Time-resolved measurements of the response of a STM tip upon illumination with a nanosecond laser pulse

Transcription

1 Appl. Phys. A 66, (998) Applied Physics A Mterils Science & Processing Springer-Verlg 998 Time-resolved mesurements of the response of STM tip upon illumintion with nnosecond lser pulse J. Boneerg, M. Tresp, M. Ochmnn, H.-J. Münzer, P. Leiderer Fkultät für Physik, Universität Konstnz, D Konstnz, Germny (Fx: /8839, E-mil: Johnnes.Boneerg@uni-konstnz.de) Received: 5 August 997/Accepted: 8 Decemer 997 Astrct. Nnosecond lser pulses re used to illuminte the very end of tip of scnning tunneling microscope in front of gold surfce. The trnsient increse of the tunneling current is mesured s function of the pulse energy density, tip retrction mplitude, polriztion of the incident light, nd is voltge. The trnsient signl hs typicl timescle of ms. Thus it cn e concluded tht the therml expnsion of the tip is responsile for this signl. The expnsion hs liner dependence on the incident light intensity with typicl vlue of the order of. Å/(mJ/cm 2 ). This result demnds criticl inspection of the interprettion of nnostructuring experiments with this technique. PACS: 66P; 657; 426K The scnning tunneling microscope nd the tomic force microscope re frequently used for the production of nnostructures on surfces. Wheres the possile use of forces or electric fields is lredy known from different pplictions [] nother quite new ide is the externl injection of lser rdition into the tip sustrte gp. It hs een shown tht y using this technique nnostructures of different forms cn e produced [2 6]. Even relile single-tom deposition is chievle [5]. The physicl resons for these possiilities re not cler. On one hnd it is proposed tht the metllic tip produces locl enhncement of the opticl rdition similr to the well-known surfce-enhnced Rmn effect [2 4, 6]. On the other hnd the possiility of therml expnsion nd therefore of mechnicl contct is lso discussed [5]. In order to clrify this sitution we hve performed STM experiments in comintion with the use of nnosecond lser pulses. Although direct ccess to nnosecond time-resolved mesurements is not possile with our present setup, we used different schemes of indirect mesurements to lern out the interction of the nnosecond lser pulse with the tip of scnning tunneling microscope. From these experiments we conclude tht therml expnsion of the tip cnnot e neglected in experiments with nnosecond lser pulses ut my even e the dominting mechnism involved. Experimentl setup All experiments were performed on thin gold films (5 nm) on mic under mient conditions with home-uilt STM using etched PtIr tips. The response time of the current mplifier system is of the order of 5 µs. The mplifier showed liner response in the mesured current rnge of up to na. The setup llows the feedck mechnism to e switched off for certin time. During this time n dditionl externl voltge cn e pplied to the z piezo in order to retrct the tip in defined mnner. After the tip hs een retrcted the lser pulse illumintes the tip surfce gp. The time-resolved response of the tunneling current is recorded. In order to improve the signl-to-noise rtio the curves shown in the experimentl prt re typiclly verged over 64 events. For the illumintion mildly focused Q-switched Nd:YAG lser pulse (.mm dimeter) ws used. The ngle of incidence ws 75 with respect to the tip xis (Fig. ). Both the fundmentl t wvelength of 64 nm s well s the frequency-douled pulse (532 nm) were pplied with qulittively identicl results. Therefore, in the following, only results otined with λ = 532 nm will e shown. The pulse width ws 7ns (FWHM). Volume sorers llowed us to choose the proper energy density rnge for different experiments, wheres the fine tuning of the energy density ws performed with the help of thin-film polrizers. The lser pulse intensity ws controlled with n energy meter efore ttenution. The pulse-to-pulse energy vrition t the tip STM-tip HWP l s Fig.. Schemtic digrm of the experimentl setup. HWP, hlf-wve plte; l, lens; s, sustrte

2 66 position ws round 5%. For this mesurement -µm pinhole ws mounted in front of PIN photodiode (FND, rise time ns) nd the whole plced t the lter position of the STM tip. The polriztion of the incident lser light with respect to the tip could e djusted with the help of hlf-wve plte. feedck loop on off 2 Results In order to verify the possiility of nnostructuring nd to define the interesting energy density rnge we incresed the pulse energy until nnostructure ppered on the surfce fter pulse. Figure 2 shows the topogrphy of gold surfce fter () illumintion with six, nd () fter the illumintion with seven lser pulses of energy density 4 mj/cm 2. In ddition to the six hillocks lredy pprent in () further hillock with dimeter of round 2 nm nd height of 5nmppered on the left side. This oservtion is similr to erlier studies [2 4] nd will not e discussed further. We focus on dynmic mesurements in the following. In order to e sure tht the tip is not modified during the experiment ll mesurements were performed t densities t lest one order of mgnitude smller thn the threshold for nnostructure formtion. In the following experiments the retrction technique mentioned ove ws used to gin more insight into the ongoing processes. Figure 3 shows the principle of the mesurement. At t =.5ms the feedck loop is switched off E lser [.u.] c d Fig. 3 d. Principle of the timing of retrction mesurement t n energy density of 2 mj/cm 2. Feedck loop, retrction mplitude, c lser pulse, nd d tunneling current Fig. 2,. Surfce topogrphy (3 nm 3 nm) ofnau film fter the ppliction of six nd fter ppliction of seven lser pulses of energy density 4 mj/cm 2. In the upper left corner new nnostructure ppered (Fig. 3). Then, y pplying voltge to the piezo, the tip is retrcted certin distnce, in this cse 5 Å[7]ndfixedthere for the next 3.5ms(Fig. 3). At t = the lser pulse (FWHM 7ns, pulse intensity 2 mj/cm 2 ) hits the surfce (Fig. 3c). Figure 3d depicts the resulting tunneling current. During the retrction the tunneling current drops from the chosen set current of na to zero. Upon illumintion of the tunneling gp t t =, the tunneling current shows steep increse towrds mximum of 2.2nAnd then slow decrese ck to zero on ms timescle. After repproching the tip the current is first slightly higher thn the set current nd relxes then towrds the set current upon rectivtion of the feedck loop. We then performed the experiment for different retrction mplitudes t the sme energy density. Some exmples re shown in Fig. 4,. A cler decrese in the current pek is oserved s the retrction mplitude is incresed. Figure 4c shows the dt of Fig. 4 on logrithmic scle. There, n lmost liner ehvior cn e recognized. The mesurements were repeted for different lser energies. Upon plotting the extrcted mximum current s function of retrction mplitude (Fig. 5) it cn e seen tht the pek current increses with lser pulse intensity. On logrithmic scle the sme dt revel liner ehvior for ll pulse energy densities (Fig. 5). Different pulse energies result in prllel shift of the curves. An dditionl energy input of 3 mj/cm 2 results in n dditionl retrction mplitude of.6 Å.

3 mj/cm mj/cm Fig. 5,. Mximum current I mx s function of the retrction mplitude for different lser intensities. In thedtofre displyed on logrithmic scle c. 2 Fig. 4 c. Trnsient tunneling current s function of the retrction mplitude t n energy density of 2 mj/cm 2 (, ). c displys the dt of on logrithmic scle with liner fit pendence ppered only s the lser focus hit the very end of the STM tip. Upon illuminting the tip some mm ove the tunneling gp, the resulting tunneling current ws considerly smller nd the polriztion dependence disppered. In the figure therml drift cn lso e oserved, the solute mplitude in the mximum is slightly incresing with time. This is explined y the fct tht it took out 3hto get these dt s ech single trce is verged over 64 events. All the mesurements shown here were repeted with other tips of different mterils. Wheres detils of the descried fetures my chnge, the overll fetures remin unchnged. The timescle of the decrese in the tunneling current chnged etween 2ms nd.5ms, nd the retrction mplitude etween.2 nd.5 Å/(mJ/cm 2 ). The mesured exmples showed correltion with the opening ngle of the tip. Nevertheless this ws not studied systemticlly. 3 Discussion Figure 6 shows 9 trces of the time-resolved tunneling current, where we studied the polriztion dependence of the effect. The ngle of the hlf-wve plte ws tuned for 36 in steps of 4 strting prllel to the tip. Thus the polriztion is chnged for 72. A strong polriztion dependence of the effect is oserved, similr to recent experiments [6]. This de- In the following we will show tht ll experimentl results fit in the scheme of therml expnsion of the tip nd the smple. The first striking result is the timescle oserved in the experiment. Wheres the current increse (Fig. 3) cnnot e nlyzed ecuse of the limited response time of the electronics, the current decrese occurs on timescle of ms, which

4 ngle HWP [ ] Fig. 6. Retrction experiment s function of the polriztion of the lser pulse for n energy density of 2 mj/cm 2 is five orders of mgnitude slower thn the lser pulse durtion (7ns). In prticulr the ms timescle is well ove the response time of the current mplifier system (5 µs). Therefore field-induced tunneling current cn e excluded for the explntion of the time response of the tunneling current. In principle the dt of Fig. 4, llow direct estimtion of the therml expnsion of the tip nd smple, which cn e estimted to e some Å for the energy density used. However, more detiled nlysis is possile. For tht purpose, first the consequences of the limited time-resolution must e considered. From Fig. 4c we cn deduce tht the tip retrcts lmost linerly upon cooling, s the current on logrithmic scle is proportionl to z. This mkes it possile to extrpolte from I mx, which is mesured t t = 5 µs, to the end of the lser pulse t t = ns, where the mximum expnsion is expected [5]. Concerning the dt it is resonle to ssume tht the devition from the mximum expnsion t t = 5 µs is elow %. Thus I mx cn e nlyzed further on. The dt of Fig. 5 include dditionl informtion. The liner dependence of z on the log I mx for ech energy density reflects the fct tht this mesurement is in principle nothing other thn modified I(z) mesurement. At the moment when I mx is mesured (t = 5 µs) the distnce to the surfce is the retrction mplitude z minus the expnsion of tip nd surfce. Thus I mx ( z) reflects the exponentil ehviour of the tunneling current. Since this is true for ech energy density it is possile to extrct the solute vlues of expnsion from these mesurements even though the solute tip surfce distnce is not known exctly. From the prllel shift of the different energy densities we get vlue of.2 Å/mJ/cm 2 for the tip/smple comintion used here. Furthermore the lines re roughly equidistnt, which llows us to conclude tht the sorption nd expnsion re liner in this energy density regime. This finding grees with theoreticl considertions y Prk et l. [9]. Consequently the expnsion cn now e deduced for ech energy density. With this vlue we cn try to get n estimtion of the temperture increse during the pulse for the mximum energy density (7 mj/cm 2 ) used in this experiment. For first pproximtion it cn e ssumed tht the therml expnsion of the sustrte is smll compred to tht of the tip [5]. The het diffusion length in pltinum is of the order of l diff =.7 µm on the timescle of the lser pulse. As lredy discussed y Ukrintsev et l. [5] not the whole illuminted re does contriute to the therml expnsion of the tip. For tip with n opening ngle of α = 3 we get n effective contriution length in the order of d eff = (l diff /2)/(tn α/2) =.3 µm. Tking n expnsion coefficient of for pltinum the expnsion of.4 Å results in n verge temperture increse of 2 K for this tip, which is surprisingly smll. For other tips we got higher vlues of expnsion. There the resulting tempertures would e fctor of 25 higher. For such smll temperture increse neither detectle chnge of the tunneling possiility (due to chnge of the Fermi function) nor detectle thermovoltge should e expected. This is confirmed y the symmetricl dependence I mx on the pplied tunneling voltge for U <.3V (not shown). The polriztion dependence of I mx cn e explined y the polriztion dependence of the sored energy. The very end of the tip cn e ssumed to e n ellipsoid nd then the sorption depends on polriztion [9, ]. Consequently the polriztion dependence should dispper lmost completely s the shft of the tip is illuminted, s oserved in the experiment. 4 Conclusion Our experiments hve shown tht even t energy densities one order of mgnitude elow the threshold for the ppernce of nnostructures, the therml expnsion of the tip cnnot e neglected. The tip is oserved to expnd linerly with the illumintion energy density nd retrcts on timescle of ms [2]. For the energy densities used in the nnostructuring experiments here s well s in the literture we estimte therml expnsion in the order of nnometers. This could imply tht mechnicl contct etween tip nd surfce is estlished upon the illumintion. Indeed our first mesurements of the gp voltge indicte quntized contct resistnce []. There-

5 69 fore we hve to question if the field enhncement t the tip is the dominting effect for the occurrence of nnostructures upon illumintion with nnosecond pulses. Hence we strted to use fs pulses where the scenrio could e expected to chnge. References. P. Avouris (Ed.): Atomic nd Nnometer-Scle Modifiction of Mterils: Fundmentls nd Applictions, NATO ASI Series e, Vol. 239 (Kluwer, Dordrecht 993) 2. A.A. Gorunov, W. Pompe: Phys. Sttus Solidi A 45, 333 (994) 3. K. Dickmnn, J. Jersch: Lser Optoelektronik 27, 76 (995) 4. J. Jersch, K. Dickmnn: Appl. Phys. Lett. 68, 868 (996) 5. V.A. Ukrintsev, J.T. Ytes Jr.: J. Appl. Phys. 8, 256 (996) 6. J. Jersch, F. Demming, K. Dickmnn: Appl. Phys. A 64, 29 (997) 7. The z-piezo ws clirted using montomic steps on the Au surfce 8. S.H. Prk, N.M. Miskovsky, P.H. Cutler, E. Kzes, T.E. Sullivn: Surf. Sci. 266, 265 (992) 9. H.C. vn de Hulst: Light Scttering y Smll Prticles (Dover, New York 957). C.F. Bohren, D.R. Huffmnn: Asorption nd scttering of light y smll prticles (Wiley, New York 983). J. Boneerg, M. Tresp, M. Ochmnn, H.-J. Münzer, P. Leiderer: to e pulished in Appl. Phys. A 2. Note dded in proof: The sme conclusions were recently given y I. Lyuinetsky, Z. Dohnálek, V.A. Ukrintsev, J.T. Ytes, Jr.: J. Appl. Phys. 82, 45 (997)