WO1997000458A1 - Optical multiplexing device and method - Google Patents

Abstract

An optical multiplexing device spatially disburses collimated light from a fiber optic waveguide into individual wavelength bands, or multiplexes such individual wavelength bands to a common fiber optic waveguide or other destination. The optical multiplexing device has application for dense channel wavelength division multiplexing (WDM) systems for fiber optic telecommunications, as well as compact optical instrument design. Multiple wavelength light traveling in a fiber optic waveguide is separated into multiple narrow spectral bands directed to individual fiber optic carriers or detectors. An optical block has an optical port for passing the aforesaid multiple wavelength collimated light, and multiple ports arrayed in spaced relation to each other along a multiport surface of the optical block. A continuous, variable thickness, multicavity interference filter (22) extends on the multiport surface (20) of the optical block over the aforesaid multiple ports. At each of the multiple ports the continuous interference filter transmits a different sub-range of the multiple wavelength collimated light passed by the optical port, and reflects other wavelengths. Multicolor light passed to the optical block from the optical port is directed to a first one of the multiple ports on an opposite surface of the optical block. The wavelength sub-range which is 'in-band' of such first one of the multiple ports is transmitted through that port by the local portion of the continuous, variable thickness interference filter (22) there, and all other wavelengths are reflected. The light not transmitted through the first port (16) is reflected to strike a second port, at which a second (different) wavelength band is transmitted and all other light again reflected. The reflected optical signals thus cascades in a 'multiple-bounce' sequence down the optical block (10) of the multiplexing device, sequentially removing each channel of the multiplexed signal. In reverse operation, individual channels are combined in the optical block and transmitted through the optical port.

Description

OPTICAL MULTIPLEXING DEVICE AND METHOD

Related Application

This application is a continuation-in-part of copending United States patent application of Michael A. Scobey, Serial No 08/490,829, filed on June 15, 1995 and

entitled Optical Multiplexing Device, and a continuation-in-part of copending United

States patent application of Michael A. Scobey, Serial No. 08/300,741 filed on September

2, 1994 and entitled Low Pressure Reactive Magnetron Sputtering Apparatus And

Method.

INTRODUCTION

The present invention is directed to an optical multiplexing device which spatially

disburses collimated multi-wavelength light from a fiber optic waveguide into individual wavelength bands, each of which can be directed to an individual fiber optic waveguide output line, light detector, etc., or multiplexes such individual wavelength bands to a common fiber optic waveguide or other destination. In certain preferred embodiments, the improved multiplexing devices of the present invention are particularly well suited for dense channel wavelength division multiplexing (DWDM) systems for fiber optic

telecommunications systems.

BACKGROUND While fiber optic cable is finding widespread use for data transmission and other

telecommunication applications, the cost of new installed fiber optic cable presents a

barrier to increased carrying capacity. Wavelength division multiplexing (WDM) would allow different wavelengths to be carried over a common fiber optic waveguide. Presently preferred wavelength bands for fiber optic transmission media include those centered at 1.3μ and 1.55μ. The latter is especially preferred because of its minimal absoφtion and the commercial availability of erbium doped fiber amplifiers. It has a useful band width

of approximately 10 to 40 nm, depending on application. Wavelength division multiplexing can separate this band width into multiple channels. Ideally, the 1.55μ wavelength band, for example, would be divided into multiple discreet channels, such as 8, 16 or even as many as 32 channels, through a technique referred to as dense channel wavelength division multiplexing (DWDM), as a low cost method of substantially increasing long-haul telecommunication capacity over existing fiber optic transmission lines. Wavelength division multiplexing may be used to supply video-on-demand and other existing or planned multimedia, interactive services. Techniques and devices are required, however, for multiplexing the different discreet carrier wavelengths. That is, the individual optic signals must be combined onto a common fiber optic waveguide and then later separated again into the individual signals or channels at the opposite end ofthe fiber optic cable. Thus, the ability to effectively combine and then separate individual wavelengths (or wavelength bands) from a broad spectral source is of growing importance

to the fiber optic telecommunications field and other fields employing optical instruments.

Optical multiplexers are known for use in spectroscopic analysis equipment and for the combination or separation of optical signals in wavelength division multiplexed fiber optic telecommunications systems. Known devices for this purpose have employed, for example, diffraction gratings, prisms and various types of fixed or tunable filters. Gratings and prisms typically require complicated and bulky alignment systems and have been found to provide poor efficiency and poor stability under changing ambient conditions. Fixed wavelength filters, such as interference coatings, can be made substantially more stable, but transmit only a single wavelength or wavelength band. In this regard, highly improved interference coatings of metal oxide materials, such as niobia and silica, can be produced by commercially known plasma deposition techniques, such as ion assisted electron beam evaporation, ion beam sputtering, reactive magnetron sputtering, e.g., as disclosed in U.S. patent No. 4,851,095 to Scobey et al. Such coating methods can produce interference cavity filters formed of stacked dielectric optical coatings which are advantageously dense and stable, with low film scatter and low absoφtion, as well as low sensitivity to temperature changes and ambient humidity. The theoretical spectral performance of a stable, three-cavity filter (tilted 12°) produced using any of such advanced, deposition methods is shown in Fig. 1 ofthe appended drawings. The spectral profile is seen to be suitable to meet stringent application specifications.

To overcome the aforesaid deficiency of such interference filters, that is, that they transmit only a single wavelength or range of wavelengths, it has been suggested to gang or join together multiple filter units to a common parallelogram prism or other common substrate. Optical filters are joined together, for example, in the multiplexing device of U.K. patent application GB 2,014,752A to separate light of different wavelengths

transmitted down a common optical waveguide. At least two transmission filters, each of which transmits light of a different predetermined wavelength and reflects light of other wavelengths, are attached adjacent each other to a transparent dielectric substrate. The optical filters are arranged so that an optical beam is partially transmitted and partially reflected by each optical filter in turn, producing a zigzag light path. Light of a particular wavelength is subtracted or added at each filter (depending upon whether the element is being used as a multiplexer or demultiplexer). Similarly, in the device of European patent application No. 85102054.5 by Oki Electric Industry Co., Ltd., a so-called hybrid optical wavelength division multiplexer-demultiplexer is suggested, wherein multiple separate interference filters of different transmittivities are applied to the side surfaces of a glass block. A somewhat related approach is suggested in U.S. patent 5,005,935 to Kunikani et al, wherein a wavelength division multiplexing optical transmission system for use in bi-directional optical fiber communications between a central telephone exchange and a subscriber employs multiple separate filter elements applied to various surfaces of a parallelogram prism. Alternative approaches for tapping selective wavelengths from a main trunk line carrying optical signals on a plurality of wavelength bands is suggested, for example, in U.S. patent 4,768,849 to Hicks, Jr. In that patent, multiple filter taps, each employing dielectric mirrors and lenses for directing optical signals, are shown in an arrangement for removing a series of wavelength bands or channels from a main trunk line.

Applying multiple separate filter elements to the surface of a prism or other optical substrate involves significant disadvantages in assembly cost and complexity. In addition, a significant problem associated with wavelength division multiplexing devices and the like

employing multiple discreet interference filter elements, arises from uncertainty as to the precise wavelength of a filter element as it is manufactured. That is, in the manufacture of multiplexing devices, wherein bandpass filter elements are produced separately, a device employing eight individual bandpass filters, for example, typically will require considerably more than eight coating lots to produce the necessary eight suitable filter elements. Bandpass filters (particularly in the infrared range) are extremely thick and require complicated and expensive vacuum deposition equipment and techniques. Accordingly, each coating lot can be expensive and difficult to produce. For this reason, devices employing, for example, eight separate interference filter elements to produce an eight channel WDM device, have been relatively costly and have not enjoyed full commercial acceptance.

Another problem associated with optical multiplexing devices employing multiple individual bandpass filter elements, involves the need to mount the elements in nearly perfect parallelism on an optical substrate. The filter elements are quite small, typically being on the order of 1 to 5 mm in diameter, and are, accordingly, difficult to handle with precision. Improper mounting ofthe filter elements can significantly decrease the optical accuracy and thermal stability of the device. A related problem is the necessity of an adhesive medium between the filter element and the surface ofthe optical substrate. The optical signal path travels through the adhesive, with consequent system degradation. In optical multiplexing devices intended for the telecommunications industry, preferably there is as little as possible epoxy adhesive in the optical signal path.

It is an object ofthe present invention to provide improved optical multiplexing devices which reduce or wholly overcome some or all ofthe aforesaid difficulties inherent in prior known devices. Particular objects and advantages of the invention will be apparent to those skilled in the art, that is, those who are knowledgeable and experienced in this field of technology, in view ofthe following disclosure ofthe invention and detailed description of certain preferred embodiments.

SUMMARY OF THE INVENTION

In accordance with a first aspect, an optical multiplexing device comprises an optical block which may be either a solid optical substrate, such as glass or fused silica or

the like, or an enclosed chamber which is hollow, meaning either evacuated or filled with air or other optically transparent medium. The optical block has an optical port for passing multiple wavelength collimated light. Depending upon the application of the optical multiplexing device, such multiple wavelength collimated light may be passed

through the optical port into the optical block to be demultiplexed, or from the optical block as a multiplexed signal to a fiber optic transmission line or other destination.

Multiple ports are arrayed in spaced relation to each other along a multiport surface ofthe optical block. As illustrated below in connection with certain preferred embodiments, the optical block may have more than one such multiport surface. Each of these multiple ports is transparent to the optical signal of one channel. Thus, each transmits a wavelength sub-range ofthe multiple wavelength collimated light passed by the optical port. In an application of the optical multiplexing device in a multi-channel telecommunication system, each ofthe multiple ports typically would pass a single discreet channel and, in combination, the channels form the aforesaid multiple wavelength collimated light transmitted by the optical port. A continuous, variable thickness interference filter, preferably a multi-cavity interference filter, is carried on the multiport surface of the optical block to provide the aforesaid multiple ports. Because this continuous interference filter extending over the multiport surface has a different optical

thickness at each ofthe multiple ports, the wavelength (or wavelength range) passed by the filter at each such port will differ from that passed at the other ports. Thus, a single film, preferably deposited directly onto the surface ofthe optical block, separately passes optical signals for each of a number of channels at separate locations, while reflecting other wavelengths. As noted above, the optical block may be either solid or a hollow chamber. In the case of a solid optical block, the multiport surface carrying the continuous, variable thickness interference filter would typically be an exterior surface of the block. As discussed further below, the individual ports ofthe multiport surface may be bandpass filters, preferably narrow bandpass filters transparent to a wavelength sub¬ range separated from the sub-range ofthe next adjacent port(s) by as little as 2 nm or even less for DWDM. Alternatively, some or all ofthe multiple ports could be dichroic, i.e., a long wavepass or short wavepass edge filter, preferably with a very shaφ transition point. The transition point of each port would be set at a slightly (e.g., 2 nm) longer (or shorter) boundary wavelength. In a demultiplexing operation, each port, in turn, would pass or transmit only optic signals in the incremental range beyond the boundary wavelength ofthe previous port, since all light at shorter (or longer) wavelengths would already have been removed. Light beyond the boundary wavelength of the new port would be reflected, in accordance with the above described principles of operation.

The optical multiplexing device further comprises means for cascading light within the optical block along a multi-point travel path from one to another ofthe multiple ports. In a demultiplexing operation, the optical signals would enter the optical block at the aforesaid optical port and travel to the multiple ports (acting in this case as output ports) along the aforesaid multi-point travel path. The signal for each individual channel is transmitted out of the optical block at its corresponding port; other wavelength are

reflected, or bounced, back to cascade further along the optical travel path within the optical block. It will be understood that at the last output port(s) there may be no remainder light to be reflected. It will also be understood from this disclosure, that the optical multiplexing device can operate in the reverse or both directions. The cascading

means preferably comprises a reflective film carried on a second surface of the optical block, either as a continuous coating spanning the multi-point travel path ofthe cascading light signals, or as multiple discreet reflector elements. The optical block would most typically be rectilinear, having the reflective film on or at a second surface ofthe optical block opposite and parallel to the multiport surface carrying the aforesaid continuous interference filter. This second film can be a broadband high reflector, that is, a film coating which is highly reflective of all wavelengths ofthe channels which combine to form the multiple wavelength collimated light, or can act as a second interference filter transparent at spaced locations (i.e., at some or each ofthe bounce points) to the optical signal of one or more ofthe channels. In either case, the interference filter and reflective film on spaced surfaces ofthe optical block operate to cascade optical signals through the optical block in a multiple-bounce sequence starting at (or finishing at) the optical port through which the multiple wavelength collimated light passes. This multiple-bounce cascading will be further described below in connection with certain preferred embodiments.

BRIEF DESCRIPTION OF THE DRAWINGS Certain preferred embodiments ofthe invention are discussed below with reference to the accompanying drawings in which:

Fig. 1 is a graph showing the theoretical performance of a high quality multi-

cavity, dielectric, optical interference filter.

Fig. 2 is a schematic illustration of a first preferred embodiment of an optical multiplexing device, specifically, a dense channel wavelength division multiplexing device for an eight channel fiber optic telecommunications system or like application;

Fig. 3 is a schematic illustration of an alternative preferred embodiment of an optical multiplexing device in accordance with the invention, specifically, a dense channel wavelength division multiplexing device for an eight channel fiber optic telecommunications system or like application;

Fig. 4 is a schematic illustration of another alternative preferred embodiment of an optical multiplexing device in accordance with the invention, specifically, a dense channel wavelength division multiplexing device for an eight channel fiber optic telecommunications system or like application;

Figs. 5, 6, and 7 are schematic illustrations, in cross-section, ofthe continuous, variable thickness, three cavity interference filter ofthe optical multiplexing device of Fig.

2; Fig. 8 is a cross-sectional schematic illustration of apparatus in accordance with a preferred embodiment of this invention;

Fig. 9 is a schematic representation in cross-section of a magnetron assembly including a source or target and an inert gas shroud in accordance with a preferred embodiment of this invention; Fig. 10 is a cross-sectional schematic illustration of apparatus in accordance with an alternative preferred embodiment of this invention, having multiple magnetron sputtering assemblies;

Fig. 11 is a graph showing the relationship between chamber pressure and chamber pumping speed assuming the magnetron pressure of 0.7 microns and a magnetron assembly conductance (C_M) of 3000 l/sec;

Fig. 12 is a graph showing the relationship between chamber pressure and chamber pumping speed assuming a magnetron pressure of 0.4 microns and a magnetron assembly conductance (C_M) of 3000 l/sec; and Figs. 13 and 14 are schematic elevation and plan views, respectively, illustrating apparatus in accordance with an alternative preferred embodiment of this invention.

It should be understood that the optical multiplexing devices and interference filter illustrated in the drawings are not necessarily to scale, either in their various dimensions or angular relationships. It will be well within the ability of those skilled in the art to select suitable dimensions and angular relationships for such devices in view of the foregoing disclosure and the following detailed description of preferred embodiments.

DETAILED DESCRIPTION OF CERTAIN PREFERRED EMBODIMENTS

The optical multiplexing device, as disclosed above, has numerous applications including, for example, in fiber optic telecommunication systems. Optical multiplexing devices of this design are particularly useful, for example, in test equipment and the like, as well as laboratory instrumentation. For puφoses of illustration, the preferred embodiments described below in detail are dense channel wavelength division multiplexing devices which can solve or reduce the above described problems of individually mounting multiple filter elements to an optical substrate for each individual signal channel, the problems of cost and complexity involved in multiple coating lots for preparing such individual filter elements, and the associated problems of filter wavelength uncertainty.

As discussed below in connection with the appended drawings, a graded wavelength all-dielectric narrow bandpass filter is placed on at least one side of an optical block, preferably a polished parallel plate of specific thickness. The filter, forming a continuous coating over at least a portion of the surface of the optical block, preferably is a multi- cavity, most preferably three cavity, film stack coating modeled after a Fabry-Perot etalon, and may be referred to simply as a cavity filter. Two dielectric thin film stacks which by themselves form a reflector for the optical wavelengths in question, are separated by a thicker cavity layer. This structure is then repeated one or more times to produce a filter with enhanced blocking and improved in-band transmission flatness. The net effect is to

produce a narrowband transmissive filter where in-band light is transmitted and out-of-band light is reflected. As noted above, dichroic filters may also be used. This improved filter performance provides commercially acceptable dense channel wavelength division multiplexing for fiber optic telecommunications applications of the optical multiplexing device. It provides low cross-talk between channels and permits a suitably high number of channels within a given band width. An excessive number of cavities will adversely affect the transmissivity of even in-band wavelengths and may increase production costs for the optical multiplexing device beyond commercially acceptable levels.

The continuous, variable thickness, multi-cavity interference filter can be produced with dense, stable metal oxide film stackes using the deposition techniques mentioned above. Such filters have been demonstrated to have excellent thermal stability, e.g., 0.004 nm/°C at 1550 nm, and ultra-narrow band widths, separated by as little as 2 nm, or even as little as 1 nm. Suitable variable thickness filters have been used in other applications such as, for example, in U.S. patent 4,957,371 to Pellicori et al. Stable Ultra-Narrowband Filters

also are shown in SPIE Proceedings 7/1994. In particular, high-quality interference filters comprising stacked layers of metal oxide materials, such as niobia and silica, can be produced by commercially known plasma deposition techniques, such as ion assisted electron beam evaporation, ion beam sputtering, and reactive magnetron sputtering, for example, as disclosed in U.S. patent No. 4,851,095 to Scobey et al., the disclosure of which is hereby incoφorated by reference. Such coating methods can produce interference cavity

filters formed of stacked dielectric optical coatings which are advantageously dense and stable, with low film scatter and low absoφtion, as well as low sensitivity to temperature changes and ambient humidity. The spectral profile of such coatings is suitable to meet stringent application specifications. Multi-cavity narrow bandpass filters can be produced

using such techniques, which are transparent to a wavelength range separated from an adjacent wavelength range by as little as two nanometers or less. One suitable deposition technique is low pressure magnetron sputtering in which the vacuum chamber of a magnetron sputtering system, which can be otherwise conventional, is equipped with high speed vacuum pumping. A gas manifold around the magnetron and target material confines the inert working gas, typically argon, in the vicinity ofthe magnetron. As the gas diffuses and expands from the area ofthe magnetron, the unusually high pumping speed vacuum removes the expanding gas from the chamber at a high speed. The inert gas pressure in the chamber, preferably in the range of 5 x 10^"5 Torr to 1.5 x IO^"4 Torr, is then a function ofthe pumping speed ofthe vacuum pump and the confinement efficiency ofthe magnetron baffle. Reactive gas enters the chamber through an ion gun which ionizes the gas and directs it toward the substrate. This has the effect of reducing the amount of gas required to provide the film with proper stoichiometry as well as reducing the reactive gas at the magnetron.

Throw distance of 16 inch and longer can be achieved. Such deposition techniques are

discussed in co-pending U.S. patent applications Serial No. 07/791,773 filed November 13, 1991 and Serial No. 07/300,741 filed September 21, 1994, the disclosures of which are hereby incoφorated by reference.

As noted above, the filter preferably comprises a multi-cavity coating in which two dielectric thin film stacks which by themselves form a reflector for the unwanted

wavelengths are separated by a cavity layer. This structure is then repeated one or more times to produce the aforesaid multi-cavity filters with enhanced blocking and improved in- band transmission flatness. The net effect is to produce a narrowband transmissive filter where in-band light is transmitted and out-of-band light is reflected. In preferred three- cavity embodiments produced by the deposition techniques mentioned above, with dense, stable metal oxide film stacks, excellent thermal stability has been achieved, for example, 0.004 nm per degree centigrade or better at 1550 nanometers, and ultra-narrow band widths separated by as little as 2 nm or even as little as 1 nm. Preferably, the interference filter is continuously linearly variable in thickness. Optionally, however, the thickness of the continuous filter may be variable non-continuously for example, having a substantially uniform thickness over each ofthe multiple ports ofthe optical block associated with the separate channels ofthe fiber optic system.

The interference filter typically comprises two materials, one of a high refractive index such as niobium pentoxide, titanium dioxide, tantalum pentoxide and/or mixtures thereof e.g., niobia and titania, etc. At 1.5 microns wavelength, the refractive index value for these materials is roughly 2.1 to 2.3. The low refractive index material is typically silica, having a refractive index of about 1.43. An interference filter has an "optical thickness" which is the numerical product ofits physical thickness times its refractive index. The optical thickness ofthe continuous, variable thickness, multi-cavity interference filters

used in the optical multiplexing devices disclosed here varies, of course, with the physical thickness ofthe filter at various points along the surface ofthe optical block. At each of the multiple ports of the optical block associated with an individual signal channel, the optical thickness ofthe interference filter is tuned to transmit the desired wavelength sub¬ range^). It will be apparent to those skilled in the art in view of this disclosure that the thickness and composition ofthe layers ofthe continuous filter can be selected to suit the spectral profile required for any given application ofthe optical multiplexing device. It will be apparent also, that the continuous filter can be continuously variable in its thickness, linearly or otherwise, or discontinuously variable in its thickness. In certain preferred

embodiments, the thickness ofthe filter at each port is substantially constant, increasing (or decreasing) in thickness only between one port and the next. The continuous, variable thickness, multi-cavity interference filters used in the optical multiplexing devices disclosed here have many advantages over prior known filtering devices. They can be produced to coat the entire operative portion of a surface ofthe optical block in a single coating step, tunable at each "bounce point" (e.g., by appropriate placement of associated lens apparatus, collimeters, etc.) to exact wavelengths of ±0.1 nm. When manufactured with durable materials to form dense layers of near unity packing density, they are stable over time and with respect to humidity. A large number of optical blocks can be coated simultaneously with the interference filters in a single coating run, thereby substantially reducing the cost ofthe optical multiplexing device. They are readily manufactured comprising multiple cavities, which are coherently coupled using a quarter wave thickness layer in accordance with known techniques. The effect of using multiple cavities, as described above, is to produce a filter with an increased slope ofthe spectral skirts, along with a wider transmission zone. As described above, both of these effects offer advantages over other types of filtering devices, such as etalons and diffraction gratings.

Since the filters can be formed by deposition directly onto a surface ofthe optical block, no epoxy need be used in the mounting ofthe filter so as to be in the path traveled by the optical signals. The stability ofthe filter is enhanced, since it is formed on the optical block, and need not be positioned and aligned in a separate mounting operation. As noted above, the center wavelength for each of the multiple signal channels can be tuned by simply moving a GRIN lens collimator or the like associated with each ofthe signal channels a slight measure in the direction of the varying thickness of the continuous filter. By so moving the associated lens apparatus, it is aligned with the desired signal wavelength. In this fashion the uncertainty of achieving the correct center wavelength in the manufacture of discreet filter elements is substantially overcome. A dense channel wavelength division multiplexing device is illustrated in Fig. 1, employing a continuous, variable thickness, multi-cavity interference filter to form an ultra- narrow bandpass filter at each of eight separate ports on an optical block. This multiplexing device has the ability to multiplex individual, separate wavelength signals into a common fiber optic carrier line and/or to demultiplex such signals. For simplicity of explanation, only the demultiplexing functionality is described here in detail, since those skilled in the art will readily understand the correlative multiplexing functionality. That is, those skilled in the art will recognize that the same device can be employed in reverse to multiplex optical signals from the individual channels. Typical specifications for an optical multiplexing device in accordance with the preferred embodiment illustrated in Fig. 2 include those provided in Table A.

TABLE A

Number of Channels 8

Channel wavelength 1544- 1560

Channel spacing 2 nm ± 0.2 nm Minimum Isolation 20 dB to 35 dB

Insertion loss (total) less than 6 dB

Fiber type single mode, 1 meter pigtail

Operating temperature range -20 ° C to +50 ° C

The optical multiplexing device of Fig. 2 meeting the specifications of Table A, is seen to include an optical block 10 which, preferably, is a stable glass substrate. A means for projecting collimated light, such as a fiber optic GRIN lens collimator 12 or the like, couples highly collimated light 14 to the optical block at a slight angle through a hole or facet in surface 16 ofthe optical block. In accordance with one preferred embodiment, the optical block has a thickness "a" of 5 mm, and a length "b" of 14.1 mm or more, and a refractive index of about 1.5. The collimated light preferably has a divergence of not more than about 0.15 ° and the tilt angle "c" at which the collimated light enters the optical block is about 15°. Thus, multicolor or multi-wavelength light carried by an optical fiber (preferably a single mode fiber) carrier is collimated by lens means 12 and directed through an optical port 18 in surface 16 ofthe optical block 10, from which it passes within the optical block to the opposite surface 20. A graded wavelength all-dielectric narrow bandpass filter 22 is carried on surface 20 ofthe optical block. Specifically, filter 22 is a continuous, variable thickness, multi-cavity interference filter as described above, and, most preferably, is a continuous linearly variable filter. Light entering the optical block at optical port 18 first strikes opposite surface 20 at output port 24. Filter 22 is transparent at output port 24 to a sub-range ofthe wavelengths included in the collimated light 14. Specifically, light 26 passes through port 24 ofthe optical block preferably to a collimating lens means

28 associated with a first signal channel. The optical signal passed by port 24 is thereby transmitted to optical fiber, preferably single mode fiber 30, as a demultiplexed signal.

The continuous filter 22 at port 24 is reflective of wavelengths which are not "in¬ band" ofthe filter at that location. This reflected light 32 is reflected from surface 20 ofthe optical block back to surface 16. Surface 16 carries a broadband high reflector film or coating 34. High reflector film 34 does not cover optical port 18, so as to avoid interfering with the passage of collimated light 14 into the optical block at that location. The reflected light 32 from the first output port 24 is reflected at surface 16 by reflector film 34 back to surface 20 ofthe optical block. The collimated light 14 enters the optical block at optical port 18 at a tilt angle of about 15°, where it refracts according to Snell's Law to an angle

of approximately 9.9° and then bounces between the opposite parallel surfaces 16 and 20 ofthe optical block. Thus, light 32 is reflected by reflector film 34 to strike surface 20 of

the optical block at a second location 36 corresponding to a second output port ofthe

optical block. At the location of output port 36, the continuous, variable thickness, multi- cavity interference filter 22 is transparent to a different wavelength or sub-range of wavelengths than it is at output port 24. For dense channel wavelength division multiplexing applications, the wavelength separation between each ofthe multiple ports

linearly spaced along surface 20 ofthe optical block is preferably about 2nm or less. Thus,

at outport port 36 an optical signal corresponding to a second channel is transmitted through the filter 22 to a collimating lens 38 and from there to fiber optic carrier 40. As at

the first output port 24, the interference filter 22 at output port 36 reflects light which is not

in-band at that location. Thus, the remaining portion 42 ofthe collimated light 14 which first entered the optical block at optical port 18 is reflected back from port 36 to the high reflector 34 on opposite surface 16 ofthe optical block, and from there it is reflected or bounced back to a third output port 44. In similar fashion, the reflected wavelengths then

continue to cascade in a zigzag or "multiple-bounce" path down the optical block, with the

optical signal for each individual channel being removed by successive bounces at surface

20 ofthe optical block. As seen in Fig. 2, therefore, the zigzag path of light travel through optical block 10

causes the reflected wavelengths to strike, in turn, the additional downstream output ports

46, 48, 50, 52 and 54. At each of these multiple ports, the demultiplexed optical signal is passed to an associated collimating lens, each communicating with a corresponding signal carrier line or other destination. While preferably the filter 22 is reflective of all wavelengths which are not in-band at each ofthe multiple output ports, in certain applications it would necessarily be reflective only of the wavelengths of optical signals which had not been

extracted at upstream output ports, that is, at output ports encountered previously in the

multi-bounce cascade sequence. Also, those skilled in the art will understand from this description that the optical multiplexing device of Fig. 2 is equally suitable for use in combining the optical signals ofthe eight individual channels. Thus, the multiple ports in surface 20 would be input ports and optical port 18 would be an output port. The cascading would then proceed downstream from the bottom (as viewed in Fig. 2) ofthe optical block toward the top. For an optical block of 5 mm thickness, as recited above for optical block 10 of Fig.

2, with a tilt angle of 15° leading to a bounce angle of 9.9° within the optical block, the linear spacing ofthe individual output ports (TAN [9.9] x 2 x 5mm) would be 1.76 mm. Thus, continuous interference filter 22 on surface 20 ofthe optical block should be at least 14.1 mm in length (8 x 1.76 mm). The total distance traveled by the optical signal associated with the last ofthe eight channels (5 mm x 8 channels x 2 bounces) would be 80 mm. The total beam spread (80 mm TAN - 1 [SIN - 1] [SIN] [0.15/1.5]) would be about

0.138 mm. The total loss, therefore, for a 0.5 mm beam would be about 1.9 dB. Accordingly, it will be appreciated by those skilled in the art that the optical multiplexing device illustrated in Fig. 2 as described above, is suitable to demultiplex numerous individual wavelength channels out of an incident Hghtbeam in a very efficient manner due to the minimal beam divergence incurred. The total beam spreading for the preferred embodiment described above would be approximately 40% for a half millimeter beam, which produces the aforesaid loss of only 1.9 dB or less than 0.25 dB per channel cascaded through the device. More specifically, those skilled in the art will recognize that the multiple-bounce cascading technique achieved with a continuous, variable thickness, multi- cavity interference filter deposited directly on the surface of an optical block provides an optical multiplexing device having performance characteristics, including cost and simplicity of construction, reliability of performance, compactness, etc., which are significantly improved over prior known devices.

In the alternative preferred embodiment illustrated in Fig. 3, collimated light 60 from a lens arrangement 62 communicating with a single mode optical fiber 64 passes into optical block 66 at optical port 68 substantially in accordance with the embodiment of Fig. 2 described above. Thus, the light passes through optical block 66 to the opposite, multi-port surface 70 ofthe optical block, striking it first at output port 72. A continuous, variable thickness, multi-cavity interference filter 74 extends over surface 70 to provide a narrow bandpass filter at each of the multiple output ports 72, 76, 78 and 80. As in the embodiment of Fig. 2, the filter 74 is transparent to a different wavelength at each such port, whereby the single optical signal associated with channels 1, 3, 5 and 7, respectively, are transmitted to corresponding lens apparatus and fiber optic waveguides. On surface 82 ofthe optical block, a reflective film 84 is provided to cooperate with interference filter 74 on surface 70 to achieve the multi-bounce cascading within the optical block. In accordance with this preferred embodiment, however, reflective film 84 also forms a narrowband filter at each bounce location. Thus, each bounce location at surface 82 ofthe optical block is an additional output port at which the optical signal associated with an additional channel is passed to an associated lens arrangement and fiber optic carrier line. More specifically, reflective film 84, which preferably is also a continuous, variable thickness, multi-cavity interference filter, and most preferably a continuously linearly variable interference filter, is transparent to the wavelength ofthe optical signal of channel 2 at output port 86 and reflective ofthe other wavelengths. Similarly, it is transparent to the optical signal of channel 4 at output port 88 and, again, reflective at that location to other wavelengths. Output port 90 is transparent to the optical signal of channel 6 and, finally, output port 92 is transparent to the optical signal of channel 8. It will be recognized by those skilled in the art that the optical multiplexing device illustrated in Fig. 3 can provide highly efficient and compact multiplexing and demultiplexing functionality. For collimated light having a divergence of 0.15 ° and entering optical port 68 at a tilt angle of about 12°, the optical block may advantageously be formed of fused silica and have a width of about 10.361 mm. Linear spacing ofthe output ports on each of surfaces 70 and 82 is preferably about 3.067 mm, yielding an overall linear dimension of approximately 15 to 20 mm for the optical block. Generally, it is preferred in devices ofthe type discussed here, to have a low entry angle or tilt angle (where zero degrees would be normal to the surface ofthe optical block) at which light passes through the optical port (measuring the angle ofthe light outside the optical block). A low entry angle reduces polarization dependent effects. It also reduces adverse effects of collimated light divergence on filter performance, since a lower entry angle results in more closely spaced bounce points within the optical block and a shorter travel path for the light.

Typically, the entry angle is less than 30°, being preferably from 4° to 15°, more preferably 6° to 10°, most preferably about 8°. Fig. 4 illustrates another preferred embodiment, wherein the reflective film on the second surface 82 ofthe optical block 66 comprises multiple separate elements 120 - 126. The other features and elements are the same as the corresponding features and elements

ofthe embodiment of Fig. 3, and are given the same reference numbers. The individual reflective film elements 120 - 126 can be deposited, e.g., by a sputtering process or the like, directly onto the surface 82 ofthe optical block or onto separate carrier substrates to be individually positioned and attached to the optical block.

Epoxy or other adhesive may be used to attach the reflector elements. The individual reflector films can be broadband reflectors, operating substantially as reflector

film 34 in the embodiment of Fig. 1. Alternatively, they may operate as multiple additional ports, i.e., as bandpass filters or dichroic filters substantially in accordance with the principles of reflective film 84 ofthe optical multiplexing device of Fig. 3.

Additional alternative embodiments will be apparent to those skilled in the art in view of this disclosure, including, for example, optical multiplexing devices wherein two (or more) solid optical substrates are coated, one or both (or all) with continuous, variable thickness interference filters to form multiple ports on a single mono-planer surface as illustrated and described above, and then joined together to form the optical block.

The film stack structure for the continuous, variable thickness, multi-cavity interference filter 22 in the preferred embodiment illustrated in Fig. 2 is illustrated in Figs. 5 and 6. Preferably, the thickness of each alternating layer (for example, of niobium pentoxide and silicon dioxide), as well as the total thickness ofthe film stack, is precisely controlled, most preferably within 0.01% or 0.2 nm over several square inches of area. In

addition, the film stack should be deposited with very low film absoφtion and scatter, and with a bulk density near unity to prevent water-induced filter shifting. Such ultra-narrow, multi-cavity bandpass filters have excellent performance characteristics including: temperature and environmental stability; narrow bandwidth; high transmittance of the desired optical signal and high reflectance of other wavelengths; steep edges, that is, highly selective transmissivity (particularly in designs employing three cavities or more); and relatively low cost and simple construction. As shown in Fig. 5, the filter is a three cavity filter, wherein one cavity, the "first cavity," is immediately adjacent the glass substrate. A second cavity immediately overlies the first cavity and the third cavity immediately overlies

the second cavity and, typically, is exposed to the ambient atmosphere. In Fig. 6 the

structure ofthe "first cavity" is further illustrated. A sequence of stacked films, preferably about 5 to 15 films of alternating high and low refractive index materials, are deposited to form a first reflector. Preferably, the first film immediately adjacent the substrate surface is a layer of high index material, followed by a layer of low index material, etc. Each ofthe high index layers 90 is an odd integer of quarter wavelengths optical thickness (QWOT), preferably one or three quarter wavelengths or other odd number of QWOTs. The low refractive index layers 92 which are interleaved with the high refractive index layers 90 are similarly one quarter wavelength optical thickness or other odd number of QWOTs in thickness. There may be, for example, about six sets of high and low refractive index layers forming the bottom-most dielectric reflector 94. Cavity spacer 96, although shown schematically as a single layer, typically comprises one to four alternating films of high and low index materials, wherein each ofthe films is an even number of QWOTs in thickness, that is, an integral number of half wavelengths optical thickness. The second dielectric reflector 98 preferably is substantially identical to dielectric reflector 94 described above.

The second and third cavities are deposited, in turn, immediately upon the first cavity and preferably are substantially identical in form. The thickness ofthe interference filter layers varies along the length ofthe multi-port surface ofthe optical block, as described above.

Thus, the physical thickness of a QWOT will vary along the multi-port surface. Various alternative suitable film stack structures are possible, and will be apparent to those skilled in the art in view of this disclosure. One alternative film stack is illustrated in Fig. 7, wherein the upper and lower reflectors 94, 98 are as described above for the embodiment of Figs. 5 and 6. The cavity spacer 97 is shown to be formed of four films, two high index films 97a alternating with two low index films 97b. Each such film is 2 QWOTs thick or one half wavelength. Various other alternative suitable film stack structures are possible, and will be apparent to those skilled in the art in view of this disclosure.

Preferred embodiments disclosed here differ and dramatically improve over the prior systems in several key areas. The ion gun used to direct ionized reactive gas towards the substrate during deposition acts to lower overall reactive gas pressure to prevent poisoning and arcing at the target, and the ionization increases the reactivity ofthe gas to improve film stoichiometry. The energy imparted by the ion source helps densify and improve film quality. Preferred embodiments also require no manifold in the substrate plane and do not require the target or sputter sources to be tilted to improve coating rate. Rather than Pond's mandatory "gas separation process" between the substrate and magnetron, preferred embodiments disclosed here rely on high speed pumping systems to reduce inert gas as well as reactive gas levels to further reduce arcing. High pumping speeds allow greater inert gas flows at the magnetron without increasing background pressure. This in turn allows increased sputtering rates at correspondingly higher power levels. Coating rates are

typically 3 - 6 ang/sec, with throw distances of 30 to 35 inches. (Effective coating rate or throughput decreases by the square ofthe distance.) Films are fully reacted and possess a fiilly dense packing structure even at this suφrisingly high deposition rate due to the energy

and reactivity ofthe ionized reactive gas.

The variable thickness filter coating on the multi-port surface can be prepared in accordance with the following preferred embodiments. The variable thickness can be achieved by positioning the substrate to be coated in the vacuum chamber at an angle to the magnetron. Partial and/or intermittent shielding ofthe substrate also can be used. The

shield or masking means can include a masking member positioned in the vacuum chamber between the multi-port surface ofthe optical block (that is, the surface to be coated) and the source of sputter material at the magnetron. Preferably the masking member is substantially closer to the multi-port surface than to the target source material at the magnetron. For example, the mask or shield can take the form of a substantially planar masking member positioned less than .5" from the multi-port surface. The planar masking member can be spun, where the term spinning or spun includes both spinning on its axis or orbital motion substantially in the plane ofthe planar member. Typically, multiple optical blocks are formed as a unitary coated substrate to be diced or divided up following the coating process. Most typically, the substrate being coated comprises a unitary optical glass disk which spins in an opposite rotational direction from the planar masking member. Alternatively, the optical block, again meaning the circular disk substrate which eventually will be divided into a number of optical blocks, can be positioned stationary in the vacuum chamber at a position laterally offset from the magnetron. As discussed further below, masking means also can be used to achieve the variable thickness desired ofthe optical block.

The preferred embodiments described below are capable of producing high quality coatings on substrates, e.g. to form mirrors which are usable in fiber optic systems, in ring laser gyroscopes, etc., using a DC reactive magnetron sputtering system instead of IBS.

Such films have comparable properties to IBS coatings in that they have extremely high packing density, as well as smooth surfaces and low scatter. Total losses for a high reflector laser mirror made in accordance with preferred embodiments of the method disclosed here, for example, are well less than 0.01% or 100 ppm.

Figs. 8 and 9 show the method and apparatus of preferred embodiments. It should be understood that the substrate referred to here is typically a flat disk of optical glass or the like, having a diameter of, for example, 8", 20", etc. One or both surfaces ofthe disk are coated to simultaneously form many (perhaps hundreds) of optical blocks. That is, the coated disk is diced or cut into many individual optical blocks, each having on one surface a variable thickness filter as described above to form the desired multi-port surface. The housing 110 forms a vacuum chamber 111 containing a low pressure magnetron assembly 112 and a planetary substrate holder 113 with a plurality of rotatable planets 114. Each planet 114 holds a substrate facing the magnetron assembly 112. In this embodiment, the distance between the top of the magnetron assembly 112 and the planets is 16". The magnetron assembly 112 is connected to a source of working gas 116 by conduit 117. In this embodiment, the housing 110 is shown spherical with a radius of 48", but other configurations are equally appropriate.

The housing 110 has a lower sleeve 118 which opens into the vacuum chamber 111 and contains a high speed vacuum pump 120 with a gate valve 121 located between it and

the vacuum chamber 111. The vacuum pump is of course used to lower and maintain the pressure in the vacuum chamber at a very low level in the inert gas pressure range of 5 x 10^"5 Torr to 2.0 x IO^"4 Torr.

In this regard the invention distinguishes shaφly from the known prior magnetron sputtering techniques and from conventional ion beam techniques. It is characterized by extremely low chamber pressures, including extremely low reactive gas pressure and extremely low inert gas pressure. The reactive gas pressure, such as O₂, N₂, NO, etc. (measured at the substratesurface being coated) is preferably in the range of 2.0 x 10 ^s to 1.5 x 10^"4 Torr, more preferably 3 x 10^'5 to 9 x 10^"5 Torr. This advantageously reduces or

eliminates arcing at the magnetron and "poisoning" ofthe source by the reactive gas. The inert gas, such as argon, krypton, xenon, etc., is in preferred embodiments introduced primarily at the magnetron. A shaφ pressure drop is established for the inert gas, preferably having a pressure (measured at the substrate surface being coated) in the range of5.0 x 10^"5 Torr to 2.0 x IO^"4 Torr, more preferably 5 x 10^"5 Torr to 1.5 x lO^Torr. Such low chamber gas pressures provide long mean free path, (MFP) and correspondingly allow advantageously long throw distances without undue collisions between the chamber gasses and the sputtered material. Advantageously good coating uniformity is achieved via long throw distance, preferably greater than 12", more preferably 20" or longer. The extremely low chamber pressures enable the use of long throw distances. That is, notwithstanding the use of such long throw distances, advantageously high coating deposition rates can be achieved with correspondingly high magnetron power levels. The loss of films or coating quality normally expected to result from higher magnetron power levels and longer throw distances is avoided by the novel use of extremely low chamber pressures. Thus, preferred embodiments of this invention duplicate several key process conditions of IBS (which, for example, operates in the same pressure range as described above), but uses a DC magnetron sputtering system. This novel system, based on magnetron sputtering substantially

improves the coating speed and correspondingly cost and throughput of depositing high film quality coatings.

Typical high speed vacuum pumps in this invention are 16" cryopumps or 16" diffusion pumps. Pumping speeds with these pumps are on the order of 5000 liters/second (nitrogen) for a 16" cryopump and 10000 liters/second for a 16" diffusion pump (ref. Leybold Product and Vacuum Technology reference book, 1993). Larger pumps can be used such as a 20" pump having pumping speeds of 10000 liters/second for cryopumps (N2) and 17500 liters/second for diffusion pumps (N2) (ref. Varian Vacuum Products Catalogue 1991-92). Pumping speeds referenced above are at the throat ofthe pump. The magnetron assembly 112 is in vertical alignment with the axis of rotation (main

center line 122) of the planetary substrate holder 113 and with a holder for monitoring witness chip 123. In this embodiment, the throw or the distance between the top ofthe magnetron assembly and the planets is 16". Each planet and its substrate rotate about their own center line 124. Such planetary holders are conventional and need not be described further except to point out that, in this embodiment, the planets are 15" and the substrates are 15" or any size less than 15" in diameter, and the center line of each planet is 14" from the center line 122 to accommodate large substrates. Larger planets can be used, for example, 24" planets, with correspondingly increased substrates sizes and throw distances, whereby even greater throughput improvements can be achieved. Masks can be used, preferably substantially planar masking members positioned, e.g., about 0.5" from the substrate. The masking member can be moved progressively during the deposition process, i.e., while the magnetron is operating, to achieve the desired variation of film thickness for establishing multiple different wavelength ports.

An ion gun 126 whose output, represented by dashed lines 127, is directed obliquely toward the substrate holder 113 and whose input in connected to a source of reactive gas mixture 128 by conduit 130. The ion gun is positioned such that its output of ions and gas mixture cover the entire substrate holder 113 and in this embodiment the top ofthe ion gun is 20" from the planets. The principal function ofthe ion gun is twofold. The first is to modify and improve film properties in a manner similar in concept to the Scott et al patent U.S. No. 4,793,908. The second function may be more important, which is to serve to maintain low reactive gas background pressure. With the ion gun, reactive gas is ionized and directed toward the substrate(s). The momentum ofthe reactive gas then carries it only

toward the substrate(s) (and not toward the magnetron where it would have the effect of

causing arcing and rate reduction). The small amount of gas which diffuses toward the magnetron does not noticeably affect its operation. Typical reactive gas pressures are in the range of 2 x IO^"5 Torr to 1.5 x IO^"4 Torr, preferably, 3 x IO^"5 Torr to 9 x IO"⁵ Torr.

A suitable hot cathode pressure gauge 131 is also connected to the vacuum chamber 111 to measure the pressure within the vacuum chamber. Also, vacuum chamber is provided with a shutter 132 oscillatible about a stem 133 blocking the output of the magnetron assembly 112, represented by dashed lines 134. The stem 133 is connected in any suitable manner to a platform 135 and to a means for oscillating the stem (not shown). The shutter is used to pre-sputter the source(s) to remove contaminates from the target which may have condensed, etc., onto the surface ofthe target while the apparatus was idle between layers being deposited on the substrate.

As shown in Fig. 9, the magnetron assembly 112 comprises a target holder 136 having a cavity 137 formed by walls 138 and target material 140. Centrally within the

cavity 136 are conventional magnets 141 which are water cooled by the circulating flow of water in and out ofthe cavity 136 through passages 142 and 143. The metallic target material 140, clamped by the holder, also is water cooled. A manifold 144, spaced slightly from the holder 136, and sealed by insulators 145, is connected to the source of working

gas 116 by conduit 117 (Fig. 8) which enables the gas to flow entirely around the top ofthe holder and over the metallic target material 140. The manifold 144 has an opening 145 substantially the size ofthe metallic target material so that sputtered target material and working gas is emitted as represented by the lines 134. The magnetron is available from Material Sciences of Boulder, Colorado and is typically 6" to 8" in diameter with high strength magnets.

When it is realized that this invention has the capability of producing extremely high quality film coatings by magnetron sputtering without the constraints of IBS or other known techniques, it will also be realized that this invention is a major advance over the prior art.

The foregoing dimensions and pressures of this embodiment— a throw distance of 16", 15" diameter planets, 15" or less diameter substrates and the distance of 20" from the top ofthe ion gun to the planets along with extremely low reactant gas pressures in the range of 2 x IO^"5 Torr to 1.5 x IO^"4 Torr and extremely low inert gas pressures of 5 x 10^"5 to 2 x IO^"4 Torr—also show the great difference between this invention and the prior art.

Compare also the throughput of preferred embodiments of this invention with the throughput of a typical IBS system in making laser quality mirrors:

This Invention IBS

Coating Rate 2-5A7sec .2-lA7sec

Substrate Area 800-1200 in ² 50-100in ²

(Total area of 5 Planets)

From the foregoing it can be seen that the throughput of this invention is 20 to 120 times faster than the throughput ofthe typical IBS system. Coating throughput is a function of coating rate and substrate area. Furthermore, the method of this invention scales easily to larger apparatus dimensions. All ofthe dimensions above can be easily increased at least by a factor of two

to allow coating of optical substrates of 30" diameter or even large with laser low loss

coatings having good uniformity. Scaling is a simple linear issue. A larger system uses larger magnetrons and more process gas (e.g., argon). The vacuum pumps need to be correspondingly increased to accommodate the larger chamber and the increase in process gas flow.

Thus, as is apparent, this invention is capable of producing, for example, laser quality mirrors which are many times greater in diameter than those known to be made by current IBS systems.

The long throw of 16" and more preferably 20" and greater, and low chamber pressures of preferred embodiments of this invention allow two or more materials to be concurrently deposited to form high optical films composed of mixtures of materials. Fig. 9 shows two sources, magnetron assembly 112 and magnetron assembly 112a in vacuum chamber 111 as an example of multiple sources. (The subscripts to the added source and the use of all other reference numerals as in Fig. 8 are to simplify the description herein).

By controlling the level of power of each source which effectively controls the

deposition rate, a layer of selected refractive index can be formed as a mixture of two or more materials. The mixture can be homogenous throughout the layer to form a film of selected index, or inhomogeneous where the layer composition and hence the refractive index varies throughout the film. One common form of inhomogeneous film is called a "rugate" filter, where the refractive index varies in a sinusoidal manner which has the effect of forming a narrow notch reflector. To maintain a low pressure for such a multi-source system, the pumping speed must be roughly increased by a factor of two for two concurrent deposition sources, or a factor of N for N sources. Given the benefit ofthe disclosure, adding pumping speed will be a

simple exercise for those skilled in the art, involving generally either increasing the size of

the pump or adding more pumps to the chamber. In practice, however, two concurrent sources need not be powered at a level equal to that used for a simple source to maintain coating rate, as the rate from the sources is additive, and hence the sources can be sized to smaller levels which use less gas.

Another device which may be used in this invention is an arc reducing electronic device sold by Advanced Energy of Boulder, Colorado under the trademark SPARC-LE.

In Fig. 8, the SPARC-LE 46 is shown connected to the magnetron assemblies 112 by an electrical conductor 147 with its own DC power supply 148. The SPARC-LE is connected

similarly to the two magnetron assemblies 112 and 112a as shown in Fig. 9. Such a device helps in reducing arcing but it is not necessary in the method and apparatus of this invention.

From the foregoing, it can be appreciated that the magnetron system operates at

extremely low pressures. The chamber pressure ofthe inert gas will be a function ofthe

magnetron pressure. Most importantly in this invention, the low total pressure region 150

(A + O_j) is always much less than the higher argon pressure region 152 as depicted in Fig. 9.

Pressure in the chamber can be modeled using the well known pressure-flow

equations (see Leybold Product & Technology Reference Book, page 18 - 5, 1993):

^pChamber ⁼ """Ar / ^CP ^PMagnetron ^=FkwAr / ^CM ^{+ p}Chamber

Where:

^pChamber is the pressure in the chamber;

"""Ar is the flow of argon into the chamber (through the magnetron);

^CP is the conductance ofthe high vacuum pump (chamber pumping speed);

''Magnetron is the pressure in the magnetron;

^CM is the conductance due to gas confinement at the magnetron (confinement

efficiency ofthe magnetron)

substituting terms, the chamber pressure can be written as:

^pChamber = Magnetron / (^cp / ^CM + 1 )

This is an important relationship, because it shows that the pressure in the chamber

is dependent upon the pumping speed of the chamber (^cp). It also shows that if the

pumping speed of the chamber were low, then the pressure in the chamber would be

approximately equal to the pressure in the magnetron. Such low pumping speed type

systems are known in the prior art, where throttle valve mechanisms are placed in front of

the pump to reduce pumping speed. See Vossen and Kern, "Thin Film Processes,"

Academic Press, New York (1978) above, at page 156. However, ifthe pumping speed

in the chamber is large, as taught by this invention, then the chamber pressure becomes low

relative to the magnetron pressure.

Using the equations above, chamber pressures can be determined approximately for

any new chamber with known pumping speed, as shown in Figs. 11 and 12. As is clearly

evident from the figures shown, any suitable desired pressure can be achieved by increasing

the pumping speed ofthe chamber. Ifthe operating inert gas pressure in the magnetron is lowered, it is possible with certain magnetron types, then the entire pressure curve is correspondingly lowered. This is shown by the comparison ofthe pressure curve of Fig. 11 for magnetron pressure of 0.7 microns and a magnetron assembly conductance (C ) of 3000 l/sec. with the pressure curve of Fig. 12 for a magnetron pressure of 0.4 microns and a magnetron assembly conductance (C ) of 3000 l/sec. The pumping speeds shown on the abscissa are quite achievable—for example, a commonly used 20" diffusion pump is rated at 17500 l/sec, and 32" diffusion pump is rated at 32000 l/sec.

An alternative preferred embodiment is illustrated in Figs. 13 and 14, incoφorating a physical mask between the source and the substrate to control and taylor the thickness of the filter coating. In other respects, the apparatus of Figs. 13 and 14 is seen to correspond to the embodiments discussed above. Specifically, a physical mask 150 is positioned between the magnetron assembly 112 and the substrate surface 115. The degree of masking varies with distance from the center ofthe substrate, such that at one radial distance the resulting filter is tuned to a first particular wavelength and at a second radial distance from the center the filter is tuned to a different wavelength. The mask can be fixed or moving, for example, spinning or rotating. For example, a mask typically can rotate about a common axis with the substrate surface, although other rotation schemes are possible and

will be readily apparent to those skilled in the art given the benefit of this disclosure. In general, it is desirable to position the mask as close as possible to substrate surface 115, preferably less than .5 inch, more preferably 0.25 inch, most preferably, for example, about

0.125 inch. Preferably, any rotation or other movement ofthe film has zero or near zero wobble, or runout, preferably less than .001 inch. In general, a greater degree of wabble is tolerable for filters having looser tolerance specifications or wider bandwidth. Preferably, the mask rotates or spins at a high rate of speed, preferably several hundred revolutions per layer, for example, about 50 to 100 revolutions per minute.

It will be apparent from the above discussion that various additions and modifications can be made to the optical multiplexing devices described here in detail, without departing from the true scope and spirit of this invention. All such modifications and additions are intended to be covered by the following claims.

Claims

I Claim: 1. An optical multiplexing device comprising an optical block having an optical port transparent to multiple wavelength collimated light, a continuous, variable thickness interference filter extending on a multiport surface of the optical block and forming multiple ports arrayed in spaced relation to each other along the multiport surface, the continuous, variable thickness interference filter being transparent at each of the multiple ports to a different wavelength sub-range of the multiple wavelength collimated light and reflective of other wavelengths thereof, and means for cascading light along a multi-point travel path from one to another of the multiple ports.

2. The optical multiplexing device in accordance with claim 1 wherem the interference filter is continuously variable.

3. The optical multiplexing device in accordance with claim 1 wherein the interference filter is continuously linearly variable.

4. The optical multiplexing device in accordance with claim 1 wherein the means for cascading light comprises a reflective coating on a second surface of the optical block.

5. The optical multiplexing device in accordance with claim 4 wherein the second surface of the optical block is spaced from and substantially parallel to the multiport surface.

6. The optical multiplexing device in accordance with claim 4 wherein the reflective coating is continuous over the second surface, being at least co-extensive with said multi-point travel path.

7. The optical multiplexing device in accordance with claim 6 wherein the reflective coating is a broadband high reflector film coating which is substantially uniformly reflective of all of said sub-ranges of the multiple wavelength collimated light.

8. The optical multiplexing device in accordance with claim 6 wherein the reflective coating forms multiple additional ports arrayed in spaced relation to each other along the second surface, the reflective coating being transparent at each of the multiple additional ports to a different wavelength sub-range of the multiple wavelength collimated light, and reflective of other wavelengths thereof.

9. The optical multiplexing device of claim 4 wherein the reflective coating comprises multiple discreet reflective film elements arrayed in spaced relation to each other along said second surface.

10. The optical multiplexing device in accordance with claim 4 wherein the means for cascading light further comprises means for directing multiple wavelength collimated light into the optical block through the optical port at an angle to the multiport surface between 4° and 15°.

11. The optical multiplexing device in accordance with claim 1 wherein each one of the multiple ports has an associated lens means for focusing collimated light passed by that one of the multiple ports.

12. The optical multiplexing device in accordance with claim 11 wherein the lens means comprises a GRIN lens communicating with optic fiber.

13. The optical multiplexing device in accordance with claim 1 wherein the optical block comprises a solid block of material substantially transparent to said multiple wavelength collimated light and selected from the group consisting of glass and fused silica, the continuous, variable thickness interference filter being on an outside surface thereof.

14. The optical multiplexing device in accordance with claim 1 wherein the optical block comprises an enclosed chamber.

15. The optical multiplexing device in accordance with claim 1 wherein the optical block is substantially rectilinear, with the optical port being at a front surface of the optical block which is opposite and parallel the multiport surface of the optical block.

16. The optical multiplexing device in accordance with claim 15 wherein (a) the means for cascading light comprises on the front surface a reflective film coating not extending over the optical port; (b) there are at least eight of said multiple ports, each being a bandpass filter transparent to a discreet wavelength sub-range separated from the wavelength sub-range of adjacent ones of the multiple ports by approximately 2 nm; (c) collimated light passes through the optical port at an angle of approximately 6° - 10° to the plane of the front surface, and (d) the multiple ports are linearly spaced from one another along the multiport surface.

17. The optical multiplexing device in accordance with claim 16 wherein the reflective film on the front surface of the optical block is a broadband high reflector film coating.

18. The optical multiplexing device in accordance with claim 15 wherein the means for cascading light comprises on the front surface a reflective film coating not extending over the optical port, the reflective film coating being a second continuous, variable thickness interference filter extending on said front surface of the optical block forming multiple additional ports, the second interference filter being transparent at each of the multiple additional ports to a different wavelength sub-range and reflective of other wavelengths of the multiple wavelength collimated light.

19. The optical multiplexing device in accordance with claim 18 wherein (a) there are at least four of said multiple ports and at least four of said multiple additional ports.

20. The optical multiplexing device in accordance with claim 1 wherein the continuous, variable thickness interference filter forms at each one of the multiple ports an all-dielectric narrow bandpass filter.

21. The optical multiplexing device in accordance with claim 1 wherein the continuous, variable thickness interference filter is a multi-cavity interference filter.

22. The optical multiplexing device in accordance with claim 21 wherein the continuous, variable thickness interference filter comprises a film stack forming at least three interference cavities.

23. The optical multiplexing device in accordance with claim 1 wherein the continuous, variable thickness interference filter comprises a film stack formed of alternating films of niobium pentoxide and silicon dioxide.

24. A method of producing an optical multiplexing device comprising an optical block having an optical port transparent to multiple wavelength collimated light, a continuous, variable thickness interference filter extending on a multi-port surface of the optical block and forming multiple ports arrayed in spaced relation to each other along the multi-port surface, the continuous, variable thickness interference filter being transparent at each of the multiple ports to a different wavelength sub-range of the multiple wavelength collimated light and reflective of other wavelengths thereof, and means for cascading light along a multi-point travel path from one to another of the multiple ports, said method comprising the steps of: positioning the optical block in a vacuum chamber having magnetron means and source means for sputtered particles, the multi-port surface of the optical block facing the source means at a long throw distance therefrom; operating the magnetron means to sputter particles from the source means for coating the multi-port surface, including introducing inert gas to the vicinity of the source means at an enveloping pressure; rapidly withdrawing and depleting the inert gas from the chamber by high speed, high vacuum pump; and directing ionized reactant gas to the multi-port surface to facilitate reactive coating, whereby the interference filter is obtained as a low-loss optical coating.

25. The method of producing an optical multiplexing device in accordance with claim 24 wherein the long throw distance between the source and the multi-port surface is at least 16".

26. The method of producing an optical multiplexing device in accordance with claim 24 wherein the inert gas pressure in the chamber is maintained less than 2.0 x 10^ Torr and greater than 5 x IO^"5 Torr.

27. The method of producing an optical multiplexing device in accordance with claim 24 wherein the masking means is positioned in the vacuum chamber between the multi-port surface and the source means.

28. The method of producing an optical multiplexing device in accordance with claim 27 wherein the masking means is substantially closer to the multi-port surface than to the source means.

29. The method of producing an optical multiplexing device in accordance with claim 28 wherein the masking means comprises a planar masking member positioned less than .5" from the multi-port surface.

30. The method of producing an optical multiplexing device in accordance with claim 29 further comprising the step of spinning the planar masking member concurrently with operating the magnetron means.

31. The method of producing an optical multiplexing device in accordance with claim 30 wherein the optical block and the planar masking member spin in opposite rotational directions.

32. The method of producing an optical multiplexing device in accordance with claim 24 wherein the optical block is positioned stationary in the vacuum chamber at a position laterally offset from the magnetron means.

33. The method of producing an optical multiplexing device in accordance with claim 24 wherein the magnetron means comprises a magnetron and a shield partially shrouding the magnetron for inhibiting diffusion of inert gas away from the source means while allowing diffusion of sputtered particles from the source means to impinge upon the multi-port surface.

34. The method of producing an optical multiplexing device in accordance with claim 24 further comprising the step of rotating the multi-port surface with respect to the vacuum chamber.