Integrated ultrafast saturable absorber - OSA Publishing

1 downloads 0 Views 832KB Size Report
May 15, 1994 - Integrated ultrafast saturable absorber. A. Villeneuve. Center for Research in Electro-Optics and Lasers, University of Central Florida, Orlando, ...
May 15, 1994 / Vol. 19, No. 10 / OPTICS LETTERS

761

Integrated ultrafast saturable absorber A. Villeneuve Center for Research in Electro-Optics and Lasers, University of Central Florida, Orlando, Florida 32826 J. S. Aitchison

Department of Electronics and Electrical Engineering, University of Glasgow, Glasgow G12 8QQ, UK J. U. Kang, P. G. Wigley, and G. I. Stegeman Centerfor Research in Electro-Optics and Lasers, University of Central Florida, Orlando, Florida 32826 Received December 20, 1993

An ultrafast saturable absorber was demonstrated experimentally in AlGaAs, operated with a photon energy below half the band gap, near 1555 nm. Both the saturation intensity and the linear transmission can be The device is based on spatial soliton emission from a tapered channel independently designed in the structure. waveguide with a corresponding increase in transmission of 50%. The experimental data compare favorably

with numerically simulated results.

AlGaAs operated below half the band gap has proved useful for many all-optical switching applications."2 The resulting devices have exhibited low switching intensities and energies as well as excellent transmission characteristics. AlGaAs has also permitted the generation of nonlinear phase shifts greater than 61r without the presence of significant nonlinear absorption and greater than 107r in the presence of some nonlinear absorption, mostly three-photon absorption.' These nonlinear phase shifts correspond to a relatively large nonlinear index change as defined by Ant' = n2 1, where n2 is the nonlinear refractive-index coefficient and I is the intensity. This index change can be used to generate spatial solitons in an AlGaAs slab waveguide at relatively low intensity when compared with previous experiments in glass and CS2 t-5 In this Letter we present a device that uses spatial solitons for switching and channel waveguides to control the switching intensity. The device exhibits behavior suitable for employment as an ultrafast, integrated saturable absorber. This saturable absorption effect of a spatial soliton has been used successfully to mode lock a Nd:YAG laser with CS2.6 The geometry of the absorber is presented in Fig. 1. The device consists of a short, 500-/am-long

input waveguide followed by an adiabatically tapered section, an open section (slab waveguide), and a capture waveguide. In this device the transmission increases when a spatial soliton is emitted from the end of the waveguide and is captured by a second waveguide after it traverses a 3-mm-long open section.

At low intensity

the

field profile

diffracts normally in the open section, causing the transmission into the capture waveguide to be low, as shown in Fig. 2(a). The capture waveguide was 1.6 mm long, for a total device length of 5.9 mm.

Simulations were performed by the standard fastFourier-transform beam-propagation method,7 with 0146-9592/94/100761-03$6.00/0

n2 = 1.1 x 10-13 cm2 /W, a

2

a (linear loss) = 0.1 cm-', (two-photon absorption) = 0, and a 3 [three-photon (3PA)] = 5 x 10-20 cm 3/GW2 1.'8 9

absorption

For

clarity, the field in Figs. 2(a) and 2(b) was allowed to propagate only a short distance in the capture waveguide. The low-intensity throughput is determined by the length of the open section, since a short open section will have a higher transmission than a longer open section. We confirmed this experimentally by comparing the output of the 3-mm device with that of one having a 4-mm open section. At high intensity, a soliton can form from the waveguide mode emitted from the adiabatic taper, as shown in Fig. 2(b). For a waveguide mode far from cutoff, the field overlap with the sech soliton shape is good, permitting low-loss emission of a spatial soliton and efficient capture by the second waveguide. The result is an increase in the I

15-..- 1, , ....

,.. 1....,1.. , ...

105-i~

050-1 5-..

0

''I

'i''I'I*

1

'I***-

2

3

4

5

6

PROPAGATION DISTANCE (mm) Fig. 1. Geometry of the ultrafast saturable absorber consisting of a 5-,um input waveguide, a 0.8-mm-long quasi-adiabatic taper, and a 3-mm open section where the field can diffract at low intensity (linear case) followed by a 1.6-mm capture waveguide. The final width of the taper and of the capture waveguide was 25 gm. The total length is 5.9 mm. © 1994 Optical Society of America

762

OPTICS LETTERS I Vol. 19, No. 10 / May 15, 1994

the propagation distance, W0 is the minimum width of the taper (here 5 pum),and a is a constant related to the adiabaticity of the taper and here is equal to 0.825, leading to a final waveguide width of 25 Am after a 0.8-mm taper. An adiabatic taper is necessary so that exchange of energy between the fundamental and higher-order modes can be avoided while the width of the waveguide increases. The 25,um waveguide reduces the intensity at which a spatial soliton is emitted into the open section, since the soliton power, Ps cc 1/(aon2 ), is inversely proportional to the soliton width a0 .3 The 25-/.tm-wide waveguide presented here has a lower saturation intensity than similar guides made on the same wafer having smaller widths of 15 and 20 ,/m. Because of the presence of 3PA, there is a trade-off, similar to that for a nonlinear directional coupler," between the adiabaticity and nonlinear loss at high intensity. We tried to keep the taper length as short as possible (here 0.8 mm), but the taper would become nonadiabatic, as evidenced by the presence of higher-order modes in the simulation (Fig. 2). The light intensity from the capture waveguide was collected with a large-area photodetector and is plotted in Fig. 3(a) as a function of the input intensity. The optical pulses were generated with an additivepulse mode-locked NaCl:OH color-center laser operating at 1556 nm and with a pulse width of 670 fs. The peak power coupled to the taper is approximately

51E

z2

(a)

(b) Fig. 2. Beam-propagation calculation of the field at (a) low intensity and (b) high intensity. The formation of a spatial soliton at high intensity leads to a high transmission into the capture waveguide. For clarity, the propagation into the capture waveguide was omitted.

transmission of the device at high intensity. It was shown numerically that in a perfect Kerr medium (no linear or nonlinear

loss) it is possible to saturate

the transmission of the device completely. When a soliton propagates with low loss, it maintains a close resemblance to the waveguide mode and can thus be captured very efficiently. When losses are present, the soliton reshapes itself to compensate for the lower energy, making capture by the second waveguide less efficient.

We believe that our sim-

ulation represents the device accurately, because the measured output field profile and the energy transmission curves match closely the results of the beam propagation. The material used for the experiments is similar to that described in Ref. 1, consisting of a guiding layer of A10. 18Gao.82As between a substrate and

0 0.5

"' 1.2 v_ >

z

0.8

(1)

where Ag is the wavelength in the waveguide, z is

|

I

I

I I

Ie

IT I,I,~~~~~~~~~~ I , r , I .

3

3.5

4

.

.

I

X

.

I

'a-,,,,.

- 0.6

H

0.4

-'9~~.

Il

-- p~~~~~

0

W = ( aAgz + W02)U12,

2.5

-- OPEN with 3PA, - CONTINUOUS

1

was selected for high coupling efficiency for a cir-

2

2

OPEN w/o 3PA

=O 0.2

cular Gaussian beam (to facilitate coupling between the laser and the integrated saturable absorber). The width of the adiabatic taper was defined as'0

1.5

INPUT INTENSITY (a.u.) (a)

a cladding of Al0.24 Gao.76 As. The channels were cre-

ated by reactive ion etching. The input waveguide

1

0

2

4

6

8

10

INPUT INTENSITY (a.u.) (b)

Fig. 3. Output versus input intensity for the 3-mm open-section sample and the continuous waveguide: (a) experiment, (b) beam-propagation method calculation.

May 15, 1994 / Vol. 19, No. 10 / OPTICS LETTERS

150 W, which is comparable with the soliton power Ps of 250 W. Experimental data points are presented in Figure 3(a) for a 3-mm open section (dashed

curve) and no open section (solid curve).

In the continuous waveguide, the open section was replaced by a 25-gm-wide waveguide. The continuous waveguide results are presented to show the influence of 3PA, evidenced by the downward curvature in Fig. 3(a), and serve as a comparison with the open section results. Even with the presence of 3PA, the open-section waveguide exhibits an increase in transmission of almost 50% (compared with the low-intensity transmission), clearly indicating that the open-section slab waveguide behaves as a saturable absorber. Figure 3(b) shows theoretical curves for the open section with and without loss (long-dashed and short-dashed curves, respectively) and the continuous waveguide (solid curve). It is clear that even with the presence of detrimental 3PA the transmission increases significantly at high intensity, although the transmission could be improved if 3PA could be avoided.

In summary, we have demonstrated an ultrafast integrated saturable absorber in AlGaAs operating at 1556 nm that uses a nonresonant nonlinear index medium and spatial soliton emission from the end of a tapered channel waveguide. This device could be used as an all-optical

router4 '1 2

with deviation caused by the interaction between solitons in a fashion similar to that described in Ref. 12. We have also shown that both the linear and the saturation intensity can be controlled independently through proper design. This technique has distinct advantages over real saturable absorbers (e.g., multiple quantum wells); since the response time is essentially instantaneous (1500 nm here) and before 3PA becomes the limiting absorption (