A fast, parallel acquisition, electron energy analyzer: The hyperbolic field analyzer

Transcription

1 REVIEW OF SCIENTIFIC INSTRUMENTS VOLUME 70, NUMBER 5 MAY 1999 A fast, parallel acquisition, electron energy analyzer: The hyperbolic field analyzer M. Jacka a) and M. Kirk Department of Physics, University of York, Heslington, York YO10 5DD, United Kingdom M. M. El Gomati Department of Electronics, University of York, Heslington, York YO10 5DD, United Kingdom M. Prutton Department of Physics, University of York, Heslington, York YO10 5DD, United Kingdom Received21October1998;acceptedforpublication26January1999 This article describes a new kind of electrostatic charged particle analyzer capable of the parallel detection of a large kinetic energy range. The main purpose envisaged is for the simultaneous detection of electrons scattered from surfaces and having energies between a few tens of ev to greater than 2000 ev. A prototype has been constructed that approximates a hyperbolic deflection fieldfortheelectronsenteringanentranceslit.itexhibitsanenergyresolutionofafewevanda collectionefficiencyof0.05%of2 sr.usefulaugerspectracanbeacquiredinatimeoflessthan 2 s. The significant improvement in spectrum acquisition time that this represents offers many possibilities to further Auger and photoelectron spectroscopy American Institute of Physics. S X I. INTRODUCTION a Electronicmail:mj13@york.ac.uk The analysis of Auger and other excited electrons in the energy range from about 50 to 2500 ev provides a nearly nondestructive means to determine the surface compositions of materials. For many applications, however, one important drawback has been the relatively long times required to make useful measurements usually many seconds or minutes. This is due to the nature of the electron energy analyzers commonly used. Their common limiting property is the sequential manner by which energy spectra are obtained. This method of data acquisition becomes a hindrance when electron spectroscopy is used in conjunction with time dependent experiments for which an entire spectrum is required at repeated short time intervals. It also becomes a considerable problem if spectrum imaging is needed. In this methodology anentirespectrummustbeacquiredforeachpixelinadigital scanned image. Sequential acquisition of each energy channelinaspectrumforeachpixelintheimagebecomesa prohibitively lengthy process. The advantages associated with simultaneous acquisition of many spectral channels during imaging with a scanned electron beam have been powerfully demonstrated by Krivanek, Ahn, and Keeney 1 and HuntandWilliams 2 usingthegatanimagingfilterforparallel detection of electron energy loss spectra in a 100 kev scanning transmission electron microscope. In general there are two approaches to improving electron analyzer design for reduced spectrum acquisition time. Oneistoincreasethesolidangleofcollection;theotheristo increase the number of energy channels which can be simultaneously measured. Electrostatic analyzers have been designed with collectionefficiencyashighas30%oftheelectronsemittedfrom asurfaceinto2 sr. 3 Nevertheless,suchanalyzersmaynot be suited to the rapid acquisition of energy spectra. For example,ifonewishestocollecta1000channelspectrumin1 s, then the 1 ms dwell time per channel must include the dead time while power supplies adjust to new values. Further, unless overlapped data processing can be used, data transfer and processing will occupy additional time. These overheads reduce the time available for collection of electrons and so may lead to unacceptable counting statistics. Finally, the large solid angle occupied by the analyzer in the proximity of the specimen may make manipulation of the sample difficult, and also the location of other components such as electron and ion columns and other detectors. Increasing the number of energy channels that can be operated in parallel has been applied to hemispherical analyzers and parallel plate analyzers. Two examples are by Weightman, 4 with15discreteenergychannelsattheoutput ofahemisphericalanalyzer,andyagishita 5 withaposition sensitive detector at the output of a 45 entry parallel plate analyzer. The limitation to this kind of development of conventional analyzers is that the range of energies that can be detected simultaneously is typically only ten or so times the energy resolution of the instrument, resulting in either a small energy span or poor resolution. To obtain an energy spectrum larger than is associated with this range requires serial acquisition. The approach adopted in the present work has been to develop an analyzer that detects the whole energy spectrum of interest, simultaneously, although somewhat at the expense of solid angle collection efficiency. The advantages of parallel acquisition far outweigh the loss in solid angle collection. The basis of this analyzer is a two-dimensional hyperbolic quadrupole field /99/70(5)/2282/6/$ American Institute of Physics

2 Rev. Sci. Instrum., Vol. 70, No. 5, May 1999 Jacka et al FIG. 1. The hyperbolic field. II. THE HYPERBOLIC FIELD AS AN ENERGY ANALYZER The electrostatic quadrupole field has long been known in the field of charged particle optics. Applications include mass spectrometry, 6 strong electrostatic lenses, 7 beam deflectors, 8 andenergyfilters. 9 In cylindrical coordinate form the potential distribution is described by V V 1 r n cosn, whereforahyperbolicfieldn 2 andv 1 isthepotentialat r 1, 0 Fig.1. The equations of motion for charged particles within this fieldarewellknownandsimpletoderive.theyare,inthe planeofthefield and rcos r 0 cos 0 cos kt 0 cos sin kt k 2 rsin r 0 sin 0 cosh kt 0 sin sinh kt, k 3 where r 0 and 0 describe the initial position; 0 the initial velocity at an angle with respect to 0, and k 2eV 1 /m, inwhicheandmarethechargeandmassof the particle. There are two similar situations in which this field can be used as a first-order focusing device that is independent of the energy of the particles. These involve sets of trajectories whichfocusontothezeropotentialplane /4 sothata detector can be placed there without disturbing the field. A further restriction allows only those trajectories passing throughornearthecenterofthefield i.e.,r 0 0.Equations 2 and 3 thensimplifyto 1 FIG.2. Variousmodesoffocusing,independentofparticleenergy. a, b, and c illustratecase1, d, e,and f illustratecase2. r 0 isthewidthof the parallel beam in units of the distance from the origin to the equipotential satisfyingv E. tan 2 tanh kt tan kt. ThefirstsolutiontoEqs. 4 and 5 gives and kt The dispersion function under these conditions is r E V 1, where E is the energy measured in electron volts. Further solutions rapidly converge to 5 tan sin kt sinh kt. A.Case1:Pointtopointfocusing In this case the first-order focusing condition r 0 is satisfied by 4 kt n 2 n odd and becomes very small. The first three solutions are showninfigs.2 a, b,and c.allbutthefirstsolutionare unsuitable for practical analyzer design as can be seen by examining the second-order term 2 r/ 2. Preferably this termshouldbesmalltoallowalargeangularacceptance without significant loss of energy resolution. For the first solution

3 2284 Rev. Sci. Instrum., Vol. 70, No. 5, May 1999 Jacka et al. 2 r 2 7.9r but for others it converges toward 2 r 2 tan kt r. Forcomparison,aparallelplateanalyzer,operatinginfirstorderfocusingmodewith 45,has 2 r/ 2 4r. B. Case 2: Parallel to point focusing is The condition for first-order focusing of a parallel beam r r 0 0. For electrons entering about the origin this is satisfied by cos kt sin cos 0, whose nontrivial solutions are kt n 2 n odd. Thefirstsolution(n 1) gives and the dispersion is described by r E V 1. Further solutions are nearly identical to those for case 1. Again, the first three solutions are illustrated in Figs. 2 d, 2 e and2 f. Somepointstobenotedfromtheresultsofthisanalysis are: 1 Inbothcasestheangles atwhichfirst-orderfocusing occurs are independent of the energy of the electrons and are similar. 2 The time of flight of the electrons within the field is independent of their energies. 3 Thedispersionisproportionaltothemomentumofthe electrons, and the constants of proportionality are similar for both cases. 4 Electrons enter the analyzing field in the region where thefieldisweakest,soinarealdeviceanentranceaperture placed there would not disturb the field significantly. Points 2 and 3 arepotentiallyinterestingfortimecoincidence electron momentum spectroscopy. III. A USEFUL ANALYZER FOR SURFACE ANALYSIS Forapracticalanalyzerdesign,andintheabsenceofa transferlens,itisusefultohaveafieldfreeregion oflength d between the source of electrons and the entrance of the analyzer. For a point source this is intermediate between cases 1 and 2 described earlier. In this case there is no singleangle forwhichthefocusingofelectronsontothe 6 TABLE I. A practical analyzer configuration d 20mm x 0 0.8mm E min 57eV 2050eV E max zero potential plane is independent of their energy, yet for even a large range of energies it can be made nearly so, especially if the entrance is offset slightly from the origin of theaxesbyanamountx 0. Furthermore, a useful analyzer must also be able to acceptarangeofout-of-plane z-component electrons.itisa fact that any two-dimensional field, which focuses particles independently of their energy, will also focus out-of-plane particles inonedimension,sincetheout-of-planecomponent of the initial velocity simply reduces the in-plane velocity and therefore the equivalent energy. Itissimpletoshowthatelectronsofconstantenergywill befocused inthexdimensiononly ontothatsectionofan ellipse x 2 A 2 z2 B 2 E V 1 bounded by z x B tan, A 8 where is the initial out-of-plane angle and A 1.313, B 1.714forcase 1 earlier,anda 1.297,B /2forcase 2 earlier. To determine the optimum parameters for a useful analyzer, electron trajectories were ray traced using the time dependent equations of motions 2 and 3 earlier, taking into account the z-component drift. A simple program was written to calculate the energy resolution and dispersion for a rangeofinputenergiesandangles.asummaryoftheresults isgivenintableiandfig.3.itisbasedonadetectorlength of51mmandadetectorspatialresolutionofatleast20line pairs per mm. By changing the strength of the analyzing field, higher or lower energy electrons can be analyzed. The energy span which can be detected simultaneously (E max /E min )isapproximately36,butthisislimitedonlyby the difficulty in putting a detector close to the entrance aperture of the analyzer. The parameters listed in Table I represent one configuration of the hyperbolic field analyzer HFA suitable for Auger electron spectroscopy. For other applications in electron spectroscopy these might be varied. For example it might be appropriate to have the source of electrons either closertotheentranceapertureorelseresemblingmoreofa parallel beam, i.e., nearer to the extreme cases of point-topoint and parallel-to-point focusing described earlier. As an aside, the properties of analyzers with fields where n 2 Eq. 1 werestudied.n 1 producesauniformfield as in a parallel plate analyzer, n 2 gives a quadrupole 7

4 Rev. Sci. Instrum., Vol. 70, No. 5, May 1999 Jacka et al FIG.3. Dispersion a andresolution b asafunctionof EfortheparameterslistedinTableI. field,n 3 asextupolefield,etc.however,forourpurposes nneednottakeonlyintegervalues,sinceitsimplygoverns theanglebetweenthezeropotentialplanes accordingto n. Rather than obtaining analytic solutions to this generalized problem, the trajectories were determined by tracking the particles through the electric field, the radial and angular components being r V 1 nr n 1 cos n, V 1 nr n 1 sin n Itwasfoundthatforalargerangeofenergiestobemeasured simultaneously, analyzer models with n 2 gave the best results Fig.4. As a further aside it is noted here that the dispersion length for charged particles passing through the origin of the axes varies as L E 1/n. It is worthwhile to specifically compare the case in which n 1 aparallelplateanalyzer withthehfa(n 2) since parallel plate analyzers are in common use due to their simple construction, and have been used in conjunction with apositionsensitivedetectorforparallelacquisition. 5 Theimportant points of comparison are the energy span and the energy resolution. The energy dispersion of a parallel plate analyzer is linearwhilethatofthehfaisproportionaltoe 1/2. Therefore, for the same length detector and distance from the entrance aperture E max E min HFA E 2 max. E min PPA ThisisclearlyanadvantagetotheHFAwhenlargeenergy spans are required. FIG. 4. The modeled performance as a function of field type, determined by n. The energy resolution is defined as the minimum difference in energy between two trajectories arriving at the same point on the detector but havingdifferentvaluesof allowedbytheentranceaperture.theaverageis calculated over the energy range ev. The energy resolution of a parallel plate analyzer with a position sensitive detector is also inferior to the HFA. In both devices the detector is placed in the zero potential plane, but for the parallel plate analyzer this is the image plane while forthehfaitisthefocalplane.asaresultelectronsofone energy will be dispersed across the detector of a parallel plateanalyzeroveraregionequaltothesizeoftheentrance aperture, and this is what limits the energy resolution, rather than the second-order dispersion term. With the HFA the focal width is typically less than one tenth of the entrance aperturesize andisdeterminedbythesecond-orderterm. IV. DESIGN AND CONSTRUCTION OF A PROTOTYPE Theprogram SIMION 10 wasusedtofindageometrythat would satisfactorily approximate a hyperbolic field, using a few simply shaped electrodes. A prototype was built to this design Fig.5.Eachelectrodeisconnectedviaasetofadjustable voltage dividers to a single power supply, allowing on-linecontroloftheshapeofthefield.asafurthercontrol, theanglethatthetopplatemakeswiththegroundedelement can be varied. The analyzer is constructed from nonmagnetic FIG. 5. The prototype HFA and electron gun, showing some electron trajectories with differing energy. The field is constructed by applying the appropriatevoltagestoelectrodese 1 E 6.

5 2286 Rev. Sci. Instrum., Vol. 70, No. 5, May 1999 Jacka et al. FIG.6. Aphotographofthepartofthephosphorscreenshowinganelastic peak.energyspectra suchasfigs.7and8 areobtainedbyverticalintegration of the signal within an area outlined by the white rectangle. materials: stainless steel 310 and molybdenum for those parts near the electron paths and aluminum for the remainder. Ultrahigh vacuum UHV compatible plastics such as polyether ether ketone PEEK andpolyimideareusedfor electrically insulating components. The detector consists of a mm microchannel plate MCP followedbyaphosphor P22 screen.thefront faceofthemcpformsoneelectrode atgroundpotential of the analyzer. The phosphor screen is mounted approximately 0.5mmbehindtheMCP.ThevoltagedropacrosstheMCP is typically V, while the phosphor screen is maintainedat 2900V.TheporesizeoftheMCP 12 m,the behavior of the phosphor, and the distance between the MCP and the phosphor limit the spatial resolution. It is thought to bebetterthan50 m. For evaluation of the prototype a charge coupled device CCD camerawasusedtorecordthepatternonthephosphor screen. This limited the number of energy channels which can be simultaneously detected to 512. A zoom lens fitted to the camera allowed closer examination of particular areas of interest. Nevertheless, the poor optics of the camera proved to be a significantly limiting factor. V. RESULTS AND DISCUSSION In order to collect Auger spectra, the analyzer is combinedwithafieldemissionelectrongun keVenergy range. 11 They are mounted together on a 200 mm conflat flange containing all necessary electrical feedthroughs, and attachedtoanuhvchamber basepressure mbar. An argon ion gun allows sputter cleaning of samples, which can be inserted and removed from the chamber via an airlock. The CCD camera is mounted outside of the vacuum system and views the phosphor screen through a glass windowontheoppositeside.anexampleofanimagecaptured bytheccdcameraisgiveninfig.6. Three types of experiment were performed to demonstrate the behavior and benefits of this analyzer. They are: 1 measurementofelasticpeaksatvariousenergiestodetermine the energy resolution of the instrument, 2 acollectionofaugerspectrafrom standard samples, and FIG.7. Elasticpeakfromcopperat a 200, b 900,and c 1800eV.The energyscaleisindicatedoneachfigure. d showsthecamera sresponseto adeltafunction obtainedbydifferentiatingacrossablack whitestep. 3 rapidcollectionofaugerspectrafromsamplessensitive to electron beam damage. A. Elastic peak measurements The elastic scattering peak from copper is shown in Fig. 7 for three different beam energies, but under identical analyzer conditions. In addition to the energy resolution of the analyzer, the energy spread of the electron beam 1 ev andtheopticsoftheccdcameracontributetothewidthof the peaks. The latter effect can be seen from the camera s responsetoadeltafunction Fig.7 d.deconvolutingthis from the measured elastic peaks results in an energy resolutionofapproximately2evat200ev,4evat900ev,and5 ev at 1800 ev. These values are slightly higher than the SIMION 10 calculatedvaluesfortheprototypedesign.itshould be noted, however, that no magnetic shielding was used. Observation of the elastic peak was used to determine the response of the analyzer to electron spot size and misalignment.itwasfoundthataspotsizeof200 mdegraded the energy resolution by approximately 15%. B. Auger spectra Auger spectra for copper and silver have been measured. TheenergycalibratedspectraareshowninFig.8 a.these spectratook2seachtoacquire.figure8 b showstheratio ofthetwospectra.thepurposeofthisistodemonstratethat muchofthestructureintherawdataissystematic,andindependent of the sample, and therefore, with suitable signal processing it can be eliminated. This structure is due to local variationsinthegainofthemcpandphosphorscreen,and

6 Rev. Sci. Instrum., Vol. 70, No. 5, May 1999 Jacka et al Preliminary results have been obtained showing rapid desorption and dissociation of surface impurities and structures,ataratewhichwouldbeimpossibletomeasureusing conventional electron analyzers without prior knowledge of the composition of the material. This work will be presented in a further article. VI. FUTURE DEVELOPMENT AspartoffurtherdevelopmentoftheHFAthenumber ofenergychannelswillbeincreasedfrom512to1024.this willbeachievedbyreplacingtheccdcamerawithaphotodiode array Hamamatsu S F, which will be mounted directly behind the MCP, i.e., in the vacuum. This will have the added benefit of making the characterization of the detector more stable. Once this has been achieved more sophisticated signal processing techniques can be developed, so that quantitative microanalysis will be possible. Further improvements will see one end of the detector placedclosertotheentranceslitoftheanalyzer,sothatthe lowenergylimitcanbereduced.itisexpectedthatthisnew version will satisfy the specifications listed in Table I and Fig. 3. FIG.8. a CopperandsilverAugerspectra eachspectrumacquiredin2s, usinga10na,7.5kevincidentelectronbeamwith 1 mspotsize. b ratio of the two spectra. also scattering of high energy electrons within the analyzer. Both phenomena are well understood and their effects will be significantly reduced in the next version of the analyzer to be constructed. It should be noted that a single spectrum with systematic variations eliminated would exhibit peak to peak noise at leastafactorof &lessthaninfig.8 b.evenso,thesignal to noise ratio in Fig. 8 b would take minutes to achieve using a conventional analyzer. C. Electron radiation damage ACKNOWLEDGMENTS The authors wish to acknowledge the financial support of the Engineering and Physical Sciences Research Council for the project to design, build, and test this analyzer, and SRL Europe Ltd.fortheircooperationwithitsfurtherdevelopment. They also thank Ian Wright in the mechanical workshop for constructing many of the components. 1 O.L.Krivanek,C.C.Ahn,andR.B.Keeney,Ultramicroscopy22, J.A.HuntandD.B.Williams,Ultramicroscopy38, K.Siegbahn,N.Kholine,andG.Golikov,Nucl.Instrum.MethodsPhys. Res.A384, P.Weightman,Phys.Scr.T41, A.Yagishita,Jpn.J.Appl.Phys.,Part125, E.W.Blauth,DynamicMassSpectrometers Elsevier,NewYork, P. W. Hawkes, Quadrupoles in Electron Lens Design Academic, New York, H.D.Zeman,Rev.Sci.Instrum.48, C.Schmidt,Rev.Sci.Instrum.41, D. A. Dahl, Proceedings of the 43rd ASMS Conference on Mass Spectrometry and Allied Topics, Atlanta, Georgia, May , p York Electron Optics Ltd., Heslington Hall Phys, Heslington, York Y0105DD, UK.