Appl. Phys. A 63, 109—115 (1996)

Femtosecond, picosecond and nanosecond laser ablation of solids

B.N. Chichkov, C. Momma, S. Nolte, F. von Alvensleben, A. Tunnermann

Laser Zentrum Hannover e.V., Hollerithallee 8, D-30419 Hannover, Germany(Fax: #49-511/2788-100)

Received: 23 February 1996/Accepted: 27 February 1996

Abstract. Laser ablation of solid targets by 0.2—5000 psTi : Sapphire laser pulses is studied. Theoretical modelsand qualitative explanations of experimental results arepresented. Advantages of femtosecond lasers for precisematerial processing are discussed and demonstrated.

PACS: 81.15

Efficient use of lasers for precise material processing isimpossible without a thorough knowledge of the funda-mental laws governing the interaction of laser radiationwith matter. For this goal systematic studies of the laser-matter interaction are necessary. Due to the recent pro-gress of laser systems, especially those based on the chir-ped pulse amplification (CPA) technique, such systematicstudies have become possible in a very broad range oflaser parameters. With CPA systems the laser pulse dura-tion can be varied from about one hundred femtosecondsto several nanoseconds leaving other laser parametersunchanged. This allows to perform a very detailed analy-sis of different nonstationary laser-matter interaction pro-cesses. As an example, recent investigations of the damagethreshold [1], of the ablation threshold [2], and of thehigh-intensity laser ablation [3] can be mentioned. Notethat such systematic studies have just begun, much moreshould be done for the understanding and demonstrationof the potential of femtosecond laser systems for precisematerial processing.

Recently several experiments [4—6] comparing laserablation of solid targets by femtosecond and nanosecondpulses have been performed. In [4] and [5, 6] advantagesof femtosecond lasers for precise material processing withdye and excimer laser systems have been demonstrated. Inthis paper, we present our results on ablation and holedrilling with a commercial, 10 Hz, femtosecond Ti : Sap-phire laser system which provides laser pulses at 780 nmwith a pulse energy of up to 100 mJ and a variable pulseduration in the range of 0.2 to 5000 ps. Experiments areperformed in a low fluence regime, slightly above theevaporation threshold. This regime is most interesting for

precise material processing, since in this case the energydeposited into the solid and the heat-affected zones areminimized. We discuss and demonstrate advantages of thefemtosecond-pulse lasers. We hope that our results canstimulate new investigations in this field. In Sect. 1, char-acteristic features of the low fluence laser ablation of metaltargets in three different pulse duration regimes: fem-tosecond, picosecond and nanosecond are considered.The experimental setup and results are presented inSect. 2.

1 Theoretical background

During the interaction of low intensity short laser pulseswith metal targets the laser energy is absorbed by freeelectrons, due to the inverse Bremsstrahlung. Then theevolution of the absorbed laser energy involves thermali-zation within the electron subsystem, energy transfer tothe lattice, and energy losses due to the electron heattransport into the target. If we assume that the thermaliz-ation within the electron subsystem is very fast and thatthe electron and the lattice subsystems can be character-ized by their temperatures (¹

eand ¹

i), the energy trans-

port into the metal can be described by the followingone-dimensional, two-temperature diffusion model [7, 8]:

Ce

L¹e

Lt"!

LQ(z)

Lz!c (¹

e!¹

i)#S , (1)

Ci

L¹i

Lt"c (¹

e!¹

i) , (2)

Q(z)"!keL¹

e/Lz , S"I(t) Aa exp (!az) . (3)

Here z is the direction perpendicular to the target surface,Q(z) is the heat flux, S is the laser heating source term, I(t)is the laser intensity, A"1!R and a are the surfacetransmissivity and the material absorption coefficient,C

eand C

iare the heat capacities (per unit volume) of the

electron and lattice subsystems, c is the parameter charac-terizing the electron-lattice coupling, k

eis the electron

thermal conductivity. In the above equations a thermalconductivity in the lattice subsystem (phonon component)is neglected. The electronic heat capacity is much less thanthe lattice heat capacity, therefore electrons can be heatedto very high transient temperatures. When the electrontemperature (in units of energy) remains smaller than theFermi energy, the electron heat capacity and thenonequilibrium electron thermal conductivity are givenby C

e"C@

e¹

e(where C@

eis a constant) and k

e"k

0(¹

i) ·

¹e/¹

i(where k

0(¹

i) is the conventional equilibrium ther-

mal conductivity of a metal) [9—11]. Expressions for thecoupling constant c are given in [7, 12] and results ofrecent measurements in [11, 13, 14].

Equations (1—3) have three characteristic time scales qe,

qiand q

L, where q

e"C

e/c is the electron cooling time,

qi"C

i/c is the lattice heating time (q

e@q

i) and q

Lis the

duration of the laser pulse. These parameters define threedifferent regimes of the laser-metal interaction which wecall femtosecond, picosecond and nanosecond regimes.

Femtosecond pulses

First we consider the case when the laser pulse duration isshorter than the electron cooling time, q

L@q

e. For t@q

e,

which is equivalent to Ce¹

e/tAc¹

e, the electron-lattice

coupling can be neglected. In this case (1) can be easilysolved. Since the general solution of this equation is quitecomplicated, we neglect the electron heat conduction termin our formulas. This can be done when the followingcondition is fulfilled D

eqL(a~2, where D

e"k

e/C

eis the

electron thermal diffusivity. In this case (1) reduces to

C@eL¹2

e/Lt"2I

aa exp (!az) (4)

and gives

¹e(t)"A¹2

0#

2Iaa

C@e

t exp (!az)B1@2

. (5)

Here I(t)"I0

is assumed constant, Ia"I

0A, and

¹0"¹

e(0) is the initial temperature. At the end of the

laser pulse the electron temperature is given by

¹e(q

L)KA

2Faa

C@eB1@2

exp (!z/d) , (6)

where ¹e(q

L)A¹

0is assumed, F

a"I

aqL

is the absorbedlaser fluence, and d"2/a is the skin depth.

The evolution of the electron and lattice temperaturesafter the laser pulse is described by (1—3) with S"0. Initialconditions for the electron and lattice temperatures aregiven by (6) and ¹

i"¹

0. After the laser pulse the elec-

trons are rapidly cooled due to the energy transfer to thelattice and heat conduction into the bulk. Since this elec-tron cooling time is very short, (2) can be written as¹

iK¹

e(q

L) t/q

i(here the initial lattice temperature is

neglected). The attainable lattice temperature is deter-mined by the average cooling time of the electronsqae"C@

e¹

e(q

L)/2c and is given by

¹i^¹2

e(q

L)

C@e

2Ci

K

Faa

Ci

exp (!az) . (7)

Note that the problem of the hot electron relaxationdynamics in metals after the excitation by a femtosecondlaser pulse has been intensively studied during the lastyears [14—16]. It has been shown that the time scale forthe fast electron cooling and a considerable energy trans-fer to the lattice is of the order of 1 ps.

The significant evaporation occurs when Ci¹

ibe-

comes larger than o), where o is the density and ) isthe specific (per unit mass) heat of evaporation. Using (7),we can write the condition for strong evaporation in theform

Fa5F

thexp (az) , (8)

where FthKo)/a is the threshold laser fluence for evapo-

ration with femtosecond pulses. Then the ablation depthper pulse ¸ is

¸K a~1 ln (Fa/F

th) . (9)

The logarithmic dependence of the ablation depth on thelaser pulse fluence is well known for the laser ablation oforganic polymers. Recently the logarithmic dependence ofthe ablation depth per pulse has been demonstrated formetal targets with femtosecond KrF-laser pulses [5].

Due to the very short time scales involved in theablation with femtosecond laser pulses the ablation pro-cess can be considered as a direct solid-vapor (or solid-plasma) transition. In this case the lattice is heated ona picosecond time scale which results in the creation ofvapor and plasma phases followed by a rapid expansion invacuum. During all these processes thermal conductioninto the target can be neglected in a first approximation.These advantages of femtosecond laser pulses allow veryprecise and pure laser-processing of metals (and othersolids) which is experimentally demonstrated below.

Picosecond pulses

Now we turn to the discussion of ablation withpicosecond laser pulses when the following condition isfulfilled q

e@q

L@q

i.

At a time tAqe

which is equivalent to Ce¹

e/t@c¹

e,

(1) for the electron temperature becomes quasistationary,and (1—3) reduce to

L/Lz (keL¹

e/Lz)!c(¹

e!¹

i)#I

aa exp (!az)"0, (10)

¹i"

1

qi

t

P0

expA!t!h

qiB ¹

e(h) dh#¹

0. (11)

Here (2) for the lattice temperature is written in the inte-gral form. These equations describe heating of metal tar-gets by the laser pulses with q

LAq

e. When the condition

t@qiis fulfilled, (11) can be simplified due to the quasi-

stationary character of the electron temperature. Neglect-ing ¹

0, we get

¹i^¹

e(1!exp (!t/q

i)) ^ (t/q

i)¹

e. (12)

As can be seen from this expression, in the picosecondregime the lattice temperature remains much less than the

110

electron temperature. This allows to neglect the latticetemperature in (10). The analysis of (10, 12) is especiallysimple when the condition k

e¹

ea2@c¹

eis fulfilled.

In this case the electron cooling is due to the energyexchange with the lattice. The electron temperature andthe lattice temperature at the end of the laser pulse aregiven by

¹e^

Iaa

cexp (!az) , ¹

i^

Faa

Ci

exp (!az) . (13)

Note that the attainable lattice temperature after the laserpulse is again determined by the electron cooling time.Since q

e@q

L, the attainable lattice temperature and

the lattice temperature at the end of the laser pulse areapproximately equal. In femtosecond and picosecondregimes (7) and (13) give the same expressions for thelattice temperature. Therefore, the condition for strongevaporation given by (8), the fluence threshold andthe ablation depth per pulse given by (9) remain un-changed.

Thus, the logarithmic dependence of the ablationdepth on the laser pulse fluence is also possible in thepicosecond range. In our derivations we have neglectedthe electron heat conduction into the target. This is a verycrude assumption for the description of laser ablation ofmetal targets in the picosecond regime. Laser ablation inthis case is accompanied by the electron heat conductionand formation of a melted zone inside the target. In spitethat at the surface we can again consider evaporation asa direct solid-vapor (or solid-plasma) transition, the pres-ence of the liquid phase inside the target reduces theprecision of laser processing of metals in this regime (seebelow).

Nanosecond pulses

Here we briefly discuss ablation with nanosecond laserpulses when the condition q

LAq

iis fulfilled. In this case

the electron and lattice temperatures are equal¹

e"¹

i"¹, and (1—3) reduce to

CiL¹/Lt"L/Lz (k

0L¹/Lz)#I

aa exp (!az) . (14)

Laser heating of metal targets by long laser pulses hasbeen a subject of many experimental and theoretical stud-ies [17—20]. In this regime the absorbed laser energy firstheats the target surface to the melting point and then tothe vaporization temperature. Note that metals needmuch more energy to vaporize than to melt. During theinteraction the main source of energy losses is the heatconduction into the solid target. The heat penetrationdepth is given by l\(Dt)1@2, where D is the heat diffusioncoefficient, D"k

0/C

i. Note that for long-pulse laser abla-

tion of metal targets the condition DqLa2A1 is usually

fulfilled. The energy deposited inside the target per unitmass is given by E

m\I

at/ol. When at a certain moment

t"tth

this energy becomes larger than the specific heat ofevaporation ), significant evaporation occurs. From thecondition E

m\) we get t

th\D ()o/I)2 (see [17]). Thus,

the condition for strong evaporation, Em') (or q

L't

th),

can be written as

I'Ith

\o)D1@2

q1@2L

, F'Fth\o)D1@2]q1@2

L(15)

for laser intensity and fluence, respectively. The thresholdlaser fluence which is necessary for evaporation with longlaser pulses grows as q1@2

L.

In case of ablation with long laser pulses there isenough time for the thermal wave to propagate into thetarget and to create a relatively large layer of meltedmaterial. In this case the evaporation occurs from theliquid metal, which makes precise material processing ofmetal targets in vacuum with nanosecond pulses verycomplicated.

2 Experimental setup and results

Now we turn to the discussion of experimental results ona low fluence laser ablation of metals and other solids. Inour experiments a commercial femtosecond Ti : Sapphirelaser system (BMI Alpha 10A) based on the chirped-pulseamplification (CPA) technique [21] is used. A detaileddescription of the principal setup is given in [22]. Thissystem provides laser pulses at 780 nm with a variablepulse energy of up to 100 mJ. Due to the CPA techniquethe pulse duration can be varied from 200 fs up to 400 ps.In this pulse duration range the spectral pulse widthremains constant (approximately 8 nm). Pulse durationmeasurements for pulses shorter than 10 ps are performedby a background free second order autocorrelator and forpulses longer than 10 ps by a picosecond streak camera(Hadland, IMACON 500). Pulse widths of 3—5 ns areobtained when the regenerative amplifier (see [22]) is notseeded by the femtosecond oscillator (Coherent, MiraBasic). In this case the regenerative amplifier operates asan oscillator, and the spectral width of the generatedpulses is significantly reduced. The time durations of thenanosecond pulses have been measured by a photodiodein combination with a fast sampling oscillograph and bythe streak camera.

Our experiments are performed in the low fluenceregime (F"0.1—5 J/cm2) with an imaging geometry. Anaperture (d"5 mm), mounted in the beam pass, is imaged(with demagnification factor of ^1/30) onto the targetsurface by a f"140 mm suprasil lense. The laser beamdiameter is approximately 20 mm. As target materialssteel, copper, AlN and silicon plates are used. These platesare attached to a computer controlled x,y,z-translationstage which is mounted in a vacuum cell at a pressurebelow 10~4 mbar. The number of pulses which are neces-sary to drill through the target plate is controlled bya photodiode mounted behind the targets.

First we present our results on the laser processing ofmetal targets. We compare holes drilled in 100 lm thicksteel foils (in vacuum) with 104 laser pulses in three differ-ent regimes: femtosecond, picosecond and nanosecond. InFig. 1 schematic of femtosecond-pulse laser ablation andthe hole drilled with 200 fs, 120 lJ, F"0.5 J/cm2 laserpulses are shown. As can be seen, there is no trace of themolten material. Only a vapor dust ring around the hole.

111

a b

Fig. 2a, b. Schematic ofnanosecond-pulse laser ablationand holes drilled in a 100 lmthick steel foil with (a) 80 ps,900 lJ, F"3.7 J/cm2; and(b) 3.3 ns, 1 mJ, andF"4.2 J/cm2 laser pulses at780 nm

Fig. 1. Schematic of femtosecond-pulse laser ablation and a SEMphotograph of a hole drilled in a 100 lm thick steel foil with 200 fs,120 lJ, F"0.5 J/cm2 laser pulses at 780 nm

In Fig. 2 schematic of nanosecond-pulse laser ablationand the holes drilled with (a) 80 ps, 900 lJ, F"3.7 J/cm2;and (b) 3.3 ns, 1 mJ, and F"4.2 J/cm2 laser pulses areshown. The trace of the molten material can be seen in

these figures. The presence of liquid phase leads to anunstable drilling process (see Fig. 2a). In case of ablationwith nanosecond pulses there is enough time for thethermal wave to propagate into the metal target andto create a relatively large molten layer. In this case thetarget material is removed both in vapor and liquidphases, since the vaporization process creates a recoilpressure that expels the liquid. In Fig. 2b a ‘‘corona’’created due to the recoil vapor pressure is clearly seen.Note that these are the best results which we obtain with80 ps and 3.3 ns laser pulses at fluences as low as possible.When the laser fluence is reduced further the drillingthrough the 100 lm steel plate in vacuum becomes im-possible.

Comparing Fig. 1 and Fig. 2, the advantages of fem-tosecond-pulse lasers for precise material processing be-come evident. Moreover, due to the absence of thermallosses in the femtosecond regime, holes in metal targetscan be drilled with much lower laser fluences.

Below we present additional illustrative informationon femtosecond-pulse laser ablation of different targetswith the second harmonic of Ti : Sapphire laser radiation(j"390). The second harmonic is used to minimize dif-fraction effects and to improve the quality of image projec-tion. In Fig. 3 the development of the ablation processwith the increasing number of 250 fs, 0.5 mJ, andF"2.5 J/cm2 laser pulses a) 10, b) 100, c) 103 andd) 5]103 is shown. The target is a 0.5 mm steel plate. InFig. 3c the creation of characteristic structures is clearlyseen. These structures appear due to the instability ofa plane evaporation front [23].

In Fig. 4 the evolution of femtosecond-pulse laserablation of a 1 mm thick copper target is shown with:a) 10, b) 100, c) 103, and d) 3]104 pulses. In Fig. 5

112

Fig. 3a–d. Femtosecond-pulselaser processing of a 0.5 mmsteel plate with 250 fs, 0.5 mJ,and F"2.5 J/cm2 secondharmonic radiation(j"390 nm) and differentnumber of pulses: (a) 10,(b) 100, (c) 1000 and (d) 5000laser pulses

Fig. 4a–d. Evolution offemtosecond-pulse laserablation of a 1 mm thick coppertarget with: (a) 10, (b) 100,(c) 103, and (d) 3]104 pulses.Laser parameters are the sameas in Fig. 3

results of laser ablation of a 0.3 mm thick silicon target aredemonstrated with: a) 10, b) 100, c) 5]103, and d) 104pulses. Redeposition of the ablated material on the targetsurface is much stronger than in the case of metal targets.

In Fig. 6 the ablation of a 0.8 mm AlN target with: a) 10,b) 103, and c) 3.5]104 laser pulses is illustrated. Thelaser parameters in Figs. 4, 5 and 6 are approximately thesame as in Fig. 3.

113

Fig. 5a–d. Laser ablation ofa 0.3 mm thick silicon targetwith: (a) 10, (b) 100, (c) 5]103,and (d) 104 pulses. Laserparameters are the same as inFig. 3

Fig. 6a–c. Ablation of a 0.8 mmAlN target with: (a) 10, (b) 103,and (c) 3.5]104 laser pulses.Laser parameters areapproximately the same as inFig. 3

Closing this section we summarize the main featuresof femtosecond laser material processing: very rapidcreation of vapor and plasma phases, negligible heatconduction and the absence of liquid phase. The

absence of the liquid phase allows better control duringthe drilling process. Therefore, the reproducibilityof our results with femtosecond laser pulses is verygood.

114

3 Conclusion

Theoretical models and experimental results on femto-second, picosecond and nanosecond laser ablation ofmetal targets are presented. Very pure ablation of metaltargets in vacuum with femtosecond laser pulses is demon-strated. Sharp well defined patterns can be ablated inmetal and other solid targets by an image projectiontechnique. These advantages of femtosecond lasers arevery promising for their future applications in precisematerial processing.

Acknowledgements. We would like to thank Hanna Jacobs for tech-nical assistance and acknowledge financial support by the GermanMinistry of Science, Education, Research, and Technology (BMBF,13N6590/0).

References

1. B.C. Stuart, M.D. Feit, A.M. Rubenchik, B.W. Shore, M.D.Perry: Phys. Rev. Lett. 74, 2248 (1995)

2. P.P. Pronko, S.K. Dutta, D. Du, R.K. Singh: J. Appl. Phys. 78,6233 (1995)

3. C. Momma, B.N. Chichkov, S. Nolte, F. von Alvensleben, A.Tunnermann, H. Welling: Opt. Commun. accepted

4. W. Kautek, J. Kruger: SPIE Proc. 2207, 600 (1994)5. S. Preuss, A. Demchuk, M. Stuke: Appl. Phys. A 61, 33 (1995)6. J. Ihlemann, A. Scholl, H. Schmidt, B. Wolff-Rottke: Appl. Phys.

A 60, 411 (1995)

7. M.I. Kaganov, I.M. Lifshitz, L.V. Tanatarov: Sov. Phys. —JETP4, 173 (1957)

8. S.I. Anisimov, B.L. Kapeliovich, T.L. Perel’man: Sov. Phys.—JETP 39, 375 (1974)

9. M.B. Agranat, A.A. Benditskii, G.M. Gandel’man, P.S. Kon-dratenko, B.I. Makshantsev, G.I. Rukman, B.M. Stepanov: Sov.Phys. —JETP 52, 27 (1980)

10. M.B. Agranat, S.I. Anisimov, S.I. Ashitkov, B.I. Makshantsev,I.B. Ovchinnikova: Sov. Phys. —Solid State 29, 1875 (1987)

11. P.B. Corkum, F. Brunel, N.K. Sherman, T. Srinivasan-Rao:Phys. Rev. Lett. 61, 2886 (1988)

12. P.B. Allen: Phys. Rev. Lett. 59, 1460 (1987)13. S.D. Brorson, A. Kazeroonian, J.S. Moodera, D.W. Face, T.K.

Cheng, E.P. Ippen, M.S. Dresselhaus, G. Dresselhaus: Phys.Rev. Lett. 64, 2172 (1990)

14. X.Y. Wang, D.M. Riffe, Y.S. Lee, M.C. Downer: Phys. Rev. B 50,8016 (1994)

15. R.W. Schoenlein, W.Z. Lin, J.G. Fujimoto, G.L. Eesley: Phys.Rev. Lett. 58, 1680 (1987)

16. W.S. Fann, R. Storz, H.W.K. Tom, J. Bokor: Phys. Rev. B 46,13592 (1992)

17. Y.V. Afanasiev, O.N. Krokhin, G.V. Sklizkov: IEEE J. QE —2 483 (1966)

18. J.F. Ready: Effects of High-Power ¸aser Radiation (Academic,New York 1971)

19. V.A. Batanov, F.V. Bunkin, A.M. Prokhorov, V.B. Fedorov:Sov. Phys. —JETP 36, 311 (1973)

20. A.M. Prokhorov, V.I. Konov, I. Ursu, I.N. Mihailescu: ¸aserHeating of Metals (Adam Hilger, Bristol 1990)

21. D. Strickland, G. Mourou: Opt. Commun. 56, 219 (1985)22. J.D. Kmetec, J.J. Macklin, J.F. Young: Opt. Lett. 16, 1001 (1991)23. S.I. Anisimov, M.I.Tribel’skii, Ya.G. Epel’baum: Sov. Phys.

—JETP 51, 802 (1980)

.

115