Electron Gun Chuanxiang Tang ! at OCPA Accelerator School (Basic), Xiuning, Anhui Province, PR China July 28, 2014 Outline • Introduction: what’s an e-gun? who needs an egun? what’s important for an e-gun? • Concept of emittance • Cathode physics and technology: photocathode • Electron guns: photocathode RF Gun • Summary What’s an electron gun? A device to generate electron? An electric component in some vacuum tubes which produces narrow, collimated electron beams with a precise kinetic energy. 1880 Classification Electric field: DC and RF Cathode: thermionic, photocathode, field emission, explosive emission, plasma source, secondary electron emission… Number of electrodes: diode, triode,… 1895 Electron Discovery: in 1897, by J J Thomson Composition: elementary particle Statistics: Fermionic Interactions: Gravity, Electromagnetic, Weak Mass: 9.109 382 91(40)x10-31kg 0.510 998 928(11)MeV/c2 Electric charge: -1.602 176 565(35)x10-19C Magnetic moment: -1.001 159 652 180 76(27)uB Spin 1/2 What can we do with electrons? as an energy carrier to generate other new particles, such as Higgs, by colliding with its antiparticle positron. —high energy physics. to generate different electromagnetic radiation (light), such as bremsstrahlung, FEL, SR, ICS and magnetron, klystron… to change properties of materials.—Irradiation. as a probe to see atomic scale structures or changes, such as SEM, UED, UEM.—material, solid state physics, IC and IT, biology… as a bullet to kill cancer or bacteria.—radiotherapy and irradiation. Electron Guns in Accelerators • Electron-Positron Collider: nanosecond DC gun, polarized electron beam (GaAs). • Synchrotron Light Source: DC or RF, thermionic or photocathode. • Free Electron Laser: photocathode or thermionic, RF or DC. • Low energy linacs or other application electron accelerators : DC, thermionic, diode or triode. • HPM: DC and plasma, explosive,or thermionic, photocathode? • Vacuum tubes: thermionic, dc Gun 200 kV High Voltage DC and RF Guns cathode and heator DC Gun RF resonance cavity and TM010 mode anode RF Gun DC Type Guns DC – always on current DC - pulsed DC – amplitude, frequency modulated time RF Type Guns CW – bunch of e- in every RF bucket, typically from 100’s of MHz to GHz, up to 100’s of pC per bunch pulsed – not every RF bucket is filled, RF frequencies of 100’s of MHz to GHz, up to ~nC per bunch, with bunch rep rates of Hz to 1 MHz What kind of electron gun do we want? • High average current—high repetition rate, high duty factor, high bunch charge, high peak current. • High peak current—high bunch charge, high current density at cathode, high QE, high gradient at cathode, large emission area. • Low emittance—high gradient at cathode, high current density, low temperature, good cathode quality, low QE, small emission area, suitable transversal and longitudinal distribution. • High reliability, small size, low cost… Why is low emittance important? Gain Length Wavelength Requirement 5/6 n 1/ 2 1/ 2 [ µm]g [cm] ( Lg ) min [m] ≅ 20 2/3 ! I [kA]λr [ A] ε 3 ε ! (λr ) min [ A] ≅ 3 × 10 λS ε< 4π E.L.Saldin et al , Optics Communications 235 5/ 4 n 3/ 4 [ µm]g [cm] 3/ 2 I [kA]LW [m] εn=βγε$ 2004 3/ 4 415-420 L∝ η RF PRF E 3/ 2 cm δ BS γ σ z η RF PRF ∝ ε n, y β y Ecm δ BS σ z ε n, y β y Some Basic Concepts of Emittance Concept: Emittance is used to describe the super-volume in the 6-dimentional phase space, as coordinates (q1,q2,q3) and momenta(p1,p2,p3) in Cartesian coordinate systems. If the particles inside any closed surface s(t) in the phase space， surfer only conservative forces, the volume v(t) enclosed by the surface s(t) will be invariant with time. Emittance is the volume v(t) occupied by the particles of a bunch. ! Without coupling between q1,q2,q3 Phase space: (q1, p1), (q2, p2), (q3, p3) Emittance is a area of the phase space RMS Emittance RMS-root of mean square RMS Normal Emittance ε n = π ⋅ me c ⋅ x 2 px 2 − x ⋅ px 2 Here mec means that px=βxγ. In most case, εn is expressed in π .mec .µm RMS Geometrical Emittance ε =π⋅ x 2 x$ 2 − x ⋅ x$ 2 x!=px/pz=βx/βz is the divergence angle, and ε is expressed in π .mm .mrad Other expressions of RMS Emittance ε n = 4 ⋅ π ⋅ me c ⋅ ε = 4 ⋅π ⋅ x 2 x 2 x$ px 2 2 − x ⋅ px − x ⋅ x$ 2 2 ε n = βγ ⋅ ε Assume all the particles have same Pz A gaussian phase-space distribution may be specified as: px with 〈 xp x 〉 = 0 To simplify the analysis, let and rms normal emittance is ε n = π ⋅ me c ⋅ ψ ( x, p x ) = 1 2π x2 px 2 x x2 px 2 1 x2 exp[ − ( 2 + )] 2 2 x px px 2 And consider an ellipse in (x, px) space defined by: The area of this ellipse is: A( K ) = π ⋅ me c ⋅ K 2 ⋅ x 2 x2 + px 2 px 2 x 2 px 2 The fraction of the bunch within this ellipse is readily computed: px For K=l, the ellipse has an area equal to the normalized RMS emittance, and contains 39.35% of the particles. The maximum x coordinate of the ellipse is the RMS value of x, while the maximum px coordinate of the ellipse is the RMS value of px. For K=2, the ellipse has an area equal to four times the normalized RMS emittance, and contains 86.47% of the particles. *M.Borland ,A High-Brightness Thermionic Microwave Electron Gun , SLAC-402(1991) ,Ph.D.Thesis = K2 Statistical Definition of Beam Emittance -From P. Lapostolle, IEEE Trans. Nucl. Sci NS-18, 1101(1971) J.Buon, CERN 91-04, 30(1991) N non-interacting particles in phase space Define distribution function ρ ( x, x") 1 xi = ∫ x ρ ( x, x#)dxdx# ∑ N 1 # x = ∑ xi# = ∫ x#ρ ( x, x#)dxdx# N 1 σ x2 = ∑ ( xi − x ) 2 N 1 σ x#2 = ∑ ( xi# − x# ) 2 N 1 σ xx# = ∑ ( xi − x )( xi# − x# ) = rσ xσ x# N x = ( x, x!) ∫ ρ ( x, x")dxdx" = 1 ε rms = σ x 2σ x#2 − σ xx#2 = σ xσ x# 1 − r 2 Assuming that particles are Uniformly distributed in an ellipse: x 2 x!2 + 2 =1 2 a b Total phase-space area: A = π ab = 4πε rms Full emittance : ε = 4ε rms A = πε RMS Geometrical & Normal Emittance dx dx dz p x = mvx = m = m ⋅ = mvz ⋅ x! = p z ⋅ x! dt dz dt The transverse momentum px & the divergence angle x! x The RMS geometrical emittance will decrease with the particales being accelerated, but the RMS normal emittance will remain the same. dx x# = = tan(θ ) ≈ θ dz px v θ" dz z x# 10MeV x x 40MeV x z x# x# x# x x x# x# x x x dx Projected Emittance y x x Real Space py px y x y Phase Space: Projected Emittance Slice Emittance and Correlated Emittance Slice Emittance: A transverse cross section of the beam is called a slice, and the emittance of a slice is called slice emittance. Slice emittance consists of two parts, one part is the thermal emittance and the other part is due to nonlinear space charge force. y x Correlated Emittance: The growth of projected emittance is due in large part to the correlation between the phase space angle and the longitudinal position of slices. Normally, this part of projected emittance is called correlated emittance. px px px x x Slice Emittance px x x Correlated Emittance Electron emitted from cathode w T=TK T=0K w0 U U xc E=0 w0! wm T=0K E=E0 x dN/dw E=0 V0! Vm E=E0 x dN/dw • Field emission: • Thermionic emission: Flower -Nordheim’s Eqn. Richardson’s Eqn. 2 J = AT e − w0 − wm kT J = CE 2 e J # = Je e kT eE 4πε 0 1952 − D E (A/cm2 ) 6.2 ×10 −6 Vm 1/ 2 C= ( ) V0 φ Schottky effect: * xc V0 D = 6.8 ×10 7 φ 3 / 2 35 103 ron e erial, ed. tric (image source: Masao Kuriki, ILC school)! 14 Photo-electron emission: photo-current is proportional to photon number (if photon energy not changed) The maximum kinetic energy of the photo-electron is proportional to photon energy QE: Quantum Efficiency ! Metal: Mg 0.4%, Cu 0.05%, at 266nm ! Semiconductor: alkali-based Na2KSb:Cs, K2CsSb, Cs2Te 266nm 6%~12%, 251nm 16% GaAs 2.55eV(486nm) 14% GaAs:Cs 2.3eV 0.26% δ Secondary emission factor δmax 1 0 E1 Emax E2 Secondary emission: When the primary electron hits the cathode, atoms will be shocked strongly. Some electrons can cross the potential well and become secondary electrons. δ r ( E0 ) = −B ε min( R ( E0 ), d ) ∫ 0 dE ( x) −α x e dx dx Ep Schottky effect and the abrupt change in electron the Fermiangle energy in Eq. (1) should be replaced by the !w % !Schottky ¼ !w % eff & FERMIII. QUANTUM AND ! THE chemical potential or Fermi level. However,EFFICIENCY at low temacross the metal-vacuum interface. peratures areFOR METALS qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ DIRAC potential MODEL Previously we developed analytic expressions forthetheFermi energy and chemical nearly and at absolute zero theyDirac are identical. In ¼ !w temperature % 0:037 947 Fa ðM function at finite he Pauli exclusion the based lackidentical of the unoccupied quantum efficiencyprinciple and thermal and emittance on this paper we only consider low fermions, temperatures and thus the Being electrons uniformly fill all energy states three-step ignoringas electron-electron scattering [4]. energy is the thethan image fielt l states, alsomodel known Pauli blocking. As a upresult, use of the Fermi is Incident photons withcharge energies above toappropriate. the Fermi level, EF , sum with noof more two opposing work [5] expanded the quantum efficiency model The electron density of occupied thestate. cath- Therefore function electron-electron are absorbed by electrons near Phys. Rev. ST Accel. Beams 12, 074201 (2009) spin electronsstates perFinside energy ctronsFurther move freely through the metal with long mean. The Schottky work !e ode and the electric potentials experienceda by a single which then migrate function, to the surface and to include electron-electron scattering, resulting in excelscattering theareFermi level suppressed duein terms of th electron immediately outside thebelow cathode plotted in is strongly The QE can be expressed lent agreement with the measured quantum efficiency of an e paths which contributes to the good electrical conofforthe Schottky potential is th Fig. 2. The occupation number is shown a Fermithese steps to occur whereand it is assumed Dirac function at finite temperature. The Schottky potential atomically clean copper surface. Here we apply the same tivity ofThe metals. The distribution ofthe occupied states as a emission potential barrier typicall is the sum of the imageof charge and field, Fermi-Dirac distribution with thefield three-step model of phodistribution occupied states asapplied a function R1 R1 Fa .toemission The Schottky work function, ! , is the peak value to give a consistent theory of the emittance and ction of energy is given by the Fermi-Dirac function, dE½1 % f ðE þ @!Þ)f ðEÞ dðcos#ÞF Schottky meters outside the cathode. The ze# FD FD e-e ðE; !; EF þ!eff %@! of energy is given by the Fermi-Dirac function: cos# max ðEÞ R R R QE ð!Þ ¼ ½1 % Rð!Þ) 1 [6] of generalizes of the the quantum Schottkyefficiency. potentialOther and recent is thework height the photodE½1 % fFD ðE þ @!Þ)fFD ðEÞ 1%1 dðcos#Þ 2" %@! E 0 d! F shifted downward at high electric this basic phenomenological process to give aa common 1 located emission potential barrier typically few nanoemission theory for thermionic, photoelectric, fFDtheðEÞ ¼ ;and field (1) work function reducing the barr meters outside cathode. The zero field vacuum state is ðE%E Þ=k T F B sorption length, $opt , and the electron-e Here Rð!Þ is the cathode optical reflectivity. The Fermi1 þ one the connection emission. Here we concentrate between shifted downward at high electric field byDirac thefunction, Schottky quantum effective path, $eThe states, and yield. systemwor is sho -e . The coordinate quantum efficiency and the photoelectric thermalfFD ðEÞ, is the density of initial % fFD ðE þ @!Þ] Unlike a semiconductor photocath is the density of final states with the reducing the and [1increasing the T function is the electron gasbarrier thermal energy, E is the ere work kBemittance. fined as F electron-phonon scattering can be igno product of these two functions giving the transition probaquantum yield. The effective work function, ! , is deeff The effective work function is defined as The derivation forthe the quantum efficiencyTo (QE)be and the electron scattering dominating. Since th bility for the excitedprecise, electron to escape. Fe-e ðE; !; #Þ is the mi energy, and E is state energy. more fined as emittance begins with the electron gas probability thermal theorythe forexcited electron reaches the surface without less than twice the work function, any scattering. This Fermi energy in Eq. (1)model should replaced byis determined the by the photon ab- scattering event eliminates both electro metals and Spicer’s three-step for photoemission as function sbe ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ !eff & !w % !Schottky ¼ ! eFa [5], at in Fig. 1.Fermi In a previous publication we low temmicalillustrated potential or level. However, !eff & !Schottky !w % eusing the work w %using calculated the!QE this ¼ approach 4""0 aturesfunction the Fermi energy and chemical potential are1. (Color) Three-step modelApplied as the only free parameter. In this model, the FIG. of photoemission. qﬃﬃ Field Potential qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ ﬃ 3-step model of photoemission 1.5 0.5 þ @!Þ)fFD ðEÞ 1 vacuum 0.5 w Voltage (volts) Energy relative to Fermi Energy (eV) Energy (arb.) rly identical and they eV: are identical. In Potential ¼ !wat % absolute 0:037 947 Fzero (2) ¼ !w Image % Charge 0:037 947 F a ðMV=mÞ 074201-1 ! 2009 The American Physical Society paper1098-4402=09=12(7)=074201(10) we only consider low temperatures and thusE the Incident photons above the effective work Fa iswith the energies applied field. Schottky Work Material of function the Fermi energy is appropriate. Incident photons with energies ab Function, φ are absorbed by electrons near the Fermi energy Work Function, φ Effective then migrate the occupied surface and escape. The which electron densityto of states inside the cathfunction are absorbed by electrons E Work Function, The Schottky potential is shown forprobabilities an applied for The QE can be expressed in terms of the φ = φ -to φ and the electric potentials experienced by a single which then migrate the surface field of 100 MV=m where the peak of the these steps to occur where it is assumed @! ' EF , ctron immediately thenmcathode are plotted in The QE can be expressed in term photoemission outside barrier is 1.9 from the surface F-D Distribution Electron Density of States Distance from Cathode (nm) . 2. The occupation number is shown forTheaenergy Fermithese steps to occur where it is ass R R FIG. 2. (Color) distribution of occupied states for a metal (left) and the electric potentials next to 0 Schottky − φ work 0.5 F0 occupied states eff 1 0.5 Electron Density of States 0 2 w Schottky 4 6 Distance from Cathode (nm) 8 10 1 2" d! dðcos#ÞF ðE; !; #Þ e e 0(right). The Schottky potential (green curve) is the sum of the applied field potential (blue) and the image charge cos#max ðEÞ R1 R2" (3) is the amount the material work function is reduced at the peak of the Schottky pote Schottky :work function fDavid ðE þ @!Þ)f ðEÞ dðcos#Þ d! is shown for an applied field of- ACCELERATORS 100 MV=m where the peak of the photoemission barrier is(2009) 1.9 nm fr FD %1F. Schmerge0, PHYSICALpotential H. DowellFDand John REVIEW SPECIAL TOPICS AND BEAMS 12, 074201 16) collimated by the surface is intriguing [12] and its impact on beam dynamics will be investigated in our future studies. sion is found to be In addition we have assumed the temp quation (9) can be integrated to obtain ndsreplacing ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ ﬃ Fermi-Dirac distrib the allow the ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ s B. Derivation of the photoelectric emittance for metals ! #2 1$ Rð!Þ inðE @! # function. "eff The rms emittance terms the moments ofEthe F þof@!Þ F þ "eff Heaviside-step The electron tem 1 $ : QE ð!Þ ¼ !pfined ; (3 ¼ as the temperature electron distribution is defined as [13] !opt ð!Þ 17) x 2 of the electro 2@! E þ @! F 1 þ !! ð!Þqﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ 3mc e"-e' hx2 ihx02 i # hxx0 i2 (22) absorption of a photon. Numerical integr the x (10) for the parameters in Table with the slope of the electron trajectory given by and therefore the normalized emittance is I the electron t be approximately 2000 K to increase the The 18) QE can be expanded as a function of @! $ ’eff in s ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ ﬃ over 7000 K to increase the QE by a factor e following Taylor series which is evaluated at photo@! 300 # "Keffwhere most ph ode temperatures near mission threshold where @! is only slightly larger than x ally "n ¼ !operated, ! px ¼ ! :QE is an app (3 x x 2 = p sin θ cos Φ p the effect on the DAVID H. DOWELL AND JOHN F.3mc SCHMERGE ff : rreincrease. The temperature has little effect $ Φ $ ω = + ) p 2m(E $ dQE $ kB T=ðh! $ "emittance unity. Fe θ $ z eff Þ approaches Eð!Þ ¼ QEj@!¼"eff $ ð@! $ "6effplots Þ $ Figure the photoelectric for copper as $ $ ene d@! @!¼"eff We have also assumed a monoenergetic function of photon energy for applied electric fields of $ du $ y the derivation of the QE. The photons ca 1 dQE $ $ 2 ver $ the parameters given in Table ð@! $ "50, Þ and $ & &100 & : MV=m (11) using þ $ eff monoenergetic unless the photon energy $ con 2 d@! $@!¼"eff The emittance approaches 0 asto the comparable the photon electron energy temperatur Dia he first twometal terms vacuum are zero verifying the well-known 255 work nm photon and transform dec proaches the effective function and thelimited reduc and 2 great adratic upon the system photon energy for the which lengthwe must be <13 fsthe to produce FIG. dependence 5. (Color) Definition of the coordinate and compobeam brightness, define as QE=! , al px int mennents of the momentum just inside the cathode surface used to antumderive efficiency near threshold result [10]: energy spread. Thus, the energy spread o the photoelectric emittance. shi be ignored even for pulses with 100 fs ris 2 abo 1 $ Rð!Þ ð@! $ "eff Þ 074201-5 QE ð!Þ ' : (12) em !opt ð!Þ 8" ðE þ " Þ eff F eff 1 þ !! ð!Þ e-e III. THE PHOTOELECTRIC EMIT the METAL CATHODES abs As described earlier, multiple assumptions are required D. Dowell, this et al analytic , Nucl. Instrum. Methods Phys. Rwe es. Aalso 622, FIG. 685， 2010. The quantum efficiency and the normalized firs 7. (Color) produce result. Therefore numeriThis section derives the thermal emit sio emittance per rms beam size plotted asAND functions of the effective H. Dowellthe andabove John F. Schmerge , PHYSICAL SPECIALof TOPICS ACCELERATORS BEAMS 12, 074201 (2009) llyDavid integrate equations to test REVIEW the accuracy electric emission by applying the same Fe Quantum efficiency and thermal emittance of metal photocathodes x total total in Figure 3. Appl. Phys. Lett. 101, 231103 (2012) emittance. Here, the measured QEs are analyzed to obtain We have shown that a QE ¼ 4.5 & 10"4 can be obtained the effective work functions for (100) copper and copper with a CsBr coated copper cathode at 257 nm with a very 1.1.Parallel analyzer with mesh. FIG. Parallel plate analyzer withAu Au mesh.The Theanalyzer analyzer measures measures the the with a FIG. 2.5 nm thickplate layer of CsBr, CsBr/Cu. These work low power density of a few Watts/cm2. The QE for higher photoemission current over a a2p2psolid angle asasaafunction ofofthe retarding photoemission current over solid angle function the retarding functions are then used to compute the intrinsic emittance. potential totodetermine potential determinethe theelectron electronenergy energyspectrum. spectrum. power density10 is >10"3 for a laser power density of The effective work function can be obtained from the 4 & 105 W/cm2. Therefore, the QE in a CsBr coated sample relation resolution for the QE metal-like photoemitters near ofofof the analyzer be eV based resolution the analyzertoto beless lessthan than0.1 0.1 eVthreshbased upon uponis increased by focusing the laser beam or increasing the 1 13 13 it is no lonvalid forwith UV CsBrfilm (since old assumed low measurements low energy energy spread spreadlaser power, an effect which is due to additional activation. measurements withaactivated adiamondoid diamondoid film ger an insulator UV activation). electronafter source. addition,ininaapaper papertotobe bepublished published elseelse-Thus, it is possible to obtain an order of magnitude QE electron source. InIn addition, where,we wehave havevalidated validatedthe the energy energy spread spread method method with with where, increase relative to UV cleaned Cu targets since the QE of 2 ð!hx films " /efwith Þ aaparallel 1 in "inRðxÞ dataobtained obtained diamondoid films parallel plate plate anaanadata diamondoid fwith with laser power : al. (2) hem-uncoated clean Cu targets does not change QE ¼ 231103-3 Maldonado et Appl. Phys. Lett. 101, 231103 ( kitto lyzercomparing comparing data obtained with a conventional optto lyzer it data obtained with a conventional hem8/ ðE þ / Þ F 1 þ !k ef f ef f e"e isphericalanalyzer. analyzer.The Theelectron electronenergy energyspectra spectra for for aa single singledensity. In addition, we14 have also observed that the experiispherical not to change energy. for TheCsBr/Cu QE curves does correspond two setswith of value crystal(100) (100)Cu Cuphotocathode photocathodeare are shown shown inin Figure Figure 22 and andmental energy spread crystal reflectivity, and the optical and the electron-electron Here, the QE is given assame the number ofcoated electrons incident density. Therefore, the increase in QE with ˚ thick ˚A compared the same samplecoated withaper a25 25A thick CsBr CsBrincreasing powerthe compared totothe sample with tering lengths have been used in the first factor of Eq photon. film. R(x) isNote thethat reflectivity for a photon of frequency film.Note thatthe theCsBr CsBr film broadens the energy energyx, spreadpower density while the energy spread remains constant film broadens the spread These curves are useful for estimating the systematic u and the electronthe 1/e somewhat optical absorption depth is k appears to be due to an increase in the electron density of somewhat as shown in the normalized data of Figure 2(b), opt as shown in the normalized data of Figure 2(b), tainty of the effective work function. Specifically, the , the Fermiasas energy isinin EFthe ,the and electronbut scattering length isQE ksubstantially than toofany decrease of butincreases increases theQE substantially shown Figurestates with activation e-e the shown Figure tematicrather uncertainty the further effective work functio 2(a)spectra spectra normalized toQE. should be mentioned mentioned thatthe work function .QE. Figure 3 shows a plot of that the effective work function is /toeff since the the energy spread is unchanged. approximately horizontal distance between the red 2(a) normalized ItItshould be FIG. 1. Parallel plate analyzer with Au mesh. The analyzer measures the when thelaser laser powerdensity density isincreased increased byphoton 5!the the change changeThus, the increase blue the QE when for a normal incidence 257 nm (4.826 eV) as inlines. QE with power density without affectthe power is by 5! photoemission current over a 2p solid angle as a function of the retarding The QE for CsBr/Cu gives an effective work functi energy spreadisisless less than 10%and andassuming operationwith with aa large largeing appreciably the energy a potential function of thespread effective work function typical spread is an interesting phenomintoinenergy than 10% operation determine the electron energy spectrum. "5 3.9þ/"0.1 eV. Copper’s QE of 2.8 & 10 correspon QE enhancement ispossible possible ataaabsorption currentdensity density greater thanenon that deserves QE enhancement is at current greater than values for the reflectivity, the optical length and more study and will be discussed in a 4.55þ/"0.05 eV for its effective work function. These 2 2. 100 A/cm A/cm the e-e 100 scattering length,toand using eVeVforbased the upon Fermi future paper. resolution of the .analyzer be less than70.1 functions along with the photon energy of 4.826 eV are 231103-2 Maldonado et al. Two analysis techniques are used extract theintrinsic intrinsic 13used Two analysis techniques are extract the lowtoto energy spread measurements with a diamondoid film used in the three-step model expression for the int emittance from the measurements. In the first method, the emittance from the measurements. In published the first method, the electron source. In addition, in a paper to be elseemittance1 widthofofthe themeasured measured spectrum used to directly directly FIG. 3.energy The Cu quantum efficiencyisphotocathodes. a function of the effective work function TABLE I.width intrinsic emittance for and CsBr/Cu spectrum is aseV. used to where, Calculated we have validated theanenergy energy spread method with rﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ for incident photon of 4.826 The QE has been computed with two obtain the intrinsic emittance. In the second method, the hﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ x " ﬃ/ef f obtain the intrinsic sets emittance. Inafor the second method, thelength, and electron eintrinsic of parameters the reflectivity, q! ﬃﬃﬃﬃﬃﬃﬃﬃﬃ data obtained in diamondoid films withq parallel plateoptical ana-absorption FIG. 2. Comparison of the energy spread of an uncoated (100) :crystal ¼uncoated h!x"/single Eexcess measuredQE QEgives givesscattering theeffective effective work function, which isthe effective ef f 2 FIG. 2. Comparison of the energy spread of an single crystal (100) length to estimate the systematic error of work (lm/ mm-rms) from (lm/ mm-rms) measured the work function, which is r 3mc 2 2 lyzer comparing it toenergy data obtained with a conventional hemx Excess (eV) 3mc 3mc copper photocathode with the single crystal (100) Cu photocathode coated function. photocathode with the single crystal (100) Cu photocathode coated thenused usedelectron withthe thethree-step three-step modelelectron compute the intrinsic intrinsic QEcopper then model totocompute the with a 25 A CsBr a 8 mm CWexpt. 257 QE nm laser spectrum expt. / film from (eV)when from illuminated expt. QE with from with ispherical analyzer. The spectrum electron energy spectra for a single with a 25 A CsBrefffilm when illuminated 2 with a 8 mm CW 257 nm laser emittance. The results of the two analysis methods are comrawintrinsic data corbeam and a power density of &3 W/cm . (a) Energy spread The experimental and the resultsspectra of the two analysis methods are com-is 0.77 crystalemittance. (100) Cu The photocathode areshow shown inexcess Figure 2 and (a) Energy spread raw data cor-em beam and a power density of &3 W/cm2. parameters #4 the energy spread eV and Cu 100in T rected by the ratio offound QE’s with of CsBr/Cu (4.5 !methods 10 #4) and Cathode pared with measurements of the energy spectrum andQE QE for for cesof these two areuncoated summarized ˚ ) and uncoated Cu 100 rected by the ratio QE’s of CsBr/Cu (4.5 ! 10 pared measurements of the energy spectrum and #5 compared to with the same sample coated with a 25 A thick CsBr eV for CsBr/Cu and copper, respectively.2.8Using these "5 ). (b) Same as (a) but with peaks normalized to unity. (c) crystal (2.8 ! 10#5 Cu & 10 4.55þ/"0.05 0.42þ/"0.04 bare copper 0.31 and for0.31 CsBr-coated copper0.45 cathodes. The exI. There is good agreement between the emittances mea crystal (2.8 ! 10 ). (b) Same as (a) but with peaks normalized to unity. (c) bare that copper forfilm CsBr-coated cathodes. The emittance exfilm. Note theand CsBr broadens the energy excess energiescopper in Eq. (1), thespread intrinsic for "4 single Error function shaped data obtained with the energy spread analyzer shown with the electron those determined with the CsBr/Cu perimental intrinsic 0.77 0.71the two methods 4.5 & 10 function shaped 3.9þ/"0.1 0.77þ/"0.05 Error data obtained spectra with the and energy spread analyzer shown emittances found using Figure 1. The energy spread of the uncoated copper sample is 0.31 eV, crystal copper is 0.45þ/"0.05 lm/mm-rms emitperimental intrinsic found using the two methods and ininthe somewhat as shown in theemittances normalized data of Figure 2(b), Figure 1. TheThis energy spread of the uncoated copper sample is 0.31 mode eV, result confirms the validity of the are in good agreement. and the energy spread of the CsBr/Cu sample is 0.77 eV, boththree-step obtained from tance is 0.71þ/"0.05 lm/mm-rms for CsBr/Cu. are in good agreement. but increases the QE substantially as shown in the Figure and the energy spread of thesimple CsBr/Cuassumptions. sample is 0.77The eV, both obtained from itsderivative rather intrinsic emittance the FWHM of the curve. In the energy spectrum analysis, the intrinsic emittance Juan R. Maldonado , et al, APPLIED PHYSICS LETTERS 101, 231103 (2012) The second analysis method uses the quantum efficiency derivative curve.copper and CsBr/Cu cathodes s article copyrighted indicated in the Reuse ofintrinsic AIP content is subject tothe theFWHM termsofat:thehttp://scitation.aip.org/termsconditions. Downloaded to Innormalized theasenergy spectrum analysis, emittance 2(a)is spectra to QE. It article. should be the mentioned that listed for bare as determ , of the emitted is given in terms of the excess energy, E Experimental results observed at magnesium cathode is different from copper and 3-step model Surface photoemission mediated by surface Plasmon (SP) Surface roughness (20 nm) enhances SP coupling from UV laser (4.66 eV) Surface oxidation (MgO) lower SP energy Strong Ez from SP field cause surface photoemission SP model: suface emission ? H. J. Qian,J. B. Murphy,Y. Shen,C. X. Tang,and X. J. Wang,, predicted by 3-step model 3-step model: bulk emission APPLIED PHYSICS LETTERS 97, 253504 (2010)! H.J. Qian,J.B. Murphy, Y. Shen, C.X. Tang, X.J. Wang , Nuclear Instruments and Methods in Physics Research A 646 (2011) 22–26 st accurate and controllable parameter in a FIB photoemission is clearly seen in the charge yield map surface. Because of the third p machine. As an experimental benchmark, we show inphotoemission is clearly seen in the charge yield mapthe laser intensi e. AsFig.an2(b) experimental benchmark, we show in density versus [Fig. 3(a)]. IR laser pulses (150 fs FWHM) were focused that by decreasing the spacing p from 745 to Selected for a Viewpoint in Physics intensity in a small local regio [Fig. 3(a)]. IR laser pulses (150 fs FWHM) were focused week ending b) that by decreasing the spacing p from 745 to to 120 !m rms and scanned around the nanopattern at the charge yield. We estimate 710 nmPRL the 110, resonance shifted P Hfrom Y S I 840 C A Lto R E V I E W L E T T E R S ! 15 FEBRUARY 2013 074801 wavelength (2013) nanostructured to 120 !m rms and scanned around the nanopattern at surface co m the813 resonance wavelength shifted from 840 to normal incidence (< 1 ) with a piezo-controlled the mirror. nm, in good agreement with the simulation prediction. equal absorbed intensity b incidence (<laser 1! ) spot withonathepiezo-controlled mirror. The position of the cathode was monitored , in good agreement with the prediction. Polarization dependence of simulation the reflectivity at normal inci-normal where the integration is per surface, and Sf and If are th a virtual cathode the beam charge was Surface-Plasmon Resonance-Enhanced Emission High-Brightness Electron The by position of the laserofscreen, spot onand the cathode was monitored was not observed in experiment to theincisymmetryMultiphoton ationdence dependence of the reflectivity at due normal flat surface, respectively. Simu by a calibrated high efficiency beam ofprofile of the nanopattern, in agreement simulations. from a Nanostructured Copper Cathode by ameasured virtualFIG. cathode screen, the beam was 3 (color online). (a) Charge and yield map of the nanopat- charge A % 14 for our nanostru was not observed in experiment duewith toBeams the symmetry terned cathode. The black square indicates the nanopatterned camera. The maximum signal, obtained when the laser observed enhancement of A ¼ The bandwidth of thewith SP resonance is also a criticalmeasured by a rms calibrated high efficiency profile nanopattern, in agreement simulations. laser spot (b) 2 The charge density 1 1 1 2 2area. The 1 size is illustrated. 2 beam the simulation R. K. Li, H. G. the Andonian, J. Feng, A. Polyakov, M.theScoby, K.nanopattern, Thompson, W. spot fully covered the isbyWan, Yaveraging # 102 prediction cons feature since it is important to To, match resonance widthcamera. exp > 1:2 $C. versus absorbed laser intensity. $ is obtained shapes the nanostructures. The maximum signal, obtained when the oflaser 2 1,beam bandwidth of the SP resonance is also a critical the measured chargethe over laser the full pattern area. $ can be * H. A. Padmore, and P. Musumeci times larger than when was only hitting the flat The increased absorption c with the bandwidth of the photocathode driver laser expressed as $ ¼the J%, where we choose % to be equal to Y the pulse> 1:2 # 102 spot fully covered nanopattern, is 1 the resonance width enhancement has an importan since("it20 is nm). important to match exp Department ofmeasurement Physics and Astronomy, UCLA, Losduration Angeles, of theCalifornia drive IRbetween laser90095, of 150 the fsUSA FWHM and Jmaximum is an surface. The transition signal and Comparison between the and threshold properties. For t 2 equivalent current density. Fitting of the low charge density hitting part Advanced Lightresonance Source LBNL, Berkeley, California 94720, USA larger than when the laser was only the flat he bandwidth of thein photocathode laser Division, simulation results Fig. 2 shows thatdriver the oftimes employed in multiphoton pho yields a slope of 3:05 + 0:07. (Received 12 November 2012; published 11 February 2013) shorter than electron-phonon 1.0 signal maximum nm). the Comparison between theandmeasurement nanopattern 1, p=745 nm and actual patterns is wider shallower thanand what issurface. 1.0The transition between the threshold is set by the absorbed nanopattern 2, p=710 nm We experimentally investigate surface-plasmon assisted photoemission to enhance the efficiency of flat surface is due to the partial illumination of the nanothe damage threshold of t ion results in Fig. 2 shows that the resonance of 0.8simulated 0.8 measured Interestingly, this large increase cannot be fully metallic photocathodes for high-brightness electron1.0 sources. pattern. A nanohole array-based copper surface was changes were observed wh 1.0 The generalized explained by the change in IR absorption. nanopattern 1, p=745 nm10 mJ=cm2 . The r ual patterns is wider designed and shallower than what is p approached to exhibit a plasmonic response at 800 nm, 0.6 fabricated using the focused ion0.6beam milling 0 -100 -200 0 (a) = J = CI 10 x A/mm2 C = 1.44±0.05 (GW/cm2 3 x = 3.05±0.07 measurement 1 2 100 200 20 2 absorbed laser intensity (GW/cm Fowler-Dubridge theory [14] predicts only an increase of nanopattern 2, p=710 on nma flat copper surfac reported p characterized and tested as a photocathode Þ3 0.8 technique, optically inð1#R a phigh power radio frequency 0.864% andphotoRf ¼ 88% into account the &4:5 times e 0.4 0.4 YFD ¼ ð1#Rf Þ3 ¼ 27 times, where Rp ¼ injector. Because of the larger absorption and localization ofarethe field intensity, theand charge yield face nanopatterning. theoptical reflectivity of the nanopattern the flat surface, p w h 0.6 0.6 0.2 respectively. It is important to distinguish (b) 0.2 hole observed under ultrashort laser pulse illumination is increased by more than square 100 times compared to a flat Gaussian hole (a)To understand the (b) in charge sion regimes from strong optic additional enhancement surface.pWe also present the first beam characterization results (intrinsic emittance and bunch length) from [16–18]. A signature of strong 0.0 0.4 0.4 0.0 emission we look into the details of the optical intensity 795 810 825 840 795 810 825 840 excess energy spread of a 2 large a nanostructured photocathode. distribution. In Fig. 4(a) we show the simulated I / jEj wavelength (nm)in contrast with the multiphot wavelength (nm) h profile at resonant wavelength on 0.2 the nanohole surface. (b) 0.2 square hole energy is typically % 1 eV. Two lineouts ofGaussian the 41.75.Ht, intensity the (b) metal-vacuum DOI: 10.1103/PhysRevLett.110.074801 PACS 29.25.Bx, 78.67.!n, 79.60.!i hole along (a) numbers: reflectivity w 100 reflectivity reflectivity reflectivity -200 -100 (b) 2 280 240 200 160 120 80 40 2 100 (pC/mm rms laser spot size charge density (a) 200 (c) the excess energy should be boundaryonline). are illustrated(a) in Figs. 4(b)0.0 and 4(c), respectively, at ofnormal FIG. 2 (color Simulated reflectivity 0.0 normalized by the intensity over the entire 825 electron source. The K 795 both 810 840 795 810 840 incidence of two825 arrays ofaverage Gaussian (solid curve) andtance square pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ week ending 2 (="½# 73:2 !½eV(=I½GW=cm wavelength (nm) wavelength (nm) P H Y S I C A L R E V I E W L E T T E R S FIG. 1 (color online). (a) (2013) Scanning electron microscopy im-fine (dashed holes with the high sameintensity p, h, and values. The ! is2013 15 wavelength, FEBRUARY PRL 110, 074801 In curve) a recent experiment, IRwlaser pulses Recent progress in nanotechnologies has enabled the work f ages of the nanohole Zoomed-in cutsurfaces view ofbythe areasoutshow results of changing h and $5%.is used to evaluate intensity, directly of a the Ti:sapphire amplifier were usedwtoby illumicontrol of the array. opticalInset: properties of metal shap- shaded (a) a) nanoholes. (b) Nanohole (c) an optical microscopy FIG. (b) 2 nate (color online). (a) of Simulated reflectivity atofstrong normal the regime (! ) (b) (a)with suboptical-wavelength rms laser array under with Measured reflectivity two nanopatterns fabricated on field surface. Because of the power the current a copper cathode [7]. Inpthird this case the scaling electrons are ing them features [1]. At nanospot size regime (! * 1). For our nano 200 visible light. (c) A nanopattern 100 off-resonance consisting ofincidence single crystal substrates with 2 nm FWHM bandwidth lasers. of twothrough arrays Gaussian (solidconcentration curve) square generated three-photon process and the and current structured interfaces, strong coupling can occur between density versusaof the laser intensity, of the laser intensity ! stays >10 after taki 280 6-by-6 25 !m square patches illuminated at resonant laser The two patterns have the same hole width and depth and areThe localization on the surface, h (color online). (a) electron microscopy im- plasmons (dasheddensity curve) holesin the same p,ofh,the and w values. 100 scales aswith third power absorbed laser light and Scanning the metal electron oscillations or surface c intensity athe small local region can significantly enhance 240 regime. x wavelength. onlyintensity. different in spacing. b experimental CI = Jby=shaded The results benefited from the (SPs). ByInset: tailoring the SP200 properties, controlled the the nanohole array. Zoomed-in of the areas show the results of changing h and w by $5%. w the charge yield. We estimate the enhancement factor of the m 160 cut view 0 In Fig. 3(b) we show p/2 10 A/mm 120 increased IR absorption, up to "85%, due to a thin MgF physical dimensions of the nanostructures, one can greatly C = 1.44±0.05 2of the es. (b) Nanohole array under an optical microscopy with (b) of surface two nanopatterns fabricated onabsorbed the reflectivity nanostructured compared toR a function flat cathode at laser (GW/cm Measured 80 -100 low extraction3 field of 25 x = 3.05±0.07 at absorption, antireflective coating. Asintensity alternative nano(c) (b) 40transmission, or reflection nance visibleenhance light. light (c) A nanopattern consisting of single crystal substrates with 2annm FWHM bandwidth lasers. equal absorbed by Aapproach, ¼ ð I3the dsÞ =ðIaf slope Sf Þ, of 3:05 074801-2 curve has nano 2 selected wavelengths and localize the optical field intensity. plasmonics concepts could be applied to engineer the measurement three-photon process is domi 1 25 !m square-200patches illuminated at resonant laser The two patterns have the same hole width and over depth andnanohole are where the of integration isandperformed the These -200 phenomena have sparked great interest in research optical response a cathode obtain high efficiency slope occurs above 40–50 2 20 -100 0 100 200 gth. only different in spacing. 2 are the area and intensity surface, and Sf and observed cathode [7] related to laser-material interactions on intensity nanometerphotocathodes, bothIfthrough a control of the metal on aofflatthe absorbed laser (GW/cm metallic cathode effects [20]. The high flat surface, respectively. Simulation results sity predict factoris closel (nm) femtosecond scales and many entailing exotic applications reflectivity and through the effect of the localization of from the aacathode in, for example, optics,(a)magneto-optic and optical This idea was explored Tsang beam brightness [21 FIG. 3 (color online). Charge yield data map storage, of the nanopatof A4 (color %intensity. 14 for (a)our nanostructure. Theby achievable experimentally FIG.field online). Intensity I distribution on the nano074801-2 ized of photoelectro hole surface. lineouts of I alongof the A metal-vacuum boundasubwavelength biomolecules [2–4]. [8,9] andThe more recently by Polyakov et FD al.& [10,11]. terned cathode. detection The blackofsquare indicates the nanopatterned et al.observed 5 isemission well within enhancement ¼ Yexp =Y z incident laser (pC/mm 2 laser polarization y x 160 140 120 100 60 40 2 20 2 3 3 normalized intenstiy normalized intenstiy charge density 80 2 1 0 0 200 400 600 800 1000 3 2 1 0 0.0 0.5 1.0 1.5 2.0

1/--pages