous preferred embodiment. For the poling
process, the electrical contact to the metal electrode is made by
probe contacts 610. A transparent electrode 605 on the back face is
created by, for example, a solution of lithium chloride in
de-ionized water. The liquid can be confined using an `o` ring 604
with a quartz cover plate 611, or simple a tape cut out or silicone
gel or grease barrier.
[0084] A typical experimental setup for the poling is shown in FIG.
8. A computer 805 controls a high voltage pulsed power supply 806
and the UV/visible light source 804. The electrical contacts 808 to
the crystal 801 are connected to the high voltage power supply, and
a current sensor 809 is preferably incorporated into the circuit.
Preferably the poling system is controlled by a PC which can
capture the voltage, current and charge flow data in real time,
allowing different voltages to be sequenced or triggered or shut
down based on time, current flow, or charge transfer values, or any
combination of these. Again preferably a photodiode or optical
monitor 807 is incorporated into the poling fixture in order to
monitor the illumination source so that the computer control
program can also sequence the required illumination exposure.
Preferably, the PC controls the voltage supply, light source
shutter and timing sequence. Alternatively some features such as
the illumination time may be controlled by independent timers and
the optical monitor used to provide process sequencing via the
PC.
[0085] The light output from the light source (in this instance
coupled via a light-pipe) 803 is arranged to provide sufficient
illumination intensity and uniformity across the electrical contact
area. Typical intensities of .about.10 W/cm.sup.2 at the output of
the light pipe (broadband, all wavelengths from a high pressure
mercury bulb) are used. A beam shaping/expanding system 802 can be
used to increase the illumination beam diameter and/or uniformity
on the substrate. Typically about 0.5 W/cm.sup.2 of total light
intensity is incident at the surface the substrate. Higher and
lower intensities may be used with the appropriate adjustments in
illumination time and applied voltage to achieve domain
inversion.
[0086] A typical sequence of voltage and UV light poling is shown
in FIG. 10. An initial voltage of about 2000V (.about.4000V/mm)
1001 is applied to the substrate in the absence of illumination,
ramping up from zero volts in about 60 ms. At this time no poling
occurs because the voltage is significantly below the coercive
field required to achieve domain inversion, and therefore no
current flows in the poling circuit. The shutter of the
illumination source is then opened, illuminating the unpatterned
face of the sample with both visible and ultra violet wavelengths
1010 (a simple broadband mercury lamp source is used in this
example. It is also possible to use a combination of one or more
narrowband light sources to achieve the same effect, as long as the
wavelength(s) and intensity(ies) are chosen so as to produce a
similar quantity and distribution of photo-excited charges within
the crystal substrate.) After the illumination begins a
photo-induced current 1020 starts to flow through the substrate
under the influence of the high voltage applied across it. This
current generally increases gradually over a time frame of 100 ths
to 10 ths of seconds, and then tends to reach a plateau. Once the
photo-current has increased to a sufficient value, which is
determined largely by the area of exposure and the applied voltage,
domain inversion can occur within the crystal substrate despite the
applied electric-field being nominally below the coercive field.
This domain inversion appears to be seeded from the patterned face
and is thought to result at least in part from the effects of the
concentration of the photo-induced current flow in the small
features of the patterned metal electrode. It is thought that
domain inversion occurs even though the applied voltage is
significantly below the nominal coercive field of the material at
least in part because the mobile charges generated in the substrate
by the illumination decrease the effective coercive field of the
illuminated material.
[0087] The function of this first illuminated voltage pulse is to
seed or initiate the domain inversion, so the illumination is
terminated before the poling is complete. This termination can be
based on an empirically determined time or a charge flow monitored
by the computer, at which point the light source shutter is closed
and the illumination is blocked. Typical values for this first
pulse are a duration of .about.0.5 to 1 sec, and a charge flow of
0.02 to 0.12 mC/cm.sup.2 at a voltage of .about.2000V (4000V/mm)
and an illumination intensity of .about.0.5 W/cm.sup.2.
[0088] Once the initiation of the domain inversion in the
illuminated voltage pulse is performed, the illumination light is
shut off, and, using the optical monitor 807 for sequencing
control, the computer applies the second voltage pulse 1002.
Preferably, this post illumination voltage is higher in magnitude
than that used during the illumination pulse, since there are no
photo-excited charges being generated to decrease the coercive
field of the material. Typically a voltage of around 3-4000V
(6-8000V per mm) may be applied post illumination. In this
un-illuminated voltage pulse, the poling current 1021 typically
increases to a well defined peak 1022, and then decreases to a
plateau value 1023. The decrease of the poling current is related
to the completion of the domain inversion. If the voltage is
removed while the current is at the peak, the poling pattern will
typically be under duty cycle and some domain features will be
incomplete. If the voltage is maintained until the poling current
has decreased to its plateau value the domain pattern will
typically be complete, with a duty cycle on the front (patterned)
face of the crystal that is dependent on the parameters of the
illumination and illumination voltage pulse. Maintaining the
voltage for a significant length of time after the current has
decreased to its plateau value typically leads to over duty cycle
domains and a larger number of merges within the domain inversion
pattern. The post-illumination voltage pulse may be controlled
using the computer control program based on either the charge flow
1030 within the circuit or the value and gradient of the poling
current 1021 or on a combination of both. Thus poling may be
terminated when a particular charge has been transferred, when the
current has fallen to a particular value, when the rate of decrease
of the current reaches a certain value or any combination of these
(and other) parameters.
[0089] After electric field poling, the insulating layer over the
metal electrode is stripped off and the metal electrode is etched
off. The substrate may then be etched in HF to reveal the poled
pattern. The inverted domains are generally observed on the back
face of the substrate. Tailoring of the dose of illumination, the
voltage applied when the illumination is applied, the
post-illumination voltage and pulse duration, etc, can be used to
adjust the duty cycle of the domain grating and the taper angle of
the domain from front surface to the back surface.
[0090] After etching of the top and bottom faces of the crystal to
reveal the domain inversion patterns, the quality of the domain
inversion grating device can be estimated using FIG. 4. The wafer
substrate containing the domain inversion gratings may be diced to
separate the individual gratings, which may have different periods
corresponding to different patterns on the photolithographic mask.
The end faces of each device may then be optically polished and,
preferably, coating with anti-reflection coatings, ready for use as
a quasi-phase-matched frequency conversion device as shown in FIG.
5. The position of the optical beam within the crystal can be
adjusted in depth to utilize the optimum 50/50 duty cycle region of
the grating which results from the tapered domain structure. In
this embodiment, the domain taper angle is reduced compared to that
of the process of the second embodiment. This means that there is
less dimension variation in the domains from the top face to the
bottom face of the crystal. This results in a wider optimal
efficiency region within the crystal (i.e. a greater depth range
over which the duty cycle is within some percentage of 50/50), but
requires more control to be exercised over the dimension of the
domain on the top face to ensure that the 50/50 duty cycle region
is centrally located within the crystal.
[0091] Optimization of the domain grating quality, duty cycle and
taper does not have to be limited to the simple sequence of one
illuminated voltage pulse followed by a second higher voltage
pulse. Any sequence of illuminated and un-illuminated voltage
pulses may be used in any order to provide the required poling
charge to realize the desired domain inversion pattern in the
substrate independent of the presence of any photo-current due to
the illumination.
[0092] Voltage pulses can be simultaneous with illumination pulses,
voltage pulses can precede or follow illumination pulses, voltage
pulses can be longer or shorter than illumination pulses. Time
delays may be applied between the termination of one illumination
or voltage pulse and the application of the next. In addition,
different illumination spectra (light wavelengths) may be used in
different illumination pulses with any combination of different
applied voltages.
[0093] The metal electrode may also be patterned on the -z face,
depending on the type of substrate. In general, adjustments of the
pulse parameters (such as the direction of the applied
illumination, and the magnitude and sequence of the illumination
and applied voltages) compared to those used for a +z face
patterned crystal will be required to achieve optimal domain
inversion patterns.
[0094] Because different wavelengths are absorbed in the material
at different depths, it is possible to use a time-varying
illumination wavelength to produce a variation with time in the
depth at which charges are generated within the crystal. In
particular, a rotating circular filter where different cut off
wavelengths are coated along the circular path may be used to
change the illumination wavelength with time during the voltage
pulse. A suitable profile of illumination wavelength versus time,
and therefore of charge generation depth, will help guide the
domain growth through the bulk of the crystal from the patterned
face to the un-patterned face.
[0095] Alternatively, a series of fixed wavelength filters may be
stepped across the illumination beam in turn to alter the
wavelength spectrum incident on the crystal substrate. Preferably,
the time taken to introduce or remove the filter from the beam
should be short in comparison to the total illumination time so
that the transition of the edge of the filter across the beam does
not affect the illumination uniformity significantly.
[0096] In another embodiment of the present invention, a dielectric
current mask with a liquid contact electrode is used. As shown in
FIG. 11 for periodic domain inversion in 0.5 mm thick .about.5% MgO
doped congruent lithium niobate 1101, a patterned insulating mask
1102 may be used as described in R8, and R9. Preferably the mask is
applied to the -Z face of the said crystal (although the +Z face
can also be used) and consists of a layer of photoresist some 2-4
microns thick (e.g. Shiply 3312 or AZ 4210). After spinning onto a
clean MgO:CLN wafer and softbaking (.about.90 C, 30 minutes) the
photoresist is exposed using standard photolithographic techniques
(contact or projection lithography) and developed to produce the
pattern desired for the domain inversion grating. After ensuring
the removal of all photo resist residue from the pattern openings
the resist layer 1102 is hardbaked, preferably at a temperature of
around 120.degree. C. or higher. The hard bake temperature is
chosen as a trade-off between crosslinking of the photoresist and
slumping or distortion of the photoresist pattern during the bake
process, which is undesirable for the subsequent electric field
poling process. It should be noted that different bake times and
temperatures will be applicable to different resist formulations
and thicknesses and different patterns, and should generally be
chosen to provide a robust and substantially electrically
insulating layer on the surface of the crystal wafer.
[0097] Electrical contact to the crystal surface during the poling
process is made using a conductive liquid, e.g. a solution of
lithium chloride in de-ionized water. The liquid conductor 1103 is
preferably applied to the patterned face first and may be confined
to the desired contact area using an o-ring and a quartz cover
plate or a simple tape cut-out. Restricting the contact area of the
liquid is preferable in order to ensure the uniformity of the
poling process. UV/visible illumination 1104 is incident from the
unpatterned (back) face of the crystal substrate. The dimensions of
the poling area where the liquid contact is made should preferably
match or be less than the dimensions of the area that can be
uniformly illuminated by the available light source. If the contact
is applied over regions that are not uniformly illuminated, the
resulting domain inversion pattern will generally be nonuniform. It
should be noted that the electrical contact areas on the front and
back faces of the crystal do not have to be the same size. For
instance, if electrical contact is made to the entire front face of
the crystal at once, the poling area may be defined to a smaller
area by confining the liquid electrode on the back face, and
preferably the illuminated area, to a small subset of the crystal
surface, for instance using a UV-opaque dicing tape to confine the
liquid conductor and cover the remaining portions of the back face
of the crystal. Electrical contact between the external circuit and
the liquid conductor of the front face of the crystal may be made
by placing the crystal front-face-down onto a metal contact plate.
Connection to the uniform liquid electrode on the back a face of
the crystal may be made with one or more probe wires, positioned to
allow uniform distribution of voltage and current to the poling
area while not obstructing the illumination of the crystal
substrate.
[0098] It is important to ensure that good electrical contact is
made to the crystal surface by the liquid electrolyte/conductor.
This may be achieved by adding a small amount of a surfactant to
the liquid to reduce the surface tension, allowing it to more
readily wet the small features in the photoresist pattern on the
front face. Alternatively the photoresist pattern may be overcoated
with a conductor, e.g. by sputtering a metal or carbon conductive
layer, so that electrical contact is maintained from the top of the
mask down to the crystal surface without the need for the liquid
conductor to completely fill each feature in the pattern.
[0099] A typical experimental setup for the poling is shown in the
FIG. 8. A computer 805 controls a high voltage pulsed power supply
806 and the UV/visible light source 804. The electrical contacts
808 to the crystal 801 are connected to a high voltage power
supply, and a current sensor 809 is preferably incorporated into
the circuit. Preferably the complete poling system is controlled by
a PC which can capture the voltage, current and charge flow data in
real time, allowing different voltages to be sequenced or triggered
or shut down at different times, current flows, or charge transfer
values, or any combination of these. Again preferably a photodiode
or optical monitor 807 is incorporated into the poling fixture in
order to monitor the illumination source so that the computer
control program can also sequence the required illumination
exposure. Preferably, the PC controls the voltage supply, light
source shutter and timing sequence. Alternatively some features
such as the illumination time may be controlled by independent
timers and the optical monitor used to provide process sequencing
via the PC.
[0100] The light output from the light source (in this instance
coupled via a light-pipe) 803 is arranged to provide sufficient
illumination intensity and uniformity across the electrical contact
area. Typical intensities of .about.10 W/cm.sup.2 at the output of
the light pipe (broadband, all wavelengths from a high pressure
mercury bulb) are used. A beam shaping/expanding system 802 can be
used to increase the illumination beam diameter and/or uniformity
on the substrate. Typically about 0.5 W/cm2 of total light
intensity is incident at the surface the substrate. Higher and
lower intensities may be used with the appropriate adjustments in
illumination time and applied voltage to achieve domain
inversion.
[0101] A typical pulse sequence is as follows: An initial voltage
of about 2000V (.about.4000V/mm) 1001 is applied to the substrate
in the absence of illumination. At this time no poling occurs
because the voltage is significantly below the coercive field
required to achieve domain inversion, and therefore no current
flows in the poling circuit. The shutter of the illumination source
is then opened, illumination the unpatterned face of the wafer with
both visible and ultra violet wavelengths 1010 (a simple broadband
mercury lamp source is used in this example. It is also possible to
use a combination of one or more narrowband light sources to
achieve the same effect, as long as the wavelength(s) and
intensity(ies) are chosen so as to produce a similar quantity and
distribution of photo-excited charges within the crystal
substrate.) After the illumination begins a photo-induced current
1020 starts to flow through the substrate in response to the high
voltage applied across it. This current generally increases
gradually over a time frame of 100 ths to 10 ths of seconds, and
then tends to reach a plateau. Once the photo-current has increased
to a sufficient value, which is determined largely by the area of
exposure and the applied voltage, domain inversion occurs within
the crystal substrate despite the applied voltage being nominally
below the coercive field.
[0102] The function of this first illuminated voltage pulse is to
seed or initiate the domain inversion, so the illumination is
terminated before the poling is complete. In general the domains
resulting from this illuminated voltage pulse are tapered, and
their width on the patterned face and depth into the substrate both
generally increase with increasing voltage, increasing illumination
intensity and increasing illumination time. This termination can be
based on an empirically determined time or a charge flow monitored
by the computer, at which point the light source shutter is closed
and the illumination is blocked. Typical values for this first
pulse are a duration of .about.0.5 to 1 sec, and a charge flow of
0.02 to 0.12 mC/cm.sup.2 at a voltage of .about.2000V (4000V/mm)
and an illumination intensity of .about.0.5 W/cm.sup.2.
[0103] The domain shape and size may be further controlled and the
quality of the domain inversion grating structure enhanced by
applying a further voltage after the illumination is removed.
Preferably, this post illumination voltage is higher in magnitude
than that used during the illumination pulse, since there are no
photoexcited charges being generated to decrease the effective
coercive field of the material. Typically a voltage of around 3500V
(7000V per mm) 1002 may be applied post illumination. During this
un-illuminated voltage pulse, the poling current 1021 typically
increases to a clearly defined peak 1022, and then decreases to a
plateau value 1023. The decreasing poling current is related to the
completion of the domain inversion. If the voltage is removed while
the current is at the peak, the poling pattern will typically be
under duty cycle and some domain features will be incomplete. If
the voltage is maintained until the poling current has decreased to
its plateau value the domain pattern will typically be complete,
with a duty cycle that is dependent on the parameters of the
illumination and illumination voltage pulse. Maintaining the
voltage for a significant time after the current has decreased to
its threshold value typically leads to over duty cycle domains and
a larger number of merges within the domain inversion pattern. The
post-illumination voltage pulse may be controlled using the
computer control program based on either the charge flow 1030
within the circuit or the value and gradient of the poling current
1021 or on a combination of both. Thus poling may be terminated
when a particular charge has been transferred, when the current has
fallen to a particular value, when the rate of decrease of the
current reaches a certain value or any combination of these (and
other) parameters.
[0104] After poling, the insulating mask layer is stripped off of
the crystal surface. The substrate may then be etched in HF to
reveal the poled pattern. The inverted domains are generally
observed on the back face of the substrate. Tailoring of the dose
of illumination (illumination time and intensity), the voltage
applied when the illumination is on, the post-illumination voltage
and pulse duration, etc can be used to adjust the duty cycle of the
domain grating and the taper angle of the domain from front surface
to the back surface.
[0105] After etching of the top and bottom faces of the crystal to
reveal the domain inversion patterns, the quality of the domain
inversion grating device can be estimated using FIG. 4. The wafer
substrate containing the domain inversion gratings may be diced to
separate the individual gratings, which may have different periods
corresponding to different patterns on the photolithographic mask.
The end faces of each device may then be optically polished and,
preferably, coating with anti-reflection coatings, ready for use as
a quasi-phasematched frequency conversion device as shown in FIG.
5. The position of the optical beam within the crystal can be
adjusted in depth to utilize the optimum 50/50 duty cycle region of
the grating which results from the tapered domain structure. In
this embodiment, the domain taper angle is reduced compared to that
of the process of the second embodiment. This means that there is
less dimension variation in the domains from the top face to the
bottom face of the crystal. This results in a wider optimal
efficiency region within the crystal (i.e. a greater depth range
over which the duty cycle is within some percentage of 50/50), but
requires more control to be exercised over the dimension of the
domain on the top face to ensure that the 50/50 duty cycle region
is centrally located within the crystal.
[0106] Optimization of the domain grating quality, duty cycle and
taper does not have to be limited to the simple sequence of one
illuminated voltage pulse followed by a second higher voltage
pulse. Any sequence of illuminated and un-illuminated voltage
pulses may be used in any order to provide the required poling
charge to realize the desired domain inversion pattern in the
substrate independent of the presence of any photo-current due to
the illumination.
[0107] Voltage pulses can be simultaneous with illumination pulses,
voltage pulses can precede or follow illumination pulses, voltage
pulses can be longer or shorter than illumination pulses. In
addition, different illumination spectra (light wavelengths) may be
used in different illumination pulses with any combination of
different applied voltages.
[0108] The insulating mask may also be patterned on the +z face,
depending on the type of substrate. In general, adjustments of the
pulse parameters (such as the direction of the applied
illumination, and the magnitude and sequence of the illumination
and applied voltages) compared to those used for a -z face
patterned crystal will be required to achieve optimal domain
inversion patterns.
[0109] In another embodiment of the present invention, patterned
current flow is generated by a combination of a patterned
illumination and a patterned applied electric field. For instance,
the substrate may be patterned with an electrically insulating and
optically absorbing or reflecting masking material. This mask can
simultaneously provide the dual roles of patterning the
illumination and the applied electric field. The substrate is
illuminated from the masked face, resulting in only the open areas
in the mask pattern being illuminated and thus mobile charges being
generated only in those areas of the substrate. Preferably the
illumination wavelength(s) are chosen such that the penetration
depth into the illuminated regions of the substrate is short, such
that no substantial diffraction or interference pattern can result
in the substrate which otherwise would allow charge generation in
unwanted areas of the crystal. Simultaneously with the
illumination, the electric field is patterned by the insulating
mask such that the areas of the substrate in the open areas of the
mask pattern are subjected to a high electric field, while the
field in the areas covered by the mask is lower. The combination of
patterned illumination and patterned electric field results in a
patterned photocurrent flow in the material which provides enhanced
seeding for domain inversion at the patterned face.
[0110] FIG. 12. depicts an example of this preferred embodiment
where a wafer 1201 of, for example, 0.5 mm thick z-cut 5% MgO doped
congruent LiNbO.sub.3 is used. Preferably, the patterned electrode
is defined on the -z face of the crystal substrate. An insulating
layer 1202 is deposited onto the -Z face of the substrate, for
example a photoresist layer of .about.2 .mu.m thickness, or a
spin-on-glass layer of .about.1 .mu.m thickness. The insulating
layer is hard-baked to crosslink the material and provide robust
physical and electrically insulating properties. Since an
insulating layer like photoresist can absorb UV light and generate
photo-induced charges, it is preferable to provide a metal
over-layer 1203 that blocks the UV and visible light from reaching
the photoresist mask. A metal layer such as Ti, NiCr, Al etc,
preferably with a high absorption and or reflection in the UV and
visible spectrum is deposited on the surface of the insulator, e.g.
by evaporation or sputtering. A layer of photoresist 1204 is spun
on the metal layer and standard photolithographic patterning
processes are used to define the desired electrode pattern. The
pattern may be transferred into the metal layer using a wet or dry
etch, e.g. reactive ion etching or sputter etching. The underlying
insulating layer of hard-baked resist or spin-on-glass is then
patterned to provide openings to the crystal surface, preferably
using a reactive ion etch process to create substantially vertical
walls and to avoid damage or removal of the metal light blocking
layer.
[0111] Electrical contact to the crystal surface during the poling
process is easily made using LiCl solution. The liquid 1207 is
preferably applied to the patterned face first and may be confined
to the desired contact area using an o-ring or a simple tape
cut-out. It is important to ensure that good electrical contact is
made to the crystal surface by the liquid electrolyte/conductor.
This may be achieved by adding a small amount of a surfactant to
the liquid to reduce the surface tension, allowing it to more
readily wet the small features in the photoresist pattern.
[0112] Electrical contact to the opposite face of the substrate
(the unpatterned or back face) is made in a similar way with LiCl
electrolyte 1205 or may be achieved with metallization of the back
side of the wafer. The wafer is oriented with the patterned face
facing the output of the UV/visible illumination source. Contact to
the electrode on the unpatterned face may be made with a simple
probe in the liquid contact or to the metal electrode, contact to
the patterned front face may be made using a probe contact to the
edge of the liquid so as not to block the illumination from
entering the crystal substrate.
[0113] The poling sequence may be described as follows and as
illustrated in FIG. 10. A voltage pulse 1001 of about 2000 volts
(.about.4000V/mm) is applied across the electrodes on the front
1207 and back 1205 surfaces of the crystal substrate. During this
pulse, a UV light pulse 1010 is applied to the patterned front
surface of the substrate. A photo-current 1020 starts to flow
through the substrate due to the photo-induced charges created by
the illumination which move in response to the applied external
field. The illumination pulse is applied for about 0.3 s, then the
UV light is shut off. The combination of illumination and applied
voltage induces seeding of the domain inversion pattern, despite
the applied voltage being considerably below the coercive field of
the bulk crystal material. The seeding occurs only in the open
areas of the mask on the -Z face where the applied field is high
and the illumination reaches the surface of the crystal, thereby
allowing a patterned photocurrent to flow in those confined
regions. In general, at the end of the illumination pulse, the
seeded domain inversion features are terminated within the bulk of
the crystal and do not extend all the way to the unpatterned
electrode on the back face of the crystal.
[0114] Once the light pulse is terminated, a second high voltage
pulse 1002 is applied to the substrate to grow the domains through
the substrate. Typically the voltage of the second pulse is around
3500V (7000V per mm. During this un-illuminated growth voltage
pulse, the poling current 1021 generally increases to a clearly
defined peak 1022, and then decreases to a plateau value 1023. The
decreasing poling current is related to the completion of the
domain inversion. If the voltage is removed while the current is at
the peak 1022, the poling pattern will typically be under duty
cycle and some domain features will be incomplete. If the voltage
is maintained until the poling current has decreased to its plateau
value 1023 the domain pattern will typically be complete, with a
duty cycle that is dependent on the parameters of the illumination
and illumination voltage pulse. The post-illumination voltage pulse
may be controlled using the computer control program based on
either the charge flow 1030 within the circuit or the value and
slope of the poling current 1021 or on a combination of both.
[0115] The choice of wavelength in this embodiment is be dictated
by the consideration that the light should be absorbed close to the
surface of the material. The absorption is preferably strong so as
to prevent significant diffraction or the creation of an
interference pattern in the bulk of the material. In the absence of
diffraction, the photo current will be well defined by the opening
in the insulating mask layer and therefore the domain inversion
seeding and subsequent growth will be similarly well defined,
creating the desired domain inversion pattern. Preferably the
absorption depth of the illumination is a few microns into the
crystal substrate.
[0116] In another embodiment of the present invention, a single
wavelength of UV or visible light is used as the illumination
source, enabling an interference pattern to be created within the
crystal substrate. The coherent light source may be a frequency
doubled diode pumped solid state laser or gas laser such as an
argon or krypton ion laser, or any other laser source operating in
the UV/visible spectral region. Absorption of light at the
constructive interference fringes within the crystal generates
localized concentrations of photo-induced mobile charges. These
charges form a photo-current in response to a voltage applied
across the faces of the crystal substrate, and this
photocurrent/voltage combination is used as previously described to
seed domain inversion in a localized manner.
[0117] In FIG. 13, an unpatterned MgO:CLN wafer 1301 of 0.5 mm
thickness is illustrated. A prism 1305 is used to split an incoming
light beam 1302, e.g. from a Krypton ion laser @ 413.1 nm, and
create an interference pattern 1303 with the desired period within
the crystal substrate. The angle of the prism 1306 for BK7 glass is
designed to be 5.043 degree to generate 4.425 um grating period. It
may be preferable to use LiNbO3 as prism material, and the angle
for LiNbO3 will be 1.891 degrees. The incident angle of the input
light beam to the prism must be well controlled, and the
orientation of the wafer must be accurately set relative to the
axes of the interference fringes such that the fringes lie
substantially along the Y-axis of the crystal.
[0118] An optically transparent conducting liquid 1304 such as LiCl
in water is used as the electrode and is introduced between the
prism and the crystal surface, preferably forming a smooth,
continuous layer with no bubbles or thickness variations that can
affect the uniformity of the interference pattern. If desired,
pressure can be applied to the prism to ensure that the liquid
layer is thin and uniform. Alternatively, a transparent conductor
such as ITO (indium tin oxide) may be deposited on the crystal
surface or the surface of the prism to act as the electrode.
[0119] Electrical contact to the liquid electrode may be made at
the edge of the prism. Typically a voltage of .about.2000V
(.about.4000V/mm) is applied to the crystal while it is being
illuminated, and the photocurrent flow through the substrate is
monitored. Preferably, the value of the photocurrent is kept low so
that the current flow is strongly localized to the narrow
constructive interference regions of the interference pattern.
[0120] After allowing the photocurrent to flow for a period of time
varying from seconds to minutes, depending on the magnitude of the
applied voltage, the intensity of the illumination, the magnitude
of the photocurrent and the poled area and material type of the
substrate being poled, a higher voltage is applied to complete the
domain inversion. It is preferable to block the illumination as the
higher voltage is applied, to prevent a dramatic increase in
photo-current flow. The higher, poling growth voltage is generally
of the order of 3500-4000V and may be applied either as a step
function or continuously ramped from the initial to final values. A
current sensor may be used to monitor the charge flow during the
poling pulse, and accounting for the photocurrent flow, the poling
pulse may be terminated when the sufficient charge has been
transferred to achieve the desired amount of poling.
[0121] When choosing an illumination wavelength for this embodiment
it is necessary to consider the dual requirements of a reasonably
strong absorption to generate the necessary photo-induced charges
while simultaneously allowing the interference fringes to extend to
a substantial depth into the crystal. Therefore the illumination
wavelength is preferably in the long wavelength UV to short
wavelength visible range around approximately 400 nm, considerably
above the band edge of .about.320 nm. As the optical absorption
will lead to a gradient in photo-charge density with depth into the
substrate, there may be a preferential illumination direction, e.g.
it may be preferable to illuminate the crystal through the +Z
face.
[0122] In the above described process the inverted domains are
seeded in the narrow illuminated regions of constructive
interference in the optical interference pattern. However, in the
arrangement of FIG. 13 there is no mechanism to prevent the domains
from growing laterally, to form the energetically favorable
hexagonal domain shape, other than the preferential seeding and
poling due to the localized photo-current and illumination. In
practice the domain confinement provided by the localized
photo-current and illumination is not sufficient to prevent the
domains from expanding laterally and merging together at short
periods.
[0123] An improvement to this embodiment is illustrated in FIG. 14,
where a patterned insulating mask layer 1404 is provided on the
opposite face of the crystal to the illuminating beam 1402. The
period of the pattern in the insulating mask is the same as that of
the interference pattern within the crystal substrate, and the mask
is preferably aligned such that the domain inversion features run
substantially parallel to the Y-axis of the crystal. The substrate
1404 is mounted on a rotation and translation stage (not shown) so
that alignment can be achieved between the patterned mask layer and
the optical interference pattern. The substrate is illuminated from
the unpatterned face to create an optical interference pattern. A
lens is placed adjacent to the patterned face of the substrate to
collect the light that is transmitted through the crystal and
transfer it to a photodetector. When the constructive interference
fringes of the interference pattern are aligned with the openings
in the insulating mask in both rotational and translation
directions, the observed transmitted light signal will reach a
maximum. Once alignment is achieved, it may be preferable to block
the coherent light to remove the interference pattern and allow the
photo-generated charges created during the alignment procedure to
dissipate.
[0124] With the insulating mask and interference pattern aligned, a
similar poling sequence to that described above for the unpatterned
sample can be performed. With the illumination source incident on
the crystal, a voltage is applied to the liquid electrodes 1405 on
the patterned face and unpatterned face, resulting in a
photo-current flow which is now confined by both the constructive
interference regions and the openings in the insulating mask
aligned to the interference pattern. The extra current confinement
effect of the insulating mask combines with a modulated electric
field to improve the definition of the domain pattern and prevent
unwanted lateral expansion of the domains.
[0125] The patterned insulating mask layer also provides a further
benefit for the short period domain inversion process. As discussed
earlier with reference to FIG. 7. there is a maximum length of
poling feature which is preferable for a given period due to
lateral domain expansion as a result of angular misalignment
between the domain feature and the crystal axes. In the present
embodiment, the optical interference pattern is composed of fringes
which are continuous across the entire illuminated area. Thus, any
slight misalignment will cause lateral expansion based on the full
width of the poled area, easily causing merging of short period
gratings. However, the patterned insulating mask layer enables the
continuous fringes to be effectively broken into shorter lengths by
adding an insulating barrier to block domain inversion at certain
points along the length of the fringe.
[0126] Thus, the length of the domain features can be
photolithographically reduced to the .ltoreq.230 .mu.m length
preferable for a 4 .mu.m period grating, or .ltoreq.180 .mu.m for a
3 .mu.m period grating, based on an angular misalignment tolerance
of .about.0.5 degrees between the grating and the crystal axes.
[0127] Another embodiment of the invention is shown in FIG. 15. An
alternative approach to generating the mobile charges of the
present invention is to use a high energy electron beam to inject
the charges into the substrate. The combination of the uniform
electron beam irradiation through the back face of the crystal with
the patterned metal or insulating mask on the front face results in
patterned current flow at the patterned mask surface and through
the depth of the substrate, resulting in patterned domain
inversion. This greatly simplifies the equipment requirements
compared to the prior art focused electron-beam approach to domain
inversion.
[0128] When using a high energy electron beam for domain inversion
according to the present invention, the energy of the electron beam
can be varied in time from high to low or low to high to vary the
penetration depth of the electrons into the substrate. This
capability in principle provides advantages over the illuminated
embodiments of the present invention since the electron beam energy
and penetration depth can be more flexibly and tightly controlled
than the absorption of light, which is limited by the available
wavelength spectrum and the absorption spectrum of the
material.
[0129] An external voltage may be applied to the substrate in a
similar manner to the illuminated embodiments, and may be applied
before, during and after the electron bombardment. It should be
noted that when the high energy electrons are stopped inside the
crystal, the kinetic energy of the electrons will be absorbed and
increase the temperature of the substrate, which may decrease the
coercive field for the domain inversion process.
[0130] FIG. 15 illustrates a substrate 1501, preferably 0.5 mm
thick MgO:CLN, with a patterned metal electrode 1502 disposed using
standard photolithographic deposition and patterning techniques on
the +Z face of the substrate. Alternatively a patterned insulating
mask may be used. A thin metal electrode 1503, e.g. .about.1000 A
of titanium, is deposited on the -Z, or back, face of the
substrate. A high energy electron beam system (not shown) e.g. a
system from HVEA Inc, is used to generate an electron beam 1505
which is incident on the unpatterned -Z face of the crystal
substrate. The electron energy and dose are controlled by the
accelerating voltage and current flow of the electron beam system
respectively.
[0131] The electron beam is collimated to provide uniform exposure
over a defined area, preferably over the entire wafer surface. For
electric field poling, the electron beam is incident through the
unpatterned -Z face while the patterned +Z face is grounded. The
electric field generated by the accumulated electrons within the
substrate is generally sufficient to cause domain inversion to
occur. An external voltage may be applied between the patterned
electrode 1502 and the unpatterned electrode 1503 to control the
flow of mobile-charge-current through the substrate and improve the
domain inversion pattern definition and quality.
[0132] The improved domain inversion structure of the present
invention is of particular value when used to construct a frequency
converter for the application of second harmonic generation for the
creation of visible laser light sources. This application has
proven very challenging for prior art devices due to the difficulty
in fabricating the very short grating periods required, .about.4
.mu.m to 6 .mu.m, with high quality and uniformity, and due to the
performance degradations due to photorefractivity and green and
blue induced infra-red absorption (GRIIRA and BLIIRA). The present
invention provides a fabrication process for high quality, high
uniformity and high efficiency quasi-phasematched frequency
converters with periods as short as 4 .mu.m in a photorefractively
robust material, MgO:CLN The present invention also provides a high
temperature annealing process coupled with a closed loop discharge
path which enables the effects of BLIIRA and GRIIRA to be
significantly reduced. In addition, the present invention provides
an optimized frequency conversion device with a tapered domain
structure which ensures that at least some portion of the bulk
crystal has an optimum 50/50 duty cycle domain grating.
[0133] In one embodiment of the present invention, an efficient
visible frequency conversion element and device is provided for the
generation of visible light using an improved domain inversion
structure. Such a device is shown schematically in FIG. 16,
indicating a single pass frequency doubled laser system. The pump
laser 1602 is coupled into the frequency converter 1601 using a set
of coupling optics 1604, e.g. a focusing or collimating lens. The
input face of the frequency converter is preferably anti-reflection
(AR) coated at the fundamental wavelength so as to minimize
efficiency losses in the optical conversion process, while the
output face is preferably AR coated for at least the second
harmonic wavelength and preferably for the fundamental wavelength
as well. The pump laser 1602 may be a semiconductor diode laser, a
diode pumped solid state laser, e.g. Nd:YAG or Nd:YVO.sub.4, a gas
laser, or any other type of laser with coherent output light 1603
of fundamental frequency which matches the conversion wavelength of
the frequency converter 1601. Preferably, for efficient frequency
conversion, the spectral bandwidth of the laser pump source is less
than or comparable to the phasematching bandwidth of the frequency
converter. The polarization of the pump beam is preferably arranged
to be parallel to the crystal Z-axis to enable the highest
nonlinear coefficient, d.sub.33 to be used for the frequency
conversion interaction.
[0134] The temperature of the frequency converter 1601 is generally
adjusted using a heated mount 1605 so as to match the operating
wavelength of the converter with the input fundamental pump laser
wavelength. Alternatively, the wavelength of the pump laser may be
tuned using a grating or an etalon so that it matches the
acceptance wavelength of the frequency converter. An advantage of
the MgO:CLN devices enabled by the present invention is that the
operating temperature is much lower than that required for CLN
devices, <100.degree. C. vs. .gtoreq.220.degree. C. The position
of the pump beam 1603 within the frequency converter 1601 should be
adjusted for maximum conversion efficiency to make use of the 50/50
duty cycle of the domain grating which is ensured by the tapered
domain structure. An optical filter 1606 may be located in the
output beam to remove the residual pump beam and transmit the
second harmonic output at visible or UV wavelengths.
[0135] In the second harmonic generation application described
here, the required period of the domain inversion grating in the
frequency converter is determined by the wavelengths of the
interacting beams as follows:
.LAMBDA. = .lamda. pump 2 ( n sh - n pump ) ##EQU00001##
where .LAMBDA. is the grating period, .lamda..sub.pump is the pump
wavelength, n.sub.sh is the refractive index at the second
harmonica wavelength and n.sub.p is the refractive index at the
pump wavelength.
[0136] A knowledge of the refractive index dispersion and
thermo-optic coefficients of the nonlinear optical crystal enables
a domain inversion grating to be designed to quasi-phasematch a
second harmonic interaction at a particular wavelength and at a
particular temperature.
[0137] The frequency conversion device of FIG. 16 finds
applications in medical instrumentation, semiconductor metrology
and laser display devices.
[0138] An alternative application for the frequency conversion
device of the present invention to yield efficient second harmonic
generation is intra-cavity frequency conversion as shown in FIG.
17. Placing the frequency converter 1702 inside the pump laser
cavity makes use of the higher circulating power within the cavity
compared to the CW or pulsed output from the same laser,
significantly increasing the efficiency of the second harmonic
generation interaction, which is proportional to the square of the
power in the pump beam. The laser gain medium 1701 may be an
electrically pumped semiconductor laser, an optically pumped
semiconductor, or an optically pumped crystal gain medium such as
Nd:YAG. One face of the gain medium 1701 is high reflection coated
at the fundamental pump wavelength, and the other is anti
reflection coated for the fundamental. Preferably this face is high
reflection coated at the second harmonic wavelength to prevent the
second harmonic visible light from damaging the semiconductor pump
source. The domain inverted frequency converter 1702 is placed
adjacent to the gain medium 1701. Optionally, there may be coupling
optics such as a focusing or collimating lens (not shown) disposed
between the gain medium and the frequency converter. Both input and
output faces of the frequency converter are anti-reflection coated
for both the fundamental and second harmonic wavelengths. A second
mirror 1703 is placed on the opposite side of the frequency
converter to the pump laser to form the resonant laser cavity. This
mirror may be a simple multilayer dielectric reflector or it may be
a distributed reflector such as a volume Bragg grating.
Alternatively the output face of the frequency converter may be
optically coated to form the cavity output mirror. Temperature
control of the nonlinear optical frequency converter using a heated
mount 1707 may be used to tune the crystal to peak efficiency for
the operating wavelength of the laser. Alternatively, the
wavelength of the pump laser may be tuned using a grating or an
etalon so that it matches the acceptance wavelength of the
frequency converter. Typically the frequency conversion crystal is
designed according to the relation described above so that it must
be held at a slightly elevated temperature to provide optimum
conversion efficiency for the design wavelength of the laser. This
is for two main reasons, firstly the slightly elevated temperature,
typically 40-90.degree. C. helps to reduce the possibility of
residual photorefraction distorting the optical beam and reducing
the efficiency, and secondly maintaining the elevated temperature
requires only a simple heater, which is generally less complicated,
failure prone and more efficient than the thermo-electric cooler
that would be required to provide a stable operating temperature
nearer room temperature. In addition, in consumer electronic
products, the ambient operating temperature is not well controlled,
so it is important to design the device to operate at a temperature
higher than that which will be experienced as an ambient
temperature, so that heating is always required and a stable
operating temperature can be maintained.
[0139] For a crystal gain medium, e.g. Nd:YAG, the laser wavelength
is defined by the crystal structure energy levels and is generally
well determined and narrow band (excepting certain crystals and
dopants such as Ti:Sapphire which show widely tunable laser
action). For semiconductor pump sources, the gain bandwidth is
typically quite broad, and a further frequency selective element
must be provided to determine the laser wavelength and bandwidth.
This frequency selective element may be placed inside the cavity
1705, where it may be an etalon or narrow band filter, or it may be
incorporated into one of the two cavity mirrors in the form of a
Bragg reflection grating. Thirdly, a Bragg reflection grating 1706
may be deployed outside the main laser cavity to provide wavelength
selective feedback to the laser pump source to determine the laser
wavelength. For optimum efficiency with the frequency converter of
the present invention, the frequency converter should be positioned
such that the pump beam travels through the optimum 50/50 duty
cycle portion of the crystal which is ensured by the tapered domain
structure.
[0140] Since the conversion efficiency of the second harmonic
generator is proportional to the square of the pump power, more
efficient energy conversion can generally be obtained from a pulsed
laser source than from a CW laser. Increasing the peak and average
powers too much however can cause crystal damage, such as surface
damage at the polished faces, or residual photorefractive effects
which are not compensated by the MgO dopant at very high optical
powers. For this reason, for some applications where high visible
powers are required, it may be preferable to provide an array of
pump beams coupled into different regions of the same frequency
conversion chip. In this way, the power in each individual beam can
be maintained well below the material damage thresholds, while the
total output power from all the beams can be scaled as high as
several watts to 10 watts of power. The frequency converter enabled
by the present invention is ideal for this application since it
provides high peak conversion efficiency and high lateral
uniformity for uniform and efficient frequency doubling performance
across an entire array of laser beams.
[0141] An example of an application which benefits strongly from
the array scalability of the frequency converter enabled by the
present invention is that of laser projection displays. In this
case, the fact that the total output power is made up of a number
of individual beams is not a disadvantage, since a single mode
diffraction limited optical beam is not generally required. In
fact, multiple beams each with slightly different wavelengths helps
to reduce the speckle effect which can otherwise render laser
displays uncomfortable to watch.
[0142] An example of a laser light source for projection display
applications is shown in FIG. 18. At present, most projection
displays are illuminated by a high pressure mercury bulb. This bulb
is inefficient at generating visible light, and the optical
components required to capture a significant proportion of the
light that is generated and project it onto the screen are complex
and relatively expensive. In addition, a color wheel or color
filters must be used to separate the three primary colors (red,
green and blue) either spatially or temporally. A laser light
source on the other hand simplifies many of these issues and offers
some advantages for the display. Firstly the laser light source can
provide a wider color gamut than the lamp by producing narrower
bandwidth light at the primary colors, leading to a richer, more
saturated and natural looking color display. Secondly the optical
coupling and projection optics for the laser beam are significantly
simpler than those for the lamp, since the laser beams are
essentially collimated and do not need fast (wide aperture)
collection optics. Thirdly the properties of the laser enable some
of the other components to be removed from the optical system
decreasing the complexity and cost. In the laser light source of
FIG. 18, laser light at the three primary colors is generated by
modules 1801, 1802, 1803, utilizing frequency doubling of
semiconductor diode lasers using the periodically poled MgO:CLN
frequency conversion device of the present invention. Collecting
optics 1804 collimate the light, overlap the multiple beams from
the array, and match the optical beam to the form factor of the
spatial light modulators 1807, 1808, 1809, which in this embodiment
are transmission LCD panels, for instance from Epson Corp. The
laser is advantageous for the use of LCD panels since the laser
light is linearly polarized matching the optimum requirement for
the LCD panel operation. The spatially modulated light at the three
primary colors is combined in an x-prism 1810 and the image
projected onto the screen (not shown) via the projection optics
1811. This embodiment has described a 3-LCD projection system
wherein the use of the laser enables the elimination of the complex
collection optics required by the lamp, as well as various color
separation filters and polarizers that are required to split the
lamp output into linearly polarized beams at the primary
colors.
[0143] The array scaling capability of the frequency converter of
the present invention is key to generating the power levels that
are required for a projection display. For instance, 2.0 W of 465
nm blue, 1.6 Watts of 532 nm green and 2.2 Watts of 635 nm red
laser light will provide 1400 lumens, which after traveling through
the typical spatial light modulators and projection optics should
yield around 400 lm on the screen. For brighter displays, even more
optical power is needed, leading to the desire to reliably produce
4-5 watts of light in each primary color.
[0144] An alternative embodiment for the projection display uses a
digital micro-mirror device (DMD, Texas Instruments) as the spatial
light modulator. In this embodiment, the light output from the
three primary-color second harmonic generation laser modules are
spatially overlapped before the spatial light modulator. Time
sequencing of the light output from the laser modules is used to
provide color-sequential operation using a single spatial light
modulator--alternatively a separate SLM can be used for each
primary color and the images superimposed after the SLM. The light
output from the SLM is projected onto the screen by the projection
optics. In this embodiment the laser light sources enable the
elimination of the rotating color wheel which is currently used to
provide color-sequential light output from the continuous wave
mercury lamp, as well as simplifying the collection and projection
optics.
[0145] It should be noted that although the above projection
display embodiments have been described with reference to full
color, 3-primary source displays, it may in some cases be
preferable to provide more or less than 3 primary colors. For
instance, by providing 4 or 5 colors, the overall color gamut of
the display can be increased and a wider range of natural colors
can be displayed. On the other hand, a simpler and cheaper display
device can be provided with only a single color, producing a
monochrome display with potentially much more compact dimensions
and lower cost. The frequency converter of the present invention is
particularly valuable for the small dimension and low cost
projection display device, often termed the pocket projector. This
is because the frequency conversion device enabled by the present
invention has the combined properties of highly efficient
operation, high uniformity, high manufacturing yield, and
fabrication in a commercially available substrate material. This
enables, for the first time, the prospect of scaling the
manufacturing cost of a precision designed and fabricated
periodically poled nonlinear optical crystal down to the few dollar
price point required for mass volume manufacturing for consumer
electronics applications.
[0146] An alternative approach to the projection display,
especially for compact and low power devices is that of directly
scanning the image over the screen, using for instance 1 or 2 axes
MEMS (micro-electrical mechanical systems) scanners. This is shown
schematically in FIG. 19. In this case, the primary color laser
modules (which generate visible light using the frequency converter
of the present invention) 1901, 1902, 1903 are imaged onto the
scanning system 1904 by the coupling optics 1905. The scanning
system 1904 may be composed of a single 2-axis scanner, or of a
single axis scanner and a rotating faceted drum, or two single axis
scanners or any other image scanning technique know in the art. The
scanner directs the color beams to the screen (not shown) and the
image is written to the screen using, for instance, raster or
vector scanning. Brightness and color information is encoded by
time domain modulation of the laser output power, either by
directly modulating the pump laser power, or by providing a
modulator integrated into the frequency converter, or located
outside the frequency conversion module. The device structure and
assembly are simplified by the lack of the spatial light modulator,
which also offers the prospect of lower device cost, albeit
accompanied by a reduced performance in terms of brightness and
image resolution.
[0147] Another application of the improved optical frequency
converter is in the generation of infra-red light for use for
example in remote gas sensing, countermeasures and Lidar. Infra-red
wavelengths are generated with a difference frequency converter or
an optical parametric oscillator. FIG. 20 shows a difference
frequency converter using the present invention. Two pump lasers
2001, 2002 are coupled into the frequency converter 2003 by
coupling means 2004 which may consist of one or more lenses to
focus or collimate the pump laser beams. The quasi-phasematched
domain inversion grating in the frequency converter 2003 transfers
optical energy from the two pump lasers into a third beam 2006
(which is actually generally collinear with the combined pump beam
2005) with a frequency equal to the difference between the
frequencies of the pump lasers. The temperature controller 2007
controls the temperature of the chip to provide maximum conversion
efficiency. In this way, a pump beam from an Nd:YAG laser at 1064
nm mixes with a diode laser at 810 nm to create a difference
frequency beam at about 3.4 .mu.m. Alternatively, the difference
frequency generator can be used for telecom wavelength conversion
and dispersion compensation applications. In this case a pump beam
at 775 nm is combined with a signal beam at around 1550 nm to
generate an idler beam also around 1550 nm. In this configuration
the signal and idler beams can range from around 1520 nm to 1570 nm
due to the slow variation of the refractive index around the 1550
nm wavelength in lithium niobate. By using a different (shorter)
grating period, the frequency conversion device can be used for sum
frequency generation. For example, a 1480 nm laser and a 920 nm
laser can be mixed to create 567 nm yellow light for medical
applications.
[0148] An alternative configuration for generating infra-red light
is the optical parametric oscillator (OPO), where a single pump
beam is used and one or both of the signal and idler wavelengths
are resonated in a cavity formed by mirrors placed around the
frequency converter as is well known in the art. Once above
threshold, (i.e. at high enough pump levels to cause oscillation)
the OPO is very efficient in transferring energy from the pump to
the signal and idler beams.
[0149] Although the present invention has been described in detail
with reference to magnesium oxide doped congruent lithium niobate,
MgO:CLN, it is equally applicable to other nonlinear optical
materials known in the art including: undoped congruent lithium
niobate and tantalate, stoichiometric lithium niobate and tantalate
materials, either grown from the melt or prepared by vapor
transport equilibrium, magnesium doped stoichiometric lithium
niobate and tantalate, Ti doped and Ti diffused CLN, SLN and SLT
and similar materials with other dopants designed to reduce the
photorefractive effect such as zinc doped congruent lithium
niobate.
[0150] The detailed discussion of the present invention has been
presented with reference to a substrate thickness of 0.5 mm since
this is a standard, commercially available substrate dimension. The
present invention is equally applicable to both thinner and thicker
substrates, in particular 0.25 mm and 1.0 mm substrates. Thicker
substrates, such as 2 mm and 3 mm, can be readily poled using this
technique for high power applications especially in infra red
generating devices. There is no limitation on the wafer diameter or
transverse dimensions of the substrate. If the illumination system
cannot uniformly illuminate the entire substrate at once, the
surface of the substrate may simply be masked off with an opaque
material, such as dicing tape, and the domain inversion process
performed in sections across the substrate.
[0151] In one embodiment, the present invention relates to the
fabrication of domain inverted structures. Generally the
applications for these domain inverted devices are in the field of
optical frequency conversion using quasi-phase matching to provide
efficient energy transfer from one wavelength to another. Whilst
the present invention has been described in detail with reference
to domain inversion gratings, it should be understood that
quasi-phasematched devices may contain periodic, aperiodic and
pseudo-random phase reversal structures as required to produce the
desired phasematching curve. In addition, whilst a grating
generally consists of a multiple number of features arrayed
periodically or aperiodically, in a domain inverted device it may
consist of as few as two domains, requiring only a single domain
boundary. Domain inversion devices may also be used for
applications other than optical frequency conversion, such as
polarization rotation (TE-TM conversion), optical switching and
optical beam deflection.
[0152] Throughout the embodiments described above, reference has
been made to the front and back surfaces of the crystal substrate.
These faces are not fixed with respect to the crystal orientation
and depend on the photolithographically applied mask layers and the
direction of the illumination applied to the crystal. In general,
the photolithographically masked face is referred to as the front
face, and the spatially uniform illumination is incident on the
back face.
[0153] No limitations have been set on the period of the domain
inversion grating that can be realized with the techniques of the
present invention. Isolated domains with submicron dimensions have
been observed, offering the prospect of domain inversion gratings
with periods of less than 2 .mu.m for quasi-phasematching of UV
interactions. The visible frequency conversion applications which
are the most promising application of the present invention require
periods ranging from .about.4 .mu.m for the blue, through .about.7
um for the green up to .about.12 .mu.m for the red. There is also
no upper limit to the period which can be fabricated, in fact
isolated domains can be reliably fabricated using the present
invention with high repeatability and precise domain size control
for use in applications such as beam deflectors or optical total
internal switches.
[0154] Although both the device fabrication and applications have
been written with reference to bulk frequency conversion
applications, the frequency conversion device fabricated by the
present invention can also be used as a substrate for highly
efficient waveguide frequency conversion applications. In this
case, the tapered domain structure does not ensure that there is a
50/50 duty cycle within the waveguide region, but by careful
control of the poling parameters the grating duty cycle at one of
the two crystal faces can be controlled to be substantially 50/50
for 1.sup.st order quasi-phasematching. Optical waveguides can be
fabricated in the MgO:CLN substrate using any of the techniques
known in the art, such as annealed proton exchange (APE), reverse
proton exchange (RPE), titanium indiffusion and zinc indiffusion.
In the fabrication sequence for APE and RPE devices the waveguide
and periodic poling steps can be performed in any order since
neither substantially affects the capability to perform the other.
With the metal indiffusion waveguides the waveguide process is
preferably performed first so that it does not disturb the short
period domain inversion during the high temperature process. In
this instance the present invention is particularly important since
the mobile charge electrode enables high quality domain inversion
to be generated even through the metal indiffused waveguide
regions. In general the same fabrication techniques are applied to
waveguide frequency converters as for the bulk embodiments
described above. The design approach for the devices is very
similar, the optical waveguide mode effective index is used to
compute the required grating period for a given wavelength rather
than the bulk crystal refractive index.
[0155] The embodiments described above serve the purpose of
demonstrating the principle of the current invention. A person with
ordinary skill-in-the art can derive more specific embodiments
beyond those described here that are in the spirit of the current
invention. Techniques described in the different embodiments can be
freely combined to produce further embodiments which enhance the
control of the domain growth.
* * * * *