Repetitively Pulsed Tunable Dye Laser for High Resolution Spectroscopy

T. W. Hänsch

doi:10.1364/AO.11.000895

I. Introduction

The present paper describes the design and characteristics of a tunable organic dye laser, repetitively pumped by a nitrogen laser, that offers bandwidths of less than 0.004 Å, convenient, reproducible wavelength tuning, and good long-term stability. This laser has made possible a recent study of the sodium D resonance lines via high resolution saturation spectroscopy.[1]

Numerous approaches to obtain tunable narrow band emission from organic dye lasers by inserting dispersive elements into the cavity have been reported previously.[2]–[6] In long-pulse flashlamp-pumped dye lasers or cw dye lasers a substantial line narrowing down to a fraction of 1 Å can be achieved with wavelength selectors of low dispersion, such as prisms or lenses with chromatic aberration.[7],[8] In dye lasers side-pumped by a nitrogen laser, on the other hand, bandwidths below 10 Å seemed to be difficult to achieve even with a high dispersion diffraction grating, due to the limited number of light passes during the short pump pulse and the small beam diameter.[9],[10] The use of a thick holographic grating reduced the bandwidth to 0.4 Å.[11] Single axial mode operation of a nitrogen-pumped dye laser with a bandwidth in the order of 0.01 Å has recently been achieved in short cavities with a holographic grating and a Lyot filter,[12] or with a grating and Fabry-Perot etalon,[12],[13] but the wavelength reproducibility from shot to shot remained unsatisfactory.

The construction of a stable, narrow band dye laser using a commercially available nitrogen laser as a pump source appeared particularly desirable, since the conveniently obtained high pulse repetition rate together with the short pulse length and high intensity open numerous promising applications in nonlinear spectroscopy. Moreover, the fast rise time eliminates the problem of triplet state absorption[14] and provides low thresholds. Numerous laser dyes, operating from the near-ultraviolet throughout the visible spectrum, can be pumped very efficiently with the nitrogen laser,[9],[10],[15]–[18] and strikingly large tuning ranges[19]–[21] are possible.

II. Dye Laser Construction

The usual way to generate very narrow band laser radiation is the isolation of one single-axial cavity mode. An alternative way was chosen, however, in the construction of the present dye laser. Since an unsaturated single-pass gain of 30 dB over a pathlength of a few millimeters is easily achieved with dye media pumped by a nitrogen laser,[22],[23] the laser cavity can be made so long that only very few (four to eight) light passes are possible during the short excitation. A single spontaneously emitted photon can still be amplified to an avalanche of, say, 10¹⁵ laser photons (0.3 mJ at 6000 Å). The laser bandwidth will then be not much smaller than that of any inserted filter, but the high gain permits the use of very narrow band wavelength selectors despite their losses. If the cavity length is in the order of the coherence length of the emitted laser light, no discrete axial mode structure is expected and the wavelength can be tuned smoothly without complicated feedback control of the cavity length. In addition, thermal refractive index changes of the liquid dye solution do not affect the generated wavelength.

A. General Description

The basic components of the dye laser, constructed according to this philosophy, are displayed in Fig. 1. A side-pumped dye cell is placed inside a cavity 40 cm long using a high dispersion echelle grating as a wavelength selective end reflector. Since only a thin layer near the inner wall of the dye cell is excited by the pump light, diffraction causes a substantial angular spread of the emerging light, which would severely limit the resolution obtainable with the angle-dependent grating. Moreover, only a small area of the grating would be illuminated by the beam. An inverted telescope for collimation and beam expansion is therefore placed inside the cavity. It ensures that the light striking the grating is well collimated and that the beam illuminates a large enough area of the grating for good resolution. In addition to the greatly improved resolution, the beam expansion has the advantage of preventing the aluminized grating surface from being destroyed by too high laser intensities. A tilted Fabry-Perot etalon is inserted in front of the grating for further line-width reduction. Fine tuning is achieved by mechanically changing the tilt angle. Due to the large diameter of the collimated beam, walk-off problems are negligible up to substantial angles. All parts of the dye laser head are mounted rigidly, with fine adjustments, on a massive metal frame.

B. Dye Laser Amplifier

The dye cell is 10 mm long and is made from Pyrex tubing of 12-mm diam. Antireflection-coated quartz windows are sealed to the ends of the cell under a wedge angle of about 10 deg to avoid internal cavity effects. The dye solution is transversely circulated by means of a small centrifugal pump at a rate of approximately 1 liter/min. The flow speed in the active region is in the range of 1 m/sec.

The ultraviolet pump light from a commercial nitrogen laser (AVCO C950, 3371 Å, 100-kW peak power, 10-nsec pulse length, 100 pps) is focused by a spherical quartz lens of 135-mm focal length into a line of about 0.15-mm width at the inner cell wall. To provide a near-circular active cross section, the dye concentration is adjusted so that the penetration of the uv pump light is also in the order of 0.15 mm. The relatively high concentrations used (typically 5 · 10⁻³M/liter of ethanol) are quite insensitive to small amounts of impurities or dye decomposition. The relative excitation density of the excited singlet state can reach 20% under the present conditions.[14]

Despite the high pump energy densities (up to several J/cm³), no noticeable thermal optical distortions are observed during the short excitation, and the amplifying medium is of very homogeneous optical quality. The thermal schlieren effects occur with a substantial time delay after the absorption of the pump light.[23] This delay is ascribed to the comparatively low velocity of the thermal volume expansion. The return to optical homogeneity before the arrival of the next pump pulse is enforced by the liquid circulation, assisted by the fast thermalization times in the small active volume.

Previous experiments were mostly performed with longer dye cells.[9],[10],[11],[16] Here it is often difficult to avoid saturation of the dye medium by amplified unfiltered spontaneous emission (superradiance). In addition, matching of the active volume to a diffraction-limited light beam becomes difficult due to the substantial angular divergence resulting from the small active cross section.

C. Cavity, Telescope

A plane outcoupling mirror is mounted at one end of the optical cavity. A broadband multilayer dielectric coating of 50% reflection is used with dyes of relatively low gain. For high gain dyes, such as Rhodamine 6G, an uncoated quartz surface with 4% reflection is sufficient.

The distance between mirror and dye cell is 50 mm. The corresponding Fresnel number of the small active cross section is less than 1, and the cavity is unstable in conventional terms.[24]

Spontaneous fluorescence light emitted into near-axial directions in the optically pumped dye medium is amplified. Part of this light, after passing an optional Glan-Thomson polarizer, is reflected by the plane end mirror back into the dye cell, where it is further amplified and emerges as a narrow light pencil at the other end. The angular divergence of this beam is essentially diffraction-limited due to the large distance of the mirror image of the active volume which acts as a light source.

The optically pumped region serves as an active pinhole and no further narrow apertures are used to confine the radiation field. Thus difficult alignment procedures are unnecessary, and optimum geometrical matching is easily achieved.

If for simplicity one assumes a Gaussian beam profile [radial intensity distribution I ~ exp(−2r²/w²)] with a waist size w₁ at the center of the active dye volume, the diffraction-limited angular divergence of the emerging light can be described by the far-field angle[25]

Δ θ_{1} = λ / π \cdot w_{1} .

At a waist size of 0.08 mm and a wavelength λ = 6000 Å, the divergence angle Δθ₁ is 2.4 mrad.

The telescope consists of two lenses L₁ and L₂ of focal lengths f₁ = 8.5 mm and f₂ = 185 mm. Both are antireflection-coated multielement systems, corrected for spherical aberration and coma within λ/8. The separation d₁ between dye cell and lens L₁ is 75 mm. The lenses are used slightly off-axis to avoid backreflection and etalon effects.

The waist diameter w₂ of the expanded beam at optimum collimation can be calculated according to[25]

{w_{2}}^{2} = ({f_{2}}^{2} / {f_{1}}^{2}) [{w_{1}}^{2} + {(f_{1} - d_{1})}^{2} \cdot {(λ / π \cdot w_{1})}^{2}] .

The initial waist size of 0.08 mm is enlarged by the telescope to 4 mm, corresponding to a beam diameter of about 10 mm, and the divergence is reduced by a factor of 50 to

Δ θ_{2} = λ / π \cdot w_{2} = 0.048 mrad .

D. Diffraction Grating

The high order echelle grating (blaze angle 63°, 73 grooves/mm) is held in a precision laser mirror mount with 0.1 sec of arc mechanical resolution. The laser tuning range is limited by the separation of adjacent grating orders, and a grating with a larger number of grooves/mm is recommended for dyes with broadband gain profiles.

At an angle of incidence ϕ, the angular dispersion of a diffraction grating in Littrow mount is given by[26]

d ϕ / d λ = 2 \cdot tan (ϕ) / λ .

Since the obtainable resolution is proportional to the number of illuminated grooves, the grating must be at least as large as the oblique beam cross section 2w₂/cos(ϕ). In order to displace the backreflected beam passing through the telescope in reverse direction by one waist size w₁ at the position of the active dye volume, and thus to prevent it from being fully transmitted and further amplified, the deflection angle at the grating must be at least equal to the beam divergence Δθ₂. The corresponding wavelength deviation, according to Eqs. (3) and (4), is

Δ λ_{G} = λ^{2} / 2 π w_{2} tan (ϕ) .

Under the present conditions, this estimated single-pass resolution is 0.08 Å.

E. Fabry-Perot Etalon

The tilted Fabry-Perot etalon, which is inserted into the collimated light field, consists of a quartz plate 6 mm thick (n = 1.458) with broadband dielectric coatings on both sides. The free spectral range is 0.57 cm⁻¹ (0.2 Å at 6000 Å); the finesse F = 20.

The etalon is mounted in an adjustable kinematic holder, and wavelength tuning is achieved by changing the tilt angle θ manually or with a motor-driven micrometer. The positions λ_m of the transmission maxima of this periodic filter satisfy the condition

m \cdot λ_{m} = 2 \cdot d \cdot cos (θ^{'}),

where m is an integer, d is the optical thickness, and θ′ = arcsin [sin (θ/n)] ≈ θ/n. The angular spread of the collimated light field contributes to the filter half-intensity width by an amount

Δ λ_{F} = {(2 \cdot ln 2)}^{\frac{1}{2}} λ \cdot θ \cdot Δ θ_{2} / n^{2} .

The small divergence permits a large tilt angle up to 70 mrad corresponding to a continuous tuning range of 7 Å, until the contribution (7) becomes comparable to the bandwith λ/mF = 0.01 Å caused by the finite finesse.

The wavelength changes by about 0.01 Å for 0.1°C temperature change due to the change in refractive index of the quartz etalon. An air-spaced etalon with Cervit spacer could reduce this temperature sensitivity substantially.

In principle, the line width could be further reduced by increasing the etalon finesse, or by using an additional longer etalon. A lower limit for the obtainable width is the Fourier-transform uncertainty corresponding to the short pulse width (60 MHz at 5 nsec).

III. Performance, Experimental Data

Experimental performance characteristics of the dye laser are given for operation near 6000 Å, using a 5 · 10⁻³M/liter solution of Rhodamine 6G in ethanol. Similar data have been obtained in most parts of the visible spectrum with numerous laser dyes and dye mixtures.

With or without the intracavity etalon, the output beam exhibits a near diffraction-limited divergence of ~2.5 mrad up to the maximum repetition rate of 100 pps. Some weak side maxima, which arise due to the nonuniform transverse gain distribution in the side-pumped dye amplifier, are easily eliminated with an external diaphragm.

For applications where high output powers are required and narrow bandwidth is of secondary importance, the laser can be operated without the etalon. Peak powers of 20 kW at 5-nsec pulse width are easily obtained with 100-kW ultraviolet pump light input. With careful adjustment of the collimating telescope, a bandwidth of 0.03 Å full width at half-maximum (FWHM) can be obtained with the grating alone. This line width is somewhat narrower than the calculated single-pass resolution, Eq. (5).

Insertion of the etalon reduces the bandwidth to 300 MHz or less than 0.004 Å, which is substantially narrower than most atomic Doppler widths. Figure 2 shows two orders of the laser output spectrum with etalon, as monitored with a scanning confocal Fabry-Perot interferometer of 2-GHz free spectral range and an instrumental bandwidth of 7 MHz. The spectrum is averaged over many thousand flashes during a scanning time of 5 min with 1 sec averaging time constant in the photodetector circuit. Tuning the laser to one of the narrow resonance lines of molecular iodine vapor and monitoring the fluorescence confirmed that the wavelength drifts in the absence of undue temperature fluctuations can be less than 0.01 Å over periods of several hours.

The output power drops to 2–4 kW when the etalon is inserted, primarily due to high losses in the available broadband coatings. The pulse amplitude is stable within 5% if the total laser output is monitored. Larger fluctuations are observed with a detector placed behind the narrow band spectrum analyzer, indicating random changes of the laser line shape within the envelope, given in Fig. 2. This spectral noise is quite insensitive to shielding of the dye laser head against mechanical vibrations or changes of the dye flow rate.

To tune the laser over a range of a few GHz, e.g., for studies of the fine structure or shape of an atomic resonance line, the grating can be left fixed and the tilt angle of the etalon alone is changed. The wavelength can be continuously tuned over a range of several Å if grating and etalon are tilted simultaneously by means of a proper gear and lever system.

For the recent sodium experiment,[1] which required a much narrower optical line width, the metioned scanning confocal Fabry-Perot interferometer was placed as an ultranarrow passband filter outside the dye laser cavity. In this way, the bandwidth is reduced to 7 MHz FWHM (≈8.10⁻⁵ Å), and the pulse length is stretched from about 5 nsec to 30 nsec FWHM. Peak powers in the order of several watts are obtained. For continuous wavelength tuning, the piezoceramic spacer of this filter is driven by a proper voltage ramp, which is generated by a precision potentiometer, geared to the mechanical drive of the tilted intracavity etalon.

IV. Conclusions

A reliable narrow band tunable dye laser operating up to high repetition rates is described. A diffraction grating and a tilted Fabry-Perot interferometer are used as wavelength selectors in a long optical cavity with inserted beam-expanding telescope. This simple basic design is made possible by the large gain of organic dye solutions pumped by a pulsed nitrogen laser. The high spectral purity, near diffraction-limited beam divergence, and convenient reproducible wavelength tunability render this laser a powerful tool for optical spectroscopy. The short pulse width makes the laser particularly well suited for dynamic studies such as measurements of lifetimes and relaxation rates. Nonlinear spectroscopic techniques, such as high resolution saturation spectroscopy of Doppler-broadened transitions, which in the past have been essentially restricted to gas laser transitions or molecular transitions in accidental coincidence, are now applicable to any atomic or molecular transition thoughout the visible spectrum.

The author—a NATO Postdoctoral Fellow—wishes to thank Arthur Schawlow for his continuous stimulating interest and generous support. The work was partially supported by U.S. Army Research Office, Durham, N.C.

Figures

Fig. 1 Basic components of narrow band tunable dye laser.

Download Full Size | PDF

Fig. 2 Two orders of the dye laser output spectrum, monitored at 100 pps repetition rate with a scanning confocal Fabry-Perot interferometer of 2-GHz free spectral range. The scan speed is 600 MHz/min.

Download Full Size | PDF

References

1. T. W. Hänsch, I. Shahin, and A. L. Schawlow, Phys. Rev. Lett. 27, 707 (1971) [CrossRef] .

2. B. H. Soffer and B. B. McFarland, Appl. Phys. Lett. 10, 266 (1967) [CrossRef] .

3. A. J. Gibson, J. Sci. Instrum. 2, 802 (1969) [CrossRef] .

4. D. J. Bradley, A. J. F. Durrant, G. M. Gale, M. Moore, and P. D. Smith, IEEE J. Quant. Electron. QE-4, 707 (1968) [CrossRef] .

5. H. Walther and J. L. Hall, Appl. Phys. Lett. 17, 239 (1970) [CrossRef] .

6. D. J. Taylor, S. E. Harris, S. T. K. Nieh, and T. W. Hänsch, Appl. Phys. Lett. 19, 269 (1971) [CrossRef] .

7. F. P. Schäfer and H. Müller, Opt. Commun. 2, 407 (1971) [CrossRef] .

8. M. Hercher and H. A. Pike, Opt. Commun. 3, 65 (1971) [CrossRef] .

9. J. A. Myer, C. L. Johnson, E. Kierstead, R. D. Sharma, and I. Itzkan, Appl. Phys. Lett. 16, 3 (1970) [CrossRef] .

10. G. Capelle and D. Phillips, Appl. Opt. 9, 2742 (1970) [CrossRef] [PubMed] .

11. H. Kogelnik, C. V. Shank, T. P. Sosnowski, and A. Dienes, Appl. Phys. Lett. 16, 499 (1970) [CrossRef] .

12. T. W. Hänsch and A. L. Schawlow, Bull. Am. Phys. Soc. 15, 1638 (1970).

13. I. Itzkan and F. W. Cunningham, IEEE J. Quant. Electron. QE-7, 14 (1971).

14. B. B. Snavely, Proc. IEEE 57, 1374 (1969) [CrossRef] .

15. H. P. Broida and S. C. Haydon, Appl. Phys. Lett. 16, 142 (1970) [CrossRef] .

16. G. Capelle and D. Phillips, Appl. Opt. 9, 517 (1970) [CrossRef] [PubMed] .

17. J. R. Lankard and R. J. Von Gutfeld, IEEE J. Quant. Electron. QE-5, 625 (1969) [CrossRef] .

18. C. E. Moeller, C. M. Verber, and A. H. Adelmann, Appl. Phys. Lett. 18, 278 (1971) [CrossRef] .

19. C. V. Shank, A. Dienes, A. M. Trozzolo, and J. A. Myer, Appl. Phys. Lett. 16, 405 (1970) [CrossRef] .

20. A. Dienes, C. V. Shank, and A. M. Trozzolo, Appl. Phys. Lett. 17, 189 (1970) [CrossRef] .

21. R. J. Von Gutfeld, B. Welber, and E. E. Tynan, IEEE J. Quant. Electron. QE-5, 532 (1970) [CrossRef] .

22. C. V. Shank, A. Dienes, and T. W. Silfast, Appl. Phys. Lett. 17, 307 (1970) [CrossRef] .

23. T. W. Hänsch, F. Varsanyi, and A. L. Schawlow, Appl. Phys. Lett. 18, 108 (1971) [CrossRef] .

24. A. E. Siegman, Proc. IEEE 53, 277 (1965) [CrossRef] .

25. H. Kogelnick, Bell Syst. Tech. J. 44, 455 (1965).

26. M. Born and E. Wolf, Principles of Optics (Pergamon, Oxford, 1970).