Optical fiber transmission of high power excimer laserradiation
Roberto Pini, Renzo Salimbeni, and Matteo Vannini
An experimental investigation of optical fiber transmission of high power excimer laser radiation is presented.Different types of commercially available UV fiber have been tested, measuring energy handling capabilitiesand transmission losses of short samples at the XeCl (308-nm) and KrF (249-nm) wavelengths by using astandard excimer laser. A power density dependent damage process has been observed over 1 GW/cm 2.Fiber losses due to different radii of curvature are also reported. Experimental results have been examined toevaluate the effectiveness of excimer laser transmission through optical fibers for such medical uses as laserangioplasty, including also a comparison between the use of KrF or XeCl emission lines for this purpose.Finally, optimum excimer laser characteristics to increase the energy coupling in fibers are discussed.
1. Introduction
The high potential of excimer laser radiation in abla-tive etching of absorbing materials has been well recog-nized. Since the early 1980s several experiments onthe interaction of excimer laser radiation with variouspolymers have been reported.1- 3 In these works aprocess called photoablative decomposition has beenused, in which material ablation is caused by directbond-breaking of polymer molecules with negligiblethermal effects.
The effectiveness of this process to produce ex-tremely precise cuts with sharp edges suggested prom-ising medical applications of organic tissue ablation(where thermal heating is undesirable) such as oph-thalmic surgery4 and laser angioplasty.5-7 Experi-mental work in these fields of research outlined thefollowing points:
(1) UV photoablative processes are energy depen-dent and show different energy thresholds at the vari-ous excimer emission lines 8 ;
(2) many applications require a suitable optical fi-ber system for local delivery of high power ultravioletradiation.
Consequently, intense activity is growing to developand study new types of fiber transmitting high UVenergies without overattenuating or damaging; on theother hand new coupling systems and suitable excimerlaser devices have to be designed to optimize energy
The authors are with CNR Institute of Quantum Electronics, ViaPanciatichi 56/30, 50127 Florence, Italy.
capabilities of these fibers. In this paper we reportsome experimental work in which a set of differentcommercial UV optical fibers was tested to withstandand transport high excimer laser energies. For thispurpose a standard excimer laser emitting at 249 and308 nm was used. Considering the relatively shortduration of the laser pulse (9-20 ns), an energy damagethreshold up to 25 J/cm 2 was observed (at 308 nm),which is already promising for laser angiosurgery.Transmission coefficients of straight and curved fibersamples with a catheter suitable length of 2 m have alsobeen measured.
11. Fibers Under Test
The fiber set consists of commercially availablestep-index fibers of 200-, 400-, and 600-pim core diam,manufactured by Fibers Optique Industries, Spec-Tran Corp., and Fiberguide Industries. All have beenspecially designed for ultraviolet light transmissionand to our knowledge can represent the state of the artin the field of UV fibers. Each type has a pure silicacore, while different materials are employed for thecladding, such as silicon or doped silica. Their generalspecifications, as reported by manufacturer's datasheets, are listed in Table I. Different fiber types arehere conventionally represented by a letter (A, B ... .)followed by a number which indicates the core diame-ter in micrometers, and an abbreviation (PC or SC) tospecify whether the fiber is plastic cladded or silicacladded.
In previous experiments91 0 one of these fibers(D.200PC) had been used to produce frequency con-version by nonlinear processes, pumping with a XeCllaser. Because of the high power levels required, someobservations of fiber damaging mechanisms had al-ready been done; furthermore, the use of an intermedi-ate medium (water) overlying the input face of the
fiber was advantageous, effectively enhancing the en-ergy capability and giving more reliable results. Fol-lowing these preliminary indications, in this test fibersamples were irradiated both in air and in a few milli-meters of distilled water.
Ill. Experimental Setup
The experimental setup utilizes an UV preionizedexcimer laser of our own design and construction, withstandard characteristics and capable of 50-Hz maxi-mum repetition rate. The laser was operated with twotypes of optical resonator: (1) a stable resonator (SR)composed of a flat rear mirror and an uncoated quartzflat; (2) an unstable resonator (UR) composed of a flatrear mirror and a 40-cm radius plano-convex lens asthe output coupler; emission parameters are reportedin Table II.
When operating with the UR, no further lens wasused to focus the laser beam into the fiber; in the SRcase, light focusing was accomplished by a set of lenseswhich gave a converging beam of N.A. = 0.1. The fiberholder was mounted on a five-axes positioner withmicrometer adjustments to match the beam cross sec-tion with the fiber input face. When underwater cou-
Table 1. Fibers Under Test; Characteristics from Manufacturer's DataSheets
Core Claddingdiam diam Attenuation
Type (um) ('m) N.A. (dB/km)
A.200 SC 200 300 (silica) -B.200 SC 200 240 (silica) 0.22 <3 at 650 nmC.200 SC 200 240 (silica) 0.25 10 at 850 nmD.200 PC 200 380 (plastic) 0.40 <8 at 820 nmE.400 SC 400 600 (silica) -
F.600 PC 600 750 (plastic) 0.40 12 at 850 nm
Table II. Laser Emission Characteristics at 308-nm (XeCI) with Stableand Unstable Resonators
pling was tested, the laser beam was sent downward bya prism while the fiber holder was set vertically underit, supporting also a water cell in which the fiber inputend was immersed. Energy and pulse duration mea-surements were performed using an energy meter(Gentec model ED200) and a vacuum photodiode (ITLmodel TF1850, 0.2-ns risetime), respectively.
IV. Damage Threshold Measurements
Preparation of fiber samples for maximum energycapability measurements has always been done ac-cording to the following procedure:
(1) All fiber samples have the same 50-cm length.(2) Input and output ends of the fiber are cut with a
cleaving machine (Radiall model F 780.005) or a rubytipped knife; no grinding and polishing techniqueshave been used because, in our experience, they couldleave a certain amount of polishing inclusions.
(3) Cut quality and face cleanliness are verified bymicroscope examination (20X objective).
(4) At each fiber end, 3 cm of plastic cladding (orjacket, for the silica cladded fibers) must be stripped toavoid deposition of ablated particles on the couplingface during laser illumination; we have noticed alsothat any residual cladding along the fiber, in closeproximity of the beam focus, absorbs laser light andafter a few shots causes a crack in the silica core. Forsome fiber types stripping operations are extremelydifficult because the protective shield, tightly glued tothe inner silica part, is not removable by the usualsolvents and mechanical stripping often damages thefiber.
(5) The sample is then checked again by microscopeobservation and finally positioned into the fiber hold-er.
Energy coupling optimization has been accom-plished by setting the fiber input end beyond the focusand approaching the higher energy density region stepby step along the beam axis, carefully centering thefiber in the transverse plane.
During the experiments, the use of stable and unsta-ble resonators for the XeCl excimer laser allowed aneasy change of the pulse duration from 9 to 20 ns. Forboth cases, the aperture of the laser beam was kept wellwithin the nominal N.A. of the fibers. Energy damagethresholds and corresponding fluences for the differ-ent fibers under test are listed in Table III.
Table I. Fiber Energy Damage Thresholds and Relative Fluences (in Brackets) at 308-nm (XeCI) In Air and In Water Coupling Conditions with Stable andUnstable Resonators, as Measured at the Output of 0.5-m Long Samples; mJ(J/cm 2 )
Fiber UR t 9 ns, N.A. = 0.04 SR t =20 ns, N.A. =0.1type In air In Water In air In Water
Fig. 1. Input face of an E.400 PC sample after damage occurred.The rectangular burnt area indicates partial illumination of the fiber
face.
Energy levels, as measured at the output of 50-cmlong samples, have been observed with a reasonableoccurrence before the damage process took place (evi-denced by a bright spark on the fiber input face).Reported results represent averages over some trials(more than three for each case; ±10% uncertainty) andhave to be considered indicative of the general trends.
Some remarkable features arising from these experi-ments are worth discussing:
(1) When using the UR, a significant number ofdamage events occurred around 5-15 mm of propaga-tion inside the fiber because of beam refocusing at thefirst internal reflection. The recognition of this pro-cess as a fiber damage mechanism was first reported byAllison et al.11 On the other hand, during SR opera-tion, this kind of damage process was never observed.Consequently it can be argued that high spatial qualityof the laser beam, as given by the UR, can be detrimen-tal to high energy coupling in fibers, while a multimodepattern in the coupling region, as produced by a SR,can prevent internal reflection-induced damage.
(2) Another drawback of the UR case is the presenceof a relatively narrow Gaussian distribution of 200-gmFWHM in the focal plane of the coupling lens (f = 40cm), which makes the fiber matching quite critical,while in the SR case a larger rectangular section of thespot (500 X 1000 ,m) permits more suitable and un-critical alignment operations. This is at the expenseof the laser coupling efficiency, defined as the ratiobetween the energy effectively coupled into the fiberand the overall energy emitted by the laser. Optimiza-tion of this parameter has not been considered a pri-mary aim of the present work, but it can play animportant role in some uses of energy delivery tech-niques through optical fibers. In this respect a futureimprovement should be faced requiring nontrivial so-lutions (as a suitable laser system design) to get a goodcompromise between fiber face illumination homoge-neity and laser energy utilization.
(3) Fluence thresholds have always been calculatedas the ratio between measured energies and core areas.While for the 200-,gm core fibers this represents areasonable approximation, for larger core fibers (400
and 600 im) it leads to an underestimation of thefluence thresholds because at the beam focus the lightspot partially illuminates the fiber face; on the other,moving away from the focus, the fiber was not dam-aged but at the expense of a lower level of coupledenergy. As an example, Fig. 1 shows a rectangularburnt area on the 400-ym fiber face which initiated thedamage process, evidenced by a crack in the silicacladding.
(4) Results obtained in underwater coupling condi-tions show unexpected and different behavior depend-ing on the type of cladding: for the two PC types,water coupling produced a significant increase of per-formances, while for all the SC types higher energylevels were obtained in air. For our purpose the use ofa medium such as distilled water, with intermediaterefractive index between air and quartz, could give abetter match and could partially relax field intensifica-tions due to surface imperfections. Furthermore thehigher dielectric constant of water would raise thebreakdown threshold. These arguments have beenverified for the PC fibers, confirming also our previousobservations reported earlier, but not for the SC fibers,where maybe the unavoidable irradiation of the outercladding ring has to be taken into account. Note thatno comprehensive explanation can be given here.
(5) For all the fibers under test, a large improve-ment of the coupled energy has been obtained usingthe SR due to the longer pulse duration with respect tothe UR. Pulse shapes are shown in Fig. 2; as can benoticed, the peak power is the same for the two caseswhile the higher energy of the SR case is given byfurther cavity round trips following the first one.
These results were obviously expected assuming thedamage mechanism induced by a surface breakdownprocess on the fiber face. In this respect, opticalbreakdown based on electron extraction and subse-quent avalanche is more likely intensity dependentrather than energy density dependent,1 2 so that theusual presentation of damage thresholds in terms offluences is somewhat misleading. Here we wouldrather evaluate a peak intensity as the parameter de-
10 ns/div
Fig. 2. Temporal pulse shapes of the XeCl emission line with (a)stable and (b) unstable resonators.
Table IV. Comparison Between Energy Damage Thresholds at the KrFand XeCI Emission Wavelengths Obtained In Air Coupling and with a
Stable Resonator
Energy damage threshold at 249 and 308 nmmJ(J/cm2 )
KrF ( = 249 nm) XeCl ( = 308 nm)Fiber SR t = 15 ns SR t = 20 nstype N.A. = 0.1 N.A. = 0.1
A.200 SC 0.7 [2.2] 7.0 [22.3]
D.200PC 1.0 [3.2] 5.0 [15.9]
F.600 PC 5.8 [2.1] 28.0 [9.9]
termining a level under which fiber damage should notconfidently occur. An estimate of the intensity dam-age threshold can be calculated considering pulseshapes and energy densities: for example, the D.200PC fiber shows damage values of 1.2 and 1.4 GW/cm2
with stable and unstable resonators, respectively (wa-ter coupling).
Energy measurements at the damage threshold werealso performed at the KrF excimer emission wave-length ( = 249 nm) by using the same laser systemoperating in SR conditions. Fiber types A, D, and Fwere tested; experimental results, compared withthose obtained at the XeCl wavelength, are reported inTable IV. As can be seen, maximum fluences for KrFemission are 5-10 times lower, showing a slight disad-vantage for the SC type, probably due to a higherabsorption of the doped silica cladding which can re-sult in damage developing earlier.
V. Transmission Loss Measurements
Transmission loss is another important parameterthat must be taken into account to evaluate the effec-tiveness of ultraviolet high power delivery throughoptical fibers.
Most of the work on reducing optical losses has so farconcentrated on the low-loss low-dispersion communi-cation window in the near infrared. In the UV, most ofthe losses are due to Rayleigh scattering, which alsodetermines the theoretical limit below which they can-not be reduced. Currently available UV fibers havelosses as high as 100-500 dB/km in the 350-250-nmregion, which hinder their use in optical signal trans-mission for communications but allow a favorable uti-lization when short transmission lengths are required,as in biomedical applications.
To obtain general information about this feature forthe set of fibers under examination, optical losses ofrelatively short samples (2 m) have been measuredboth in straight and curved geometries simulating op-erating conditions of typical applications, such as laserangioplasty.
Transmission losses of straight samples have beenmeasured at both excimer wavelengths (308 and 249nm) by the cutback technique. For each fiber type, a2.5-m long sample was prepared, according to the pre-viously described procedure. The laser was operatedwith the SR, and the coupled energy level was kept at-1 mJ. After the energy output of the 2.5-m sample
Table V. Fiber Transmission of 2-m Long Samples of the KrF (249-nm)and XeCI (308-nm) Excimer Emission Wavelengths
Fiber KrF XeCltype (X = 249 nm) (X = 308 nm)
A.200 SC 80 >90
B.200 SC 85 >90
C.200 SC 85 >90
D.200PC 60 90
E.400 SC 85 >90
F.600 PC 60 90
Table VI. Curvature Losses Measured on Decreasing Radii of Curvature(dB/m)
Fibertype R,= 10cm R,=5cm R,=2.5cm R,= 1cm
A.200 SC 0.70 1.99 4.85 16.90(1.44) (3.00) (6.28)
B.200 SC 0.30 0.45 0.73 1.36(0.34) (0.77) (1.06)
C.200 SC 0.05 0.16 0.26 2.37(0.31) (0.62) (0.88)
D.200 PC 0.05 0.07 0.14 0.43(0.09) (0.13) (0.16)
Note: Values reported in parentheses have been obtained start-ing from the minimum radius and reaching the maximum; a hystere-sis effect is evident.
was measured, the fiber was cut at 0.5 m from the inputend, taking care to preserve the same coupling condi-tion. Energy output of the 0.5-m piece was consideredas the input of the removed 2-m long fiber piece, allow-ing evaluation of the percentage transmitted.
Measured values are shown in Table V; the averageover a relatively small number of trials gives an esti-mated error of -5%. The most remarkable results are(1) transmission at 308 nm is reasonably good for allthe fibers with levels of 90% or more; (2) SC typesperform quite well also at 249 nm, while the two PCtypes present a markedly lower transmission factor.
Curvature losses have been measured comparing theenergy output of a 2-m long sample in a straight pathwith the output from the same sample after a full turnon a circle of fixed diameter. Table VI reports curva-ture losses in dB/m as measured starting from largerradii of curvature, and reaching the minimum of 1 cm(this value was chosen to fit the curvature of the endsegment of a typical catheter for angiography); lossvalues reported in brackets were observed coming backto the larger radii. As expected, SC fibers experiencehigher curvature losses due to the smaller N.A.; thisfeature is particularly evident for the A type: forexample, an easy calculation for this case indicatesthat a full turn on the 1-cm radius reduces light trans-mission by more than 20%. Besides different behaviorof each fiber type, this presentation also points out aninteresting hysteresis effect, as can be noted compar-ing loss values referred to the same radius, obtained in
the roundtrip set of measurements. Fiber bending ona radius of curvature as small as 1 cm produces anirreversible worsening in the local transmission. Thiseffect can be caused by microcracks in the silica clad-ding, induced by mechanical stress overimposed on arelatively high field intensity, and represents in anycase an aging process reducing the lifetime of thesefibers in similar operating conditions.
VI. Conclusions
In this experimental investigation on energy han-dling capability of UV optical fibers transmitting highpower excimer laser radiation, the following pointshave been outlined:
(1) Maximum fluences are limited by a surface opti-cal breakdown damaging the input face of the fibers.This intensity dependent process has been observed tooccur in the 1.0-1.4-GW/cm 2 range at 308 nm.
(2) Fluences as high as 25 J/cm 2 have been mea-sured at the damage threshold for a XeCl laser pulseduration of 20 ns. Higher fluences could be obtainedusing longer pulse durations13 and shaping the tempo-ral evolution to avoid high power peaks, which couldlikely damage the fiber.
(3) High spatial quality beams obtained from un-stable resonators can easily produce by internal reflec-tions very high intensity regions inside the fiber, dam-aging the inner core even below the input face damagethreshold. To prevent this effect, coupled energy op-timization must use suitably focused multimodebeams, determining top-hat intensity distributions inthe coupling region. In other words, the above-men-tioned homogeneous temporal and spatial distributioncan be recognized as the typical requirements allowingbetter control of a strongly nonlinear process such assurface optical breakdown.
(4) Transmission losses of 2-m long straight sam-ples have to be considered reasonably low for the SCtypes at both XeCl and KrF wavelengths, while PCtypes present unsatisfying performances at 249 nm.In this respect, considering UV laser angioplasty as atypical use of this energy delivery technique, the com-parison between fluence levels transmitted at the twowavelengths could favor the XeCl excimer choice. Infact the damage thresholds of fibers for the XeClcase-which are -10 times greater than the ablationthreshold of artery walls8 (5 times for KrF)-and thelower transmission losses result in a higher level ofconfidence for fiber safety and system reliability.
(5) The operative lifetime of UV fibers exposed tohigher power excimer laser radiation (even at 1/5 of thedamage level) will likely be shortened by aging effectsas reported by Taylor.13 In this experiment irrevers-ible worsening of the transmission coefficient wasdemonstrated by bending the fibers around increasingradii of curvature; in this test PC fibers have shownnegligible hysteresis compared with SC fibers.
In conclusion, while the general performance ofcommercial UV fibers for high power delivery of ex-cimer laser radiation can be considered promising forthe previously mentioned applications, further workrequiring long pulse, high repetition rate excimer la-sers is necessary to determine the ultimate limits ofoperating fluences, average power and fiber lifetime,and consequently to demonstrate the effectiveness ofthis technique.
The authors acknowledge the partial support of thiswork offered by the Finalized Project "Tecnologie Bio-mediche e Sanitarie" of the Italian National ResearchCouncil.
References1. R. Srinivasan and V. Mayne-Banton, "Self-developing Pho-
toetching of Poly(ethylene Terephthalate) Films by Far-ultravi-olet Excimer Laser Radiation," Appl. Phys. Lett. 41, 576 (1982).
2. J. E. Andrew, P. E. Dyer, D. Forster, and P. H. Key, "DirectEtching of Polymeric Materials Using a XeCl Laser," Appl.Phys. Lett. 43, 717 (1983).
3. G. J. Bishop and P. E. Dyer, "Polymer Film Cutting and Abla-tive Etching Using a 1-kHz XeCl Laser," Appl. Phys. Lett. 47,1229 (1985).
4. S. L. Trokel, R. Srinivasan, and B. Braren, "Excimer LaserSurgery of the Cornea," Am J. Ophthalmol. 96, 710 (1983).
5. W. S. Grundfest et al., "Pulsed Ultraviolet Laser and the Poten-tial for Safe Laser Angioplasty," Am. J. Surg. 150, 220 (1985).
6. W. S. Grundfest et al., "Laser Ablation of Human Atherosclerot-ic Plaque Without Adjacent Tissue Injury," J. Am. Coll. Cardiol.5, 929 (1985).
7. T. J. Bowker, F. W. Cross, P. T. Rumsby, M. C. Gower, A. F.Rickards, and S. G. Bown, "Excimer Laser Angioplasty: Quan-titative Comparison in vitro of Three Ultraviolet Wavelengthson Tissue Ablation and Haemolysis," Lasers Med. Sci. 1 91(1985).
8. D. L. Singleton et al., "Excimer Laser in Cardiovascular Sur-gery: Ablation Products and Photoacoustic Spectrum of Arte-rial Wall," Appl. Phys. Lett. 48, 878 (1986).
9. R. Pini, R. Salimbeni, M. Matera, and C. Lin, 'Wideband Fre-quency Conversion in the UV by Nine Orders of StimulatedRaman Scattering in a XeCl Laser Pumped Multimode Fiber,"Appl. Phys. Lett. 43, 517 (1983).
10. R. Pini, R. Salimbeni, M. Vannini, A. F. M. Y. Haider, and C.Lin, "High Conversion Efficiency Ultraviolet Fiber Raman Os-cillator-Amplifier System," Appl. Opt. 25, 1048 (1986).
11. S. W. Allison, G. T. Gillies, D. W. Magnuson, and T. S. Pagano,"Pulsed Laser Damage to Optical Fibers," Appl. Opt. 24, 3140(1985).
12. See, for example, Y. R. Shen, "The Principles of NonlinearOptics,"( Wiley, New York, 1984), Chap. 27.
13. R. S. Taylor, K. E. Leopold, S. Mihailov, and J. Seregelyi,"Optical Fiber Transmission in Excimer Laser Angioplasty," inTechnical Digest, Conference on Lasers and Electro-Optics(Optical Society of America, Washington, DC, 1986), paperTUA4.
