Spectral and Temporal Study of Picosecond-Pulse Propagation in a ...

40 downloads 0 Views 2MB Size Report
Prince Consort Road, London SW72BZ, D.K.. Received 15 May 1985jAccepted 11 ... played on a chart recorder. In order to vary the power coupled into the fibre ...
Appl. Phys. B 39, 43-46 (1986)

Applied

physICS Phot~-

Physics B ~~~:;~ © Springer-Verlag

1986

Spectral and Temporal Study of Picosecond-Pulse Propagation in a Single-Mode Optical Fibre A. S. L. Gomes, W. Sibbett, and J. R. Taylor Photonics Group, Optics Section, The Blackett Laboratory, Prince Consort Road, London SW72BZ, D.K. Received 15 May 1985jAccepted 11 September

Imperial College,

1985

Abstract. The power-dependent pulsewidth variations for input 85 ps pulses from a cw mode-locked Nd: YAG laser propagating through 125m of single-mode optical fibre have been examined using a Synchroscan streak camera. Simultaneous spectral and temporal measurements provide information as to the optimum parameters for pulse compression in fibres.

PACS: 42.10,42.65

The availability of low-loss single-mode optical fibre has made possible the study of many nonlinear effects because high power densities can be propagated over long interaction lengths [1-3]. A recent application is the optical pulse compressor [4, 5J where the pulsewidth broadening and the frequency chirp produced in a fibre due to self phase modulation and group-velocity dispersion is subsequently compressed by a grating pair dispersive delay line. Light pulses as short as 12 fs have been obtained using this technique [6J, and the method has also been successfully applied to longer duration (~100 ps) pulses [7,8]. In this letter we report on time-resolved measurements of the power dependent pulsewidth variations for 85 ps pulses launched into a 125m long single-mode fibre. Together with simultaneously recorded spectra the information enables the optimum power/fibre length to be determined for pulse-compression experiments, and it is shown that stimulated Raman generation limits the available bandwidth for compression. A schematic of the experimental arrangement is shown in Fig. 1. The source of picosecond pulses was a cw mode-locked Nd: YAG laser (Quantronix model 116). We have previously described this modified laser system [9J which typically produced pulses of 85 ps at 100 MHz repetition rate with an average power of 7 W (peak power 820 W) in a lasing bandwidth of 0.03 nm [10]. The output from the laser was partially reflected

using beamsplitter BSI into a pulse calibration and delay arrangement (BS4/M) and the pulse durations were continually monitored using a Photochron 11 Synchroscan streak camera (S 1 photocathode) which was driven in synchronism with the modulator of the laser. After suitable attenuation the remainder ofthe laser output was focussed into the fibre using an uncoated 20 X microscope objective (Ld. The fibre used was single-mode non-polarisation preserving with a 7!lm core diameter and a 1dB/km loss at 1.06 !lm. Although several different fibre lengths were investigated the results reported here primarily relate to those obtained with 125 m of the fibre type described above. The pulses exciting the fibre were partially reflected off beam splitter BS2 and directed colinearly with the monitored input pulse beam via the calibration delay line into the streak camera, thereby facilitating continuous measurement of the input and output pulse widths. A 1 m scanning grating monochromator (1800 l/mm) with a resolution of 0.05 nm was used to simultaneously record the bandwidth of the remainder of the beam emerging from the fibre. The signal from the SI type photomultiplier on the output of the monochromator was directly displayed on a chart recorder. In order to vary the power coupled into the fibre the focus position of the microscope objective LI was varied. A maximum coupling efficiency into the fibre of 50% was achieved.

A. S. L. Gomes et al.

44

PMT Monochromator

Fig. 1. Schematic of experimental

(a)

LOUT

=

60 ps

-t-

1ouT=127 ps

tiN = 85 ps

Fig. 2a-c. Synchroscan streak camera recorded profiles of the pulses transmitted through 125 m of single mode fibre for an average input power of (a) 200 mW, (b) 800mW, and (c) 1.5W. The pulse to the rhs in each case is the input pulse which is shown for calibration purposes

arrangement

The variation of the output pulsewidth with average (peak) power in the 125m fibre for a constant input pulsewidth of 85 ps can be seen in Fig. 2a-c, which are Synchroscan traces of the input (rhs) and its respective output pulse (lhs) for average (peak) powers of200 mW (23.5 W), 800 mW (94 W), and 1.5W (175.5 W), respectively. The corresponding spectra of the output pulses are shown in Fig. 3a-c, respectively. It can be seen that at the low peak power of 23.5 W the pulse emerging from the fibre had been reduced in duration from 85 to 60 ps while its spectrum had increased from 0.3 to ~ 3 A (Fig. 3a). It is most likely that the decrease in the pulse duration was due to a negative chirp in the output spectrum ofthe mode-locked laser pulses which was compensated by the positive dispersion in the fibre. A similar observation has previously been made for somewhat longer pulses from a semiconductor laser in substantially longer lengths of fibre [11]. Further increase in the peak power led to increased bandwidth while the output pulsewidth remained relatively constant or slightly decreased, up to an average power of 400mW (47W peak) where significant temporal broadening commenced. For peak powers of 94 W the emergent pulse had increased to 127 ps (Fig. 2b) and the corresponding spectra (Fig. 3b) had broadened to ~ loA, which corresponded to an overall bandwidth expansion of 33 times. The temporal form of the pulse had changed markedly to take the form of a square profile with a noticeable selfsteepened edge. It has been shown theoretically by Grischkowsky and Balant [12] that the combined action of self phase modulation and positive group velocity dispersion (the so-called dispersive self phase modulation) can lead in certain circumstances to pulses of this form, in which case the output pulse can be described by a single linear chirp which can contain as much as 90% ofthe pulse energy. The characteristics

Picosecond-Pulse

Propagation

45

in a Single-Mode Optical Fibre Peak Power (W) 100 150

50



1.5

-

(a) \

t-:J

10625

t···. ·I



I

tl0.5

200



L = 125 m



L,N=

85 ps

• ••

1.0



0.5 Average

1.0 Power (W)



1.5

Fig. 4. Variation of output pulsewidth with average (peak) power entering the fibre, for a fixed fibre length of 125 m and input pulsewidth of 85 ps

(b)

10655

I

10625 I

10625

I

Fig. 3a-c. Simultaneous recorded spectra corresponding to the output pulses in Fig. 2 (a) 200mW, (b) 800mW, and (c) 1.5W average power in the fibre

of the output pulse shown in Figs. 2b and 3b therefore have the best-suited temporal and spectral forms for passage through a linear dispersive delay line to temporally compress the pulses to the transform limit of the supporting bandwidth (10 A at 1.06/lm corresponds to 1.6ps pulsewidth assuming Gaussian pulseshapes). Both the temporal and spectral shapes of Figs. 2b and 3b are in good qualitative agreement with the theoretical predictions of Grischkowsky and Balant [12]. The variation with power of the output pulsewidth relative to the 85 ps into pulse is shown in Fig. 4. Up to an average power of 800 mW the increase in pulsewidth, as described above, can be seen. Above 800 mW the output pulsewidth decreased markedly, reducing

from 120 ps at 900 mW to ~ 60 ps at 1W. For powers in excess of 1W a slight increase in the pulse duration occurred probably due to self phase modulation, followed by a regime (1.2-1.5 W) offurther decreases in pulse durations. The very noticeable decrease in the pulse duration in the region 900mW-1 W was due to significant stimulated Raman generation which was spectrally detected at 1.12 /lm (1st Stokes) and showed a threshold in the region of 900mW (106W peak). Theory predicts the Raman threshold peak pump power (given by Pth = 30A/GL where A is the effective core area 49 x 10- 8 cm 2, G the Raman gain 0.92 x 10- 11 cm/W at 1.06 /lm, and L the effective length of the fibre 123 m [2, 3, 13]) to be 102 W which agrees closely with the experimental value. The reduction in the pulsewidth above this threshold value was due to significant depletion of the pumping fundamental pulse at 1.06 /lm, and this can be seen in Fig. 2c where the output pulsewidth had decreased from 85 to 43 ps. The lower intensity pulse leading the output pulse by 195ps (Fig.2c) is the first Stokes Raman component at 1.12/lm. Depletion of the fundamental pulse is evident in the corresponding spectrum of Fig. 3c. As a pulse passes through a fibre, self phase modulation gives rise to a frequency down shift on the leading edge and an up shift on the trailing edge [1]. From Fig. 3c it can be seen that the down shifted region of the ~ 16 A wide spectra was severely depleted with increased power in the fibre and this corresponded to a loss on the leading edge of the pulse due to Raman generation. A similar behaviour was observed for different lengths of fibre and input pulsewidths. For pulsewidths greater than 85 ps and 125m of fibre results similar to those in Fig. 4 were obtained only with the maximum in pulsebroadening shifted to higher average powers. For

A S. L. Gomes et al.

46

greater lengths of fibre and 85 ps pulses the maxima shifted to lower powers. In all cases the marked decrease in pulse duration due to Raman generation was present. In conclusion, we have shown the power dependent effects on pulse width and spectra for relatively long (85 ps) pulses propagating through 125 m of single mode fibre. Spectral and temporal results agree relatively well with theoretical predictions. The loss of power to the Raman process with its corresponding reduction in the available bandwidth represents a severe limitation to the maximum power in the fibre for effective pulse compression experiments. The optimum region for maximum compression is where the temporal form of the pulse emerging from the fibre exhibits a square profile (Fig.2b) and where a maximum symmetrical spectral broadening is observed (Fig. 3b) without depletion of the low frequency component to the Raman process. We have carried out compression experiments [14J and have successfully used the above criterion to obtain compressed pulses of 2.1 ps for 85 ps input pulses. Changes in average power ~ 100mW above or below the optimum regime described above resulted in much broader and noisier compressed pulses. Acknowledgements. A post-graduate scholarship for one of us (ASLG) from the CNPq (Brazilian Agency) is gratefully acknow-

ledged. Supply of the monomode optical fibre by Drs. Doran and Blow of British Telecom Research Laboratories, Ipswich is gratefully acknowledged. The overall funding of the work was from SERC.

References I. RH. Stolen, C. Lin: Phys. Rev. A 17, 1448-1453 (1978) 2. L.G. Cohen, C. Lin: IEEE J. QE-14, 855-859 (1978) 3. Y. Ohmori, Y. Sasaki, M. Kawachi, T. Edahiro: Appl. Opt. 21,3496-3501 (1982) 4. C.V. Shank, RL. Fork, R. Yen, RH. Stolen, WJ. Tomlinson: Appl. Phys. Lett. 40, 761-763 (1982) 5. B. Nikolaus, D. Grischkowsky: Appl. Phys. Lett. 43, 228-230 (1983) 6. J.M. Halbout, D. Grischkowsky: Appl. Phys. Lett. 45, 1281-1283 (1984) 7. AM. Johnson, RH. Stolen, W.M. Simpson: Appl. Phys. Lett. 44,729-731 (1984) 8. JD. Kafka, RH. Kolner, T. Baer, D.M. Bloom: Opt. Lett. 9, 505-506 (1984) 9. MD. Dawson,AS.L. Gomes, W. Sibbett,J.R Taylor: Optics Commun. 52, 295-300 (1984) 10. MD. Dawson: PhD. Thesis, University of London (1985) 11. J.V. Wright, RP. Nelson: Electron. Lett.13, 361-363 (1977) 12. D. Grischkowsky, AC. Balant: Appl. Phys. Lett. 41, 1-3 (1982) 13. RG. Smith: Appl. Opt. 11,2489-2494 (1972) 14. AS.L. Gomes, U. Osterberg, W. Sibbett, J.R. Taylor: Opt. Commun. 54, 377-382 (1985)