Light Source, MEMS Optical Switch, Sensor and Methods for Manufacturing the Same

Abstract

The present invention relates to a light source for generating an optical frequency comb. The present invention further relates to a method for manufacturing the optical resonator used in this light source. The present invention additionally relates to microelectromechanical systems, MEMS, optical switch and system comprising the same. The present invention also relates to a sensor and to a method for manufacturing a suspended silicon nitride structure comprised in the sensor. According to the present invention, a single-step LPCVD deposited monolithic stoichiometric Si.sub.3N.sub.4 layer is used on a mono-crystalline aluminum oxide substrate such as sapphire. The thickness of the Si.sub.3N.sub.4 layer exceeds 500 nm. This layer can be realized with relatively low residual stress.

Claims

1. A light source for generating an optical frequency comb, comprising: an optical resonator, comprising: a mono-crystalline aluminum oxide substrate; an input waveguide; an output waveguide; a closed-loop waveguide arranged on the substrate, and optically coupled to the input waveguide and output waveguide, wherein the closed-loop waveguide is configured for: receiving at least a part of a beam of light from the input waveguide; accumulating optical energy inside the closed-loop waveguide using the received beam of light; generating an optical frequency comb using the accumulated optical energy; coupling at least a part of the generated optical frequency comb to the output waveguide; wherein the closed-loop waveguide is a monolithic silicon nitride waveguide having a thickness of 500 nm or more, and which is deposited on the substrate.

2. The light source according to claim 1, further comprising a laser source for transmitting a beam of light into the input waveguide.

3. The light source according to claim 2, wherein the laser source is a continuous-wave laser.

4. The light source according to any claim 2 or 3, wherein laser source is configured to generate a light beam at a first frequency, and wherein the closed-loop waveguide is configured to generate said frequency comb to have equidistantly arranged frequency components around the first frequency.

5. The light source according to claim 1, wherein the closed-loop waveguide is configured to generate a Kerr optical frequency comb.

6. The light source according to claim 1, wherein the silicon nitride waveguide has a thickness of 750 nm or more.

7. The light source according to claim 1, wherein the silicon nitride waveguide comprises a SiXNY layer, wherein 0.71<=(X/Y)<=0.76.

8. The light source according to claim 1, wherein the mono-crystalline aluminum oxide substrate comprises a sapphire substrate.

9. The light source according to claim 1, wherein the monolithic silicon nitride waveguide is deposited directly on the mono-crystalline aluminum oxide substrate.

10. The light source according to claim 1, wherein the monolithic silicon nitride waveguide is deposited on the mono-crystalline aluminum oxide substrate via an intermediate layer, wherein a thickness ratio between a thickness of the mono-crystalline aluminum oxide substrate and the intermediate layer exceeds 100:1.

11. The light source according to claim 1, wherein the closed-loop waveguide, the input waveguide and/or the output waveguide, is a ridge waveguide.

12. The light source according to claim 1, wherein at least one of the input waveguide and the output waveguide is a silicon nitride waveguide formed during the same process as the silicon nitride waveguide of the closed-loop waveguide.

13. The light source according to claim 1, wherein the input waveguide and output waveguide are part of a same waveguide.

14. The light source according to claim 1 any of the claims 1-12, wherein the input waveguide and output waveguide are arranged on different and preferably opposite sides of the closed-loop waveguide.

15. A method for manufacturing the optical resonator according to claim 1, the method comprising: providing a mono-crystalline aluminum oxide substrate; depositing a monolithic silicon nitride film of at least 500 nm thick on the substrate in a single-step low-pressure chemical vapor deposition, LPCVD, process at a temperature between 750 and 950 0C; providing a masking layer on top of the deposited silicon nitride film; patterning the masking layer; etching the silicon nitride film using the patterned masking layer to thereby form at least the closed-loop waveguide, and preferably all, among the input waveguide, output waveguide, and closed-loop waveguide.

16. The method according to claim 15, wherein the deposited silicon nitride layer has a thickness of 750 nm or more.

17. The method according to claim 15, wherein the silicon nitride layer, SiXNY, has a composition in which 0.71<=(X/Y)<=0.76.

18. The method according to claim 15, wherein the mono-crystalline aluminum oxide substrate comprises a sapphire substrate.

19. A microelectromechanical system (MEMS) optical switch, comprising: a mono-crystalline aluminum oxide substrate; a monolithic silicon nitride optical waveguide having a thickness of 500 nm or more, and which is deposited on the substrate, said optical waveguide comprising: a base part deposited onto the substrate, wherein the first base is configured to receive a beam of light, and a suspended part having a first end at which the suspended part is integrally connected to the base part and a second end configured to emit said beam of light; a light reception unit comprising an optical waveguide; an actuator configured for displacing the second end relative to the light reception unit in response to an actuation signal to allow or prevent the light beam emitted by the second end to enter the optical waveguide of the light reception unit.

20. The MEMS optical switch according to claim 19, wherein the light reception unit comprises a plurality of optical waveguides, wherein the actuator is configured to, in response to the actuation signal, select one optical waveguide among the plurality of optical waveguides in which the light beam emitted by the second end is allowed to enter.

21.-56. (canceled)

Description

[0058] Next, the present invention will be described in more detail by referring to the appended drawings, wherein:

[0059] FIG. 1 illustrates a method for realizing Si.sub.3N.sub.4 microstructures on sapphire in accordance with the present invention;

[0060] FIG. 2 illustrates a first method for realizing suspended Si.sub.3N.sub.4 microstructures on sapphire in accordance with the present invention;

[0061] FIG. 3 illustrates a second method for realizing suspended Si.sub.3N.sub.4 microstructures on sapphire in accordance with the present invention;

[0062] FIG. 4 illustrates a third method for realizing suspended Si.sub.3N.sub.4 microstructures on sapphire in accordance with the present invention;

[0063] FIG. 5 illustrates a fourth method for realizing suspended Si.sub.3N.sub.4 waveguides on sapphire in accordance with the present invention;

[0064] FIG. 6 illustrates a fifth method for realizing suspended Si.sub.3N.sub.4 waveguides on sapphire in accordance with the present invention;

[0065] FIG. 7 illustrates an embodiment of an optical switch in accordance with the present invention;

[0066] FIG. 8 illustrates an embodiment of a sensor in accordance with the present invention;

[0067] FIGS. 9A-C show spectra generated by four-wave mixing;

[0068] FIGS. 10A-C show top down views of possible configurations of optical resonators; and

[0069] FIG. 11 shows a graph indicating transmission and types of dispersion of light in a waveguide.

[0070] FIG. 1 illustrates a method for realizing Si.sub.3N.sub.4 microstructures on sapphire in accordance with the present invention. In this method, as a first step, a sapphire substrate 1 is provided that is cleaned using HNO.sub.3, ozone-steam, or a Piranha solution, i.e. a mixture of sulfuric acid H.sub.2SO.sub.4, water, and hydrogen peroxide H.sub.2O.sub.2. Additionally or alternatively, an RCA cleaning step can be performed.

[0071] As a next step S1, a single-step low-pressure chemical vapor deposition, LPCVD, process is used to deposit a stoichiometric Si.sub.3N.sub.4 layer 2 on sapphire substrate 1. Typically, NH.sub.3 and SiH.sub.2Cl.sub.2 are used as precursors in a flow ratio of 3:1, and a deposition temperature between 750 and 950? C., preferably between 800 and 850? C., and more preferably around 825? C., is used at a pressure of around 200 mTorr. Thicknesses of Si.sub.3N.sub.4 layer 2 can be in the range between 10 nanometer and 10 micrometer, and are preferably in excess of 750 nm.

[0072] Instead of stoichiometric silicon nitride, a silicon nitride layer Si.sub.xN.sub.y may be deposited that has a composition in which 0.71<=(X/Y)<=0.76.

[0073] As a further step, a photoresist layer 3 is applied onto the deposited Si.sub.3N.sub.4 layer 2. As a next step S2, photolithography techniques are used to define patterns in photoresist layer 3. This is shown in top view and cross-sectional view in FIG. 1. Next, in step S3, the pattern in photoresist 3 is transferred into Si.sub.3N.sub.4 layer 2 by means of plasma etching, for example using reactive ion etching using a CHF.sub.3/O.sub.2 plasma. As a next step, photoresist layer 3 is removed.

[0074] FIG. 2 illustrates a first method for realizing suspended Si.sub.3N.sub.4 microstructures on sapphire in accordance with the present invention. Compared to FIG. 1, the process of FIG. 2 differs in that suspended Si.sub.3N.sub.4 microstructures are formed.

[0075] The first step of this method is identical to that of FIG. 1. A sapphire substrate 1 is provided and cleaned. As a next step, a single-step LPCVD process is used to deposit an amorphous silicon layer 4 using SiH.sub.4 as a precursor. The thickness of this layer can be in the range between 10 nm and 10 um. After this deposition, a cleaning step may be used that is identical or similar to the cleaning step used for cleaning sapphire substrate 1.

[0076] As a next step S4, a stoichiometric Si.sub.3N.sub.4 layer 2 is deposited on sapphire substrate 1 using the LPCVD process of FIG. 1. Furthermore, similar to FIG. 1, a photoresist layer 3 is applied and patterned in step S5, and the pattern is transferred into Si.sub.3N.sub.4 layer 2 in step S6.

[0077] The method continues, in step S7, by etching the amorphous silicon underneath Si.sub.3N.sub.4 layer 2 at predefined positions on substrate 1. In this respect, the amorphous silicon acts as a sacrificial layer. This etching is typically performed using a wet-chemical etching step through the created openings in Si.sub.3N.sub.4 layer 2, for example using Tetramethylammonium hydroxide, TMAH. This etching step causes an under-etch that should be accounted for when designing the mask layers to be used during the lithography step. Furthermore, the composition of Si.sub.3N.sub.4 layer 2 may be non-stoichiometric as explained in connection with FIG. 1.

[0078] As shown in FIG. 2, final figure, Si.sub.3N.sub.4 layer 2 comprises a suspended part 2A that is spaced apart from sapphire substrate 1.

[0079] FIG. 3 illustrates a second method for realizing suspended Si.sub.3N.sub.4 microstructures on sapphire in accordance with the present invention. As a first step, a sapphire substrate 1 is provided and cleaned similar to that in FIG. 1. As a next step S7, an amorphous silicon layer 4 is deposited onto sapphire substrate 1 using a similar or identical process as explained in connection with FIG. 2. In step S8, a photoresist layer 3 is applied and patterned. The pattern in transferred in step S9 into amorphous silicon layer 4 using plasma etching, such as reactive ion etching using an SF.sub.6/CHF.sub.3/O.sub.2 plasma. Subsequently in step S9, photoresist layer 3 is removed and a cleaning step is performed similar or identical to the cleaning step for cleaning substrate 1. After having performed step S9, small bodies of amorphous silicon layer 4 remain on substrate 1.

[0080] Next, in step S10 a single-step LPCVD process is used for depositing a Si.sub.3N.sub.4 layer 2 similar or identical to the deposition process in FIGS. 1 and 2. Next, in step S11 a further photoresist layer 5 is applied and patterned. More in particular, a pattern 5A is formed for realizing a meandering Si.sub.3N.sub.4 track. Other shapes would be equally possible. This pattern is transferred in step S12 into Si.sub.3N.sub.4 by means of plasma etching step similar or identical to the etching step in FIGS. 1 and 2. After etching, further photoresist layer 3 is removed in step S12.

[0081] As a next step S13, amorphous silicon layer 4 is removed by means of wet and dry etching. First, a wet-chemical etching step is performed using a wet-chemical etching process using TMAH as illustrated in FIG. 2. This process is continued until the Si.sub.3N.sub.4 structures are nearly suspended or released. Thereafter, a dry-etching step, such as gas phase etching, is performed using XeF.sub.2 to obtain fully suspended Si.sub.3N.sub.4 structures.

[0082] FIG. 4 illustrates a third method for realizing suspended Si.sub.3N.sub.4 microstructures on sapphire in accordance with the present invention. As a first step, a sapphire substrate 1 is provided that is cleaned similar or identical to that shown in FIG. 1. As a next step S14, a single-step LPCVD deposition process is performed for depositing an amorphous silicon layer 4 similar or identical to that of FIG. 2. In addition, a SiO.sub.2 layer and a Si.sub.3N.sub.4 layer are deposited in the order mentioned (not shown). These layers are then covered with a photoresist layer 3 that is patterned in step S15. The pattern in photoresist layer 3 is transferred into the Si.sub.3N.sub.4 layer using a dry-etching technique after which photoresist layer 3 is removed. Thereafter, a localized oxidation of silicon, LOCOS, process is applied to locally oxidize amorphous silicon layer 4 through the SiO.sub.2 layer. Due to the presence of the SiO.sub.2 layer, oxygen will laterally diffuse causing a vertical tapering in the oxidized amorphous silicon layer. After oxidation, one or more selective etching steps will be performed in step S16 for removing the Si.sub.3N.sub.4 layer, and the deposited and formed SiO.sub.2 thereby leaving the remaining amorphous silicon layer 4. After this step, a tapered amorphous silicon layer 4 is obtained that tapers in the vertical direction.

[0083] As a next step, a cleaning step will be performed similar or identical to the step of cleaning substrate 1 in FIG. 1. Then, in step S17, a single-step LPCVD deposition process is performed for depositing a Si.sub.3N.sub.4 layer similar or identical to that of FIG. 1. In step S18, a further photoresist layer 5 is deposited and patterned for forming a meandering structure 5A similar or identical to that of FIG. 3. This pattern is transferred into Si.sub.3N.sub.4 layer 2 in step S19 using a plasma etching step similar or identical to that of FIG. 3 and photoresist layer 5 is removed. As a next step S20, amorphous silicon layer 4 is removed using a combination of wet and dry etching steps similar or identical to that of FIG. 3.

[0084] In FIGS. 1-4 above, an optional cladding layer (not shown) can be deposited onto the formed Si.sub.3N.sub.4 layer. Such cladding layer may for example comprise a SiO.sub.2 layer that is deposited using a single-step LPCVD process using a tetraethyl orthosilicate, TEOS, precursor. The thickness of such layer may be in the order of 1 um. Furthermore, the cladding layer may be subjected to an annealing step at 1100? C. for 1-3 hours in an N.sub.2-environment. A cladding layer may also be deposited using atomic layer deposition or sputtering of Al.sub.2O.sub.3. Compared to SiO.sub.2, Al.sub.2O.sub.3 has the advantage of being transparent in the mid-infrared spectral range.

[0085] In addition, instead of amorphous silicon, sacrificial layers of different sacrificial material can be used.

[0086] Furthermore, if needed, sapphire substrate 1 can be diced into individual dies after completing the above and optionally other processing steps.

[0087] FIGS. 5 and 6 illustrate a fourth and fifth method for realizing suspended Si.sub.3N.sub.4 microstructures on sapphire in accordance with the present invention. Here, FIG. 5 corresponds to the process depicted in FIG. 3 and FIG. 6 to that shown in FIG. 4. More in particularly, FIGS. 5 and 6 illustrate the consequences of depositing a Si.sub.3N.sub.4 waveguide over a sacrificial layer for the transmission of an optical signal indicated by the arrows. As shown in FIG. 5, due to the strong vertical boundaries in amorphous silicon layer 4 at the positions indicated by B, little to no light will be outputted in FIG. 5. On the other hand, due to the vertical tapering in FIG. 6, most if not all of the inputted light will be outputted.

[0088] It should be noted that the tapering illustrated in FIGS. 4 and 6 is not true to scale. More in particular, a typical tapering would be in the range between 1 and 50 nm vertical offset per um in the horizontal direction.

[0089] FIG. 7 illustrates an embodiment of an optical switch in accordance with the present invention.

[0090] As shown, the optical switch comprises a sapphire substrate 1. On sapphire substrate 1, a sacrificial SiO.sub.2 layer has been deposited, which has been patterned and selectively removed. In region II, this sacrificial layer is completely removed, whereas regions I indicate a transitional region in which the sacrificial layer has a vertical tapering as discussed in connection with FIG. 6.

[0091] The optical switch comprises an input Si.sub.3N.sub.4 ridge waveguide 10 which outside of regions I and II has a part 10A that is directly formed on sapphire substrate 1. Inside region I, waveguide 10 has a part 10B that is gradually lifted away from substrate 1 by an increasingly thicker sacrificial layer. Inside region II, the sacrificial layer has been etched away. Consequently, waveguide 10 comprises a suspended part 10C.

[0092] An output waveguide 11 is arranged opposite to waveguide 10, albeit at a given horizontal clearance. The end facets of waveguides 10 and 11 are shaped to allow a proper optical coupling provided that waveguides 10, 11 are aligned.

[0093] Similar to input waveguide 10, output waveguide 11 comprises a part 11A that is directly contacting sapphire substrate 1 and a part 11B that gradually moves away from sapphire substrate 1 to bring the ends of parts 11B and 10B in vertical alignment.

[0094] Suspended part 10C is fixedly connected to a support beam 12 that is formed of a suspended Si.sub.3N.sub.4 beam. On another end, support beam 12 is connected to a suspended electrode 14A. This electrode comprises a Si.sub.3N.sub.4 base on which an electrode metal layer has been deposited. On opposite sides, suspended electrode 14A is connected to sapphire substrate 1 using Si.sub.3N.sub.4 springs 13A, 13B, which have, similar to input waveguide 10, a part that is directly connected to sapphire substrate 1, a part that is gradually lifted away from substrate 1 by the sacrificial layer, and a part that is suspended above substrate 1.

[0095] Non-suspended electrode 14B comprises a Si.sub.3N.sub.4 base that is arranged on the sacrificial layer, which in turn is deposited on sapphire substrate 1. On top of the Si.sub.3N.sub.4 base, an electrode metal layer has been deposited.

[0096] By applying a voltage between electrodes 14A, 14B, these electrodes can either be pulled together or be pushed apart. The applied voltage thereby forms an actuation signal in dependence of which suspended part 10C performs a lateral movement as indicated by the arrow. In this manner, the optical coupling between input waveguide 10 and output waveguide 11 can be established or broken.

[0097] By arranging a plurality of output waveguides adjacent to one another, a 1?n switch can be realized. More in particular, an optical coupling between the input waveguide and one among the n output waveguides can be established in dependence of the actuation signal.

[0098] A possible drawback of using a support beam 12 as illustrated in FIG. 7 is related to optical loss at the point at which support beam 12 is connected to suspended part 10C. To address this issue, it is possible to provide a clearance between Si.sub.3N.sub.4 support beam 12 and suspended part 10C and to fill this clearance with a cladding layer, such as a SiO.sub.2 layer. In this manner, a cladding layer or a material suitable to be used as a cladding layer is used to connect suspended part 10C and support beam 12.

[0099] In even other embodiments, suspended electrode 14A is formed directly on, below or to the side of suspended part 10C. To prevent optical losses, a cladding layer such as a SiO.sub.2 layer can be provided in between suspended part 10C and the metal layer of electrode 14A.

[0100] FIG. 8 illustrates an embodiment of a general sensor in accordance with the present invention. This sensor comprises a reference arm 20 comprising a Si.sub.3N.sub.4 waveguide on a sapphire substrate 1 and a sensing arm 21 also comprising a Si.sub.3N.sub.4 waveguide on a sapphire substrate 1.

[0101] Sensing arm 21 comprises a part 21A that is directly arranged on sapphire substrate 1 and a part 21B that is suspended above sapphire substrate 1 by using a sacrificial layer that has been etched away inside region 22. A tapering (not shown) may be used to provide a transition between parts 21A and 21B.

[0102] Light inputted into input Si.sub.3N.sub.4 waveguide 23 is split, preferably equally, over reference arm 20 and sensing arm 21. At output Si.sub.3N.sub.4 waveguide 24, the light from these arms is combined and fed to a light intensity meter (not shown).

[0103] Suspended part 21B of sensing arm 21 may deform. For example, suspended part 21B may bend towards or away from substrate 1. Alternatively or additionally, suspended part 21B may bend in a direction parallel to sapphire substrate 1.

[0104] Regardless the direction of movement, most if not any deformation of suspended part 21B will change the effective refractive index of suspended part 21B. Consequently, the effective optical path length through sensing arm 21 will change whereas the optical path length through reference arm 20 will essentially remain constant.

[0105] In some embodiments, particularly but not necessarily those in which suspended part 21B has a more complex shape than shown, such as a spiral, reference arm 20 can also comprise a suspended part similar to the one described in relation to sensing arm 21 and/or sensing window 22 can extend to both arms of the sensor. This places the reference arm in a more similar environment and therefore allows for it to provide a better reference.

[0106] At output waveguide 24, light from both arms 20, 21 will interfere. Depending on the relative phase offset, the interference can be constructive or destructive. This difference can be determined using the light intensity meter that is connected to output waveguide 24.

[0107] The sensor of FIG. 8 can be used to determine bending of suspended part 21B by looking at the measured light intensity value and comparing this value to a known reference value or a previously determined value, preferably a value representative for a state of sensing arm 21 in which the arm was not bent. More in particular, by comparing these values, the amount of bending can be determined or estimated.

[0108] The bending of sensing arm 21B can be caused by various factors depending on the type of sensor. For example, at least suspended part 21B may be provided with a capturing agent or coating layer for capturing specific species, particles, molecules, pathogens, or the like. Once these species, particles, molecules or pathogens adhere to suspended part 21B, this part may bend as a result of the accumulated weight. In this manner, the sensor can be used as a detector for detecting the presence and/or quantity of adhered particles. In some embodiments, the capturing agent can also be applied onto reference arm 20. In addition, in some embodiments in which the capturing agent is only applied to sensing arm 21, reference arm 20 may also have a suspended part similar to sensing arm 21.

[0109] Suspended part 21B may also bend as a result of an acceleration of the sensor as such. It is therefore possible to use the sensor of FIG. 8 as an accelerometer. Alternatively, suspended part 21B may also bend as a result of a collision with an ultrasonic wave. It is therefore possible to use the sensor of FIG. 8 as an ultrasound sensor.

[0110] In FIG. 8, a sacrificial layer was used for allowing part 21B to suspend above substrate 1. Alternatively, suspended part 21B can be realized by arranging a recess in or through hole through substrate 1. Such through hole is etched from the backside of sapphire substrate 1 using a suitable wet-etchant such as a H.sub.2SO.sub.4-H.sub.3PO.sub.4 mixture and a suitable mask layer such as SiO.sub.2 deposited on the back side. Such etchant should preferably not or hardly etch suspended part 21B once it reaches the front side of sapphire substrate 1. When using wet-etchants for realizing through holes, C-plane oriented sapphire is preferably used. An advantage of using a through hole is that the sensor could be placed inside a stream of particles to be detected without completely blocking the particles inside this stream.

[0111] Sensing arm 21 is depicted as a straight waveguide. Other shapes, such as suspended membranes, suspended interdigitated structures, and suspended spirals are not excluded.

[0112] Although the sensor depicted in FIG. 8 is an optical application, the present invention is not limited thereto. For example, instead of relying on interference to detect bending of sensing arm 21, non-optical means could be employed such as a strain gauge or other electrical means. In such case, reference arm 20 may be omitted.

[0113] Next, an application of monolithic silicon nitride on sapphire substrates for non-linear optical applications will be described in more detail by referring to FIGS. 9-11.

[0114] Optical resonators using closed-loop waveguides rely on four-wave mixing to generate the optical frequency comb. FWM is the non-linear optical process by which two photons are annihilated and two new photons are created in a non-linear optical material, as illustrated in FIGS. 9A and 9B. Specifically, FIG. 9B shows degenerate FWM, where the two annihilated photons are of identical energy (w.sub.0), whereas FIG. 9B shows non-degenerate FWM, where the two annihilated photons are of unequal energy (w.sub.0, w.sub.1). The difference in frequency between the original photons and the photons newly generated (D.sub.w) corresponds to the distance between adjacent resonator modes, or an integer times this distance.

[0115] Because FWM is subject to conservation of energy without loss to the material, the energy splitting needs to be symmetric in both cases. The modes will be approximately equally spaced in a resonator with low integrated dispersion, as shown in FIG. 9C. In this case, FWM will fill all the cavity modes, creating a frequency comb. The dynamics of comb formation are highly complex, and involve both degenerate and non-degenerate FWM of modes that are not necessarily adjacent.

[0116] FIGS. 10A-C each show an optical resonator that could be part of a light source according to the present invention. Specifically, each of these figures show an input waveguide 102A having a coupling section 102C; 102C, an output waveguide 102B also having a coupling section 102C; 102C, as well as at least one closed-loop waveguide 101; 101A, 101B. The coupling sections of the input and/or output waveguide each optically couple the respective waveguide with the closed-loop waveguide.

[0117] Closed-loop waveguide 101 in the embodiments shown is a ring resonator or, what is also referred to as a ring cavity. However, waveguide 101 can also take on other shapes such as that of a racetrack waveguide. A racetrack waveguide is a type of closed-loop waveguide comprising a plurality of semi-circular sections and a plurality of linear sections integrally connected in a manner such that the waveguide as a whole loops back on itself.

[0118] The closed-loop waveguide is a stand-alone structure which interacts with its surroundings by optical coupling. This can be achieved in a number of ways, such as directing a laser onto the waveguide directly, via free space or via a fiber directly connected to the waveguide. Such coupling can be enabled by providing waveguide 101 with a prism for in-coupling said laser, or by providing waveguide 101 with coupling gratings. In the embodiments shown, optical coupling is enabled by arranging the coupling sections 102C; 102C, 102C adjacent to closed-loop waveguide 101. When arranged sufficiently close to each other, the coupling sections and the parts of the closed-loop waveguide adjacent thereto together form a directional coupler. Therefore light will jump from coupling sections 102C; 102C, 102C to closed-loop waveguide 101 and vice-versa.

[0119] Closed-loop waveguide 101 is a silicon nitride waveguide deposited onto a mono-crystalline aluminum oxide substrate 100, which preferably comprises a sapphire substrate. Input waveguide 102A and output waveguide 102B can also be fabricated on this material and are preferably ridge type waveguides. However it is also possible for these waveguides to be made of other materials. Closed-loop waveguide 101 has a thickness of 500 nanometer or more. The top down views shown should not be interpreted as providing insight to the height of any of the waveguides. While the other waveguides are preferably as thick as closed-loop waveguide 101, these can also have thicknesses different from that of closed-loop waveguide 101, as well as different from each other. Likewise, widths and lengths of waveguides, as well as distances between them, as shown in FIGS. 10A-C, are not true to scale.

[0120] FIG. 10A shows a specific embodiment in which input waveguide 102A, coupling section 102C, and an output waveguide 102B are all integrally connected and are collectively embodiment by a singular linear waveguide. Further embodiments are possible in which the input waveguide and/or the output waveguide are bent and/or curved. The direction in which light travels though the in linear waveguide dictates the direction in which light coupled to the closed-loop waveguide will travel in said closed loop waveguide. Similarly, the direction in which light travels in the closed-loop waveguide dictates the direction in which light coupled to the output waveguide will travel in said output waveguide. In this particular embodiment, light travels from left to right in input waveguide 102A and consequently will travel from left to right in the part of closed-loop waveguide adjacent to the coupling section of input waveguide. In closed-loop waveguide 101, light travels clockwise. In this embodiment the coupling sections of the input and output waveguides are embodied by the same piece of waveguide. Light will therefore also travel from left to right in the coupling section of the output waveguide.

[0121] FIG. 10B shows a specific embodiment in which input waveguide 102A and output waveguide 102B are arranged on different sides, specifically opposite sides, of the closed-loop waveguide 101. In this particular embodiment, light travels from left to right in input waveguide 102A and consequently will travel from left to right in the part of closed-loop waveguide adjacent to the coupling section of input waveguide. In closed-loop waveguide 101, light travels clockwise. Some light will leak through the directional coupler formed by the closed loop waveguide and the coupling section of the input waveguide. In this embodiment the coupling sections of the input and output waveguides are embodied by different pieces of waveguide. Because the coupling section 102C of the output waveguide 102B is adjacent to a part of the closed-loop waveguide in which light travels from right to left, light coupled from that part of the closed loop waveguide that is coupled into the output waveguide will also travel from right to left. Consequently, light will be emitted from the optical resonator in the same direction as it was received.

[0122] In some embodiments, not shown, input waveguide 102A extends beyond coupling section 102C and loops back to integrally connect to the coupling section 102C of the output waveguide 102B. Such an embodiment allows for any input light that leaks beyond coupling section 102C to be part of the ultimately emitted beam of light while in-coupling and out-coupling behavior of the closed-loop waveguide can be configured separately. In some embodiments, not shown, the coupling sections 102C and 102C can be sufficiently long for multiple closed loop waveguides to be arranged in between them, and wherein each of the plurality of closed loop waveguides are optically coupled to the input waveguide and the output waveguide and not optically coupled to one another.

[0123] FIG. 10C shows a specific embodiment in which two adjacently arranged closed-loop waveguides 101A, 101B are optically coupled to each other. The loop dimensions need not be identical. However, when the loop dimension are identical, a steeper filter effect can be obtained. Similar to the FIG. 10B embodiment, input waveguide 102A and output waveguide 102B are arranged on different sides, specifically opposite sides, of the closed-loop waveguide 101A, 101B.

[0124] In this particular embodiment, light is coupled between the closed-loop waveguides 101A, 101B by the same directional coupling mechanism as described earlier. Consequently, whereas light travels clockwise in closed-loop waveguide 101A, it travels counterclockwise in closed loop waveguide 101B.

[0125] FIG. 11 shows types of dispersion that can occurs for light having a wavelength of 1550 nm in a silicon nitride waveguide. Specifically, the horizontal X-axis shows a number of possible widths that such a waveguide can have, in particular between 0.5 and 2 micrometer. The vertical Y-axis shows a number of possible heights that such a waveguide can have, in particular between 0.1 and 2 micrometer. In waveguides of which the width/height combination puts said waveguide in area C, light of this wavelength is not guided at all. In waveguides of which the width/height combination puts said waveguide in area B, light of this wavelength will experience normal dispersion. In waveguides of which the width/height combination puts said waveguide in area A, light of this wavelength will experience anomalous dispersion. As can be seen, a waveguide with a thickness of approximately 900 nm is required for light having a wavelength of 1550 nm to experience anomalous dispersion at all, which shows the earlier assertion that relatively thick silicon nitride waveguides are required to achieve anomalous dispersion for wavelengths in the infra-red spectrum. In the above, the present invention has been explained using detailed embodiments thereof. However, it should be apparent to the skilled person that various modifications are possible to these embodiments without deviating from the scope of the present invention, which is defined by the appended claims and their equivalents.