U.S. Patent Sep. 20,1988 Sheet 3 of 3 4,773,063
1 2
ixm range using a commercially available grating mirror OPTICAL WAVELENGTH DIVISION and homogeneous lenses. MULTIPLEXING/DEMULTIPLEXING SYSTEM None of the foregoing is as effective as the Applicants' invention which, besides, has economies in fabriFIELD OF THE INVENTION 5 catjon and US6j and enlarged channel processing capaThe advent of single-mode optical fibers and lasers bilities which completely surpass expectations, for optical signal transmission has begun to make wave- SUMMARY OF THE INVENTION length-division multiplexing/demultiplexing economically advantageous. Thus far, research into increased 10 This invention comprises a method and apparatus for fiber bandwidth and wider ranges of operating wave- wavelength-division demultiplexing and, with approprilengths for both laser transmission sources and optical ate alterations, multiplexing.
receiver diodes has not been paralleled by adequate In the preferred demultiplexing embodiment it utimethods for both multiplexing and demultiplexing opti- lizes an integrated optical planar waveguide formed on cal signals. State-of-the-art experimental wavelength- 15 a substrate and provided with a periodic radiation transdivision multiplexers/demultiplexers are capable of missive diffraction grating and sensing means for the multiplexing only about twenty, or fewer, channels, output of segregated optical signals to individual recepwhile being relatively complex, difficult to fabricate, tors. ^ grating bars must be sufficiently long for suband expensive. stantially Bragg diffraction to occur.
BACKGROUND OF THE INVENTION 20 The transmissive diffraction grating can be fabricated
by conventional techniques, including but not limited
Insofar as the Applicants are aware, there are pres- .„ /n t u- »• u • * *u -j c
. , *T , , • to: (1) etching grating bars into the waveguide surface;
ently three types of wavelength-division-multiplexing , ... ° , .. , „ ■ . X. ..
J. J • \ J /i\ i J- • /-.\ i (2) depositing metal or dielectric bars on the waveguide
methods in use today: (1) angular dispersion: (2) optical » TM • . . - • r
filters; (3) optical absorbers. Our invention utilizes the 25 furfaf '> <3> >nducmS asurface acoustIC wave of Prese"
first of these methods. lected Penod Vla a P'ezoelectnc or magnetostnctive
The closest prior art known to the Applicants is con- transducer in an acoustooptic waveguide material; (4)
tained in the literature articles "An Integrated-Optical employing the electro-optic effect to produce a prese
Approach to the Fourier Transform" by D. B. Ander- lected Periodic variation of the refractive index of the
son, J. T. Boyd, M. C. Hamilton, and R. R. August in 30 waveguide by applying an electrical potential across
IEEE Journal of Quantum Electronics, Vol. QE-13, interdigitated contacts or across finger contacts using
No. 4, April 1977, and "Wavelength multiplexing in the substrate as a ground plane; (5) other methods capa
multimode optical fibers" by W. J. Tomlinson, Applied b,e of producing a periodic variation of the refractive
Optics, Vol. 16, No. 8, August 1977, the latter being a index of the waveguide.
general treatment of the state-of-the-art at the time of 35 The utilization of wavelength-division multiplexer/publication. demultiplexer units in conjunction with a single fiber There exist a number of wavelength-division multi- optical link enlarges markedly the information transmisplexers which are, however, different from the Appli- sion capacity of the system, thereby drastically reduccants' invention. The best of these utilize mirrored dif- ing the cost per information channel in both materials of fraction gratings in several configurations to achieve construction and installation. In comparison to frequendiffering angular dispersion according to wavelength. cy-division multiplexing which encodes several inforThus, M. Seki et al. disclosed a "20-Channel Micro- mation channels electronically onto a single waveoptic Grating Demultiplexer for 1.1-1.6 ^.m Band Using length, wavelength-division multiplexing offers ina Small Focusing Parameter Graded-Index: Rod Lens," 45 creased transmission speed and significantly reduced Electronics Letters, Vol. 18, No. 6, pp. 257-258 (1982). complexity of associated electronic support systems. It Their design was produced as a twenty-channel demul- should be noted that wavelength-division and frequentiplexer utilizing a preferentially-etched silicon cy-division are complementary multiplexing methods, echelette grating butted against a Selfoc rod lens. Wata- and) thus> a w h information capacity flber.optic com. nabe and Nosu reported a Slab waveguide demuhi- 50 munication system would utilize both wavelength-diviplexer for multimode optical transmission in the UM.4 sion and ... mutiplexing. jam wavelength region in Applied Optics, Vol. 19, No.
21 pp. 3588-3590 (1980), which utilized a ground con- THE DRAWINGS
cave grating mirror attached to the end of a planar ~ . , . ... , „ . ,. ,
•j , x. ■ The following drawings constitute part of this disclo
waveguide, normal to the plane of the guide. With this 55 ,. , ° e r
system the authors were able to demultiplex ten chan- SU~i,n W. :, . .
nels. They claim the advantages of (1) stable, rugged , FIG" 1 18 a schematic representation of a complete
construction, (2) no antireflection coatings, and (3) re- electro-optional fiber signal transmitting/receiving sys
duced dimensions. Watanabe et al. reported an "Optical tem'
grating multiplexer in the 1.1-1.5 urn wavelength re- 60 FIG- 2 is a PartiallV schematic perspective represen
gion" in Electronics Letters, Vol. 16, pp. 108-109 tation of a preferred embodiment of Applicants'demul
(1980), which could multiplex ten channels in the tiplexer (multiplexer) in its relationship to the optical
1.1-1.5 jam wavelength range using a planar diffraction signal transmitting and receiving apparatus of FIG. 1,
grating mirror made from anisotropically etched sili- FIG. 3 is a schematic representation of the diffraction con. Aoyama and Minowa, in "Optical demultiplexer 65 grating of FIG. 2, and
for a wavelength-division multiplexing system," Ap- FIGS. 4A, 4B, 5-7, inclusive, are ray diagrams repre
plied Optics, Vol. 18, No. 8, pp. 1253-1258 (1979), re- sentative of accepted principles believed applicable to
ported a demultiplexer for five channels in the 0.8-0.9 Applicants' invention.
DETAILED DESCRIPTION OF THE INVENTION
Referring to FIG. 1, a conventional unidirectional fiber optic communication system utilizing wavelength- 5 division multiplexing (WDM) comprises signal inputs 5a to 5n (received, respectively, from channels 1 to n) supplied to wavelength-division multiplexer (WDM) 6a which, conventionally, can be a known arrangement of beam splitters and lasers, the multiplexed signal from 10 which is transmitted via a single optical fiber 7 to a receptor WDM device 6b (conventionally a prism, reflection grating, or multilayer thin-film dielectric filter arrangement) which resolves the composite signal into its component signal outputs 9a to 9n directed into 15 individual detectors 10a to lOn which deliver information to their predetermined destinations, channels V to n'.
Our invention is described as particularly applicable to a demultiplexing apparatus shown in FIG. 2. Here 20 the composite signal input transmitted via optical fiber 7 is received by Applicants' radiation transmissive planar waveguide 14 which embodies, preferably in monolithic construction, an input collimating lens 15, Applicants' radiation transmissive periodic diffraction grating 25 16, an output focusing lens 17, a multiplicity of signal receptors denoted generally at 18, which can be a conventional photodetector array comprising individual detectors connected with preselected information receptor channels 1' to n' (not shown). The collimated 30 input radiation spectrum beam is denoted 19, whereas two examples of the output radiation spectra are denoted 20a and 20*.
The conventional optical ray approach is taken in the following explanation of operation, it being understood ^ that the complexities encountered in multiple optical signal processing cannot be shown adequately in drawing representation. However, Applicants' radiation transmissive diffraction grating 16 does operate in a manner which is understandable by application of the 40 principles elucidated, thus corroborating in design what was certainly not predictable in advance by theory.
Referring to FIG. 3, Applicants' radiation transmissive diffraction grating generally comprises a multiplicity of optically reflective elements 16a to 16m which are spaced apart at regular intervals A (typically 1 or 2 jxm) with the grating radiation interaction length 1, and width, w, very much greater than the input radiation beam 19 width b (typically 0.5 to 1.0 cm).
Grating 16 must be sufficiently long for Bragg diffraction to occur at the center wavelength in the material, Xc. The required condition is given by the so-called thick-grating approximation:
assuming that the medium is optically isotropic. The same effect occurs in an optically non-isotropic material, although equation (2) would then be in tensor form to reflect the directional dependence of the medium. Such a case will not be considered further, as it is not instructive with respect to this invention.
When the Bragg Condition is satisfied, the output beam 19a will be diffracted through an angle of 26 B, or 6r=0B with respect to the grating bars. Any deviation from the Bragg Condition reduces the power of the radiation in the first order diffraction, as some of the power is directed into higher orders of diffraction. However, such losses can be tolerated to a substantial degree in the apparatus of this invention.
Any shift AX0 of input wavelength from the wavelength in the material, Xc=Xoc/n, for which the Bragg Condition is satisfied will result in a shift A0r in output angle, as portrayed in FIG. 4A, and some decrease in output beam intensity.
Referring to FIG. 4B, the radiation reflected from any two adjacent grating bars C and D must add up in phase along the direction of the diffracted beam. For this to occur, the path difference AO +OB of a given optical wavefront resulting from reflection from any pair of adjacent grating bars separated by a distance A must be equal to an integral number of wavelengths, m. From FIG. 4B:
![[blocks in formation]](http://www.google.fr/patents?id=CysxAAAAEBAJ&hl=fr&ie=ISO-8859-1&output=text&pg=PA7&img=1&zoom=3&hl=fr&q=&cds=1&sig=ACfU3U1-D9d0FomlT4-qvUaWSxJXLnyU-w&edge=0&edge=stretch&ci=508,632,396,211)
5 6
where the trigonometric relation sin2+cos2= 1 has which can be multiplexed by a given design of the Apbeen used. Substituting equation (2) into equation (7) plicants' invention. A second figure of merit for the the relation multiplexer constitutes the angular channel separation
as the magnitude of the ratio of the incremental angular A\0 2AX„ (8) 5 dispersion of the spots differing in wavelength by an
A9'= nAcos0£ = \|Ao Tf7 amount of AX0 to the angular half-power width of the
spots:
can be derived. It should be noted that equation (8) is
valid only for conditions which satisfy or approxi- \o \ ^0 \ { b\ \/ \ mately satisfy the Bragg Condition as presented in equa- | 1=« I If Ac^gB J = tion (2). 1 ^ 0 \ /
Fundamental diffraction theory (refer to R. G. Hunsperger, Integrated Optics: Theory and Technology, / A\„ V 2nb
Springer-Verlag (1982), pp. 151-152) teaches that, if the 15 \K~ )[ M 2 —7 j
beam width b is much greater than the grating period A, («)-<>then the far-field pattern consists of a set of diffraction
maxima with an angular half-power width given by: This is illustrated in FIG. 7. The greater the ratio given
in equation (13) the less the diffraction spots will overx lap, and, hence, the cross-talk between channels will be
Afli = -j- = -jj- 20 reduced. A value of 5 or greater should be sufficient for
most designs.
where b is the width of the input beam and X0 is the The dimensions of the individual photodetectors, or wavelength of the radiation in free space. One example their equivalent, in array 18 is determined by the spotof a diffraction maximum is shown in FIG. 5. A set of 25 size P and the spatial separation S of the output chandiffraction maxima is illustrated in FIG. 6. nels. Both P and S are functions of the focal length F of
The radiation will undergo several orders of diffrac- the focusing output lens 17. Assuming the system uses a tion by the grating resulting in a series of peaks angu- near-diffraction-limited lens, the output spots will be larly separated by: separated by an amount:
30
The first and second-order diffractions are illustrated in , Ka ■ ■ _r -m. * * » • n n
_ , ... - . , . . , . where A0ris in radian units. The output spot size P will
FIG. 5, where the first-order diffraction is diffracted at 35 , r r
an angle of dr\ with respect to the grating and the se- e'
cond-order diffraction is diffracted at an angle of dr2
with respect to the grating and at an angle of A&2 with p _ g^8lF _ gAoF *!5)
respect to the first-order diffraction. Also shown in nb FIG. 5 is one diffraction spot for the first-order diffrac- 4Q
tion and one diffraction spot for the second-order dif- where A0/ is in radian units and g is a constant deterfraction, mined by the definition of the optical focal spot resolu
Most of the power of the incident beam will go into tion. For a Gaussian output beam truncated at the 3 dB the first-order diffraction when the Bragg Condition is points, g=l. For a Gaussian beam truncated at 1/e exactly satisfied. The input angle 0,- is chosen to satisfy 45 points in the Fourier transform plane, g= 1.21. At the this condition at the center wavelength \oc- As the 1/e2 points, g= 1.64, and if the beam is truncated at the wavelength of the input signal varies from the center full main lobe spot size, g=2.48. wavelength by a small amount AX0, the Bragg Condi- The photodetectors or their equivalents for a demultion is no longer exactly satisfied, thereby increasing tiplexer should be spaced at regular intervals, with cenamounts of optical power diffracted into other orders of 50 ter-to-center spacing determined by the value of S. The diffraction, concurrently reducing the intensity of the detectors or their equivalents should be larger than the spot diffracted into the first-order diffraction. value of P, so as to capture the maximum amount of the
As shown in FIG. 6, the output radiation will be signal available, and smaller than the value of S so that angularly scanned according to equation (8) and the the output devices do not overlap. For a multiplexing intensity delivered will follow a bell-shaped pattern 55 device, the appropriate optical sources would be placed having an angular half-width given by: on centers determined by S.
In designing a planar transmission grating wave4fl3 2A (11) length-division multiplexer (or demultiplexer), the fol
1 lowing considerations apply:
60 (1) A large Bragg Angle, Ob, for the center waveOne figure of merit, the number of resolvable spots, is iength( x^, yields the maximum angular scan for a given by: change in wavelength, AX0, according to equation (8).
(2) A small angular half-power width of diffraction _ A#3 _ in\b ('2) maxima, A#i, yields well-defined spots and thus in
_ A0| _ x0i 65 creases the number of resolvable spots according to
equation (12).
and is respresented diagramatically in FIG. 6. N repre- (3) A large angular half-power width of intensity sents the uppermost limit on the number of channels maximum of the angularly scanned beam, A03 (refer to
FIG. 6), permits a large scan range which increases the number of resolvable spots and hence the maximum number of channels a given multiplexer/demultiplexer design can support.
(4) A large angular separation of peaks (which are coincident with the beam centers) for higher-order diffractions, A02 (refer to FIG. 5), decreases the optical power lost from the first-order diffraction, thereby decreasing device insertion loss.
To build the system, it is feasible to butt-couple to the planar waveguide 14 (see FIG. 2) an input optical fiber 7 polished at its distal end to provide a lens effective to cause radiation divergence to the beam width desired. An optical waveguide consists of a material with low attenuation at the wavelengths of interest surrounded by materials of lesser refractive index to produce total internal reflection of the optical signal, thus guiding the optical wave (refer to T. Tamir, Integrated Optics, Springer-Verlag (1979), pp. 15-61). Waveguide 14 is preferably designed to be single-mode for the radiation wavelengths to be transmitted, although this is not essential to device operation. The waveguide 14 is formed on the substrate 10 which provides mechanical support. Waveguide 14 can be sputter deposited or deposited from solution on an optical flat. Using glass optical flats for the substrate 10, dopant diffusion, ion exchange, and ion implantation can be used to form a thin region with index of refraction greater than that of the substrate to form the waveguide 14. Using semiconductor materials for the substrate 10 (which permits monolithic integration of electronic devices into the multiplexer system), waveguides can be fabricated by both sputter and solution deposition of multilayer structures having the required change of refractive index to cause waveguiding, as well as carrier concentration reduction via proton bombardment and epitaxial growth. The passive semiconductor (incapable of photon generation through stimulated emission) silicon presents a unique advantage for use as a substrate in that a waveguide can be produced on a layer of thermally grown silicon dioxide which has low refractive index and serves as an isolation layer between the waveguide and the silicon substrate, reducing optical power loss by reducing the penetration of the evanescent tail of the optical signal into the silicon substrate.
Waveguide 14 should possess low attenuation at the optical wavelengths at which it is to function (typically at wavelengths ranging from 0.8 to 1.5 u-m). Operative waveguide materials include optical glasses, lithium niobate, lithium tantalate, silicon dioxide, silicon oxynitride, tantalum pentoxide, niobium pentoxide, polymethylmethacrylate (PMMA), photoresist, polyurethane epoxy, the polymerized organic monomers vinyltrimethylsilane (VTMS) and hexamethyldisiloxane (HMDS), the passive (incapable of photon generation) semiconductor materials silicon, zinc selenide, and zinc telluride, and the active (capable of photon generation) semiconductor materials gallium arsenide, gallium aluminum arsenide, gallium indium arsenide, gallium indium phosphide, gallium arsenide phosphide, and gallium indium arsenide phosphide.
Lenses 15 and 17 can be of a number of conventional varieties. The most appropriate lens types are geodesic and Luneburg lenses. The geodesic lens is produced by grinding an appropriately shaped depression into the substrate 10 surface prior to formation of the waveguide 14. Geodesic lenses are not appropriate for use with semiconductor substrates because of the grinding in
,063
8
volved. Luneburg lenses are deemed most appropriate for use with semiconductor substrates as they are formed by depositing a circularly symmetric protrusion of high refractive index material on the surface of the 5 isolation layer prior to deposition of the waveguiding layer. It is also possible to fabricate by conventional means (ion implantation, proton bombardment, dielectric strip-loading or topographic inhomogeneity) acceptable geometric lenses by forming an appropriately 10 shaped region on or in the waveguide having an effective index of refraction sufficiently different from the waveguide 14 for optical refraction to occur at the lens/waveguide boundary. Grating 16 can likewise be fabricated by different 15 conventional methods, e.g., (1) etching grating bars into the waveguide surface either chemically or via ion milling, (2) depositing metal or dielectric bars on the surface of the waveguide, (3) inducing a surface acoustic wave of preselected period via a piezoelectric or magneto20 strictive transducer, (4) using the electrooptic effect to produce a periodic variation of the refractive index of the waveguide 14 by applying an electric potential across interdigitiated contacts deposited on the waveguide surface or across finger contacts deposited on the 25 waveguide surface using the substrate material as a ground plane, or (5) other methods effective tc cause a controlled periodic variation of the refractive index of the waveguide. As hereinbefore mentioned, the apparatus of this 30 invention can be optionally employed to either multiplex or demultiplex optical signals via wavelength-division.
For multiplexing optical sources onto a single optical fiber, it is practicable to either couple lasers or LED
35 optical sources (appropriately focused using lenses) or integrate these optical sources monolithically onto the planar substrate 10. This latter option requires that the substrate on which the waveguide 14 is formed be an active semiconductor material (capable of stimulated
40 photon emission) such as gallium arsenide, gallium aluminum arsenide, gallium indium arsenide, gallium indium phosphide, gallium arsenide phosphide, gallium indium arsenide phosphide, or other Group III-V and II-VI (binary, ternary, or quaternary) direct (or near
45 direct) bandgap materials. It is also practicable to couple optical fibers from remote optical sources to the waveguide as inputs, although this may result in unacceptable optical power loss. Coupling of input or output devices to the waveguide can be achieved in several
50 ways (refer to R. G. Hunsperger, ibid, pp. 88-106). Common methods include (1) butt-coupling directly to the exposed edge of the waveguide; (2) coupling via a prism through the planar waveguide surface; (3) coupling via a grating through the waveguide surface; (4)
55 coupling via a tapered film coupler.
Operation of the system of the Applicants' invention as a multiplexer is essentially the reverse of operation for the demultiplexing embodiment as hereinbefore described. Referring to FIG. 2, optical sources are sub
60 stituted for the photodetector receivers 18, placed on centers S, as determined by equation (14). The optical sources are amplitude modulated by conventional means according to the information to be transmitted. Semiconductor laser diodes are deemed to be particu
65 larly applicable to our invention due to their narrow spectral response and potential for single-mode operation. This does not, of course, preclude the use of other types of laser sources. LED optical sources would re
|
