Local Electrode Atom Probesarc.nucapt.northwestern.edu/refbase/files/MatChar44p059...60 T. F. Kelly...

MATERIALS CHARACTERIZATION 44:59–85 (2000)© Elsevier Science Inc., 2000. All rights reserved. 1044-5803/00/$–see front matter655 Avenue of the Americas, New York, NY 10010 PII S1044-5803(99)00055-8

59

Local Electrode Atom ProbesThomas F. Kelly* and David J. Larson†

*Department of Materials Science and Engineering, University of Wisconsin, Madison, WI 53706; and †Metals and Ceramics Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831

The historical developments leading to the advent of Local Electrode Atom Probes (LEAP)are reviewed. An assessment of the state of the art is made, and the major advantages ofLEAPs over conventional atom probes are described. The best implementations of theseconcepts and the remaining challenges for realization of LEAP’s potential are also de-scribed. It is concluded that LEAPs should be an important tool for materials characteriza-tion at the atomic scale. Modern materials-dependent industries as diverse as steel and mi-croelectronics should benefit from this technology. © Elsevier Science Inc., 2000. All rightsreserved.

INTRODUCTION

Much of the current activity and excite-ment in materials science involves process-ing and understanding materials at theatomic scale. Accordingly, it is necessaryfor materials scientists to control and char-acterize materials at the atomic level to op-timize a variety of materials systemsfrom ferrous metallurgy to multilayerthin films. There are only a few mi-croscopies that are capable of providinginformation about the structure of mate-rials at the atomic level. The atom probefield ion microscope (APFIM), particu-larly the three-dimensional atom probe(3DAP), is the only instrument that de-termines the 3D location and elementalidentity of atoms in a sample at theatomic scale.

WHO NEEDS LEAP?

As most of the articles in this volume attest,the atom probe is one of the most spectacu-lar and uniquely capable analytical instru-ments ever developed. Given its extraordi-nary capabilities and a burgeoning need forthis capability, it is curious that atom-probe

technology has not enjoyed more wide-spread adoption. There are many ways toexplain this but, in the authors’ opinion, itis a problem of applicability. If the resultsavailable from the atom probe were moregenerally useful, there would have beenmore eager customers for the instrument.Three-dimensional atom probes havegreatly improved the situation, but the un-derlying problem remains. The tedium andconstraints of sample preparation could becited as the reason, though the situation isno worse than for transmission electron mi-croscopy. Nonetheless, it is very true thateasier and more powerful sample prepara-tion methods are the key to application ofeither technique to a broader spectrum ofspecimen types. The problem then, as wesee it, is twofold: today’s atom probes donot analyze very large volumes of material(limited to about 106 atoms), and the speci-men geometries that are readily accessible(principally needle-shaped specimens ofelectrically conducting material) limit thetechnique. As a result, atom probes can beapplied only to those problems for whichknowledge of the limited analysis volume isuseful. Based on today’s capabilities, thatset of problems is not large.

60 T. F. Kelly and D. J. Larson

The Local Electrode Atom Probe (LEAP)has the potential to greatly expand thereach of the atom-probe technique tomuch larger volumes (e.g., 109 atoms) andmore general specimen geometries (e.g.,planar structures, where good electricalconductivity is not a requirement). The ar-guments for this conclusion are many andare described in detail in this paper. Theessential ideas are as follows. (1) With alocal electrode mediating the extraction ofions from a particular tip (Fig. 1) the spec-imen may consist of many tips rather thana single tip. These tips may be formed ona planar surface. (2) With the extractionelectrode close to the tip, much lower ex-traction voltages are required comparedwith conventional atom probes. Pulsingsystems can achieve much higher repeti-tion rates at lower voltages, and so theextraction process may achieve several or-ders of magnitude higher rates. (3) Be-cause primary acceleration of ions occursin a spatially and temporally small region,the spread in ion times of flight is smallinitially. With appropriate design, if the

spread is kept small relative to the totaltime of flight, the mass resolution of theinstrument can be excellent. (4) Because ofthe low extraction voltage in LEAP, sec-ondary acceleration of the ions to higherfinal energies is one strategy that can beapplied to achieve improvement in themass resolution.

Taken together, these advantages add upto major improvement in the key operatingparameters of atom probes. They can changethe way that the instrument is perceivedand utilized. Thus, LEAPs could greatly ex-pand the reach and commercial viability ofatom-probe technology.

RECOGNITION OF THE NEED FOR THE TECHNOLOGY

Recently, a U.S. National Science Founda-tion Panel Report on Atomic ResolutionMicroscopy was commissioned to accessthe opportunities in materials research uti-lizing atomic resolution microscopies [1].This report states “The continuing minia-turization of electronic devices requires theinvention of manufacturing tools and pro-cesses to operate on scales that are muchbelow current micron-size technology . . .At length scales of 20 to 100 atoms, there isan urgent need for instruments with reso-lution at the atomic level and with a capa-bility for chemical identification for pro-cessing and characterization.” This need isevident today, as several industries are de-veloping products where subnanometerlayers are utilized and must be understood.With regard to how atomic resolution mi-croscopies are needed to bring about a rev-olution in materials processing, the reportstates, “The holy grail of semiconductoranalysis is the ability to identify, anywherein a crystal and with full three-dimensionalatomic resolution, an impurity or dopantatom. Such analysis will require sub-Å res-olution utilizing tomographic techniques.”This statement sounds like an endorsementof 3DAP, in general. A LEAP performing3D analyses, as described in this paper, canresult in a user-friendly instrument that de-livers this needed capability.

FIG. 1. Schematic illustration of the LEAP geometrywith a local electrode that is positioned relative to aplanar microtip specimen.

Local Electrode Atom Probes 61

HISTORICAL DEVELOPMENTS

THE SCANNING ATOM PROBE

As described in the following sections,there were several research efforts which,in retrospect, appear as natural precursorsto the breakthrough that established theconcept of the LEAP or scanning atomprobe (SAP). Clearly, however, it was thework of Nishikawa, with the help of Ki-moto, who first presented their work at the1993 International Field Emission Sympo-sium in Nagoya, Japan [2] that establishedthe idea of using a positionable small ex-traction electrode relative to multitude ofsharp points. Nishikawa reviews his workin this area separately in this volume [3].

By applying a high voltage between planarmicrotip specimen and a flat screen, the fieldat any one tip will not be high enough forfield evaporation. In addition, control overwhich tip is evaporating could not be con-trolled even if the field could be made highenough. Nishikawa and Kimoto [2] sug-gested that a small aperture could be scannedover the surface at high potential until ionswere extracted from one of the protrusions;hence, the name, “scanning atom probe.”They evaluated the electrode geometriesneeded to achieve the electrostatic fields re-quired for field evaporation of the anodewhile avoiding field electron emission fromthe cathode. They suggested that the SAPmight be used to analyze any type of (planar)specimen that had protrusions on its surfacesuch as fracture surfaces and particulate ma-terials. They also suggested that specimens ofplanar materials might be made by cuttingmany parallel grooves in each of two perpen-dicular directions on a surface.

Nishikawa continued development of theSAP, and in 1995, his group presented ini-tial results concerning a needle-shaped spec-imen that was positioned relative to a flatcircular aperture of 10

�m diameter [4].They showed that the field electron emis-sion current could be used to align the tiprelative to the aperture. They have per-formed one-dimensional atom-probe anal-yses of protrusions in diamond-like coat-ings [5, 6] and silicon emitter tips [7, 8]. The

instrument uses a local aperture, and isconfigured as a one-dimensional atomprobe. Thus far, the instrument operates atroom temperature. The extracted ions aresent through a reflectron [9, 10] to improvethe mass resolution of the instrument.

THE SCANNING ATOM PROBE ANDTHE LEAP

At that 1993 IFES meeting, one of the au-thors of this paper (T.F.K.) was particularlyimpressed by the SAP concept and the pos-sibilities that this new specimen geometrymight open. This engendered a lot of think-ing about the ramifications of the small ex-traction electrode. T.F.K. was familiar withthe work on field emitter arrays wheresmall extraction apertures are patternedaround arrays of sharp tips. It was wellknown at the time that the voltage neededto achieve field electron emission from fieldemitter arrays in much lower than the volt-age needed for field electron emission fromisolated needles. It is the proximity of theextraction electrode to the tip that localizesand enhances the field. The major advan-tage of the scanning atom probe is that itmakes it possible to work with planar mi-crotip specimens. However, the local extrac-tion aperture makes performance improve-ments possible. Thus, the name “LocalElectrode Atom Probe” is used to empha-size the instrumental technique where prox-imity effects play a major role in definingthe instrument. Note that the advantagesthat result from a local electrode apply to allatom-probe specimen types including con-ventional needle-shaped specimens.

It is worth considering the developmentof the SAP/LEAP in an historical context.Several avenues of development have con-tributed to this technology, and specimenpreparation is paramount. The followingsections describe some of the most impor-tant contributions to specimen fabricationwork for SAP/LEAP.

DEVELOPMENT OF MICROTIP SPECIMENS

To investigate a materials system by anycharacterization technique, whether it is


transmission electron microscopy, scan-ning electron microscopy, field ion micros-copy, etc., an absolute requirement is theability to fabricate a good specimen. This isparticularly important for field ion micro-scopy and atom probe microscopy, wherethe specimen must be in the form of asharply pointed “tip” having a radius ofcurvature at the apex of about 10 to 50nm.Although the fabrication of a specimenfrom a metal wire is relatively straightfor-ward and is generally performed by electro-chemical polishing methods, preparing afield ion specimen from a material thatdoes not electropolish well can be difficultor impossible. If a particular physical orcrystallographic orientation of the speci-men is required as well, the chances of suc-cess diminish further. This is especiallytrue for the case where the specimen mustcontain thin films on a planar structurewith tips normal to the surface. The planarspecimen requirement has either limited orprecluded the application of field ion andatom probe microscopy to a number of ma-terials systems. As stated above, the mainadvantage of an atom probe employing alocal electrode is the ability to analyze mi-crotips on such a specimen.

The planar microtip geometry ideallyconsists of small cones formed from a flatsubstrate in a direction normal to the planeof the substrate (Fig. 2). Some of the advan-tages of this type of specimen include thecapability to: (1) fabricate multiple speci-mens from a single small piece of material,(2) form specimens with tip axes orientedalong a specific direction, (3) form tips atspecific positions on the sample surface (atinterfaces, precipitates, etc.), and (4) takeadvantage of methods not commonly usedfor specimen fabrication, such as ion beamtechniques.

Ion Sputtering of Surfaces

Ion sputtering may be defined as the re-moval of material from a surface due tothe impact of ions. Sputtering was report-edly first observed in a gas discharge asthe removal of a cathode impacted by en-

ergetic ions from a plasma [11]. (For a briefhistorical overview of sputtering, see [12].)Ion sputtering has many applications,which may be divided into two general cat-egories: (1) those that remove or exposematerial for the purpose of analysis, and (2)those that remove material with the goal ofmodifying the structure of the substrate.Examples of the former include secondaryion mass spectrometry, sputtered neutralmass spectrometry, Auger electron spec-troscopy, X-ray photoelectron spectroscopy,low-energy ion scattering spectrometry,and Rutherford backscattering spectrom-etry. Examples of the latter include sub-strate cleaning, specimen preparation forvarious microscopy techniques, thin-filmdeposition, and various forms of lithogra-phy. The specific sputtering applicationthat is of interest here is the structuring of amaterial to form cones from a surface foruse as field ion specimens. It should benoted that the concept of fabricating “mul-tipoint” emitters using ion beam tech-niques for use in field emission or field ionmicroscopy was proposed as early as 1981by Auciello [13], who pointed out that ionbombardment-induced arrays of cones might

FIG. 2. SEM image of silicon surface with microtipscreated by ion milling at normal incidence through3�m-diameter diamond particles.


be used as an alternative field ion specimengeometry.

Cone Formation

It has been known for quite some time thatit is possible to form “cones” on a surfaceby the use of ion sputtering under certainconditions (Fig. 3). Many reviews discuss-ing this type of cone formation have beenpublished including those by Auciello [12,13], Witcomb [14], Navinsek [15], Hauffe[16], and Ghose and Karmohapatro [17]. Ifforeign particles are present on a surfaceduring ion sputtering, these particles willmask the substrate from sputtering andcones will be formed at these locations.This process is known as ion beam masketching (IBME). Sputter yield is known todepend on the angle of incidence and to gothrough a critical angle (

�max) somewherebetween 60

� and 80

� [15], where

� is definedas the angle between the surface normaland ion beam incidence. At angles greaterthan

�max, the ions are reflected from thesurface and do not contribute to sputtering[18]. Ions that are reflected from the surfaceof a forming cone will impact at the conebase and form a “moat”-type structure, asillustrated in Fig. 4. The critical angle forsuch ion reflection is given by [19] [Eq. (1)]:

(1)

where a0 is the Bohr radius,

� is the sub-strate atom density, Ei is the incident ionenergy, Er is the Rydberg constant, and Z1

and Z2 are the atomic numbers of the ionand substrate, respectively. As material iseroded away from the substrate, a structureforms that exposes those faces having themaximum sputtering rate. A cone developswith its axis parallel to the direction of theincident ion beam, and has a shank anglegiven by [18] [Eq. (2)]:

(2)

Note that this analysis assumes a ran-dom polycrystalline orientation, that is,no crystallographic dependence (channel-

θmax π 2⁄( ) 5πa0ρ2 3⁄ Z1Z2Er(( )Z1 Z2+( )Ei( ) )1 2⁄

⁄=

α π 2θmax–=

ing is ignored). Polycrystalline surfacesthat are sputtered will result in height con-trast due to different sputtering rates fromrandom grain orientations [20]. The size ofthe particle and the relative mill rates of theparticle and substrate will determine the fi-nal length of the cone.

With respect to the formation of conesfor use as a field ion specimen, it is of inter-est to examine a typical shank angle of sucha structure formed by IBME. Using copperas an example substrate, and Ar ions of4keV energy at normal incidence (

�

� 0),the equation given above predicts a shankangle of about 32

�. This compares favor-ably to an experimentally measured angleof about 30

� [21]. Although this angle ishigher than that normally observed in fieldion specimens (about 1 to 10

�), it is ex-pected that the use of a local electrode ge-ometry for electric field production willminimize the detrimental effects of thelarger shank angle [22]. In addition, smallershank angles may be obtained using IBMEwith angles other than normal incidenceand a rotating specimen (Fig. 5) [23, 24]. Ingeneral, milling conditions must be se-lected such that a final geometry suitable

FIG. 3. Image of tin microtips formed by ion beammask etching with 5keV Ar ions. From the work ofStuart and Thompson [18].


for field evaporation is obtained. Ion sput-tering of the mask must produce tips thathave an apex radius of curvature of

�100nm,be at least a few microns long, have ashank angle

�45

�, and be spaced such thatthe nearest neighbor tips do not interfere

with the extraction electrode used for fieldevaporation.

Mask Material and Positioning

Particles of many types including polymerspheres [25], metallic spheres and ceramic

FIG. 4. (top) Schematic illustration of microtip cone formation with a moat. (bottom) An unfinished microtipformed in silicon by ion beam mask etching with diamond-particle masks. The moat is clearly illustrated in thisimage. The remnants of the diamond particle are visible at the top of the microtip.


particles [26] can serve as masks for IBME.Diamond [21] works especially well for thisapplication, due to its very low sputteringrate [Figs. 2 and 4(bottom)]. The type ofmask material used will determine whetherthe mask can be placed on the sample inrandom positions, in arrayed positions, orin specified locations. Random particle mask-ing is relatively straightforward, but doesnot offer specific positioning capability.Another possibility is to use electron-beam–fabricated carbon contamination spikes toform a mask [21, 23]. This type of spike hasbeen successfully fabricated for use as a

scanning tunneling microscopy tip [27]. Itshould also be possible to deposit othermask materials using an electron beam. Forexample, aluminum carbide can be depos-ited from the breakdown of trimethyl alu-minum gas by an electron beam (T. F.Kuech, University of Wisconsin, privatecommunication, 1995). Similarly, a metallicmask can be produced using a focused ionbeam instrument, and this technique hasbeen used to protect specific regions in thefabrication of samples for transmissionelectron microscopy (D. T. Foord, Univer-sity of Cambridge, private communication,1998). These approaches would make itstraightforward to control the mask posi-tion and to interactively select features ofinterest to be contained in the field ionspecimen.

It is also possible to deposit masks litho-graphically. These masks can be positionedboth with and without regard to the under-lying microstructure. In this manner, tipscould be fabricated, for example, from aparticular metallization region in a deviceor in a linear array across a grain boundary.This method holds promise for the preciseplacement of tips on specific features of in-terest if the required instrumentation isavailable.

Field Ion Specimen Preparation by IBME

Preparation of planar microtip specimens Thefeasibility of fabricating microtips by IBMEfor LEAP has been investigated by Larsonet al. [21]. In this work, tips were formed onplanar samples using 3 and 6

�m diamondparticles as masks for ion beam sputteringat normal incidence. Samples of cooper, 304stainless steel, a metal–oxide semiconduc-tor structure, and a BiSrCaCuO supercon-ductor were studied. It was found that tipscould be formed from all materials exam-ined. The tips were many microns tall witha radius of curvature at the apex of lessthan 100nm and shank angles down toabout 20

� (see Fig. 2).A random-placement IBME can be uti-

lized in a larger area to create multilayerspecimens from even complex structures.

FIG. 5. SEM image of iron microtip created by a two-stage ion-beam mask-etching process. Initial millingwas at normal incidence through 3�m-diameter dia-mond particles. For the second stage, the specimenwas tilted 20� away from normal incidence. The nearlyparallel shank section was created during the secondstage.


Figure 6 shows the results of ion beammask etching of an Intel 286 processor [28].Many tips from a large area of the speci-men are shown in Fig. 6(top). This image il-lustrates that within the same specimenfabricated by IBME, there can be a range oftip outcomes from unfinished to overmilled.Some tips may also be poorly shaped foranalysis. Figure 6(bottom) shows a sharpmicrotip with many layers of the structurevisible in the image. Note also that the coneangle,

�, changes slightly at each layer. Arange of tip outcomes can be due either tovariations in the size and shape of the maskparticles or to variations in the ion beamcurrent. This fact can be used to advantage.

In Fig. 6(top), it is apparent that the upperright-hand side of the image has beenmilled more than the lower left-hand side.The ion beam creates this circular region,and the center of the region corresponds tothe center of the beam. A typical instru-ment used for ion milling of transmissionelectron microscopy specimens producesan ion beam of about a millimeter diameterwith a Gaussian current profile. When us-ing the IBME technique, an effective ap-proach is to ion mill until the region at thecenter of the beam is slightly overmilled.Under these conditions, there is a gradientaway from the center in the amount of ionmilling exposure. Far from the center of thebeam will necessarily be undermilled.There will be a ring about the center where“perfect” microtips will be found. The in-herent range of tip exposure can be used tostudy material at different depths in astructure. Using mask particles with arange of size can create the same effect. Inprinciple, if the particles are attached to thesubstrate with a binder, this techniquecould be used to analyze the particlesthemselves.

In all of this work, it is clear that the abil-ity to image the microtips can be essential toproductive analysis of a specimen. As dis-cussed in the Current State of Developmentsection, an SEM can be an important integralpart of a LEAP, and it can serve the impor-tant function of tip imaging for preview.

Preparation of specimens for conventional atomprobe geometries Liddle et al. [25] employedthe IBME technique in attempts to preparefield ion specimens from a III–V epitaxial-layered material. This work resulted in thefabrication of cones that were about 1

�mtall. Due to the small height of these tips, nosuccessful atom probe analyses of thesematerials was obtained. It was determinedthat a combination of small tip height andlarge substrate area resulted in a reductionof the electric field formed at the apex ofthe tips. Silicon tips of height

� 50

�m werelater fabricated by Larson et al. [24], andthis work was partially successful as it suc-cessfully lead to field ionization and field

FIG. 6. (top) SEM image of an Intel 286 chip after ion-beam mask etching with randomly placed 3�m-diame-ter diamond particles. (bottom) SEM image of one mi-crotip from the top panel that shows multiple layers ofmaterial in the image.


evaporation from the silicon specimens.Kvist et al. [29] have also used a multistageprocedure involving IBME to prepare spec-imens with 10

�m of the surface in WC–TiC–Co-based materials and a CVD multi-layer-coated cutting tool.

One of the primary applications of aLEAP is the analysis of thin-film structures.The large amount of research in this areaindicates its importance and the potentialadvantages associated with successful thin-film field-ion specimen preparation. Initialwork on preparation of field-ion specimensfrom thin-film structures was done byMelmed [30] and Camus et al. [31] on homo-geneous thin films. Further work has beenperformed by various researches includingHono et al. [32], Pundt and Michaelsen [33],Cerezo et al. [34], and Al-Kassab et al. [35].The typical method for fabricating multi-layer-film field-ion specimens is to evapo-rate individual layers onto a prepared spec-imen of the proper needle geometry. Thedisadvantage of this technique is that it isnot directly possible to correlate structureand composition results from this type ofsample with magnetic measurements car-ried out on “flat substrate”-type films. An-other preparation method for field-ionspecimens containing a single-layer filmhas been reported by Hasegawa et al. [36].This technique involves a combination ofphotolithography and pulsed micropolish-ing techniques, and results in a specimenwith the film parallel to the tip axis. Al-though the Hasegawa technique results inthe fabrication of a field-ion specimen froma flat-substrate multilayer film, it still re-quires that the material used to form the filmscan be easily electrochemically polished. Dif-ficulties arise when the multilayer film mate-rials do not polish well or have highly differ-ent polishing rates or if the final blankgeometry is of nanoscale proportions. In suchcases, another method of sample preparation,such as ion-beam milling, must be used.

Focused Ion-Beam Milling

Needle-shaped specimens by FIB Recently, pro-gress has been made in the fabrication of

field ion specimens from thin film struc-tures using focused ion-beam milling. Theapplication of a focused ion-beam instru-ment that allows concurrent imaging andmodification of a specimen was initiallyproposed by Waugh et al. [37]. Advances inthe capabilities of focused ion beams haverecently enabled their successful use in thefabrication of scanning probe tips [38] andfield ion specimens from various materialsincluding nanoscale thin film structures[39–41]. In this work, focused ion-beammilling has been used to fabricate field-ionspecimens from a multilayer-film structurecontaining alternating layers of 2nm-thickcopper and 2nm-thick cobalt. This methodof specimen preparation has made it possi-ble to observe these layers by both field-ionimaging and three-dimensional (3D) atom-probe analyses [42–44]. The importance ofthis result is that it is now possible to corre-late directly the 3D morphology and com-position data obtained from atom-probeanalyses with magnetic measurements fromthe same film.

In addition, atom-probe analysis of speci-mens fabricated by focused ion-beam mill-ing has made it possible to quantify the im-plantation and damage induced by preparingsamples in this manner [41, 42]. Field evap-oration through the region damaged by gal-lium ion implantation (

�30nm thick) hasbeen demonstrated. This work shows thefeasibility of using focused ion-beam mill-ing for field-ion specimen preparation andthis technique could be applied to the fabri-cation of samples for LEAP. A disadvan-tage of this technique when used to fabri-cate needles, however, is that it does notreadily lend itself to imaging the layers withthe layer normal parallel to the tip axis.With multilayer films in cross-section, it ispossible to get local magnification effectsthat distort the ion trajectories significantlyfrom different layers and limit the spatialresolution to a nanometer or more. This lim-itation is especially deleterious in thin mul-tilayer films because they are typically onthe order of a few nanometers thick, andthe degree of interlayer mixing is of partic-ular interest. In atom probe analyses, the tip


axis direction is currently the only directionwhere atomic layer resolution is achieved,and thus, it is preferable that the layer nor-mals are parallel to the tip axis. AlthoughFIB could be used to fabricate needle-shapedspecimens with their layer normals parallelto the tip axis, planar microtip specimensanalyzed with LEAP should be much sim-pler and more expeditious. Planar microtipspecimens of multilayer materials could bemade by random-placement IBME if thearea of interest is large, or site-specific FIBcould be used as described below.

Site specific planar microtip specimens by FIBWith the amount of control available froma FIB instrument, it is possible to fabricatemicrotips from a specific location on a flatsurface in a relatively short time like 1 h orless [45]. Higher energy ions like 50keVwill shorten the milling time markedly.However, the higher beam energies canlead to significant ion implantation anddamage [39]. For this reason a dual-beamFIB, which uses an electron beam for imag-ing, is preferable. A hollow-cone illumina-tion pattern of the focused ion beam is usedto sputter away material around a centraltip. It is be possible to locate the center ofthe microtip to within a few hundred na-nometers. If this technique is to be used forspecimen fabrication for LEAP, then suffi-cient material around the tip must be re-moved to allow an extraction electrode tobe positioned above the microtip. Sharptips can be fabricated with this technique.The possibility of using site-specific FIB tofabricate blunter tips may be especially ap-pealing for analysis of very thin layers asdescribed in the next section. A potentialdifficulty with this technique is that ion im-plantation and damage to the underlyingmaterial may intersect the analysis volumeand must be considered.

The effect of counter electrodes on field The useof an extraction electrode to localize andenhance an electric field is certainly notnew. In 1968, the first results from micro-fabricated field emitter arrays were pub-lished [46]. These devices consist of many

sharp tips (cathodes), each having its ownmetallic extraction electrode or “gate.” Theradius of these tips is comparable to APspecimens, and the gate opening is about 1to 2�m. Spindt et al. [47] found that fieldemission currents of 50 to 150�A can beachieved from single tips of radius 50nmusing 100 to 300 volts compared to 1 to30kV for etched wire emitters. In this case,the voltage is approximately a factor of tenless than what is normally required for aconventional field emitter with the anodeessentially at infinity. Spindt et al. [47] alsoconcluded that when the tip radius is in the50 to 150nm range, rather than an inverserelation between tip radius and field, theradius appears to have only a second-ordereffect on the field. This fact will be consid-ered further with regard to analysis ofblunt tips in the next section. These struc-tures have also been the subjects of theoret-ical modeling [48, 49]. It should be notedthat there is a vast amount of work and on-going developments in field emission ar-rays, and we have not attempted to assem-ble a complete review. The interestedreader is referred to [50] for more details.

Spindt has also experimentally demon-strated field ionization using an extractionelectrode geometry at voltages on the orderof 1kV [51], which is much lower than theusual 5 to 30kV. Again, as might have beenexpected, this value is about a factor of tenless than what is normally required forfield ionization. Even though, in this case,the field ionization is not used for FIM orfield evaporation, this proves that not onlydoes an extraction electrode localize theelectric field, but it also leads to a signifi-cant enhancement in the local electric fieldso that low extraction voltages can be em-ployed.

Extraction electrodes have been used inconventional APFIM [52, 53], and have re-cently been employed frequently by severalgroups [54, 55]. Although these electrodesare not “local” by the definition employedherein (see below), they still provide a sig-nificant field enhancement and an im-provement in mass resolution when com-pared to the case where there is no counter


electrode. Huang et al. [56] have recordedfield ion images and AP data from a singletip and an array of FE tips where the ap-plied voltage was less than 300 volts. Thegroup at the Université de Rouen, France,have measured reductions of a factor ofthree in the applied voltage needed forfield evaporation with a counter electrodethat is on the order of a millimeter in diam-eter relative to a remote electrode [57].

The effects of the proximity of a localelectrode on the electric field near a tip canbe examined analytically using a simpleparaboloidal model [58]. For example, atypical arrangement for AP evaporation ofRh2 consists of a tip of radius 45nm, anelectrode at 1mm from the tip, a detector atsome distance from the electrode, and anapplied voltage of 11kV between the tipand the electrode [59, 60]. The electric fieldat the specimen apex calculated with theparaboloidal model is 45.6V/nm. A simpleapproximation for the field in a local elec-trode case can be made by taking the pa-raboloidal detector to be 1�m from thesame tip. To duplicate the field of 45.6V/nm, only 3.95kV needs to be applied to thespecimen. Hence, the proximity of the localelectrode results in a field enhancementfactor of slightly less than 3 for this particu-lar case. This model only approximates thereal case of a local extraction electrode. Inour experience with detailed numericalmodeling based on finite element methods,which is expected to be much closer to ac-tual values, the actual field is always greaterthan this parabolic model by as much as an-other factor of 3 [61], which agrees with theexperimental observations noted above.

CURRENT STATE OF DEVELOPMENT

EXTRATION ELECTRODE DESIGN AND FABRICATION

Because field ion evaporation from the an-ode requires a field that is about a factor of8 greater than the field required for fieldelectron emission from the cathode, the ex-traction electrode must be designed care-

fully to prevent field emission from occur-ring. Nishikawa and Kimoto [2] paidspecial attention to this issue in their firstpaper on SAP. Actually, if field electronemission does occur during a voltagepulse, the emitted electrons will be at-tracted to the specimen tip, which may helpstimulate field evaporation. This processcould be utilized as a means to accomplishpulsed-field evaporation. However, thecreation of extraneous electron signals inthe vacuum chamber during the pulse is aconcern, because they can make it more dif-ficult to detect ion arrivals at the detector.Constructing the extraction electrode withno protrusions that might act as a source offield electron emission will help minimizethis possibility. In addition, the extractionelectrode should have a high work functioninherently, or it should have a coating thatsuppresses field electron emission. For ex-ample, Mousa [62, 63] has experimentallyshown that field electron emission will besuppressed from aluminum and magne-sium by their native oxides as long as theoxide is at least 2nm thick. This may bereadily accomplished by native oxide for-mation, but could be enhanced with an oxi-dizing treatment. In addition, Mousa hasexperimented with epoxies [64] that arealso successful at minimizing field emis-sion. The issue of field electron emissionfrom the cathode is one that will require at-tention on any instrument that is built, butit appears that this issue may be solved sat-isfactorily with existing technology. Wenote that Nishikawa [5–8] has successfullyproduced results with the Scanning AtomProbe with no obvious difficulty from fieldelectron emission.

Nishikawa and Kimoto [2] suggested theuse of silicon microfabrication techniquesfor fabrication of the extraction electrodes.In this approach, a hole would be createdlithographically in the apex of a square pyr-amid of silicon. A limitation of this ap-proach is that the height of the pyramid islimited to about 25�m. Bajikar et al. [65] de-veloped a technique based on machining ofa conical form on the end of a metal rod us-ing a lathe and then replicating this form


with a removable coating. An example ofthis type of electrode is shown in Fig. 7 [66].The machined rod is first coated with apolymeric release layer that also serves tosmooth out the machining marks. A metalcoating is then applied, and is released us-ing a solvent. A hole is cut in the apex ofthe cone by a focussed ion beam instru-ment [65–67]. Aluminum metal was used inthis case for the electrode material.

The size of the aperture that is requiredfor a SAP/LEAP has been considered.Nishikawa and Kimoto [2] looked at partic-ular aperture diameters with electrostaticmodeling and found several geometriesthat would work. Bajikar et al. [61] usedelectrostatic modeling to seek an optimalaperture size and position for a given tip(Fig. 8). They found that for a given aper-ture diameter (in this case, 1.5�m), the fieldon the tip went through a maximum as acentered tip was moved axially (the z-direc-tion in Fig. 1) relative to the aperture [Fig.8(a)]. The maximum field in this case corre-sponds to a tip location just inside the aper-ture (position � 0.7�m). The maximumin the tip-field-to-electrode-field ratio, how-ever, occurs when the tip is just outside theaperture (position � 0.2�m) [Fig. 8(b)].This latter curve is slowly changing near

the maximum, and taking these two figurestogether; they suggest that an optimal posi-tion for this case is just about where the tipenters the aperture (position � 0). the effectof aperture diameter on the field was alsoconsidered. Figure 8(c) shows that the fieldon the tip increases monotonically as theaperture diameter decreases. However,according to Fig. 8(d), the tip-field-to-aper-ture-field ratio goes through a maximum ataround 2�m diameter in this case. Becauseboth a high tip field and a high tip-field-to-aperture-field ratio are needed, this partic-ular set of electrostatic models indicatesthat a 1.5�m-diameter aperture positionedwith the tip just at the entrance plane of theaperture will achieve the best results forLEAP. Further detailed electrostatic model-ing is warranted, but the general conclu-sions from this work should hold up.

EXTRACTION ELECTRODE POSITIONING

To position an extraction aperture abovethe tip of the specimen, it is necessary thatthe positioning stage has three axes of mo-tion; lateral centering in two dimensions,and height adjustment. This positioningcould be accomplished with a stage outside

FIG. 7. (left) Light micrograph of conical extraction electrode mold prior to coating with aluminum and drillingthe aperture. (right) SEM image of the apex of the aperture after a 2�m-diameter hole has been drilled by FIB. Thefocused ion beam work was performed by M. Coy and V. Dravid of Northwestern University, Evanston, IL.


the vacuum system to place an aperture rel-ative to a tip for insertion as a prealignedunit. Alternatively, a stage may be built foroperation inside the microscope vacuum.The former approach would offer, at least,greater simplicity of the UHV instrument,while the latter would offer the opportu-nity to examine multiple tips on a singlespecimen in rapid succession. The follow-ing discussion will consider primarily thecase of a stage that is internal to the vac-uum system. This case is the more chal-lenging, and such a stage can be used out-side of a vacuum.

If a 1�m aperture is to be positioned pre-cisely with respect to a tip of 0.1�m radiusof curvature, then the positioning precisionshould be at least 0.1�m or finer. Considerfurther that a planar microtip specimen

may be several millimeters across, and itmay be necessary to change extraction ap-ertures on occasion, which may be a fewmillimeters across. Thus, the demands onthe stage that positions the tip relative tothe extraction aperture are that it must beable to cover large distances (millimeters)in a short amount of time (tens of seconds)with high precision (tens of nanometers).Rotary motor-based stages can meet thesemotion requirements, especially the speedrequirements, but for ultrahigh vacuum ap-plications, they will be bulky and complex.Most of the technologies that are used to re-alize this positioning precision in ultrahighvacuum are based on piezoelectric actua-tion. Because piezoelectric actuators achievevery precise motion in submicron steps, aslip/stick mechanism is used in almost all

FIG. 8. Calculated dependence of the field at the apex of a microtip 10�m tall with 100nm radius of curvature andVextraction � 3.5kV (from Bajikar et al. [61]). (a) Dependence of the field at the tip on axial position of the aperture(electrode) for a 1.5�m-diameter aperture. (b) Dependence of the ratio of the tip field to the electrode field on axialposition of the aperture for a 1.5�m diameter aperture. (c) Dependence of the field at the tip on aperture (elec-trode) diameter. (d) Dependence of the tip-field-to-electrode-field ratio on aperture diameter. The solid line in (c)and (d) corresponds to the position where the field is maximum in (a), while the dashed line corresponds to theposition where the tip-field-to-electrode-field ratio is maximum in (b).


of the actuators that meet these require-ments. Inchworm actuators [68] may beused to move linear slides [69] on multipleaxes at 250�m per second with 1nm preci-sion. Compact stages are available [70] thathave the slide and actuator assembled as aunit, and they achieve 1nm precision over5mm travel with 100�m per second speed.Linear micropositioners [71] are a similartechnology that can achieve very high lin-ear speeds (10mm per second) with 50nmprecision. These later actuators may be a bittoo coarse for final positioning but could beuseful, nonetheless, for aperture changingor course specimen movement.

Some form of feedback is needed for po-sitioning the tip relative to the aperture.Specimen/aperture sets that are prealignedoutside the vacuum would need to addressthis issue. Light microscopes have beenshown to work adequately for this purpose[65]. This solution is difficult to implementinside the vacuum. Once the tip is close tothe aperture, Nishikawa et al. [4] haveshown that field electron emission may beused to center the tip in the aperture. Inprinciple, the field ion image may also beused for fine alignment. However, prior tofine alignment, it is necessary to find a use-able tip and achieve at least-course align-ment. A visual feedback mechanism ap-pears to be the most expeditious solution atthis time. Light microscopes might be used,but they cannot readily provide the submi-cron resolution needed at long workingdistance to align and inspect the tips. Scan-ning electron microscopy offers submicronresolution at long working distances andappears to be a preferable solution. Withresolution on the order of 20nm, the SEMmay be used to select the best tips, based ontheir image, in addition to providing infor-mation for positioning the aperture. A sin-gle SEM imaging from the side perpendicu-lar to the tip axis will make it possible toalign laterally relative to one axis of mo-tion. The second axis of motion, along theoptic axis of the SEM, could be scanned in

the field-electron emission mode until thesignal indicates that the tip is aligned. Al-ternatively, a second SEM column locatedat 90� to the first and in the plane of thespecimen can be used for alignment of thesecond axis. This adds additional cost andcomplexity, but should make the alignmentprocess very simple. These various scenar-ios will be tested both technologically andeconomically in the years to come.

MASS RESOLUTION

From its first inception, there has been in-terest in improving the mass resolution1 ofthe atom probe. Pulsing of the voltage ap-plied to the tip is used to achieve thepulsed-field evaporation required for time-of-flight measurements. The voltage pulsesapplied must have a subnanosecond risetime and a few nanoseconds duration. Thispulsing introduces a time-varying electricfield during ion acceleration. Because themass spectrometry in atom probe is basedon time of flight, mass resolution is im-proved by ensuring that ions of the samemass-to-charge ratio have a small spread intheir time of flight. Ions that evaporateearly in the pulse as the voltage is rising (a)begin their journey earlier than others, and(b) are accelerated by an electric field that isincreasing initially. After the peak in thevoltage pulse, ions that evaporate get alater start, experience a decreasing electricfield, and achieve a lower speed than theearly ions. They continually fall further be-hind in the race to the detector. Thus, themeasured timing resolution, and hence,mass resolution, suffers in voltage-pulsedatom probes due to aspects that are inher-ent to the pulsing. The spread in time offlight for ions of the same mass-to-chargeratio can be 1 to 3%, and so the mass resolu-tion can be as poor as 1 part in 50. One partin 500 is commonly considered adequatemass resolution for a large fraction of mate-rials analyses, and one part in 1000 or betteris preferred for a general-purpose instru-

1The term mass resolution will be used here instead of the technically more correct “mass resolving power” be-cause the former is widely used. The mass resolution cited here is defined as m/m.


ment. Better resolution is always desiredfor separating closely spaced isotopicpeaks.

If the timing electronics of a given instru-ment cannot fully resolve the time spreadof identical ions, then higher mass resolu-tion can be achieved by increasing theflight distance (longer total time of flight) atthe cost of a decreased solid angle in bothlinear AP and 3DAP. Conventional atomprobes may be fitted with devices that com-pensate for the kinetic energy spread of theions such as a Poschenrieder lens [72] or areflectron [9, 10]. The resulting mass reso-lution can be as high as 1 part in 2000.Good mass resolution for 3DAP is a topicthat received attention at a workshop on3DAP [73]. Cerezo et al. [74] have applied areflectron to a 3DAP, and achieved verygood mass resolution while passing a 3Dimage with minimal distortion. Further im-provement of the mass resolution, perhapswithout the need for a reflectron, remains agoal for 3DAP.

To realize high-mass resolution in aLEAP, it is important to either minimize orreduce (ideally to zero) the time spread oflike ions. Thus, two approaches to improv-ing mass resolution in LEAP are suggested:one based on minimization, and one basedon reduction. Both paths to higher massresolution are considered below. Note thatthe entire discussion below applies regard-less of whether the ions actually departduring the same pulse or during differentpulses.

Minimization of the Time Spread

Kelly et al. [22] looked at mass resolution ofthe LEAP with a first-order model based onion energies. Assuming that the energyspread could be taken as comparable forLEAP and conventional APFIM, their con-clusions were that secondary or post accel-eration of the ions could be used to achievea relative improvement in mass resolution.That is, with the extraction field as the pri-mary acceleration field, a second accelera-tion field is used to bring the ions up to ahigh total energy so that the small energy

differences between like ions becomes neg-ligible relative to the total. In Fig. 9(a), acontour plot of the mass resolution of aconventional atom probe is shown as afunction of flight path length and total volt-age. At longer flight times and higher totalvoltages, the mass resolution is maximizedat about one part in 200. In Fig. 9(b), thissame plot is shown for a LEAP with sec-ondary acceleration. In this case, the massresolution is significantly higher under allconditions. The calculated mass resolution

FIG. 9. Calculated mass resolution [22] of (a) a con-ventional atom probe, and (b) a LEAP with secondaryacceleration. These calculations are based on energyspread arguments assuming an extraction voltage of300 volts and a 1-ns timer resolution.


for LEAP is severely limited in this case bythe timer resolution of 1 ns, which has beenused in the calculation.

This first-order study of the potential forsecondary acceleration to improve the massresolution of LEAP was followed by bothfurther analytical study [75] and experi-mental work [76]. Bajikar et al. [75] devel-oped an analytical model that treated twoseparate sections of the LEAP indepen-dently. The results of this work are similarto the earlier work of Kelly et al. [22]. Theyshow explicitly, however, the effect oftimer resolution on mass resolution for agiven energy spread (in this case 0.5%). ALEAP with a 155mm flight length operat-ing at a total voltage of 10kV will be se-verely limited by timer resolution of 1 ns to

a mass resolution of about one part in 245[Fig. 10(a)]. With a timer resolution of 156ps, the calculated mass resolution wouldimprove to about one part in 600 [Fig. 10(b)].Commercially available multihit timershave historically offered 1-ns timing resolu-tion [77]. Faster timers are available withtime resolution on the order of 100 ps, butthey are not capable of properly encodingmultiple hits with a pair resolution on theorder of nanoseconds or better. Thus, de-velopment of a 100-ps multihit timer wouldbring significant benefits to LEAP technol-ogy. In addition, these results suggest thatthe time spread for like ions that is desir-able in a LEAP is on the order of 100 ps orless.

Bajikar et al. [75] found that it is best toaccelerate the ions to high speed as early aspossible after the extraction electrode. Asecondary acceleration electrode such atthe one shown in Fig. 1 could be used to ac-complish this effect. This work also consid-ered image magnification for the casewhere secondary acceleration is used. Theypointed out that the high magnification ofan atom probe, which operates in a pointprojection mode, is reduced by secondaryacceleration. Fortunately, it appears thatadequate image magnifications can be ob-tained in a LEAP. When the field distribu-tion between the extraction electrode andthe detector retains a high divergence, thishelps retain a high magnification.

Bajikar et al. [76] performed an experi-ment with secondary acceleration usingself-aligned apertures in a silicon array.This geometry was not ideal, because thesecondary acceleration field penetrated thetip/aperture region so that as the second-ary acceleration field was increased, thefield on the tip also increased. In addition,the capacitance of this geometry was highenough that it was not possible to applytemporally sharp pulses to the tip. None-theless, evidence was obtained that themass resolution was improved by applica-tion of a secondary acceleration field.

More recent work [78] has focused onhow to reduce the time spread of like ionsto a minimum. Because it is the time spread

FIG. 10. The mass resolution of a LEAP calculated byBajikar et al. [75]. The curve labeled S is the sum of thecontributions from the timer limitations (T) and theenergy-dispersion limitation (E). A timer resolutionof 1 ns is used in (top) and a timer resolution of 156 psis used in (bottom). The extraction electrode length isboth cases is 5mm, and the total flight length is 155mm.The extraction voltage was taken as 3.5kV, and a nor-malized energy spread of 0.5% is used.


that ultimately matters in time-of-flight-spectrometry, this is considered a better ba-sis for analyzing the performance of aLEAP. The time spread of ions originatesfrom at least four sources: (1) the timespread in the time of departure, (2) the timespread that develops during primary accel-eration of the ion to the extraction elec-trode, (3) the time spread that developsduring free flight inside the extraction elec-trode, and (4) the time spread that developsduring secondary acceleration toward thedetector. More segments could be added asnecessary to treat, for example, additionalcontrol electrodes. The interest in this workis to find how to minimize the overall timespread.

The total time that ions spend in the pri-mary acceleration zone in a LEAP is on theorder of 50 ps, so the time spread from seg-ment (2) will generally be negligible. Fur-thermore, a short extraction aperture willminimize the time spread that develops insegment (3) due to differences in speed ofions in the field-free zone. The shortest thatthe aperture can be and still perform as alocal aperture is the length where the fieldspillage from the secondary accelerationzone becomes significant. We have foundfor a 1�m aperture that about 10�m lengthis needed to ensure against field spillage.Nishikawa and Kimoto [2] found the sameresult in their initial study. Assuming a 1kVextraction voltage, a typical ion of mass-to-charge ratio of 30 will be traveling at 8 �104m/s through the extraction aperture,and will spend only about 100 ps in a10�m-long extraction aperture. This doesnot leave much time for time spread to de-velop. However, if the extraction apertureis 5mm long, then the resident time insidethe extraction aperture is 50 ns, and a timespread of several hundred picoseconds candevelop.

Construction of a very short extractionaperture might be accomplished by form-ing a dielectric material into a cone andthen coating only the apex region with ametal. A secondary acceleration electrodecould also be introduced along the remain-der of the extraction electrode.

The time spread that develops in seg-ment (4) is minimized by minimizing thetotal flight time after the extraction aper-ture. This says that the flight distance mustbe as short as possible, and the ions mustbe accelerated to as high energy as possible.However, remember that it is the relativetime spread, �t/t, that is to be minimized. Ifwe treat this segment as a region of con-stant acceleration throughout, then theequation of motion is [Eq. (3)]:

(3)

where L is the flight distance, a is the con-stant acceleration, t is time, and vo is the ini-tial speed of an ion entering the segment.The relative time spread is [Eq. (4)]:

(4)

For a given speed spread, �vo, and initialspeed, vo, the relative time spread is mini-mized in segment (4) by making the prod-uct, at, as large as possible. So, a high accel-eration and a long flight time (distance) aresuggested. The length of the flight path isconstrained by the desire to make the solidangle of the position-sensitive detector aslarge as possible. For a given timer resolu-tion, the acceleration magnitude should beas large as possible such that the actualtime spread is matched to the timer resolu-tion. For the position-sensitive detector de-scribed below, a flight length of about200mm is chosen.

Finally, there is the question, What is theoptimal voltage pulse shape? In the idealcase, ions would all evaporate at the sameinstant and would experience a constantprimary acceleration field. This suggeststwo extreme cases: a short impulse, and along square pulse. In the first case, a sharpvoltage impulse [100 ps full-width half-maximum (FWHM)] can result in a shortspread in time of departure (about 33 ps).(For a 100-ps FWHM impulse, the rise timeand fall time are each about 100 ps. Becauseions evaporate within only the top 3% ofthe applied field, for a pulse fraction of20%, this corresponds to the top one-sixthof the pulse. In time, this is one-sixth of the

L 1 2⁄ at2 vot constant+ +=

δtt-----

δvo–

at vo+----------------=


rise and one-sixth of the fall or 33-ps total.)However, ions will depart over a range ofvoltages near the peak. Furthermore, pri-mary acceleration of the ions takes onlyabout 50 ps in a LEAP. Thus, the extractionvoltage is still changing rapidly during pri-mary acceleration and ions with differentdeparture times are accelerated differingamounts. Thus, in this first case, the time-of-departure spread is low but the speedspread is high.

In the second case, a square pulse wouldbe such that ions could evaporate any timeduring the pulse, but they would all see thesame constant acceleration field. In thiscase, the time-of-departure spread is largebut the speed spread is small. However, aperfectly square pulse does not exist, andthe ions that depart on the rise or the fall ofthe voltage will be accelerated differingamounts. Furthermore, if we are trying tokeep the total time spread to about 100 ps,then a longer time-of-departure spread willdefeat the whole purpose. From this stand-point, it appears that it is best to use a fastimpulse for voltage pulsing the LEAP anduse secondary acceleration to minimize theoverall relative time spread and thus maxi-mize the mass resolution. Voltage pulsersare available that can deliver sub-200-psFWHM pulses of up to several hundredvolts (S. Fulkerson, Lawrence LivermoreNational Laboratory, private communica-tion, 1996) [79, 80]. Note that many com-mercially available pulsers are designed todeliver very clean pulses with no ringing orother structures. For pulsing a LEAP, animpulse need only be clean on its top one-third or less of the pulse height. Fasterpulses can be produced when it is not nec-essary to design against pre- or postringingof the pulse.

Reduction of the Time Spread

Once a finite time spread exits between twoions, this time spread can be reduced onlyby changing (a) their relative speeds, or (b)their relative path lengths to the detector.Poschenrieder lenses and reflectrons oper-ate on both (a) and (b) simultaneously.

Nishikawa [6] has used a reflectron on hisSAP to reduce the time spread of the ions.

Because there is a field-free zone early inthe flight path of a LEAP (inside of the ex-traction aperture), there is a possibility forreducing the time spread with a tailoredpulse shape. If an impulse were to evapo-rate two ions at different start times, thenthe first ion has a head start and a higherspeed. If the primary acceleration field iscontinuously increasing, then once the firstion enters the field-free zone of the aper-ture, the second will be accelerated tohigher energy without affecting the firstion. If the voltage rise is fast enough, thiswould boost the energy of the second ionrelative to the first, and provide time focus-ing at some point in the path. Ideally, theions should be time focused at the detector.The challenge here is to achieve this timefocusing with realistic aperture lengths andpulse shapes. Because the transit time ofthe ions to the aperture is on the order of 50ps, this scenario may only be feasible if theaperture, or a separate correction field, ismoved further away to increase the timeavailable for the rising pulse to act on theions. Furthermore, it is not clear that thiscan be made to work on more than one iontype at a time. If this time focusing can beachieved, then the other approaches to im-proving mass resolution become less im-portant.

Data acquisition rate Higher data collectionrates are of interest for atom probes be-cause the sensitivity of the technique is di-rectly improved as the number of atomscollected increases. Because the pulse mag-nitude in a conventional atom probe needsto be on the order of 20% of the standingvoltage [81a], a pulse generator must pro-duce on the order of 3 to 4kV pulse magni-tudes. Historically, the prevalent device forproducing these pulses with the requiredperformance is a mercury-wetted reedswitch. Reed switches are mechanical de-vices that produce up to 7kV [82], 1 ns risetime, 5 ns-long pulses at frequencies up toabout 200 pulses per second. Recently, Be-hlke switches [83] have been used in com-


mercial high voltage solid-state pulsers [84,85]. These latter pulsers can operate at a1.7-ns rise time, 8-ns long pulses at fre-quencies up to 2 � 103 pulses per second.Because ion flight times do not normallyexceed 20�s (2�s in 3DAP), atom probescould be operated at much higher speeds,like 105 pulses per second, if higher repeti-tion rate were available.

Because the voltage needed to create thehigh field for evaporation in a LEAP ismuch smaller than in a conventional atomprobe, the pulse voltage required for a 20%pulse fraction is also much smaller. Withthe technology available today, as the volt-age magnitude of a pulse decreases, it ispossible to make the temporal length of thefastest pulse smaller and the maximumrepetition rate higher. Solid-state voltagepulsers are commercially available (S. Fulk-erson, Lawrence Livermore National Labo-ratory, private communication, 1996) [79,80] that can produce pulses of several hun-dreds of volts with a 100 ps rise times, a 100ps FWHM, and repetition rates up to 105

pulses per second. Note that some attentionmust be paid to minimize the capacitancebetween the tip and the extraction elec-trode if 100-ps impulses are to be realizedin practice at the tip. At repetition rates of105 per second and greater, the power sup-plies for solid-state voltage pulsers becomelarge and expensive. It is reasonable to ex-pect, however, that solid-state pulsers canbe pushed to provide pulsing rates up to106 per second if necessary (S. Fulkerson,Lawrence Livermore National Laboratory,private communication, 1996).

Detector technology In a conventional atomprobe, the needle-shaped specimen is ma-nipulated on a goniometer through twoaxes so that, for example, a feature of inter-est in a FIM image may be rotated into co-incidence with the probe hole for analysis.The requirements on eucentricity of this go-niometer are low because movement of thespecimen apex on the order of a fraction ofa millimeter is insignificant. If we considerthat the extraction aperture in a LEAP is onthe order of 1�m in diameter, then it is

clear that an unrealistic degree of eucentric-ity would be required in a LEAP stage toallow real-time rotation of the specimenrelative to the aperture. There are four al-ternatives apparent: (1) use a multistepprocess where the tip is retracted, rotated,then repositioned in the aperture; (2) ro-tate/translate the entire aperture and tipassembly as a unit to sweep the imageacross the detector; (3) rotate/translate thedetector to sweep the image across the de-tector; or (4) make the detector large enoughto capture the entire field ion or field-des-orption image. Alternatives 1 through 3 donot require any new technology, and can beimplemented based simply on which is thebest option in a given circumstance. Alter-native 4 offers several significant advan-tages if it can be realized. They are exploredbelow.

Making a detector large enough meansthat it must have enough pixels to fullyrecord the field ion and field-desorptionimages. If we assume that a wide-anglefield ion image is 500 atoms across, then adetector with about 2,000 � 2,000 pixels isrequired to record such an image with highquality. There are several detector technol-ogies that can reach this level of resolutionincluding charge-couple devices (CCDs),and some single-particle detectors [86–89].A detector based on delay-line technologyhas the potential to provide this level ofresolution. One advantage of delay-linetechnology is that the same type of timerused in the position determinations canalso be used for measuring ion times offlight. The projected performance of a de-tector under construction by the lead au-thor is 3 � 106 properly encoded hits persecond at 2,000 � 2,000 resolution. Thehigh data rate is needed if the large numberof pixels is to be populated with data in ashort amount of time. It can be made to anyphysical size from 75 � 75mm active areato 200 � 200mm active area.

Consider the advantages of this detectorfor a LEAP. First, the high data rate meansthat it can be used to record real-time fieldion images in addition to field desorptionimages. This greatly simplifies the design


of the instrument by eliminating the needto switch specimen positions between twoor more detectors. Furthermore, becausethere is no need to rotate the specimen intoa final analysis position, a simple fixedmounting can be used. This greatly facili-tates the connection of cryogenic cooling tothe specimen (see below). The specimencould be mounted directly on the cold fin-ger of the cryo-cooler if a vibration-freecooler is available (see below).

The high data rate of the detector makesit possible to consider high pulsing rates onthe field evaporation process, which, asmentioned above, should be possible withLEAP. With the high resolution of the de-tector, a large swath may be recordedthrough the sample in the field desorptionmode. A large swath diameter means highsensitivity per atomic plane, high sensitiv-ity to localized 3D concentrations, and highefficiency of specimen utilization. The com-plexity in this case goes into the detectorand the rest of the instrument is, therefore,greatly simplified.

Cryo-cooling of the specimen To freeze diffu-sive movement of atoms on the surface of aspecimen, it is necessary to cool the tip tocryogenic temperatures in an atom probe[81b]. The resolution and contrast in FIMimages are both markedly improved at lowtemperatures relative to room temperature.Different materials require different tem-peratures for best operation. Aluminum al-loys typically require very low tempera-tures like 20 to 40K, while steels aretypically analyzed at 60K. A LEAP shouldbe no different in this requirement. It maybe slightly easier to reach the low tempera-tures with a LEAP geometry because thespecimen is planar, but this is countered bythe fact that the nearby extraction apertureis at room temperature.

Vacuum requirements for LEAP LEAP, like theconventional atom probe, is an ultrahighvacuum technique. Because surface atomsare analyzed, it is necessary to keep thatsurface clean from contamination. Basepressures of low 10 9Pa (low 10 11mbar)are desirable to keep tramp gas atoms off

the specimen. Actually, during a typicaloperation, an imaging gas like neon or ar-gon is admitted into the main chamber to apressure of about 10 3Pa (10 5mbar) forfield ion imaging. Upon completion of theimaging, the imaging gas is pumped out toa pressure of about 10 7Pa (10 9mbar). Thislatter pressure is best for atom probe analy-sis because it has been found that the fieldneeded to evaporate atoms decreases slightlyas the pressure rises from 10 9 to 10 7Pa(10 11 to 10 9mbar). The lower field helpsreduce the occurrence of catastrophic me-chanical specimen fractures.

In a LEAP, these same considerations ap-ply for the most part. There are two possi-ble exceptions. First, if the data rates of aLEAP are much higher than conventionalatom probes, then the relative rate of con-tamination of the data by tramp elementsin the vacuum is decreased proportion-ately. By this argument the base pressurecould afford to be higher for a given con-tamination rate, but it seems more prudentto keep the base pressure low and realize alower contamination rate in the data. Sec-ond, the shorter shank length of tips andthe different field distribution along theshank due to the local extraction apertureshould lead to different mechanical stressdistribution in microtip specimens relativeto needle-shaped specimens in a conven-tional atom probe. Whether these stressesare lower or higher has yet to reported inthe literature. If they are lower, then thepressure during analysis could be lower,perhaps at base pressure. The shorter shanklength of microtips will mean a lower prob-ability of a fatal mechanical flaw in the tip.At this point, however, it remains to beshown whether microtip specimens in aLEAP have a higher or lower propensityfor mechanical failure than needle-shapedspecimens in a conventional atom probe.The conclusion at this time is that a LEAPshould be designed as an ultrahigh vacuumsystem with the same considerations as aconventional atom probe.

Mechanical vibration Conventional atom probesare insensitive to mechanical vibration.


This is not the case for LEAPs, however.The relative vibration of the tip and the ex-traction electrode has the potential to alterthe magnitude of the field at the tip, alterion trajectories, and in the extreme, makeoperation impossible. Indeed, one methodof pulsing the field evaporation wouldhave the tip mechanically oscillating alongthe tip axis direction such that the peakfield encountered is sufficient to induceevaporation. Furthermore, in any three-di-mensional atom probe, the field distribu-tion must be axially symmetric along thetip axis. Otherwise, asymmetric fields willlead to anisotropic magnification of the de-sorption image. It is, therefore, essentialthat relative vibration of the tip and aper-ture, which would introduce field asymme-tries, be eliminated in the instrument.

The main concerns are isolation from am-bient vibration sources and prevention ofvibration from vacuum pumps and cryo-coolers. There are three prominent technol-ogies for isolation of the vacuum chamberfrom vibration: pneumatic systems [90, 91],mechanical systems [92], and active systems[93]. Pneumatic isolation systems are proba-bly adequate for the job. They have beenused extensively with electron microscopesthat achieve subnanometer resolution. Themain advantages of mechanical systems arethat they achieve greater isolation of thesystem from ambient and they never needservicing. Mechanical systems can be moreexpensive than pneumatic systems, how-ever. Active systems are the most complexand tend to be the most expensive solution.Their performance is excellent in general.

The stage should be designed to be insensi-tive to external vibration to the maximum ex-tent possible. The fixed specimen design men-tioned above is desirable from this standpointas long as the cryogenic cooling system doesnot introduce vibration. In general, the aper-ture and tip assembly should be mounted onthe same basic mount assembly so that theyvibrate as a unit and not differentially.

Magnetically levitated turbomolecularpumps have been used on electron micro-scopes for over a decade, and should be ad-equate for this application. Ion pumps on

the main chamber can be used, but theyshould be coupled with a throughputpump such as a turbomolecular pump topump the inert gases often used for fieldion imaging.

Convenient cryogenic cooling of thespecimen without vibration is perhaps thebiggest challenge. Multistage helium com-pressor cryopumps are able to cool to about14K, and are the standard in conventionalatom probe technology. However, theyproduce far too much vibration for theLEAP application. Vibration-isolated he-lium compressor cryopumps have beenbuilt by individual research teams fromcommercially available components (R.Dykhuizen, Sandia National Laboratory,private communication, 1998; and H-O An-drén and A. Kvist, Chalmers University ofTechnology, private communication, 1998).This solution looks promising for this ap-plication. It requires that the cryohead bemounted independently of the chamber,and compliant connections to the vacuumchamber are used. Heat is transferred byhelium gas between the vibrating cryoheadand the vibration-isolated specimen holder.A complete commercial system of this typeis available from Advanced Research Sys-tems [94].

In those cases where liquid helium isavailable, gaseous helium boil-off from thehelium bottle can be directed through thespecimen holder. This approach createsminimal vibration and excellent cooling,but requires a supply of liquid helium. Sim-ilarly, liquid neon cooling is a straightfor-ward and inexpensive method of coolingthat will reach about 27K. Liquid nitrogenis even less expensive, but it will reach only80K, which is not a sufficiently low temper-ature for most AP analyses.

Electromagnetic fields Conventional atom probesare insensitive to low-level electromagneticfields in the environment. This is not neces-sarily the case for LEAPs, however, if a SEMis used to image the tip and aperture region.A typical SEM is sensitive to time-varyingelectromagnetic fields that are on the orderof a few milligauss or greater in magnitude.


Depending on details of the shielding anddesign, a LEAP using SEM imaging willmost likely have similar constraints.

PROSPECTS FOR FUTURE DEVELOPMENTS

In addition to instrumental developments,there are several specimen geometry devel-opments particular to a LEAP that areworth considering.

BLUNT TIPS

In a conventional APFIM, the specimenmust have a radius of curvature at the apexthat is less than about 100nm to achieve thehigh electric field needed for field evapora-tion. The electric field on the tip is generallyrecognized as being inversely proportionalto the apex radius of curvature [81c]. Witha local electrode, the dependence of thehigh field on the radius of curvature at theapex is reduced, and the high electric fieldscan be produced even at modest voltages.Not only does this make it possible to con-tinue analysis of tips to a blunter conditionto get more material removed, but it alsomakes it possible to consider utilizing blunttips purposefully, as illustrated in Fig. 11.

For example, using numerical modeling,Bajikar [95] showed that within the limitsof expected LEAP operation, a tip radius ofcurvature up to 0.5�m will still allow elec-tric fields high enough for evaporation tobe achieved.

The consequences are important for workon interfaces and other 2D structures. At a0.5�m radius, the field desorption imagewill contain about 5,000 atoms across thediameter. This corresponds to about 25 mil-lion atoms per atomic plane. Note that atthis image size, the detector resolutionwould not be sufficient to detect atomicspacings laterally (x-y directions in Fig. 1).Furthermore, a reconstruction of the three-dimensional image over such a wide areamay not achieve atomic resolution in thelongitudinal direction (z direction in Fig. 1)either. However, this tradeoff of resolutionfor sensitivity can be beneficial in many ap-plications. With this sort of 3D image, thesensitivity of the technique would exceedparts per million per atomic plane. A goodexample of where this sensitivity is neededcan be found in the semiconductor industrytoday. The dopants in shallow implantstructures are only about 100 atom layersdeep. The total number of atoms detectablein a volume of this depth is about 2.5 bil-lion. If we take 10 atoms as the minimum

FIG. 11. Schematic illustration of the concept of using blunt tips to realize a greater number of atoms analyzed peratomic layer.


number that represents a statistically sig-nificant sample, then this translates into avolume sensitivity of 4 atomic parts per bil-lion or 2 � 1014 dopant atoms per cm3 of sil-icon. This is a very low concentration.

Another possible application of blunt tipswill be for analysis of submicron structures.As device size diminishes, characterizationof these structures, especially in three di-mensions, becomes a harder problem. 3DAPcould be used to study such structures buta lateral image size of hundreds of atomswould make it very difficult to obtain thedata. However, large sections of a submi-cron device can be contained within a blunttip. If tip formation is accomplished at aspecific location as mentioned above, thentargeted regions of a very large integratedcircuit could be analyzed. These regionsmight be either deliberately fabricated testsections on a wafer or sections of a devicethat has been shown to possess a defect.

SPECIMENS OF LOW ELECTRICAL CONDUCTIVITY

Except in a few cases, atom probe analysishas been performed almost exclusively onspecimens that have a high electrical con-ductivity. This is primarily because the

high-voltage pulse that initiates field evap-oration must travel down the length of aslender needle. In poor electrical conduc-tors, the pulse is heavily attenuated beforeit reaches the specimen apex and it is inad-equate to cause field evaporation. Twomethods have been used to overcome thisproblem for poor conductors. In the first, apulsed-laser atom probe was developed byKellogg and Tsong [96] to thermally pulselow-conductivity materials like silicon. Inthe second, specimens of poor electricalconductors have also been treated with anelectrically conducting coating like carbon[97, 98]. During analysis, the coating is fieldevaporated away from the analyzed region,and generally does not influence the re-sults. This latter approach should alsowork for LEAP specimens. Figure 12 illus-trates this concept schematically. After aplanar microtip specimen is prepared, athin coating of an electrical conductor canbe applied to the top surface. One potentialadvantage of the planar microtip in this re-gard is that it may be easier to achieve thecoating on any standard coating system,but this is a minor point. More importantly,because the thickness of a planar microtipspecimen is on the order of a millimeter orless, the voltage pulse through the speci-

FIG. 12. Schematic illustration of the concept of coating an electrically insulating material with a conductive layerso that LEAP analysis may be performed.


men will be much less attenuated than in aneedle-shaped specimen. Thus, for thesereasons, it should also be possible to ana-lyze uncoated specimens in LEAP that arelower in electrical conductivity than maybe analyzed in a conventional APFIM. Han-dling of weak or brittle materials shouldalso be easier with planar microtip speci-mens than with needles.

LEAP AS A METROLOGY TOOL

For an instrument to qualify as a metrologytool, it is necessary that analyses be com-pleted in a time frame that allows the infor-mation to be used as feedback for furtherprocessing. In a conventional APFIM or3DAP, no methodology has yet been devel-oped where this is a realistic possibility.Complete analyses that do not destroy theoverall viability of an entire wafer must beperformed in a matter of hours or less. Itappears, however, that a LEAP could beconfigured to suit this task. An FIB may beused to create microtips from specific loca-tions on a wafer. These locations may betest sites that have been engineered into thewafer design. Facilities for handling andcooling wafers in ultrahigh vacuum havebeen developed by the semiconductor pro-cessing industry. A FIB could be incorpo-rated in the metrology station. Dependingon the information that is sought, 3D im-ages could be recorded and analyzed fromseveral sites on the wafer within the sev-eral-hour constraint. There would be con-siderable development work needed tomake this a routine task, but it is no moredaunting than similar metrology tasks thatare currently performed by analytical in-struments.

CONCLUSIONS

All of the scientific bases of LEAPs havebeen established. Although dedicated sys-tems have not yet been completed, LEAPhas the potential to reach:

1. pulsed evaporation rates of 106 pulsesper second;

2. mass resolution of one part in 500 or bet-ter; and

3. billion atom, three-dimensional images.

LEAP makes innovative new specimengeometries possible that could push thesensitivity limits of two-dimensional struc-tures to parts per million per atomic plane.Part per billion analyses should be possiblein volumes of 1-�m diameter by hundredsof atomic layers thick. With this type ofperformance, LEAPs will compete favor-ably with other analytical techniques likesecondary ion mass spectrometry, bulkchemical analysis, electron probe microanal-ysis, and scanning auger microprobe.

LEAP can play a crucial role in character-ization of materials with nanometer-scalecompositional structure in three dimen-sions. In a user-friendly package, a LEAPcould become a routine analytical instru-ment. The semiconductor and thin-film in-dustries especially need a means for imag-ing structures in three dimensions at theatomic scale. 3D LEAP may be the onlytechnique that can reach that goal in thenear future.

One of the authors (T.F.K.) has been supported bygrant number DMR 97-03932 from the NationalScience Foundation under the supervision of Dr.Lorretta Hopkins, while the other author (D.J.L.)was supported by the Division of MaterialsSciences, U.S. Department of Energy, undercontract DE-AC05-96OR22464 with LockheedMartin Energy Research Corp. We would like tothank Richard Martens for his work inpreparation of the images in Fig. 6, and MichaelCoy and Vinayik Dravid of NorthwesternUniversity for the FIB work shown in Fig. 7. Wewould like to thank Dr. M. K. Miller for helpfuldiscussions.

References

1. National Science Foundation Panel Report onAtomic Resolution Microscopy, (April 1993).

2. O. Nishikawa and M. Kimoto: Toward a scanningatom probe-computer simulation of electric field.Appl. Surface Sci. 76/77:424–430 (1994).

3. O. Nishikawa, Y. Ohtani, K. Maeda, M. Masafumi,


and K. Tanaka: Development of the scanning atomprobe. Mater. Char. 44:29–57 (2000).

4. O. Nishikawa, M. Kimoto, and Y. Ishikawa: Devel-opment of a scanning atom probe. J. Vac. Sci. Tech-nol. B 13:599–602 (1995).

5. O. Nishikawa, T. Sekine, Y. Ohtani, K. Maeda, Y.Numada, and W. Watanabe: Atomic investigationof individual apexes of diamond emitters by ascanning atom probe. J. Vac. Sci. Technol. B 16:836–840 (1998).

6. O. Nishikawa, T. Sekine, Y. Ohtani, K. Maeda, Y.Numada, and W. Watanabe, M. Iwatsuki, S. Aoki,J. Itoh, and K. Yamanaka: Development of a scan-ning atom probe and atom-by-atom mass analysisof diamonds. Appl. Phys. A 66:S11–S16 (1998).

7. O. Nishikawa, K. Maeda, Y. Ohtani, W. Watanabe,K. Tanaka, T. Sekine, M. Iwatsuki, S. Aoki, J. Itoh,and K. Yamanaka: Atomic level analysis of elec-tron emitter surfaces by the scanning atom probe.Appl. Surface Sci. (in press).

8. O. Nishikawa, M. Watanabe, Y. Ohtani, K. Maeda,K. Tanaka, T. Sekine, and J. Itoh: Atomic-by-atomanalysis of micro tip emitter surfaces by the scan-ning atom probe. J. Vac. Sci. Technol. B (1999).

9. V. I. Karataev, B. A. Mamirin, and D. V. Shmikk: Anew principle of focusing of ion packets withtime-of-flight mass-spectrometers. Sov. Phys. Tech-nol. Phys. 16:1177 (1972). [Translated from Zhurnal-Tekhnicheskoi-Fiziki 41:1498–1501 (1971)].

10. B. A. Mamirin, V. I. Karataev, D. V. Shmikk, andV. A. Zagulin: The mass-reflectron, a new non-magnetic time-of-flight mass spectrometer withhigh resolution. Soviet-Phys. JETP 37:45–48 (1973).[Translated from: Zhurnal-Eksperimental’noi-i-Teo-reticheskoi-Fiziki 64:82–89 (1973)].

11. W. R. Grove: On some anomalous cases of electri-cal decomposition. Philos. Mag. 5:203–209 (1853).

12. O. Auciello: Historical overview of ion-inducedmorphological modification of surfaces. In. Ion Bom-bardment Modification of Surfaces, O. Auciello and R.Kelly, eds., Elsevier, Amsterdam, pp. 1–25 (1984).

13. O. Auciello: Ion interaction with solids: Surfacetexturing, some bulk effects, and their possible ap-plications. J. Vac. Sci. Technol. 19:841–867 (1981).

14. M. J. Witcomb: The development of ion bombard-ment surface structures on stainless steel. J. Mater.Sci. 9:551–563 (1974).

15. B. Navinsek: Sputtering—Surface changes in-duced by ion bombardment. Prog. Surface Sci. 7:49–70 (1976).

16. W. Hauffe: Production of microstructures by ionbeam sputtering. In Sputtering by Particle Bombard-ment III, R. Behrisch and K. Wittmaack, eds.,Springer Verlag, Berlin, pp. 318–333 (1991).

17. D. Ghose and S. B. Karmohapatro: Topography ofsolid surfaces modified by fast ion bombardment.Adv. Electron. Electron. Phys. 47:73–154 (1978).

18. A. D. G. Stewart and M. W. Thompson: Microto-

pography of surfaces eroded by ion-bombard-ment. J. Mater. Sci. 4:56–60 (1969).

19. J. Lindhard: Influence of crystal lattice on motionof energetic charged particles, Kgl. Danske Viden-skab. Selskab, Mater.-Fys. Medd. 34:14–88 (1965).

20. H. Wilson and M. W. Kidd: A study of cones de-veloped by ion-bombardment of gold. J. Mater.Sci. 6:1362–1366 (1971).

21. D. J. Larson, C.-M. Teng, P. P. Camus, and T. F.Kelly: Fabrication of microtips on planar speci-mens. Appl. Surface Sci. 87/88:446–452 (1995).

22. T. F. Kelly, P. P. Camus, D. J. Larson, L. M. Holz-man, and S. S. Bajikar: On the many advantages oflocal electrode atom probes. Ultramicroscopy 62:29–42 (1996).

23. J. Linders, H. Niedrig, and H. Koch: Developmentof microcones induced by contamination lithogra-phy. Nucl. Instrum. Methods Phys. Res. B 13:309–312(1986).

24. D. J. Larson, M. K. Miller, R. M. Ulfig, R. J. Matyi,P. P. Camus, and T. F. Kelly: Field ion specimenpreparation from near-surface regions. Ultrami-croscopy 73:273–278 (1998).

25. J. A. Liddle, A. Norman, A. Cerezo, and C. R. M.Grovenor: Pulsed laser atom probe analysis of ter-nary and quaternary iii-iv epitaxial layers. J. Phys.49-C6:509–514 (1988).

26. A. J. Little: The investigation of MOCVD layers ofcompound semiconductors by atom probe. D.Phil. Thesis, University of Oxford (1989).

27. Y. Akama, E. Nishimura, A. Saki, and H. Mu-rakarni: New scanning tunneling microscopy tipfor measuring surface topography. J. Vac. Sci.Technol. A8:429–433 (1990).

28. T. F. Kelly and R. J. Martens: unpublished re-search. (1998).

29. A. Kvist, H.-O. Andrén, and L. Lundin: A speci-men preparation technique for atom probe analy-sis of the near-surface region of cemented car-bides. Appl. Surface Sci. 94/95:356–361 (1996).

30. A. J. Melmed: The method of sharp-pointedshards. J. Phys. 50-C8:547–548 (1989).

31. P. P. Camus, R. D. Shull, and A. J. Melmed:APFIM analysis of composite magnetic thin films.J. Phys. 50-C8:343–345 (1989).

32. K. Hono, N. Hasegawa, S. S. Babu, H. Fujimori,and T. Sakurai: Atom probe analysis of a nano-crystalline Fe-C-Ta sputtered soft magnetic thinfilm. Appl. Surface Sci. 67:391–397 (1993).

33. A. Pundt and C. Michaelsen: APFIM study of thecompositional inhomogeneity of sputtered Co-Crmagnetic thin films. Appl. Surface Sci. 87/88:264–270 (1995).

34. A. K. Petford-Long, A. Cerezo, and J. M. Hyde:Atom probe analysis and modeling of interfaces inmagnetic multilayers. Ultramicroscopy 47:367–374(1992).

35. T. Al-Kassab, M.-P. Macht, V. Naundorf, H. Wol-


lenberger, S. Chambreland, F. Danoix, and D.Blavette: Characterization of sputter-depositedmultilayers of Ni and Zr with the APFIM/TAP.Appl. Surf. Sci. 94/95:306–312 (1995).

36. N. Hasegawa, K. Hono, R. Okano, H. Fujimori,and T. Sakurai: A method for preparing atomprobe specimens for nanoscale compositionalanalysis of metallic thin films. Appl. Surface Sci. 67:407–412 (1993).

37. A. R. Waugh, S. Payne, G. M. Worrall, and G. D. W.Smith: In situ ion milling of field ion specimensusing a liquid metal ion source. J. Phys. 45-C9:207–209 (1984).

38. M. J. Vasile, D. A. Grigg, J. E. Griffith, E. A.Fitzgerald, and P. E. Russel: Scanning probe tipsformed by focused ion beams. Rev. Sci. Instrum. 62:2167–2171 (1991).

39. D. J. Larson, D. T. Foord, A. K. Petford-Long, T. C.Anthony, I. M. Rozdilsky, A Cerezo, and G. D. W.Smith: Focused ion beam milling for atom probefield ion microscopy specimen preparation: Pre-liminary experiments. Ultramicroscopy 74:147 (1998).

40. D. J. Larson, D. T. Foord, A. K. Petford-Long, A.Cerezo, and G. D. W. Smith: Focused ion-beamspecimen preparation for atom probe field-ion mi-croscopy characterisation of multilayer film struc-tures. Nanotechnology 10:45 (1999).

41. D. J. Larson, D. T. Foord, A. K. Petford-Long, H.Liew, M. G. Blamire, A. Cerezo, and G. D. W.Smith: Field-ion specimen preparation using fo-cused ion-beam milling. Ultramicroscopy 79:287(1999).

42. D. J. Larson, A. K. Petford-Long, A. Cerezo, G. D. W.Smith, D. T. Foord, and T. C. Anthony: Three-dimensional atom probe field-ion microscopy ob-servation of Cu/Co multilayer film structures.Appl. Phys. Lett. 73:1125–1127 (1998).

43. D. J. Larson, R. J. Martens, T. F. Kelly, K. M. Miller,and N. Tabat: Atom probe analysis of planar mul-tilayer structures. J. Appl. Phys. (in press).

44. D. J. Larson, A. K. Petford-Long, A. Cerezo, andG. D. W. Smith: Three-dimensional atom probestudies of metallic multilayers. Acta Mater. (inpress).

45. D. J. Larson and D. T. Foord: unpublished re-search (1998).

46. C. A. Spindt: A thin-film field emission cathode. J.Appl. Phys. 39:3504 (1968).

47. C. A. Spindt, I. Brodie, L. Humphrey, and E. R.Westerberg: Physical properties of thin-film fieldemission cathodes with molybdenum cones. J.Appl. Phys. 47:5248–5263 (1976).

48. K. K. Chin and R. B. Marcus: Field emitter tips forvacuum microelectronic devices. J. Vac. Sci. Tech-nol. A8:3586–3590 (1990).

49. L. Yun-Peng and Z. Mao-Sheng: The boundary el-

ement algorithm for the electric field of the Spindtdevice. Surface Sci. 246:75–78 (1991).

50. I. Brodie and C. A. Spindt: Vacuum microelectron-ics. Adv. Electron. Electron Phys. 83:1 (1992).

51. C. A. Spindt: Microfabricated field-emission andfield-ionization sources. Surface Sci. 266:145–154(1991).

52. J. A. Panitz: The 10 cm atom probe. Rev. Sci. In-strum. 44:1034–1038 (1973).

53. J. A. Panitz: Imaging atom-probe mass spectros-copy. Prog. Surface Sci. 8:219–262 (1978).

54. J. M. Sarrau, F. Danoix, B. Deconihout, M. Bouet,A. Menand, and D. Blavette: The Rouen energy-compensated atom probe. Appl. Surface Sci. 76/77:367–373 (1994).

55. S. J. Sijbrandij, A. Cerezo, and G. D. W. Smith: Im-provements in the transmission of voltage evapo-ration pulses in the atom probe. Appl. Surface Sci.87/88:414–420 (1995).

56. M. Huang, R. A. D. Mackenzie, T. J. Godfrey, andG. D. W. Smith: Characterization of gridded fieldemitters. J. Vac. Sci. Technol. B 12:713–716 (1994).

57. B. Deconihout and D. Blavette: Presented at 19983DAP Workshop at Oak Ridge National Labora-tory, Oak Ridge, TN (July 17, 1998).

58. R. Smith and J. M. Walls: Ion trajectories in thefield-ion microscope. J. Phys. D: Appl. Phys. 11:409–419 (1978).

59. E. W. Müller and S. V. Krishnamurty: Energy defi-cits in pulsed field evaporation and deficit com-pensated atom probe designs. Rev. Sci. Instrum. 45:1053 (1974).

60. S. V. Krishnamurty and E. W. Müller: Prematurefield evaporation in an atom probe. Rev Sci. In-strum. 45:1049–1052 (1974).

61. S. S. Bajikar, T. F. Kelly, and P. P. Camus: Electro-static analysis of local-electrode atom probes.Appl. Surface Sci. 94/95:464–471 (1996).

62. M. S. Mousa: Characteristics of tungsten substratewith AI2O3 coatings under U.H.V. conditions.Vacuum 38:835–838 (1988).

63. M. S. Mousa and M. Al Sharé: Study of the MgO-coated W emitters by field emission microscopy.Ultramicroscopy 79:195 (1999).

64. M. S. Mousa: Cold cathode field emission usingboth Al-resin and Au-resin coatings on tungstensubstrate. Surface Sci. 231:142–148 (1990).

65. S. S. Bajikar, T. F. Kelly, C.-M. Teng, and P. P. Ca-mus: Microfabrication of extraction electrodes forlocal-electrode atom probes. Appl. Surface Sci. 87/88:438–445 (1995).

66. T. T. Gribb, T. K. Kelly, M. Coy, and V. Dravid:Unpublished research (1998).

67. D. J. Larson, M. Huang, and D. T. Foord: Unpub-lished research (1998).

68. Burleigh Instruments, Inc.: http://www.burleigh.com/.


69. Schneeburger Linear Technololgy: http://www.schneeberger.com/default.htm.

70. Omicron Vakuumphysik GmbH at http://www.omicron-instruments.com/products/ms5/.

71. Micro Pulse Systems, Inc . : ht tp ://www.micropulsesystems.com/index.html.

72. W. P. Poschenrieder: Multiple-focusing time offlight mass spectrometers. I. TOFMS with equalmomentum acceleration. Int. J. Mass Spectrom. IonPhys. 6:413–426 (1971).

73. Summary of discussion at the first workshop onthree-dimensional atom probe analysis—Itspresent and future, held at the 40th Annual Inter-national Field Emission Symposium, Nagoya, Ja-pan. Appl. Surface Sci. 76/77:xxi–xxvi (1994).

74. A. Cerezo, T. J. Godfrey, S. J. Sijbrandij, G. D. W.Smith, and P. J. Warren: Performance of an en-ergy-compensated three-dimensional atom probe.Rev. Sci. Instrum. 69:49–58 (1998).

75. S. S. Bajikar, D. J. Larson, T. K. Kelly, and P. P. Ca-mus: Magnification and mass resolution in local-elec-trode atom probes. Ultramicroscopy 65:119–129 (1996).

76. S. S. Bajikar, D. J. Larson, P. P. Camus, and T. K.Kelly: Mass resolution enhancement in local-elec-trode atom probes: A study using field emitter ar-rays. J. Phys. IV C5-6:303–308 (1996).

77. LeCroy Corporation at http://www.lecroy.com/lrs/default.htm. See model 4208.

78. T. T. Gribb and T. F. Kelly: Unpublished research(1998).

79. Avtech Electrosystems: www.avtechpulse.com.80. Picosecond Pulse Labs: www.picosecond.com.81. M. K. Miller, A. Cerezo, M. G. Hetherington, and

G. D. W. Smith: Atom Probe Field Ion Microscopy,Oxford University Press, UK (a) p. 89 (1996); (b)pp. 72–73; (c) p. 2.

82. Part number RP18803 G2, made by C. P. ClareCorporation, http://www.cpclare.com.

83. Distributed in the US by Eurotek, Inc., http://www.eurotek.com/

84. Model HVP30, Kindbrisk Ltd., 8 Tigarsley Road,Eynsham, Oxfordshire OX8 1PP, UK.

85. Cameca S. A., Paris, France. http://www.microanalysis.org/sm/cameca/cameca.html.

86. P. P. Camus, D. J. Larson, L. M. Holzman, and T. F.Kelly: A new electronic detector for atom probefield ion microscopy. J. Phys. IV C5-6:291–295(1996).

87. C. Martin, P. Jelinsky, M. Lampton, and R. F. Ma-lina: Wedge-and-strip anodes for centroid-findingposition-sensitive photon and particle detectors.Rev. Sci. Instrum. 52:1067–1074 (1981).

88. C. Martin and A. Rasmussen: Mosaic wedge-and-strip arrays for large format microchannel platedetectors. IEEE Trans. Nuc. Sci. 36:836–840 (1989).

89. M. Lampton, O. Siegmund, and R. Raffanti: Delayline anodes for microchannel-plate spectrometers.Rev. Sci. Instrum. 58:2298–2305 (1987).

90. Newport Corporation: http://www.newport.com/catalog.html.

91. Technical Manufacturing Corporation: http://www.techmfg.com/body.htm.

92. Minus K Technology: http://www.mimusk.com/index.htm.

93. BBN Corporation: http://www.bbn.com/products/monitoring/asvc.htm.

94. Advanced Research Systems: http://www.arscryo.com

95. S. S. Bajikar: Local electrode atom probe studiesusing silicon field emitter arrays, Ph.D. Thesis,The University Of Wisconsin, Madison, WI (1996).

96. G. L. Kellogg and T. T. Tsong: Pulsed-laser atom-probe field-ion microscopy. J. Appl. Phys. 51:1184–1193 (1980).

97. G. L. Kellogg: Field ion microscopy and pulsed la-ser atom-probe mass spectroscopy of insulatingglasses. J. Appl. Phys. 53:6383–6386 (1982).

98. A. J. Melmed, W. A. Schmidt, J. H. Block, M.Naschitzki, and M. Lovisa: Effects of a gold shank-overlayer on the field ion imaging of silicon. J.Phys. (Paris) 47:333–336 (1986).

Received January 1999; accepted February 1999.

Date post:	10-Jan-2020
Category:	Documents
Upload:	others
View:	2 times
Download:	0 times

Local Electrode Atom Probesarc.nucapt.northwestern.edu/refbase/files/MatChar44p059...60 T. F. Kelly...

Documents