yellow conversion of copper vapour laser output

Volume 75, number

3,4

1 March

OPTICS COMMUNICATIONS

EFFICIENT GREEN/YELLOW

1990

CONVERSION OF COPPER VAPOUR LASER OUTPUT

D.W. COUTTS, M.D. AINSWORTH and J.A. PIPER andApplications. Macquarie University, N.S. W. 2109. Australia

Centrefor Lasers Received

2 June 1989; revised manuscript

received

I5 November

1989

The design and operating characteristics of a high average power yellow dye amplifier pumped by the green (1~ 510 nm) and injected by the yellow (A= 578 nm) outputs of a CVL as the bases for efficient green-to-yellow conversion are reported. A simple longitudinal pump and injection geometry was employed to simultaneously couple the CVL pump and injection signals in the gain medium. Amplifier efficiency as functions of pump and injection focal power densities, dye concentration and gain length were investigated. Green-to-yellow amplifier efficiencies in excess of 50% are reported corresponding to approximately 7036 initial CVL laser power becoming available in the yellow. A simple rate equation model for the amplifier is also described and used to evaluate the limitations of the scheme

1. Introduction The technology of copper vapour lasers (CVL) is now well established, laboratory scale systems giving tens of watts average power at 510.6 nm and 578.2 nm currently being available in a variety of commercial packages. Though many applications for the CVL are based on the higher-power green output, several important applications are now emerging for the yellow output for which there are few practical alternative laser sources. Of particular interest is the technique of vascular photocoagulation for the treatment of vascular malformations such as port wine stains and which relies for its effect on selective absorption in the yellow by oxyhaemaglobin [ 11. For normal operating conditions of the CVL the fraction of the total output power available at 578.2 nm is typically one third. This fraction may be enhanced by operating the laser at elevated plasma tube temperatures, however, this effect is generally due to reduction in the green ( 5 10.6 nm) output rather than any significant increase in the 578.2 nm output. Given that the population densities of the 2P,,2 and 2P ,,2 upper laser levels are little affected by the intracavity radiation field for a wide range of discharge conditions appropriate to self-heated discharge CVL’s [ 2 1, it is not surprising that attempts to enhance the CVL output at 578.2 nm by suppressing lasing ac0030-401 S/90/$03.50

0 Elsevier Science Publishers

tion at 510.6 nm have proved unsuccessful. Use of a dye amplifier pumped by the 510.6 nm output and simultaneously injected with the 578.2 nm yellow output of the CVL is an alternative approach to enhancing the 578.2 nm output power available from these systems, however to date only one brief report describing the operation of such an amplifier can be found in the literature [ 31. In this paper we report details of an extensive investigation of a practical green/yellow converter for the CVL based on such a simple dye amplifier arrangement. Green-to-yellow conversion efficiencies exceeding 50% giving average powers in the yellow of approximately 70% of the total available power from the CVL are reported.

2. Experimental

method

A schematic diagram of the experimental set up is shown in fig. 1. The output from a conventional discharge heated CVL is focused directly into a dye cell using the spherical lens Ll giving good spatial overlap of the pump (5 10.6 nm) and injection (578.2 nm) signals in the gain medium efficiently and conveniently. A second lens (L2) is positioned at the exit of the cell to recollimate the output. Apertures are positioned before Ll and after L2 to reduce the

B.V. (North-Holland

)

301

Volume 75. number

3.4

Aperture

Fig. I. Schematic output.

diagram

showing the experimental

configuration

undesirable ASE components of both the CVL and amplifier outputs at the detectors and a dichroic beam splitter (DBS) is positioned at the output of the amplifier to remove the pump radiation component of the amplifier output. All average power measurements were made with a black disc thermopile detector ( Scientech 36000 1 ) and pulse shapes monitored using a high-speed vacuum photodiode (Hamamatsu Rl 19311-02) and displayed on a fast oscilloscope. The dye amplifier cell was of demountable structure incorporating two optical quality silica flats separated by a Teflon spacer and “O”-ring sealed using stainless steel pressure plates. The thickness of the dye region could be varied from 0.6 mm to several millimeters by choosing appropriate thickness of the Teflon spacer. The dye, rhodamine 590 (R590) in solution in ethanol or other solvents was circulated rapidly through the cell giving transverse flow velocities of I-10 m/s. High transverse flow velocities are necessary to minimise photochemical and thermal decomposition of the dye in the active region and to reduce optical distortions and losses through refractive index variations caused by localised heating of the solution by the focused pump beam. A heat exchanger was included in the dye flow circuit to maintain the dye solution at a constant temperature. Note that neither the lenses Ll or L2 or the windows of the dye cell were anti-reflection coated resulting in significant avoidable reflection losses, measured to be 25% overall. The copper vapour laser (CVL) used in these experiments was a conventional self-heated longitudinal device giving average power output up to 7.6 W (total green plus yellow power in the ratio 2 : I ) at a pulse repetition rate of 6 kHz. Corresponding single pulse energies were 0.84 mJ and 0.42 mJ for the green and yellow respectively; peak powers were 302

I March

C)PTIC‘S (‘OMMUNI<‘ATIONS

IWO

Aperture

used in the investigation

of green-to-yellow

converson

of copper laser

approximately 24 kW and I5 kW for roughly triangular pulses of duration 35 ns and 27 ns respectively. Note the onset to lasing for the 578.2 nm pulse occurs some 8 ns after that for the 510.6 nm pulse. and the peak power at 578.2 nm is delayed some I4 ns from that at 510.6 nm. In these experiments no attempt was made to spatially separate the 5 10.6 and 578.2 nm pump and injection signals, or to vary thei] relative temporal positions. Green-to-yellow conversion efficiency has been investigated as a function of the R590 concentration. solvent properties. gain length and pump and injcction power densities. Gain lengths investigated were 0.6 mm, 2.6 mm and 5.0 mm. Pump and injection power densities were most readily varied by changing the focal length of the lens Ll (focal length values of 62.5. 120. 160 and 200 mm were used). The effect of the CVL beam quality on amplifier efficiency has also been investigated. For these experiments the conventional flat-flat resonator of the CVL was replaced with an edge coupled confocal unstable resonator cavity having a magnification of 16. CVL output beam divergence was reduced from 3 mrad ( l/e full angle) for the flat-flat cavity to 0.2 mrad for the unstable cavity with only a small ( d 10%) reduction in overall output power and no change in the ratio of green to yellow output power. Note for the range of focal lengths employed for L I 1 focal spot diameters varied from 0.2 to 0.6 mm for the flat-flat cavity and from 20.02 to 0.06 mm for the M= I6 unstable cavity.

3. Experimental results Green-to-yellow conversion efficiency is shown as functions of the R59O/ethanol concentration and cell length in fig. 2 for fixed focal length for Ll. The data

OPTICS COMMUNICATIONS

Volume 75, number 3.4

x

50

3 .h

40

.g ‘;’

30

.!? e w 20 $ G

10

0

4

2 Dye

Concentration

a

6

10

(1 04U)

Fig. 2. Green-to-yellow conversion effkiency as a function of the R59O/ethanol concentration and cell length for fixed focal length for Ll of 120 mm.

takes account of reflection losses (25%) at the lenses and cell windows and relates to the CVL operated with the flat-flat cavity. For the 0.6 mm cell thickness a maximum effciency of 38% was obtained for a (R590) dye concentration of 7 x 1Oe4 M in ethanol. At higher dye concentrations the correspondingly short optical absorption depth resulted in intense localised heating of the dye mixture leading to the formation of vapour cells and a resultant large scattering loss at both pump and injection wavelengths. This problem could only be overcome by defocusing the CVL beam in the amplifier cell with an accompanying reduction in amplifier efficiency. For the 5.0 m cell a maximum efficiency of 43% was obtained at a substantially reduced R590 concentration of 1 x 10e4 M. With amplifier thickness 2.6 mm a maximum efficiency of 52% was obtained for a R590 concentration of 2 x 1O-4 M. For this configuration the 2.2 W injection signal at 578.2 nm was amplified to 4.2 W for an average pump power (at 5 10.6 nm) of 4.0 W. Thus approximately 70% of the total 6.2 W available from the CVL was ultimately available at 578.2 nm. These conversion efficiencies compare well with reported efficiencies for other green pumped yellow dye amplifiers, for example Hargrove and Kan [4] report conversion efficiencies of 30% at an injection wavelength of 572 nm in a R590 dye amplifier transversely pumped at 5 10 nm by a CVL. A comparison of amplifier performance for the

I March 1990

CVL fitted with flat-flat or a M= 16 confocal unstable resonator is presented in table 1. In table 1 the data for the flat-flat resonator is presented as a function of both gain length and focal length of lens Ll, while for the unstable cavity data is presented for the optimum cell length (flat-flat configuration) of 2.6 mm only. All efficiency values quoted are for optimum dye concentration. From inspection of table 1 it is clear that for conditions of optimum cell length and dye concentration, amplifier efficiency appears to be approximately independent of CVL output beam quality and choice of Ll. This result may at first seem surprising however it is explained by our observation that with improved beam quality and subsequent tighter focusing in the amplifier, losses due to thermal effects in the dye quickly nullify any gains made through higher focal power densities; in fact, to obtain data for the unstable cavity it was necessary to somewhat defocus the pump and injection beams in the gain volume. Under some conditions modification of the spectral power density of the 578.2 nm CVL output in passing through the amplifier was observed. Lineshapes of the CVL and amplified 578.2 nm outputs were measured using a Fabry-Perot interferometer (Burleigh, FSR 30 GHz) and boxcar integrator (ORTEC 94 15/9425). For the flat-flat cavity, of the three components to the 578.2 nm line due to the hyperfine splitting of the 2P,,z and 2D3,2 resonance and metastable levels of copper, that with the shortest wavelength has a magnitude 1.2 times that of either of the other two, components (see also ref. [ 31). After passing through the amplifier this component was enhanced to a magnitude 1.9 times that of the other two components. For the CVL fitted with the unstable resonator all three components had equal magnitude and no spectra1 enhancement of the components of the 578.2 nm line was observed. For normal operation the output of the CVL was only weakly polarised ( polarisation ratio 3 : 1 ). No change in amplifier performance was observed if the CVL output was strongly polarised (polarisation ratio 100: 1 ) using an intra-cavity polarizing cube. The amplifier was incorporated at the output of a commercial 10 W air cooled CVL (metalaser Technologies MC-lo) fitted with standard flat-flat optics. For average output power of 12 W, the green to yellow ratio was 60: 40 and output energy 1.2 mJ ( 10 303

Volume 75, number Table I Optimum

conversion

OPTICS

3.4

efficiencies

.f(mm)

62.5 120 160 200

for different

amplifier

COMMUNICATIONS

I March

configurations.

Cell length (mm

)

0.6 (f-f)

2.6 (f-f)

5.0 (f-f)

2.6 (unstable W= 16)

37.9% 38.5% 37.6% 37.3%

52.0% 52. 1% 52.3%

43.0% 41.9% 41 .9% 40.2%

51.5% 52.9% 49.5%

f-f

1990

means flat-flat

kHz pulse repetition rate) corresponding to peak powers of 35 kW and 25 kW for the green and yellow respectively in roughly triangular pulses of fwhm 20 ns. A maximum amplifier efficiency of 53% (accounting for avoidable optical losses) was measured at a dye concentration of 2~ 1O-4 M R590/ethanol for ,f= 160 mm for L 1 corresponding to an average power available from the CVL/dye amplifier at 578.2 of ethanol with nm of ~9 W. The replacement deionized water had only a small effect on amplifier efficiency: for mixtures consisting of 60-70% water balance ethanol, laser efficiency was reduced by only 3-4%. The effect of water on the lifetime of the dye has not been investigated to date.

4. Discussion

volume (at a fixed wavelength in the emission band ). With only the pump signal incident on the amplifier the fluorescence was observed to follow the temporal evolution of the 5 10.6 nm pulse as in fig. 3. Injection of the 578.2 nm signal into the amplifier reduces the fluorescence for the duration of the 578.2 pulse. but has no effect during the first 8-10 ns. From this observation WCestimate that approximately 15% of the pump energy was not accessed by the injected signal

“: 20

IIS

-

Maximum green-to-yellow conversion efficiency is 85% based on a quantum efficiency of unity. For the amplifier configuration used here there are a number of reasons which may account for measured efficiencies of only 50%. Firstly, there is incomplete temporal overlap between the pump and injection signals from the laser. For the CVL of the present study the 578.2 nm pulse was delayed some 8-9 ns after the onset of the 5 10.6 nm pump pulse due to the different cavity build up times of the two copper laser lines. Incomplete pulse overlap and the fact that there is little upper level storage in the dye (first excited singlet state lifetime for R590 is z 4 ns [ 51 ) results ,in the loss of a substantial fraction of the pump energy through fluorescence or ASE. This effect is easily demonstrated by monitoring the temporal evolution of the singlet fluorescence from the pumped 304

PUMP

SIGNAL

FLUORESCENCE

x PUMP

AND

INJECTION

FLUORESCENCE

SIGNALS

Fig. 3. Comparison ofdye cell fluorescence without the 578.2 nm injection signal.

(at 558 nm) wth and

Volume 75, number

OPTICS COMMUNICATIONS

3,4

at 578.2 nm due to temporal mismatch. Secondly, there is incomplete spatial overlap of the pump and injection signals since the beam quality of the injection signal is rather better than that of the pump as a result of lower gain at 578.2 nm in the CVL. This further reduces extraction efficiencies as the cross sectional area of the gain volume is larger than that of the injection signal. Finally, the short upper level spontaneous emission lifetime results in spontaneous emission competing with stimulated emission. Estimates based on saturation flux for the amplifier suggest that at least 10% of the upper level population is lost through spontaneous emission even in the presence of the intense injection field. A simple rate equation mode1 for the green-to-yellow amplifier has been developed and used to predict amplifier performance for a wide variety of pump and injection powers, pulse overlap (temporal and spatial), dye concentration and gain length. For a longitudinally pumped dye amplifier with collinear pump and extraction beams. the system can be fully described by the following three rate equations (see for example Hargrove and Kan [4] and Nair [6])

c-‘al,/at+al,/a,y=-((T~,nO+(T~Zn,) c-‘aIi/at+aIi/ax=(a,n,

-a~,no)

dn,/dt=aoP,no/p+~o,noI,-n,/7,-a,n,I,,

I~, I,,

1 March

species to relax. Consequently circulation of the dye solution is required solely to avoid dye heating in the pump region. Spatial overlap issues can be dealt with by simple geometric means. The comparatively long CVL pulse duration and rapid relaxation mechanisms of the dye allow steady state solution of eqs. ( 1 )-( 3). Eqs. ( I)-( 3) are integrated along the dye cell length using measured pulse shapes for IP and 1, and ground state density no. Time averaged amplifier outputs were calculated for a range of beam diameters, cell lengths and dye concentrations. In these calculations Z, and Ii were assumed to propagate through the dye solution with constant beam diameter, the value of which is determined by the focal length of Ll (this is a good approximation for the case of the relatively high divergence CVL beam). Calculated green-to-yellow conversion efficiencies for conditions relevant to the present experiments (see table 2 ) are shown in fig. 4. For the three cell lengths investigated there is good agreement between predictions of the mode1 and experimental results. Table 2 Parameters

(1)

used in rate equation

(2)

os1 ob, aP2

(3)

G r,

where no, and n, are the dye ground and first excited singlet states respectively, IP and Z, the pump and injection intensities (expressed in photons cm-2 s-i); a;, , a;, and ay2 are the pump and injection ground state and pump wavelength excited state absorption cross sections respectively. The stimulated emission cross section is denoted by a, and the upper level spontaneous lifetime by 7,. A term for excited state absorption is not included in eq. (3) since the radiationless decay time for the second excited state (7512%IO-l2 s) is much StW]]er than 7,~ lop9 s. In addition singlet to triplet intersystem crossing is ignored as the inverse rate constant of lo-’ s [ 71 is greater than the pump pulse width and comparable to the triplet state lifetime [ 7 1. For an interpulse time of z 3000 triplet state lifetime (CVL operated at 6 kHz) there is sufficient time between successive CVL pump pulses for any triplet

1990

analysis.

1.78X10-‘6cm2 [S] 0.02 x IO-l6 cm* [ 91 0.37 x 10-‘6cm2 [8] 2.07 x 10-‘6cm* [8] 4.4 x Io-9s [4,5]

60

5.0mm cell 0.6mm

=

2

4 6 Dye Concentration (1 04M)

u

cell

”

8

10

Fig. 4. Comparison of the calculated (-) and experimental conversion efficiencies as a function of R59O/ethanol concentration and cell length for fixed focal length for L I of 120 mm.

305

For incident pump pulse intensity of 3.8~ IO” photons SC’ cmp2 the calculated saturation intensity for the amplifier is 4.37~ 1014 photons SC’ cm-‘. This photon flux is reached within the first 10% of the amplifier cell length so that the amplifier is driven into saturation. Thus the 578.2 nm output scales linearly with focal beam diameter with the result that at optimum dye concentrations, the incident pump power density is the most important factor in determining the conversion efficiency of the amplifier. The model predicts an output intensity that is functionally dependent on the product of the cell length and dye concentration: the maximum 578.2 nm output occurring when the product is = 4.5 x 10e8 mol cm-‘. At higher values of dye concentration complete absorption of the pump radiation leaves an unexcited length of dye solution which acts a yellow absorber reducing the amplifier efficiency. In practice efficiencies at high dye concentrations (e.g. > 2 x 10e4 M for the 5.0 mm cell) were lower than predicted by the model due to thermal effects previously noted and the optimum cell length is that for which a uniform high optical power density can be maintained throughout the cell without overheating the dye solution. Under the present experimental conditions this was best achieved using the 2.6 mm cell with J= 160 mm for lens Ll. Note that the maximum conversion efficiency predicted by the model is 60% for conditions similar to those employed in the present experiments but where there is complete temporal and spatial overlap of the pump and injection signals. The validity of the steady state approximation was verified by solving the full time dependent rate equations ( 1 )-( 3). Amplifier output powers determined by this model agreed with the time independent model to within 4%.

5. Conclusion The results of a detailed

306

1 March 1990

OPTICS COMMUNICATIONS

Volume 75. number 3.4

investigation

of a prac-

tical green-to-yellow dye amplifier for the copper vapour laser have been presented. For a simple longitudinal pump and injection geometry and for conditions of optimal gain length, dye concentration and pump and injection focal power densities a maximum amplifier conversion efficiency of just ovet 50% was obtained. This corresponded to 9 W a\erage power at 578.2 nm from a 12 W CVL having a green/yellow ratio of 60: 40. Modelling of the amplifier suggested that a maximum conversion efficiency of 60% is possible and is limited by spontaneous emission losses of the dye and the different focal beam diameters of the pump and injection signals.

Acknowledgement The authors wish to acknowledge the National Kesearch Fellowship Scheme for supporting the position of M.D. Ainsworth and the Commonwealth Special Research Centre Scheme for funding the research.

References [ I ] R.R. Anderson and J..A. Parrish, Lasers Surg. Med. I ( I98 I ) 263. [2] D.J.W. Brown. Ph.D. Thesis, University of New England (1988). [3] V.I. Kravchenko. A.Ya. Ltvinenko and 4.A. Smirnov. Sov. Tech. Phys. Lett. 5 ( 1979) 277. [4] R.S. Hargrove and T. Kan. IEEE J. Quantum Electron. QE16 (1980) 1180. [5] A. Penzkofer and Y. Lu. Chem. Phys. 103 (1986) 399. [6] L.G. Nair. Proc. Quantum Electron. 7 ( 1982) 153. [7] B.B. Snavely. Proc. IEEE 57 (1969) 1374. [ 8 ] P.R. Hammond. IEEE J. Quantum Electron. QE- I5 ( I979 ) 624. [9] A.A. Hnilo. O.E. Martinez and E.J. Quel, IEEE J. Quantum Electron. QE-22

( 1986)

20.