Table Of Content

Evidence of vectorial photoelectric effect on Copper E. Pedersoli,1 F. Banfi,1 B. Ressel,2 S. Pagliara,1 C. Giannetti,1 G. Galimberti,1 S. Lidia,3 J. Corlett,3 G. Ferrini,1 and F. Parmigiani1 1Dipartimento di Matematica e Fisica, Universita` Cattolica, Via dei Musei 41, 25121 Brescia, Italy 2Sincrotrone Trieste S.p.A., Strada Statale 14, km 163.5, 34012 Basovizza (TS), Italy 3Lawrence Berkeley National Laboratory, One Cyclotron Road, Berkeley, CA 94720 USA Quantum Efficiency (QE) measurements of single photon photoemission from a Cu(111) single crystal and a Cu polycrystal photocathodes, irradiated by 150 fs-6.28 eV laser pulses, are reported 2 overabroadrangeofincidenceangle,bothinsandppolarizations. ThemaximumQE(≃4×10−4) 1 for polycrystalline Cu is obtained in p polarization at an angle of incidence θ = 65◦. We observe 0 a QE enhancement in p polarization which can not be explained in terms of optical absorption, 2 a phenomenon known as vectorial photoelectric effect. Issues concerning surface roughness and n symmetry considerations are addressed. An explanation in terms of non local conductivity tensor a is proposed. J 4 PACSnumbers: 79.60.Bm,61.80.Ba,41.85.Ar 1 Keywords: Vectorialphotoelectriceffect, femtosecondphotoemission,copper photocathodes. ] ci The advent of the 4th generation free electron lasers -60 -40 -20 0 20 40 60 80 s (FEL)sources[1–3]triggeredseveralimportanttechnical Polycrystal - 4.0 l questions. A fundamental issue regards the photocath- r t odematerialforthelaser-drivenphotoinjectordevices,to m 3.0 obtain short electron bunches with high charge density ) . and low emittance. Metal photocathodes are good can- -40 t 12.0 a x didates, havinga high reliability,long lifetime anda fast ( m Y time response (1-10 fs). However, two major drawbacks y 1.0 - limit their usefulness,the smallquantumefficiency (QE) c d n n andthe high workfunction (Φ), requiring lightsourcein e0.0 i o the ultraviolet (UV) for efficient linear photoemission. fic4.0 Single crystal p-pol. Fresnel c f In this Letter we study the experimentalconditions to E [ maximize the QE of Cu photocathodes using UV short m 3.0 p-pol. 1 laser pulses from the quadrupled output of an ampli- u s-pol. 6v fied Ti:Sapphire laser. The QE for linear photoemis- ant2.0 u 4 sion in the femtosecond regime is measured as a func- Q 0 tion of the angle of incidence θ in the angular range 1.0 3 −55◦ ≤ θ ≤ +80◦, both in s and p polarizations. The 1. maximum quantum efficiency Y ≃ 4× 10−4, obtained 0.0 0 with p polarizationat θ =65◦, is four times the value at -60 -40 -20 0 20 q40 60 80 Angle of incidence (deg) 2 normal incidence. 1 The QE dependence on angle of incidence and light : v polarizationisalongstandingproblem[4–8]thatlargely FIG. 1: Measurements of quantum efficiency dependence on i the angle of incidence θ for a Cu polycrystal and a Cu(111) X remains to be understood. Our data are well fitted by single crystal for p (circles) and s (triangles) polarized light. r a phenomenological model [6] that keeps into account Fits, based on Eq. 1, are reported as solid lines. The dashed a only light absorption, without any explanation in terms lines are calculated taking into account Fresnel absorption of microscopic quantum physics. A justification of the only. phenomenologicalmodelbasedonthecalculationsofthe conductivity tensor for a jellium model is proposed. The photoemission from a polycrystalline Cu sample crystals output with a MgF2 prism, with minimal tem- and a Cu(111) single crystal is investigated with 150 fs poral and pulse front tilt distortions. laserpulseswithaphotonenergyof6.28eV,obtainedby We do not use a more efficient third harmonic con- twosuccessivedoublingprocessesoftheTi:Sapphirefun- version to obtain linear photoemission from Cu (3hν = damental frequency (hν = 1.57 eV) in β-barium-borate 4.71 eV, Φ=4.65 eV for polycrystalline Cu [9]) because (BBO)crystals. Theseconddoublingprocessisobtained of the onset of multiphoton regime upon a work func- outphase-matchinginathin(200µm)BBOcrystal. The tion increase due to sample contamination. Moreover, fourth harmonic is selected by dispersing the doubling an effective laser-induced oxide removal and contami- 2 nants chemical-bond breaking obtained with UV short QE, normalized with respect to its value at normal inci- laser pulses [10, 11] improves with shorter wavelengths dence Y(0), is: [10]. Working with a 6.28 eV photon energy should thus Y(θ) ε (θ) ε (θ) help to increase the duty time of machines based on Cu = k +r ⊥ , (1) photocathodes. Y(0) εk(0) εk(0) The quantum efficiency Y is the ratio between the whereε =ε andε =ε +ε aretheelectromagnetic ⊥ p⊥ k pk s number of photoemitted electrons, obtained from the energies inside the sample due to the fields components photocurrent measuredfrom the sample with a Keithley indicated by the suffixes. A value r = 1 means that 6485 Picoammeter, and the number of incident photons, photoemission is proportional to the absorbed intensity, detectedmeasuringonaTektronixTDS3054Bdigitalos- whereas r > 1 implies that E is more efficient than ⊥ cilloscope the output ofa HamamatsuR928 photomulti- E in producing photoelectrons. Eq. 1 specialized for p k pliertube. ThemeasurementsareperformedwiththeCu polarization is: photocathodeskeptinaultrahighvacuumchamberwith a base pressure of 2×10−10 mbar at room temperature. Yp(θ) = εpk(θ) +rεp⊥(θ), (2) During the total yield measurements, the photoemission Y (0) ε (0) ε (0) p pk pk spectrum is acquired using a time of flight spectrometer whereas for s polarization (E =0): in order to measure the sample work function and mon- ⊥ itor possible onset of sample contaminations and space Y (θ) ε (θ) s s chargeeffects. The samples arecleaned by cycles ofAr+ = . (3) Y (0) ε (0) sputtering followed by annealing at 500◦C. This proce- s s dure is continued until the proper value of the measured Once the electromagnetic energies ε (θ), ε (θ) and pk p⊥ work function (4.65 eV for the polycrystal and 4.94 eV ε (θ) are calculated from classical electrodynamics, as- s for the single crystal) is obtained. In these conditions a suming volume absorptionas in Refs. 6, 13, the parame- clear low energy electron diffraction (LEED) pattern for ter r is obtained fitting the experimental data for p po- theCu(111)sampleisobtained. Thelaserpeakintensity larization with Eq. 2. The best fit values are r = 13 for on the target is I ≃5×105 W/cm2. the polycrystalline Cu and r = 9 for the Cu(111) single The QE measured for both samples are reported in crystal(seeFig.1). TheQEdependenceexpectedonthe Fig.1. AnenhancementoftheQEisevidentforppolar- basis of Fresnel laws only, setting r =1, is also reported ization as compared to what would be expected taking as a dashed line in Fig. 1. The data for s polarization intoaccountonlytheelectromagneticabsorptionprocess. are in agreement with Eq. 3. The maximum QE do not occur at the pseudo-Brewster At the light of our data, it is important to investigate angle of incidence θ =57◦ [12](see Fig. 1), where there thephysicalmechanismsthatenhancesthephotoelectron B is maximum absorption, but is shifted by ∼ 8◦ toward yield due to E⊥ over Ek. the normal. The crystalline symmetry, important when dealing with polarization dependent photoemission, play no role Our experimental results can be rationalized in the in the present experiment. The photoemission process frame of a phenomenological model proposed in Ref. 6. dueto E isabout10times moreeffective thanE both The electric field transmitted inside the sample can be ⊥ k written as E = E +E = E +E , where E and E in the Cu(111) single crystal, where symmetry consider- p s k ⊥ p s ations could apply, and in the polycrystalline Cu, where are the p and s polarized field components respectively, E =E +E and E =E are the components par- any symmetry-related contribution is cancelled by the k pk s ⊥ p⊥ random orientations of the single crystals domains com- allel and perpendicular to the surface respectively. The posing the sample. electricfieldvectorcomponentsaredefinedinFig.2. The Photoemission enhancement due to surface roughness has been recently investigated [14–16]. In the present case surface roughness enhancement can be ruled out. Several atomic force microscopy (AFM) scans of the samples surface, with sizes ranging from 1×1 µm2 to 60 × 60 µm2, give values of the root mean squared roughness h ∼ 20 nm for the Cu polycrystal and rms h ∼ 2 nm for the Cu(111) single crystal, see Fig. 3. rms Theobservedvectorialphotoelectriceffectiscomparable on both samples, despite their surface roughnesses differ FIG. 2: Representation of incidence angle θ, wave vectors k byanorderofmagnitude. The comparativestudy ofthe andktforincidentandtransmittedlightandfieldcomponents single crystalCu and polycrystalline Cu cathodes allows addressedinthetext. Arealindexofrefractionnisassumed toclarifythatourexperimentisnotdependentonsample for the present figure. morphology. 3 at an incidence angle of θ =65◦. Investigationof both a Cu(111) single crystal and a Cu polycrystal allows us to ruleoutamicroscopicprocessesbasedonsymmetrycon- siderations and surface roughness to explain our data. An explanation in terms of a rapidly varying effective field, due to the non local character of the conductivity tensor, is suggested. This work was supported by the U.S. Department of Energy,OfficeofScience,underContractNo. DE-AC03- 76SF00098. FIG.3: AtomicForceMicroscopy imagesofthetwosamples’ surfaces. Measured route mean squared roughness is 20 nm fortheCupolycrystaland2nmfortheCu(111)singlecrystal. [1] R. W. Schoenlein, S. Chattopadhyay, H. H. W. Chong, Therefore, we seek for an explanation in terms of a T. E. Glover, P. A. Heimann, C. V. Shank, A. A. Zho- moregeneralmechanism. SolutionsoftheMaxwellequa- lents, and M. S. Zolotorev, Science 287, 5461 (2000). tionsonanidealjellium-vacuuminterfaceforanimping- [2] R.Neutze,R.Wouts,D.vanderSpoel, E.Weckert,and ing plane electromagnetic wave of frequency ω, evidence J. Hajdu, Nature406, 752 (2000). an electromagnetic field spatially varying on the length [3] H. C. Kapteynand T. Ditmire, Nature429, 467 (2002). scale of ∼ 1 ˚A on the jellium side [17]. The spatially [4] D. W. Juenker, J. P. Waldron, and R. J. Jaccodine, J. Opt. Soc. Am.54, 216 (1964). varyingelectromagneticfieldisduetothenonlocalchar- [5] R. M. Broudy,Phys. Rev.B 1, 3430 (1970). acterof the conductivity tensor. This is calculatedusing [6] R. M. Broudy,Phys. Rev.B 3, 3641 (1971). free-electron like wave functions, so it does not depend [7] J.P.Girardeau-Montaut,S.D.Girardeau-Montaut,S.D. on the symmetry of the crystal. The matrix element Moustaizis, and C. Fotakis, Appl. Phys. Lett. 63, 699 entering the differential cross-section for photoemission (1993). is composed of two terms. The first is the usual electric [8] T.Srinivasan-Rao,J.Fischer,andT.Tsang,Appl.Phys. dipolecontribution,thesecondisduetotherapidlyvary- Lett. 63, 1596 (1993). [9] R.C. Weastand M.J. AstleHand,Hand Book of Chem- ing electric field. The second term prevails for ω < ω , p istry and Physics (CRC Press, Boca Raton, Florida, where ω is the plasma frequency, and leads to an en- p 1982). hancementofthe photocurrentforthe electricfieldcom- [10] M. Afif, J. P. Girardeau-Montaut, C. Tomas, M. Ro- ponents perpendicular to the sample surface [18, 19]. In mand, M. Charbonnier, N. S. Prakash, A. Perez, the present experiment, ~ω = 6.28 eV and ~ω ∼ 19 eV G. Marest, and J. M. Frigerio, App. Surf. Sci 96, 469 p [20]. This mechanism explains an enhancement of the (1996). QE for p polarized incident radiation while not affecting [11] C. Beleznai, D. Vouagner, J. Girardeau-Montaut, and L. Nanai, App.Surf.Sci 138, 512 (1999). theresultsforspolarizedlight. Furthermore,itdoesnot [12] An index of refraction n = 0.98+1.49i [? ] is assumed dependonsurfaceroughnessoraparticularsymmetryof for Cu. the crystal. We therefore propose it as the main micro- [13] H. Y.Fan, Phys. Rev.68, 43 (1945). scopicmechanismtoexplainourexperimentalevidences. [14] G. Banfi, G. Ferrini, M. Peloi, and F. Parmigiani, Phys. In this Letter quantum efficiency measurements on Rev. B 67, 035428 (2003). Cu photocathodes, irradiated with 150 fs laser pulses at [15] J. Zawadzka, D. J. ad J.J. Carey, and K. Wynne, Appl. 6.28 eV, are reported over a broad range of incident an- Phys. Lett. 79, 2130 (2001). [16] M. Kupersztych and P. Monchicourt, Phys. Rev. Lett. gles in both s and p polarizations. A QE enhancement 86, 5180 (2001). is found for light with electric field perpendicular to the [17] P. J. Feibelman, Phys. Rev.B 12, 1319 (1975). sample’ssurface,showingavectorialphotoelectriceffect. [18] P. J. Feibelman, Phys. Rev.Lett. 34, 1092 (1975). The maximumvalue ofquantumefficiency Y ≃4×10−4 [19] H. J. Levinson, E. W. Plummer, and P. J. Feibelman, isfourtimesbiggerthantheQEatnormalincidenceand Phys. Rev.Lett. 43, 952 (1979). isachievedwithppolarizedlightimpingingonthesample [20] D.L.MisellandA.J.Atkins,Philos.Mag.27,95(1973). Copyright (2005) American Institute of Physics. This article may be downloaded for personal use only. Any other use requires prior permission of the author and the American Institute of Physics. The article appeared in Appl. Phys. Lett. 87, 081112 (2005); doi:10.1063/1.2031949and may be found at http://link.aip.org/link/?apl/87/081112.