The present mvention relates to the creation of nonlmearities in optical materials and in particular optical waveguides and devices for the use of the nonlmearities, such as switching devices and wavelength converters. Background of the Invention There are known significant advantages in utilising an all optical communication system. Being all optical, such a system requires all optical switching and routing functionalities to allow a continuous optical path through the system. Such optical switches are preferably operated across a range of input (switching) powers and in a fibre optic, planar waveguide or solid-state form Often switching speeds of the order of milliseconds are sufficient to set up a transparent path for a communications link. Previous all silica fibre optical switching devices have relied on the weak Kerr nonlinearity in silica fibres to create the switching action. For example, see B. K. Nayer, N. F layson, N. J. Doran, S. T. Davey, D L. Williams, and J. W. Arkwright, "All-optical switching a 200m twin-core fiber nonlinear Mach-Zehnder interferometer", Optics Letters, Vol. 16, No. 6, pp 408- 410, March 15th, 1991. Although this form of nonlinearity has produced good results, with switching speeds the order of femtoseconds, it requires high power-length products of greater than 50W.m to achieve full switching.

An alternative approach is to utilise the resonantly enhanced nonlinearity observed rare earth doped fibre, the mechanism for which is described in R.H. Pantell, M. J. F. Digonnet, R. W. Sadowski, and H. J. Shaw, "Analysis of nonlinear optical switching in an erbium-doped fiber" , Journal of Lightwave Technology, Vol ll, No ₉ , pp. 1416-1424, September 1993 This type of nonlinearity is

inherently slow (of the order of milliseconds) even though it has the advantage of requiring a significantly lower power length product than devices utilising the Kerr effect method. The speed of this nonlinearity is sufficient for the all-optical switching and routing functionalities as mentioned above.

Further, it is desirable to produce a controllable non-linearity that is suitable for incorporation to a solid-state form and particular, suitable for use in switching devices.

Summary of the Invention

It is an object of the present invention to provide a method for controlling the magnitude of nonlmearities induced in optical materials and further to provide for a number of advantageous devices utilising the nonlmearities, such as optical switching and wavelength conversion devices, in which the output state is held fixed for a wide range of input powers.

In a first aspect of the present invention there is provided a method for producing a refractive index change in an optical material, the method comprising the steps of: defining a resonant laser cavity m the optical material ; pumping the cavity with first electro-magnetic radiation of first predetermined wavelengths which induces a refractive index change in the optical material and which, at a predetermined intensity, causes the cavity to lase; injecting second electro-magnetic radiation of second predetermmed wavelengths through the cavity, the light of a second wavelength being influenced by the change refractive index.

Preferably, the refractive index of the optical material is altered in accordance with a predetermmed relationship to the intensity of the electro-magnetic radiation of the first wavelength Preferably, the relationship includes a clamped portion where an increase

the intensity of the first electro-magnetic radiation results m substantially no variation the refractive index.

In accordance with a further aspect of the present invention there is provided a method for producing an interference between two coherent beams of electro¬ magnetic radiation wherein the method includes: transmitting one of the coherent beams through an optical material defining a resonant laser cavity, pumping the cavity with a first pumping electro¬ magnetic radiation of first predetermmed wavelengths which induces a refractive index change in the optical material and which, at a predetermmed intensity, causes the cavity to lase. Again, preferably the refractive index change s clamped such that increases in the intensity of the pumping electro-magnetic radiation causes substantially no change in the refractive index.

Further, the clamping is advantageously utilised such that the spatial difference m the intensity of the interference pattern is maximised so as to produce as large an extinction ratio as possible. In this way, the interference pattern can be utilised to allow the first electro-magnetic radiation to modulate the second electro-magnetic radiation thereby producing an optical switch effect. Preferably, the optical material comprises a Ytterbium doped waveguide and the first electro-magnetic radiation can comprise laser light substantially at 980nm. The second electro-magnetic radiation preferably comprises laser light substantially at 633nm, 1310nm or 1545nm. The interference effect is preferably induced by a Mach-Zehnder Interferometer which can be utilised to cause the first signal carried by the first electro-magnetic radiation to modulate the second electro-magnetic radiation to produce an optical switching or wavelength conversion effect .

In accordance with third aspect of the present invention there is provided a wavelength converter for

converting a first electro-magnetic signal having a first frequency into a corresponding second electro-magnetic signal having a second frequency, said converter comprising: an mterferometπc arrangement having two optically transmissive arms adapted to receive coherent probe electro-magnetic signals of the second frequency, the arms having output portions adapted to proαuce an interference between the coherent probe electromagnetic signals, one of the arms further forms a resonant laser cavity and is adapted to receive the first electro¬ magnetic signal, and the first electro-magnetic signal causing the resonant laser cavity to undergo a change in refractive index m an amount having a predetermined relationship to the intensity of the first electro¬ magnetic signal .

Preferably, the laser cavity includes a dichroic mirror at one end and has a cleaved end face at the other end of the cavity. The optical material is preferably adapted to undergo a clamped refractive index change upon lasing of the laser cavity. BRIEF DESCRIPTION OF THE DRAWINGS Notwithstanding any other forms which may fall with the scope of the present invention, preferred forms of the invention will now be described, by way of example only, with reference to the accompanying drawings in which:

Fig. 1 is a schematic illustration of a first arrangement of the preferred embodiment; Fig. 2 is a data plot of the observed phase shift versus launched pump power for the arrangement of Fig. 1;

Fig. 3 illustrates a prior art method of wavelength conversion utilising an all optical semiconductor optical amplifier; Fig 4 illustrates various graphs illustrating the operation for the arrangement of Fig. 3 ;

Fig 5 illustrates the relationship between probe output power and phase difference for a Mach-Zehnder

arrangement;

Fig. 6 illustrates correspondmg input values for the output values of Fig. 5; and

Fig. 7 illustrates an adaptation of the arrangement of Fig. 3 so as to utilise the effect demonstrated by the arrangement of Fig. 1 m a wavelength converter. DESCRIPTION OF THE PREFERRED AND OTHER EMBODIMENTS

Referring now to Fig. 1, there is illustrated an apparatus utilised for creating a clamped nonlinearity an optical waveguide where the nonlinearity was utilised to induce a mterferometric phase change between two interfering sources.

At the centre of the arrangement 1 of Fig 1 is a twin core optical fibre 2 which was approximately 1.3 metres in length, having dual cores 3, 4 separated by 24μm which is large enough to ensure that no optical coupling occurred between the cores. Both cores 3,4 were doped with approximately 0.2 wt% Yb ³⁺ (hereinafter called "ytterbium") , and had a higher mode cut off of approximately 940nm. The absorption of the fibre was found to be 2.7dB/cm at 980nm and 0.84db/m at 1310nm and 1545nm. The end 6 was utilised as a dual point source producing an interference or fringe pattern 7 A TV camera 8, able to detect in the mfra red and visable, was utilised to detect the fringe pattern 7 and display the results on monitor 9.

Ytterbium is an advantageous dopant for providing nonlinear effects in the important communications standard windows around 1300nm and 1550nm, because the absorption peak at 980nm used for inducing the nonlinearity is far removed from these windows.

In a first experiment, a laser 11 was utilised to proηect mfra red or visable radiation via lens 12 to dichroic beam splitter 13 to further project mfra red or visable radiation simultaneously mto the two cores 3 and 4. The detected fringe pattern was detected utilising a television camera 8 adapted to detect the infra-red and visable, and displayed on TV monitor 9

Subsequently, light from a 980nm laser 14 was projected through lens 15 and mirror 13 and launched co- directionally mto the core 3 only. It was found that the use of the laser 14 induced a nonlinear phase change in the core 3.

Referring now to Fig. 2, there is illustrated a graph 20 with a first series of plots eg. 22 relating the phase shift of a signal from a 1310nm laser 11, as detected on TV monitor 9, for various values of launched power for laser 14 (Fig. 1) . The values measured e.g 22 are those denoted to be "without lasing" .

It is theorised that, due to the large Stark splitting of the energy levels present in the ytterbium doped fibre, the core 3 acts as a pseudo three level lasing system in which the decay from an excited state is dependent on the long life time of the metastable level . This three level operation allowed significant population inversions to build up m the pumped core. Lasing in the fibre was suppressed by butting the fibre end 10 against a microscope slide (not shown) and refractive index matching the abutment. This removed the usual 4% reflection from the cleaved fibre end 10 preventmg feedback into the pumped fibre 3. By measuring the observed fringe shift induced by pumping the core 3 of the twin core fibre 2, the induced phase shift (Fig. 2) and hence the nonlinear refractive index change was inferred. This was repeated for lasers 11 having wavelength 633nm, 1310nm and 1545nm and similar effects noted. The microscope slide placed at end 10 was then removed and a dichroic mirror (not shown) was substituted in its place. This created a lasing cavity between the dichroic mirror placed at end 10 and the 4% reflection due to the cleaved fibre end at the output end 6 of the optical fibre 2. The dichroic mirror placed at end 10 allowed the core 3 to lase, but still transmitted enough of the signal wavelengths to enable a fringe pattern to be observed on monitor 9. The observed measurements e g

21 are plotted in Fig. 2 and denoted "with lasing".

As is evident from the slope of a line joining the "with lasing" plots, the fringe phase shift grew steadily with increasing launched pump power until the fibre core 3 started to lase approximately at point 23. At this point, the gradient of the resulting fringe phase shift curve slowly reduced until it reached a steady state value at the launched pump power of approximately 13mW. As can be seen from the graph of Fig. 2 it appears that the core 3 started to lase at about lOmW lauched pump power which corresponds to the point 23 where the slope of the fringe pattern for the "with lasing" points starts to significantly diverge from the fringe shift pattern of points 22 recorded for the "without lasing" system. By observing the direction of the phase shift, it was possible to determine the sign of the refractive index change . The results indicated that the pump induced refractive index change was positive for all tested signal wavelengths. For a 1.3 metre length of fibre, the phase changes recorded for 633nm, 1310nm and 1545nm radiation from laser 11 for a maximum launched pump power of 18mW were 9.4τr, 2.17T and 1.337T radians respectively. This corresponds to an average refractive index change along the core 3 of 2.3 x 10 ^"6 , 1.1 x IO ^"6 and 0.8 x IO ^"6 respectively.

While not wishing to be bound by theory, it is theorised that when the pump power is sufficient for the system to start to lase (the threshold condition) , the population inversion between the upper and lower laser levels is clamped at a constant value, with further increases in pump power increasing the rate of stimulated emission. It is theorised that this is the reason for observing clamping of the phase shift at larger pump powers. As the fringe shift is clamped when the fibre starts to lase, it is highly likely that this effect is due principally to changes in absorption rather than thermal effects induced by the pumping. Once the phase

shift has been clamped, the phase was found to be stable with ± O.ITT which indicates that any thermal effects present are most probably small compared to the population induced changes. The induced refractive index change increases towards shorter wavelengths and is of the same sign for all signal wavelengths used. It is therefore theorised that the effect may be due to changes m absorption in the far ultra violet, as, if the effect was due solely to the localised absorption at 980nm, the induced refractive index change would be the opposite sense for signal wavelengths on either side of the 980 nm absorption peak.

The clamped effect illustrated in the arrangement 1 of Fig. 1 can have significant utility in telecommunications, and, in particular, in all-optical switching and in an all optical wavelength to wavelength converter. One such system will now be described.

Referring now to Fig. 3, there will now be explained the operation of a known form of wavelength to wavelength converter arrangement 30 which utilises a semiconductor optical amplifier (SOA) 31 in a Mach-Zehnder (MZ) interferometer arrangement 32. In prior art type arrangements, the SOA 31 is normally made up of a semiconductor waveguide with a pair of electrical contacts which allow current to be injected into the waveguide region. This current, otherwise known as the bias current, supplies energy to electrons in the semiconductor, exciting them from their unenergised states in the valence band to high energy states in the conduction band, which creates so called "carriers". Light is launched mto one end of the SOA 31 waveguide through a lens or the like and collected from the other end of the waveguide through another lens. If the wavelength of the photons launched mto the SOA 31 is within the so called gain (or emission) bandwidth of the SOA 31, then each input photon can cause an excited carrier in the SOA waveguide to decay back from the conduction band to the valence band, emitting a

corresponding photon by stimulated emission. In this way, the input light will be amplified.

If an input optical signal to the SOA 31 is within the gain bandwidth of the SOA 31, and is of sufficient intensity, it will cause enough stimulated emission to deplete the carrier density in the SOA 31. If the optical signal is then turned off or reduced in intensity, the carrier density in the SOA 31 recovers to its undepleted value. In thiε way, an intensity modulated data signal launched mto the SOA 31 causes a modulation m the SOA carrier density.

Referring to Fig. 4, there is illustrated a number of timing diagrams illustrating the operation of the SOA

31 of Fig. 2. First timing diagram 40 illustrates an example data intensity at input wavelength λ^^ input to the SOA along optical waveguide 33. The input intensity is assumed to be between the two extremes of PdataO and Pdatal . A corresponding SOA carrier density timing diagram is shown 41 with the carrier density taking an inverse form to the data input intensity 40.

It is a known property of semiconductor materials that their carrier density is in turn related to their refractive index. This property can then be utilised in a wavelength to wavelength converter. A continuous (unmodulated) signal, called a probe, at a wavelength λ _probe , is launched 34 (Fig. 3) into the MZ interferometer

32 and SOA 31 via couplers 35, 36. The probe is launched into the SOA 31 along with the modulated input signal λ _data . The λ _pr0D e signal passing through the SOA 31 experiences a retardation or phase modulation 42 (Fig. 4) between two values Φ _Q and Φ _j accordance with the change m the refractive index as a result of changes in the carrier density of the SOA 31 due to the modulation of the SOA 31 by the input data intensity modulation signal 40. The λ _probe signal passing through the second arm 37 of the MZ interferometer 32 does not experience this change m refractive index. Therefore, when the probe signals are recombined in the coupler 38 at the output of the MZ

interferometer 32, interference between the phase modulated and unmodulated probes results in intensity modulation 43 (Fig. 4) of the output. Hence, the intensity modulation 40 of the λ _data signal causes a phase modulation 42 of the probe λ _pro b _e . Timing diagram 43 further illustrates the ideal case of probe intensity modulation as a result of the interference effects between the two beams at the MZ interferometer output .

The wavelength converter 30 (Fig. 3) is completed by providing a bandpass filter 39 at the output of the MZ interferometer 32. The filter passes the modulated probe signal λ_ _ro b _e but blocks the input data signal X^^.

The wavelength converter arrangement 30 of Fig. 3 has a number of advantages including high speed operation (> 20 Gbit/s) .

Further, it is a desirable feature of a wavelength converter that it should have as high an output extinction ratio as possible across the possible range of input and output wavelengths. The extinction ratio of a digital signal can be defined as a ratio of the power output of a "one" or high level to the power output of the "zero" or low level. A high extinction ratio is important as, for a given signal power, the lower the extinction ratio, the more difficult it is to distinguish the "ones" from the "zeroes". A low output "zero" power is obtained from the MZ mterferometric arrangement 32 when the phase difference between the two is an odd integer number of half cycles (ie. Aθ = π , 3τr, ...) .

Referring to Fig. 5, there is illustrated a graph of probe output power P _QUt versus the phase difference between the two arms of the MZ interferometer 32. It can be seen that the maximum contrast between the two output values P _out ι and P _out ø will occur when the phase difference resulting from the operation of the SOA 31 results a phase difference increase from, for example, the point 50 to the point 51. Hence, a phase modulation of π causes the output power of the MZ interferometer to change from a minimum to a maximum intensity level when,

- li ¬ as illustrated Fig. 6, the input data power P _data goes from a "zero" or low level 60 to a "one" or high level 61.

Further, to obtain a high output extinction ratio, it is necessary that the "visibility" of the interference from the two branches of the MZ interferometer 32 be as large as possible High visibility interference requires that, when the phase modulated and unmodulated probes are recombined at point 38, they have substantially the same power and the same state of polarisation.

As mentioned previously, the input wavelength _jata for the wavelength converter 30 can typically lie anywhere with the gam bandwidth of the SOA 31. This is typically approximately 60nm wide when operatmg in the 1550nm region. For a given data input wavelength and data "on" power value, the magnitude of the phase change seen by a particular probe wavelength λ _Dro tø _e is relatively independent of the probe wavelength across a similarly broad range of wavelengths . Known wavelength converters can therefore achieve a high output extinction ratio across a range of input and output wavelengths so long as the data modulation is sufficient to cause a probe phase modulation of π radians.

The arrangement 30 also has a significant advantage m that the data input power of signal λ _data required to cause a probe phase modulation of π radians can be as little as O.lmW. High output powers however can be produced as the insertion loss of the wavelength converter at the probe wavelength λ _p robe ^1S small . Further, it is possible to make the arrangement 30 insensitive to polarisation effects

However, one significant problem with the arrangement 30 is that, as can be shown from Fig. 5, the probe output power P _om and the extinction ratio can be very sensitive functions of the input power Pda _t a* Although it is possible to "lock" the MZ interferometer 32 so that P _outυ , correspondmg to an input "zero" P _fj a _t aO (Fig. 5, Fig. 6) is very small, the output value P _outl ,

corresponding to an input data "one" of power P _data ι, is very sensitive to the level of ■ When the value of ^p da _t al i* ³ <J ^reater than or less than the value corresponding to ΔΦ = 2nπ, (n = 1,2,3...) the probe output "one" power ^p outl ^w ϋl fall below the maximum value at point 51.

Referring now to Fig. 7, there is shown an alternative wavelength converter arrangement 70 which utilises the clamped lasing effect of the arrangement of Fig. 1 to produce a "clamped cross-phase modulation wavelength converter". The arrangement 70 of Fig. 7 is similar to the arrangement 30 of Fig. 3 except that a lasing device 71 is substituted for the optical amplifier 31 of Fig. 3. The laser 71 can comprise the waveguide 3 enclosed in a resonant cavity as previously described. The most important difference in the operation of the two devices is that the phase shift when the laser is lasing can be "clamped" as illustrated by the "with lasing" set of data points as illustrated in Fig. 2. When the semiconductor laser 71 is lasing, its carrier density, and hence its refractive index, can be clamped at a value determined by the threshold carrier density (the point 23 in Fig. 2) . The clamping point can be set to it radians by adjusting the bias current of the semiconductor laser. If the pumping supplied to the laser is changed, the output power at the lasing wavelength changes, but the carrier density and the refractive index of the laser do not change and hence the clamped phase shift can be utilised advantageously.

In the arrangement of Fig. 7, the input data wavelength λ _data is chosen with respect to the lasing wavelength of the laser 71 so that the input signal 73 acts as the laser pump. That is, when the input signal is a "one" the input signal photons are absorbed by the laser 71 and increase the carrier density (population inversion) in the -laser.

The arrangement 70 of Fig. 7 is to be distinguished from the arrangement 30 of Fig. 3 in that in the arrangement 70, the input data signal photons 73 are

absorbed by the laser arrangement 71 and so they increase the carrier density (population inversion) , whereas in the conventional arrangement 30, as previously herein described, the input data signal photons cause stimulated emission in the semiconductor optical amplifier 31 and so decrease the carrier density (population inversion) The input signal wavelength must be chosen carefully with respect to the emission and absorption spectra of the semiconductor laser 71 so that pumping will occur. In accordance with Fig. 2, if the input signal power is sufficiently high ⁽ P _£ j _ata = ^ _t hreshold ' ^tne carrier density the laser will be increased to a threshold, as the laser begins to lase. Increasing the input signal further (P _data > ^p threshold^ will not change the laser carrier density or refractive index because, in accordance with the (with lasing) plot 21 of Fig. 2, there is no substantial change the laser carrier density or refractive index with increasing pump power as they are clamped at a particular level. By choosing the design parameters of laser 71 and P ^um P input 73 to be such that, firstly, the threshold results in a phase difference of TT between points 50 and point 51 of Fig. 5 and, secondly, the phase difference for zero input pump power is "locked" at a point equivalent to the point 50 (Fig. 5) , then the output power P _out o correspondmg to an input data "zero", P _fj a _t aO' will be very small and the output power for all values of ^p da _t a ^a b° ^{ve a} given threshold will be at the maximum output power point 51, with the threshold phase difference being maintained. The result will then be a wavelength converter arrangement which produces a high output extinction ratio for all input data powers and which produces maximum output "one" power for all data input powers greater than or equal to a predetermmed threshold ^(p da _t a > P _t hreshold ⁾ ' ¹ * ^e * ^a d ⁱ gital response.

The embodiment of Fig. 7 could be alternatively configured as an integrated silicon optical waveguide and operate as an integrated silicon optical device

As a further example embodiment, a ytterbium-doped fibre, similar to the fibre 3 in Figure 1 but with a single core, could be substituted for the semiconductor laser 71 in the MZ interferometer of Figure 7. Further, the resonant cavity necessary for laser clamping could be formed either by dichroic mirrors or by two in-fibre Bragg gratings at the lasing wavelength (approximately 1020nm) , and the clamping point set to π radians by adjusting the cavity losses. The slow switching speed of this example embodiment would still be applicable to signal routing applications.

The foregoing describes only various preferred embodiments of the present invention and its applications to optical switching and wavelength conversion. Further embodiments and refinements and further applications, obvious to those skilled in the art, can be made thereto without departing from the scope of the present invention.