Integrated three-dimensional optical multilayer usingfree-space optics

Manfred Jarczynski, Thomas Seiler, and Jürgen Jahns

An integrated three-dimensional optical multilayer system for optical data communications is presented.It is based on the use of free-space optical light propagation and combines two integration principles,namely, planar and stacked integration. The combination of both integration schemes aims at a maximaldesign flexibility for complex geometric layouts. On the other hand, packaging issues that stem fromassembly and tolerance have to be considered. Here we describe the basic concept and demonstrate theimplementation of an optical interface module in a processor-memory bus. © 2006 Optical Society ofAmerica

OCIS codes: 200.2000, 200.2610, 200.4650, 130.0130, 130.2790, 130.3120.

1. Introduction

Optics is on the way to becoming an alternative in-terconnection technology for computer communica-tion. As for reasons why, one can state, e.g., its largebandwidth, its low latency, and its use of the thirddimension. Optics allows one to overcome the crucialbottleneck of communication speed between chip andonboard devices.1 The strengths of fiber optics are theareas related to point-to-point interconnects with arelay distance from several kilometers down toseveral tens of centimeters in communication and inrack-to-rack networks, respectively. In these fields,they are an established technology, for onboard oron-chip interconnection fibers are no longer the bestchoice. At this interconnect level, the channel densityincreases and tasks imply fanning operations. Alter-native optical concepts are embedded waveguideand integrated free-space optics. Currently, polymer-based waveguides reach the level of low-cost devicesin optoelectronic-printed circuit boards (OE-PCB).2,3

However, the achievable interconnect density andthe quantity of fan-out is currently not comparablewith those on electronic chips. Especially the geomet-ric and topological limits are a critical issue. Here, the

strengths of free-space optics (FSO) such as a high-space–bandwidth product, flexible beam shaping, andsplitting are the advantages that help to overcomethe geometric and topological limitations. The inher-ent property of free-space optics is the use of localizedoptical elements, e.g., lenses, that overtake the taskof the waveguide that is completely localized with thelight path. In FSO, signal paths cross each otherwithout any interference, and the medium betweentransmitter and receiver can be used for arbitraryinterconnection. By use of integrated FSO, thesefeatures are kept. Additionally, similar microfabri-cation concepts help to match optics and electronics.Several integrated free-space optical modules, e.g.,for clock distribution4 or data communication,5–7

were demonstrated. The typical interconnect dis-tance is of the order of 3–30 mm. Having these fea-tures in mind, it is a small step to use integrated FSOas an interface between OE and fiber or waveguideoptics in optical-bus systems.

The European Union-funded project, High-speedOptoeLectronic Memory Systems (HOLMS) was initi-ated to show the feasibility and advantages of such anoptical-bus system.8 The application is a JEPG2000decoder that runs on an optical processor-memory busin a multiprocessor system. The bus concept is basedon three pillars: a board-to-board fiber bus, an on-board waveguide bus, and a free-space optical inter-face for coupling signals from–to OE components, e.g.,laser diode or photodetector arrays. To keep systemssmall and to match OE and optical input�output (I�O)ports, integrated FSO is required. Two viable ap-proaches to integrate FSO have been investigated ex-tensively: stacked planar optics9 (SPO) and planar-

The authors are with the Lehrgebiet Optische Nachrichtentech-nik, Fern Universität in Hagen, Universitätsstrasse 27�PRG,Hagen 58084, Germany. M. Jarczynski’s e-mail address is manfred.jarczynski@fernuni-hagen.de.

Received 13 January 2006; revised 6 June 2006; accepted 7 June2006; posted 8 June 2006 (Doc. ID 67227).

0003-6935/06/256335-07$15.00/0© 2006 Optical Society of America

1 September 2006 � Vol. 45, No. 25 � APPLIED OPTICS 6335

integrated free-space optics10 (PIFSO). We discussthese concepts and introduce their combination as anintegrated 3D free-space optical multilayer (Section 2)that performs the tasks of the interface module. Sec-tions 3 and 4 describe the investigations into the de-sign, fabrication, and assembly, as well as, the testsetup and the experimental results, respectively. Thepaper closes with a conclusion and an outlook (Section5) about the potential of the concept of a 3D free-spaceoptical multilayer.

2. Merging Planar and Stacked Free-Space Optics

Originally, integrated optics was exclusively associ-ated with waveguide optics. Miller11 introduced thisterm in 1969. However, with the works of Iga et al.9in 1982 and Jahns and Huang10 in 1989, the idea ofintegration was also introduced to FSO. As for elec-tronic chips, integrated optics is generally connectedto the ideas of (1) low-cost mass fabrication, (2) re-duced need for optical alignment, and (3) stable set-ups that are (4) widely immune to external influencesand provide (5) a platform for various materials andtechnologies.

The proposal of Iga et al.9 describes an integrationof optical components on several planar substratesthat are stacked together, which is illustrated in thetop part of Fig. 1. They introduce the term SPO tospecify this integration method. The optical elementsare integrated on the surface or in a small depth of thesubstrates. Several of these substrates are stacked to-gether to form the optical system. The strengths ofSPO are found especially for high parallel on-axis op-tical interconnects with passive components, e.g., fiberinterconnects with gradient-index optical elements,see Refs. 9 and 12. SPO is less appropriate for complexoptical systems that have to perform a complex topol-ogy or a large number of fan-out or fan-in. The impor-tant challenge of the SPO concept is the assembly ofthe planar substrates with high accuracy. This means,the macroscopic substrates have to be aligned andfixed with microscopic or submicroscopic tolerance. Incomparison with the conventional free-space opticswhere air usually serves as the transfer medium, thisconcept uses the medium of the substrate. WhereasFSO in air is mostly limited by the thickness of theoptical elements (or their surface profile), the thicknessof the planar substrate is the key parameter for theSPO concept. According to the stack of planar sub-strates and the placement of optical elements on theirsurface all optical subsystems (or channels) have thesame properties, in general.

Planar-integrated free-space optics (PIFSO) con-siders the above-mentioned integration features, too.PIFSO is based on folding the 3D optical setup into a2D layout, where the optical elements are placed onboth main surfaces of a planar substrate (center partof Fig. 1). Due to the folded optical path, the optical-path length between two elements is not so stronglycorrelated with the substrate thickness as it is in theSPO concept. Besides, along the axis of propagation,the signals are accessible at every surface for miscel-laneous interaction with passive or active optical or

electronic elements. Herewith, the PIFSO concept in-creases the degree of design freedom in comparison tothe SPO concept. For fabrication, standard process-ing methods such as lithography and etching, injec-tion molding, casting, or embossing, as well asbonding are suitable. The compatibility of the PIFSOconcept with surface-mounting techniques allows forthis concept to be used as a platform for electronicand OE components integration. The monolithicsetup is the important advantage of this concept andreduces the packaging task to a minimum. Micro-optical systems that are based on the PIFSO conceptwere realized for several applications, e.g., clock dis-tribution,4,13 planar-optic-disk pickup,14 matrix multi-plier,15 ultrathin-folded imager,16 or cryptographicdecoder.17

The integrated 3D free-space optical multilayermerges the two concepts described above. In Ref. 18,the idea of coupling a small number of substrateswith optical elements placed on outer and interimsurfaces and assembled into one package is already

Fig. 1. Schematic to point up the merging of stacked planar andplanar-integrated FSO. Stacked planar optics is based on an as-sembly of stacked substrates with optical components integratedon their surfaces (top part). In planar-integrated FSO, the opticalpath is folded inside a substrate (center part). This leads to amaximized degree of integration and a minimized packaging effort.A 3D free-space optical multilayer results from a combination ofboth concepts (bottom part). Herewith, an optimized relationshipbetween design freedom and monolithic integration can beachieved.

6336 APPLIED OPTICS � Vol. 45, No. 25 � 1 September 2006

mentioned. By following and extending this idea, wearrive at a 3D free-space optical multilayer. The mainfeature of the PIFSO concept, the folded optical path,is retained (see bottom part of Fig. 1). In common,this concept allows an increased flexibility for theoptical design and the topology. In fact, this meansthat more complex geometric layouts can be realized.The advantages are exemplified in the schematic bymeans of the optics shown in the schematics for SPOand PIFSO. However, the concept of an integrated 3Dfree-space optical multilayer reinserts the necessityof assembly. It is the task of the designer to balancethe convenience for optical design and the challengeof packaging to find the optimal trade-off. The con-cept includes all features of integrated optics, espe-cially if it can be seen as a modular platform forvarious materials and technologies. Of course, ther-mal and material issues have to be considered.

3. Design and Implementation of the Optical Interface

In the optical-bus concept of the HOLMS project, in-tegrated FSO was intended to serve as an interface(module) among waveguides, fibers, and OE compo-nents. To ensure an overall power budget of 21.8 dB,VCSEL diode arrays (ULM-850-05-TT-A0112B) withan optical output power of 1.5 mW and photodetectorarrays with 10 �W of incident optical power are used.The planar free-space optical modules are the carrierfor MT connectors and the multichip modules19

(MCM) that contain fiber-ribbon cables for board-to-board interconnection and electronic and OE chips,respectively. These populated modules are insertedvertically into the OE-PCB. Geometric layout con-straints resulting from electronic processing and as-sembly of the MCMs and the OE-PCBs challenge thedesign of the free-space optical modules (see Fig. 2).The task gets more complicated by the fact that ar-rays of multimode transmitters with large cross sec-tions (waveguide core with up to 70 �m � 70 �m)

have to be relayed with high efficiency. The 3D opticalmultilayer appears as a promising approach to solvethese challenges. The optical system design of theprototype is shown in Fig. 2(c). This picture displaysthe bottom part of the interconnects that is presentedin the 3D visualization in Fig. 2(a). The optical sub-systems for the long and the short distances are la-beled as subsystems A and B in the interconnectschematic of Fig. 2(b), respectively. The optics of sub-systems A and B are based on a hybrid imaging20 anda multichannel system,21 respectively. The intercon-nection to fiber bundles is also achieved by a similarhybrid imaging system.

The hybrid imaging optics used for subsystem Aconsists of an incoupling and an outcoupling lensletarray with an interim central imaging lens. The cou-pling elements are placed on the outer surfaces ofsubstrate 1 or 4, whereas the central imaging lens isfabricated on the right surface of substrate 2 [see Fig.2(c)]. The optics are designed for point sources at awavelength of 850 nm and with a NA of 0.1. Theeffective interconnect distance is 5 mm, whereas thelength of the optical path is 43.65 mm, because ofthe module thickness of 12 mm. The coupling dis-tance between the OE elements on the MCM and thePIFSO is optimal with 750 �m in quartz glass�n � 1.465�. A microlens implemented directly abovethe laser diodes on the MCM22 reduces the trans-mitter NA from 0.24 to 0.1. The coupling distance be-tween a waveguide or fiber and the PIFSO is 900 �m.The theoretically achievable efficiency, consideringFresnel, geometric, and diffraction loss, is �5.3 and�9.5 dB for coupling from laser diodes and wave-guides or fibers, respectively. The calculation is basedon a homogeneous illumination of the incoupling el-ement and the first-order intensity equation forquantized phase functions: � � sinc2 �1�N�, where �is the diffraction efficiency, and N is the number ofphase levels. The geometric loss was considered by a

Fig. 2. (Color online) Schematics to demonstrate tasks and implementation of a 3D optical interface module in a processor-memory-businterconnect. (a) Three-dimensional visualization of the PIFSO module packaged with MCM MT-housed fiber bundles, and PCB-embeddedwaveguides. (b) x–z cross sections showing the specification of the free-space interconnect. (c) One part of several similar optical interconnectsto explain the arrangement of optical elements. Keys A, B, optical subsystems; LD, laser diode; MT, multitermination; PCB, printed circuitboard.

1 September 2006 � Vol. 45, No. 25 � APPLIED OPTICS 6337

ray-tracing analysis tool.23 A ray-tracing-based toler-ance analysis of a similar optical design was pub-lished in Refs. 24 and 25. The analysis resulted in aloss of rays of less than 6% or 27% for 5 �m or for10 �m lateral transmitter misalignment, respectively.For less than �112 �m, the longitudinal transmittermisalignment geometric loss is lower than 1%.

The subsystem design (A) is based on a multichan-nel approach. The design consists of seven opticalelements that relay the light through the substrate.The incoupling element L(1), an off-axis lens, imagesthe transmitter (a laser diode–microlens combinationon the MCM) via grating G(2) on the surface of L(3).Lens L(3) images the surface of L(1) onthe surface of L(4) with a magnification of �1�2,while L(4) images the plane of L(3) to L(5). L(4) is thesymmetry axis of the design, so that G(6), L(5), andL(7) are the conjugated elements to G(2), L(3), andL(1), respectively. The necessary deflection angle isachieved by a cascade of two elements, off-axis lensL(1) and grating G(2). By splitting the two tasks, i.e.,reduction of the divergence and achieving a largedeflection angle among the two elements, an opticalsystem with highly efficient multilevel �N � 8� ele-ments is feasible. Theoretically, an efficiency of�5.5 dB can be achieved. A simulation, based on raytracing, shows that the design allows a light transferwithout geometric and NA loss for incoupling from theabove-described laser diode arrays.

The optical elements were fabricated by means ofmultimask binary photolithography and reactive ionetching upon fused-silica substrates.25 For the reflec-tive coating, we used a thin layer of silver that iscovered by aluminum. Both coatings were e-beamevaporated. For the alignment and assembly proce-dure, two different techniques were applied: (a) vi-

sual adjustment of marks, e.g., dark cross over brightcross, where consecutive surfaces have to be alignedand (b) an imaging alignment method.26 In principle,

Fig. 3. (Color online) Assembled3D optical multilayer module.Four planar-cemented substratesof 3 mm thickness are shown. Thecomplete stack is saved by protec-tive substrates (0.5 mm thick).Surfaces with functionality arelabeled with slf, slr, etc. The fin-ished module is an interface for192 optical channels. The OEMCM will be fixed on the visibletop side.

Fig. 4. (Color online) (a) Experimental setup to test the 3D opticalmultilayer by use of MT-housed fiber bundles and MIPs. The pho-tographs in (b) and (c) show the front MIP with four fiber bundlesand aligned to the 3D optical multilayer, respectively. Key: A, 3Doptical multilayer module; B, chuck for module; C, MIPs; D, gimbalmount for MIP alignment; E, CCD cameras; F, fiber bundles; G, lasertransmitter with ST pigtail; H, sensors of powermeter with FC�PCpigtail.

6338 APPLIED OPTICS � Vol. 45, No. 25 � 1 September 2006

the latter method uses a mark consisting of conju-gated quadrants and a reflective lens on differentsubstrates. The mark and the reflective lens areplaced relative to the known coordinate system of thesubstrates. Both substrates are optimally aligned, ifthe opaque quadrants of the conjugated mark areimaged to the bright areas. The alignment processwas assisted by the precise alignment mechanic of amodified mask aligner (Suess, MA4). The cementa-tion of the substrates is accomplished by UV-curingglue (Bohle, B-665-0) with low viscosity that allowsfor very thin layers. After assembly of the four sub-strates that form the 3D free-space optical multi-layer, thin glass substrates were attached to protectthe optical elements on the outer surfaces. The fin-

ished 3D optical multilayer with a capacity of 192channels is shown in Fig. 3.

4. Experimental Results

Before the 3D optical multilayer is assembled withthe MCM and the PCB-embedded waveguides asshown in Fig. 2(a), all optical I�Os were tested by useof multitermination (MT)-housed fiber ribbon cables.Therefore metal interface plates (MIP) with MT sock-ets for all I�Os were fabricated by micromachining inbronze alloy. The size and the position were achievedwith an accuracy that is better than 5 �m. The topside of the front MIP with MT sockets and four MT-

Fig. 5. First experimental results for the hybrid imaging system(subsystem A).

Fig. 6. (Color online) Incoupling and outcoupling sections of the multichannel optics (subsystem B). Pairs of active channels are visiblein the photographs on the right-hand side.

Fig. 7. Incoupling and outcoupling sections of the hybrid imagingoptics (subsystem A). Pairs of active channels are visible in thephotographs on the right-hand side.

1 September 2006 � Vol. 45, No. 25 � APPLIED OPTICS 6339

housed fiber bundles is shown in Fig. 4(b). Figure 4(c)shows the MIP aligned to the 3D. optical multilayer.The complete setup for the efficiency test of the 3Doptical multilayer module is shown in of Fig. 4(a).

For the experimental test, a single-mode fiber bun-dle (F-SBA, Newport, 0.17 NA) that is terminatedwith 12 ferrule connector/physical contact (FC/PC)connectors and one 12�-MT connector is used as atransmitter. A similarly terminated multimode fiberbundle (gradient index, 62.5�125, 0.27 NA, 12�FC�PC, 12� MT) constitutes the target. A lasertransmitter [BCP 400, 50�125, straight tip (ST) pig-tail] and an optical powermeter (Ando AQ 2140 withdetector AQ 2741) serve as a light source and fordetection, respectively. Due to the use of single-channel connectors, a defined measurement of theefficiency and the cross talk for each channel is avail-able. The efficiency of the module is measured bydetermining the difference from a calibration mea-

surement (with butt-coupled MT connectors) and ameasurement with the inserted module.

First results for subsystem A are presented in Fig.5. The values are in the range of �7.37 and�8.31 dB, which is approximately 2 dB over the the-oretical feasible efficiency. Channels 1 and 10 aremismeasurements due to a damaged fiber and afailed calibration, respectively. Except for these chan-nels, the homogeneity is given with less than 1 dB.The cross-talk suppression is larger than 31 dB. Thedifferences between experimental and theoreticalvalues can be explained by the mismatch between thedesign and the experimental parameters: First, theoptics are optimized for a NA of 0.1, and the trans-mitter fiber has a NA of 0.17. Second, the microlenson the MCM that could decrease the NA to 0.1 is notin use. Obviously, channel 10 shapes up as an outlier.A visual impression of the interconnect is given bysome observation pictures showing pairs of activechannels for subsystems A and B in Figs. 6 and 7,respectively. For this, the outcoupling MT connectorswere removed and additionally illuminated to makethe surrounding area visible.

The concept for the precise assembly of PIFSOmodules and OE-PCBs was demonstrated by useof another PIFSO prototype. Figure 8(b) shows thePIFSO prototype inserted in the OE-PCB. The test ofthis interconnect yielded to a stable assembly withadequate alignment. From experimental results, theinterconnection via the complete system exhibits anaverage coupling efficiency of �8.4 dB in which aportion of the waveguide is �3.4 dB.

5. Conclusions and Outlook

We have presented an integrated 3D free-space opticalmultilayer system that combines two FSO integrationconcepts: stacked planar and planar-integrated free-space optics. On the one hand, this combination fea-tures flexible designs for complex topologies and highinterconnect density. On the other hand, it implies amore complex assembly and packaging process. For anoptimum implementation of such systems, these trade-offs have to be balanced.

The integrated 3D free-space optical multilayermodule serves as an interface in an optical processor-memory bus. The optical design has been presented,and the steps of implementation have been described.Finally, we have tested the function and efficiencyexperimentally. The first results confirm a good ho-mogeneity, an average efficiency of approximately�7.73 dB and a good cross-talk suppression. The ef-ficiency can be further increased by using refractive–reflective elements with lower individual losses.Various efforts are under way to demonstrate an im-proved system. Improvements in efficiency of up to�4.43 dB by the use of refractive–reflective couplingelements (fabricated by gray-scale lithography andreplication) for an interconnect module with a similaroptical design have been achieved.27 However, it isobvious that an optical processor-memory-bus inter-connection with a power budget of 21.8 dB as speci-fied during the course of the HOLMS project is

Fig. 8. (Color online) (a) Laboratory setup to check the opticalinterconnection between PIFSO prototype and PCB-embeddedwaveguides. (b) PIFSO prototype with alignment structures in-serted into the OE-PCB. (c) Picture of the OE-PCB abutting sur-face with one active channel. The light signal was transferred viathe PIFSO prototype and the waveguides.

6340 APPLIED OPTICS � Vol. 45, No. 25 � 1 September 2006

achievable. By use of a previous PIFSO prototype, wehave also demonstrated that a precise alignment andassembly of PIFSO modules and OE-PCB is feasible.A similar assembly is also planned for the 3D free-space optical multilayer.

The concept of a 3D free-space optical multilayercan also be seen from a more general point of view.With several I�Os on the surfaces and potentially onall surfaces in the geometric 3D space, one can intro-duce a free-space optical cube that interconnects thelevels of PCB-embedded multilayer waveguides andOEs. The result can be a free-space node that servesas a passive connector in the optical data communi-cation bus.

The authors thank all partners in the HOLMSConsortium and Matthias Gruber for helpful discus-sions and hints as well as for the congenial collabo-ration. This effort was sponsored by the EuropeanCommission during the Fifth Framework Programmeof the Information Technology Society under agree-ment IST-2002-35235.

References1. D. A. B. Miller, “Rationale and challenges for optical intercon-

nects to electronic chips,” Proc. IEEE 88, 728–749 (2000).2. C. Berger, R. Beyeler, R. Dangel, L. Dellmann, F. Horst, T.

Lamprecht, T. Morf, B. J. Offrein, F. Yamada, M. Hasegawa,H. Numata, and Y. Taira, “Optical interconnect demonstratorwith embedded waveguides and butt-coupled optoelectronicmodules,” in Proceedings of the OSA Topical Meeting on In-formation Photonics, IWD II (Optical Society of America,2005).

3. H. Albrecht, A. Beier, P. Demmer, M. Franke, R. Mödinger, K.Pfeiffer, P. Beil, J. Kostelnik, J. Bauer, F. Ebling, H. Schröder,and E. Griese, “New-generation interconnect,” in Proceedingsof the OSA Topical Meeting on Information Photonics, IWD I(Optical Society of America, 2005).

4. S. J. Walker and J. Jahns, “Optical clock distribution usingintegrated free-space optics,” Opt. Commun. 90, 359–371(1992).

5. A. G. Kirk, D. V. Plant, M. H. Ayliffe, M. Chateauneuf, and F.Lacroix, “Design rules for highly parallel free-space opticalinterconnects,” IEEE J. Sel. Top. Quantum Electron. 9, 531–547 (2003).

6. C. Debaes, M. Vervaeke, V. Baukens, H. Ottevaere, P. Vynck,P. Tuteleers, B. Volckaerts, W. Meeus, M. Brunfaut, J. VanCampenhout, A. Hermanne, and H. Thienpont, “Low-costmicro-optical modules for MCM level optical intersconnec-tions,” IEEE J. Sel. Top. Quantum Electron. 9, 518–530 (2003).

7. G. Li, D. Huang, E. Yuceturk, P. J. Marchand, S. C. Esener,V. H. Ozguz, and Y. Liu, “Three-dimensional optoelectronicstacked processor by use of free-space optical interconnectionand three-dimensional VLSI chip stacks,” Appl. Opt. 41, 348–360 (2002).

8. P. Lukowicz, J. Jahns, R. Barbieri, P. Benabes, T. Bierhoff, A.Gauthier, M. Jarczynski, G. A. Russell, J. Schrage, W. Süllau,J. F. Snowdon, M. Wirz, and G. Tröster, “Optoelectronic inter-connection technology in the HOLMS system,” IEEE J. Sel.Top. Quantum Electron. 9, 624–635 (2003).

9. K. Iga, M. Oikawa, S. Misawa, J. Banno, and Y. Kokubun,

“Stacked planar optics: an application of the planar microlens,”Appl. Opt. 21, 3456–3460 (1982).

10. J. Jahns and A. Huang, “Planar integration of free-space op-tical components,” Appl. Opt. 28, 1602–1605 (1989).

11. S. E. Miller, “Integrated optics: an introduction,” Bell Syst.Tech. J. 48, 2059–2068 (1969).

12. R. Klug and K.-H. Brenner, “Implementation of multilensmicro-optical systems with large numerical aperture by stack-ing of microlenses,” Appl. Opt. 38, 7002–7008 (1999).

13. M. Gruber, R. Kerssenfischer, and J. Jahns, “Cascadable fan-out module for parallel optical short-haul interconnects,” inProceedings of the OSA Topical Meeting on Optics in Comput-ing 2003 (Optical Society of America, 2003).

14. T. Shiono and H. Ogawa, “Planar-optic-disk pickup with dif-fractive micro-optics,” Appl. Opt. 33, 7350–7355 (1994).

15. M. Gruber, J. Jahns, and S. Sinzinger, “Planar-integrated op-tical vector-matrix multiplier,” Appl. Opt. 39, 5367–5373(2000).

16. E. J. Tremblay, J. Rutkowski, I. Tamayo, R. A. Stacko, R. L.Morrison, D. Combsc, J. Maderc, H.-C. Kim, U. Levy, Y.Fainman, and J. E. Ford, “Ultra-thin folded imager,” in Pro-ceedings of the OSA Topical Meeting on Computational Opti-cal Sensing and Imaging (Optical Society of America, 2005).

17. V. R. Daria, P. J. Rodrigo, S. Sinzinger, and J. Glückstad,“Phase-only optical decryption in a planar integrated micro-optics system,” Opt. Eng. 43, 2223–2227 (2004).

18. A. Huang and J. Jahns, “Mask controlled coupling of inter-substrate optical components,” U.S. patent 4,966,446 (30October 1990).

19. M. Wirz, R. Barbieri Carrera, P. Lukowicz, G. Troester, M.Jarczynski, and J. Jahns, “Combining optical and electricaldesign constraints in the HOLMS opto-electronic MCM com-ponents,” in Proceedings of the OSA Topical Meeting on Opticsin Computing (Optical Society of America, 2003).

20. A. W. Lohmann, “Image formation of dilute arrays for opticalinformation processing,” Opt. Commun. 86, 365–370 (1991).

21. S. Sinzinger and J. Jahns, Microoptics (Wiley–VCH, 1999),Chap. 7.2.

22. J. J. Casswell, S. Kumpatla, J. F. Snowdon, M. Jarczynski, andM. Wirz, “Board integrated micro-optics for divergence man-agement in opto-electronic interconnect systems,” in Proceed-ings of the OSA Topical Meeting on Information Photonics(Optical Society of America, 2005).

23. R. Bathel, S. Sinzinger, and J. Jahns, “Ray-tracing analysis ofplanar optical systems,” in OSA Topical Meeting on DiffractiveOptics and Microoptics (Optical Society of America, 2000),pp. 145–146.

24. M. Jarczynski and J. Jahns, “Planar free-space optical inter-face module based on a hybrid imaging system,” in EOS Top-ical Meeting, Optics in Computing 2003 (European OpticalSociety, 2004), pp. 21–23.

25. M. Jarczynski and J. Jahns, “Planar-integrated free-space op-tics for optical interconnects and fan-out�in operations,” inPhotonic Devices and Algorithms for Computing VI, K. M.Iftekharuddin and A. A. S. Awwal, eds., Proc. SPIE 5556,15–26 (2004).

26. M. Gruber, D. Hagedorn, and W. Eckert, “Precise and simpleoptical alignment method for double-sided lithography,” Appl.Opt. 40, 5052–5055 (2001).

27. R. Heming, J. Jahns, M. Gruber, L. Wittig, and E. Kley, “Com-bination of binary and analog lithography to fabricate efficientplanar-integrated free-space optical interconnects,” in Pro-ceedings of the Deutische Gesellschaft für angewandte optic,Wroclaw, Poland (German optical Society, 2005), Vol. 106.

1 September 2006 � Vol. 45, No. 25 � APPLIED OPTICS 6341