GB2272434A - Electro-optic modulators and optical frequency convertors in glass waveguides - Google Patents

Abstract

Second order non-linearities in glass are created by exposing glass to an electron on beam. Electrons from the beam penetrate the glass to set up a space charge field to give rise to a second order non-linearity. As electron beams, in particular, can be readily and accurately controlled, the location and size of non-linearities can accurately be determined. The beam may also be used to modify, for example, to erase, non-linearities formed by conventional poling methods.

Description

LOW-COST ELECTRO-OPTIC MODULATORS AND OPTICAL FREQUENCY CONVERTORS IN GLASS WAVEGUIDES The present invention is a new method for creation of a second-order nonlinearity (SON) in glasses. Glass is an extremely important optical material because of low fabrication costs compared to crystalline materials and superior optical properties such as high damage threshold and low losses. Low losses also make it the material of choice for optical fibres, which are extensively used in communication systems. Unlike many crystalline materials, however, glass possesses a centre of inversion which rules out a second-order nonlinearity. A second-order nonlinearity is needed to make devices such as parametric frequency convertors, linear electro-optic modulators and switches, and electric field sensors.Availability of SON in glasses will also allow monolithic integration of the above devices into optical fibres at a low cost. Hence, it is extremely desirable to create SON in glasses.

A host of poling techniques have already been investigated to produce a permanent SON in glasses. Examples of such techniques are to be found in (i) A. Okada, K. Ishii, K.

Mito, and K. Sasaki, "Phasematched second-harmonic generation in novel corona poled glass waveguides," Appl. Phys. Lett. Vol. 60, No. 23, 2853 (1992), (ii) R.A. Myers, N. Mukhejee, and S.R. Brueck, "Large second-order nonlinearity in poled fused silica," Optics Letters, Vol. 16, No. 22, 1732 (1991), (iii) M.-V. Bergot, M.C. Farries, L. Li, L.J. Poyntz-Wright, P.St.J. Russell and A. Smithson, "Generation of permanent optically-induced second-order nonlinearities in optical fibers by poling," Optics Letters, Vol. 13, No. 7, 592 (1988) and (iv) L. Li, P. J. Wells, E. R. Taylor, and D. N. Payne, "Generation of permanent second-order susceptibility in lead-silicate glass fibers," Technical digest on Integrated Photonics Research (Optical Society of America, Washington, D. C., 1990), paper MJ5.

All of these involve application of high electric fields under special conditions of high temperature or UV illumination. The most successful poling technique to date has been the corona poling technique of Okada et al (see (i) above). They were able to pole a thin film of Corning 7059 glass by applying a voltage of 5 kV to a tungsten needle placed 1 cm above a ground plane on which the thin film was placed at an elevated temperature of 100 OC or more. A SON of 1 pm/V was achieved. Myers et al (see (ii) above) have also reported a nonlinearity of 1 pm/V using thermal poling, which again involves the application of an electric field at an elevated temperature. In their calculation of the nonlinearity, however, they fail to take into account all the contributions to the nonlinear polarisation.Thus they arrive at a value which is a factor of 2 higher than the actual nonlinearity.

The problem with the techniques discussed above is that they are not easily applied to practical devices since phasematching, which often requires periodic encoding of the nonlinearity, is difficult to obtain. Phasematching is needed to compensate for dispersion in the refractive index, which can limit useful interaction lengths to 30-40 zm. The present invention claims a new technique - electron implantation - that allows for periodic encoding of a large SON in a practical way. Two separate methods are claimed. In the first (Method I), a large SON is created in the first place by electron implantation; and in the second (Method II), an existing high spatially uniform SON (created by one of the above techniques) is selectively erased with high spatial resolution by electron implantation.

In Method I, a suitable glass is exposed to an electron beam instead of being subjected to a high poling field. A large fraction of the bombarding electrons penetrate into the glass sample, where they get trapped and set up a space charge field which destroys the inversion symmetry of the glass to give rise to a SON. The magnitude of the secondorder nonlinear susceptibility is then given by x, < X(j3) E1 , where %(j3) iS the third- order nonlinear susceptibility and El is the macroscopic space charge field set up by the implantation of electrons.

In Method II, electron implantation is used to neutralise an existing high SON created by one of the previously reported thermal poling techniques. This has the advantage that the observed problematic lateral spreading of the SON created by thermal poling, which renders the technique of limited use for realising the fine pitch periodic structures needed for device development, can be side-stepped; these structures can be created by selective erasure, in a tightly focused electron beam, of an existing uniform high SON.

The invention will be described by way of example with reference to the figures, in which: Fig. 1 Second harmonic signal as a function of electron beam energy.

Solid line is the best linear fit on the log-log scale. The electron beam energy was varied from 5 keV to 40 keV while the current was kept constant at 3 nA.

Exposure time was 5 minutes for all the measurements.

Fig. 2 Second harmonic signal as a function of current. Solid line is the best parabolic fit to the data. The electron beam current was fixed at 3 nA.

Fig. 3 Square root of the SH signal versus the distance from the centre of the positive electrode Fig. 4 Second harmonic signal from thermally poled fused silica after electron-beam irradiation as a function of electron energy. The electron current and exposure time were fixed at 3 nA and 1 min respectively We have demonstrated Method I in bulk lead-silicate glass. 1 mm x 1 mm squares were marked with a permanent marker on an optically polished 3 mm x 25 mm x 25 mm sample. The sample was placed in the vacuum chamber of a scanning electron microscope so that the marked 1 mm x 1 mm square was in focus on the entire screen of the TV monitor of the scanning electron microscope.This led to the exposure of this square to a focused scanning electron beam of spot size less than 0.5 im in TV mode, which corresponds to a horizontal scanning rate of 20 ms/line and a vertical scanning rate of 0.017 s/frame. One minute exposure time was found to be sufficient. The typical energy of the electron beam and the total current were 40 keV and 3 nA, respectively.

The SON generated by exposure to the electron beam wascharacterised by the well known technique of second harmonic (SH) generation [2]. Using this technique we measured a second-order susceptibility of 0.7 pm/V. The dependence of the SH signal on electron energy and total current were also studied. Fig. 1 is a plot of the SH signal as a function of energy for a fixed current of 3 nA and Fig. 2 shows the dependence on current for fixed electron energy of 40 keV. Exposure times were limited to 5 minutes for all the measurements reported in Fig. 1 and to 1 minute for the measurements in Fig.

2. Separate measurements with fixed energy and current but longer time exposures were also tried. No significant difference in the SH signal for samples irradiated for 1, 5, 15, and 30 minutes was observed. There was some evidence of surface charging effects for longer exposures. The image of the sample in the electron microscope disappeared from the monitor screen after a few minutes and instead, a very bright spot appeared. Such a phenomenon can happen when the surface of the sample under examination becomes sufficiently charged to repel primary electrons back before they can reach the sample.

The 0.7 pm/V value of the SON represents an unoptimised figure. The following improvements may be able to enhance this figure substantially: Figs. 1 and 2 show that SH signal increases with increasing electron energy as well as increasing current. While the increase in SH signal due to higher electron energies may be partially related to deeper penetration depths and longer interaction lengths, any increase due to higher values of current are only likely to be related purely to the magnitude of the nonlinearity. Since currents as high as 1 yA are possible, much higher nonlinearities are expected.

Although the technique was tried with lead-silicate glass, other glasses are not precluded.

In fact, glasses with large values of third-order nonlinearity, which is not forbidden in glasses, may also yield high values of SON.

An ultra-thin metallic electrode deposited on the surface facing the incoming electrons will eliminate the surface charging which may be limiting the number of electrons which actually penetrate the glass material. Also, it will lead to a more spatially uniform nonlinearity.

In the situation when a metallic electrode is deposited on the surface opposite to the one exposed to the electron beam, it may be possible to transfer the high field region to the opposite side of the implanted layer and thus eliminate any chance of erasure of the nonlinearity due to exposure to optical radiation during the reading process. ( We have investigated Method II in fused silica and lead-silicate glasses. The samples were heated (following the technique of Myers et al) to about 3000C in an oven while applying a voltage of 4 kV across 1.3 mm thick silica discs with about 20 mm diameter.

After some 20 minutes of poling the samples were cooled to room temperature. The anode and cathode dimensions were 2mm x 20mm and 35mm x 75 mm respectively, and they were pressed to the sample. After cooling to room temperature they were removed.

Q-switched (1 kHz repetition rate, 200 ns envelope duration) and mode-locked (76 MHz repetition rate, 3 ns pulse duration) Nd:YAG laser pulses at 1064 nm were used to probe the second order nonlinearity, with average powers of about 1.2 W. The pump laser beam (polarised in the plane of incidence) was focused by a lens (focal length 10 cm) on the anodic surface. The angle of incidence (about 60 ) was chosen to lie close to the Brewster angle. No second harmonic (SH) signal was observed in the poled lead silicate samples, while a strong signal (compared to the value observed in electron beam irradiated lead silicate samples) was observed in samples of fused natural quartz.We were surprised to discover, while scanning the focused laser beam along the surface of the sample in the direction perpendicular to the anode, that a second harmonic signal was observed outside the electrode region (Fig. 3). The poled second-order nonlinearity extended beyond the anode, covering an area some 1.8 times wider than the anode width.

The use of focused electron beams to erase the SON in the thermally poled regions provides a high-resolution means of circumventing the problem of the lateral spreading out of the SON in a simple and practical manner. In our experiments a scanning electron microscope was used for irradiation of the samples. The beam current used was 3 nA and the electron energy ranged between 5 and 40 keV. The TV scanning mode of the electron microscope (horizontal scanning rate 0.064 ms/line and vertical scanning rate of 0.017 s/frame) was used. Areas of about 1 mm x 1 mm on the surface of the samples were irradiated for about 1 min. The thermally poled SON was observed to decrease in strength with increasing electron-beam energy on the fused silica samples (Fig. 4), being fully erased at around 40 keV.

Both Method I and Method II could be implemented using ion beams in place of electron beams; ion beams, however, are more likely to damage the sample.

In addition to the relatively large SON achievable both alone and in combination with thermal poling techniques, the electron implantation technique has several fundamental advantages over other poling techniques. These are listed below: 1. An extremely important advantage of the technique is the ease with which phasematching can be achieved. Phasematching for frequency doublers in optical fibers for example requires a periodic modulation in nonlinearity. With this method 100% modulation may be obtained by simply programming the electron beam machine to expose the material in steps of the required period. Programmable electron beam "directwrite" machines are quite ubiquitous since they are commonly used in the microelectronics industry. Unlike thermal poling there is no need to deposit electrode patterns with the required period.Deposition of metallic electrodes on optical fibers is possible but it is a major technological complication.

2. Since the penetration depth of the electrons can be controlled with the electron energy, it is possible to control the location of the nonlinearity by choosing the correct energy. Penetration depths of over 1 mm are possible with 1 MeV electrons. This obviates the need to use specially fabricated D shaped fibres for fabricating practical devices.

3. Another advantage of direct poling with electron beams is the high resolution (10 nm) which is possible. An electro-optic Bragg grating for example requires a fine pitch of 300 nm or less. This will not be possible with any other poling technique, but will be possible with electron beam poling which has resolution of 10 nm or better.

4. Using the thermal poling technique alone, realisation of the fine pitch SON gratings needed for device development will be severely hampered (if not prevented completely) by the extensive spreading of the SON beyond the boundaries of the positive electrode. Although this effect can be eliminated by using identical upper and lower plate electrodes, it cannot be avoided when periodic patterning of the SON is required, such as in quasi-phase-matched second harmonic generators. It will have a deleterious effect on the performance of such structures. The Method II circumvents this problem, combining the greater thermal stability of the SON in thermally poled silica with the high resolution of the electron implantation technique. In glasses where Method I produces a highly stable SON, it can be used alone to produce high resolution patterning of the SON.

The above advantages confer on electron implantation huge potential for developing practical all-glass devices. Devices which become feasible with this technique may include: 1. Intracavity as well as external parametric frequency converters for fibre and planar waveguide lasers 2. Linear electro-optic modulators and switches in optical fibres and silica based planar optical chips 3. Second-order nonlinear mirrors for passive mode-locking of fibre and planar waveguide lasers 4. Electrically controllable Bragg gratings, exploiting the high resolution attainable with with this technique

Claims (5)

1. CLAIMS 1. A method of creating a second-order non-linearity in glass, the method comprising exposing an area of the glass in which the non-linearity is to be formed to an electron or ion beam.
2. 2. A method according to claim 1 in which the energy of the electrons in the electron or ion beam is controllable to vary the location of the non-linearity in the glass medium.
3. 3. A method according to claim 1 or 2 in which the direction and position of the electron beam is variable to permit exposure of a plurality of different predetermined areas of the glass medium to the beam.
4. 4. A method according to claim 1 or 2 in which the glass medium is subject to poling prior to exposure to the electron or ion beam, the said exposure acting to modify the non-linearity created by the poling.
5. 5. A method according to claim 4 in which the exposure to said beam acts to erase, selectively, the non-linearity created by poling.