atom probe tomography 2012 - ایران...

MR42CH01-Kelly ARI 31 May 2012 13:30

Atom Probe Tomography 2012Thomas F. Kelly and David J. LarsonCAMECA Instruments, Inc., Madison, Wisconsin 53711; email: [email protected],[email protected]

Annu. Rev. Mater. Res. 2012. 42:1–31

First published online as a Review in Advance onMay 1, 2012

The Annual Review of Materials Research is online atmatsci.annualreviews.org

This article’s doi:10.1146/annurev-matsci-070511-155007

Copyright c© 2012 by Annual Reviews.All rights reserved

1531-7331/12/0804-0001$20.00

Keywords

local electrode atom probe, atomic-scale analysis, LEAP, APT

Abstract

In the world of tomographic imaging, atom probe tomography (APT) occu-pies the high-spatial-resolution end of the spectrum. It is highly complemen-tary to electron tomography and is applicable to a wide range of materials.The current state of APT is reviewed. Emphasis is placed on applicationsand data analysis as they apply to many fields of research and developmentincluding metals, semiconductors, ceramics, and organic materials. We alsoprovide a brief review of the history and the instrumentation associated withAPT and an assessment of the existing challenges in the field.

1

Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


INTRODUCTION

If the purpose of tomography is to obtain a three-dimensional view of the internal structure ofan object, then atom probe tomography (APT) sits alone at the high-resolution end of spatialscales as the analytical technique with the highest spatial resolution. If every atom were identifiedand positioned in a volume, then APT may be called the ultimate tomography. The techniquecannot yet perform the latter, but it is close to doing so, and there are plans to achieve to this goal(1–3). In the meantime, the technique offers analysts the opportunity to examine inorganic andorganic materials with a three-dimensional image made up of detected atoms or small molecules(up to 60% detection efficiency) with high spatial resolution (better than 0.3 nm achievable inall directions) and high analytical sensitivity (as good as 1 appm) over volumes of greater than106 nm3 (100 nm × 100 nm × 100 nm). These attributes are a powerful combination.

As with any analytical technique, APT has limitations. Most notably, not all specimens will runsuccessfully, and for some complex material structures, the reconstruction of the three-dimensionalimage can have severe distortions (4, 5). For those structures that do not have these problems, theresults can be awe inspiring. The examples chosen for this review provide a good sampling of thekinds of successes the technique has achieved.

History of Atom Probe Tomography

The history of APT is one of successive invention by an individual, Erwin W. Müller. In 1935,he invented the field emission electron microscope (6), which was further adapted into the fieldion microscope (7), which in 1955 produced the first human-generated images of atoms (8, 9). By1967, to identify those atoms, Müller and coworkers (10) had invented the atom probe field ionmicroscope, which has undergone numerous major improvements, including the imaging atomprobe in 1973 (11) and the position-sensitive atom probe in 1988 (12). The current most commongeometry is a local electrode atom probe (LEAP) (13), which first appeared commercially in 2002.This instrument has a focused small-spot laser beam (14) that first appeared commercially in 2005with a 10-μm-diameter beam and is now in its second generation with a 50-μm-diameter) laser beams. Historical summaries of the field can befound in Miller et al. (18) and Miller (19). Kelly & Miller (20) provided a summary of the hardwareconfigurations, and Kelly & Larson (21) recently published a more recent and personalized historyof the development of the LEAP.

Current Activity

With the introduction of the LEAP and other commercial non-LEAP configurations in the pastdecade (15, 22), major progress has been made for hardware configurations so that present-daycommercial atom probes have become practical instruments. There have been major increasesin data-collection rates of four orders of magnitude (to 5 × 106 atoms min−1), in field of viewof two orders of magnitude (to 3 × 104 nm2), and in mass-resolving power (MRP) of an orderof magnitude (to >1,500) since 2000. These hardware and software developments have led to aburgeoning of applications as described below. By various estimates, the number of people activelyusing atom probes has quadrupled in the past decade. Perhaps most importantly, the number ofatom probes being used for industrial research and development has increased by an order ofmagnitude in that same time. This latter trend is taken as a sign that the technique is achievingthe maturity needed for commercial adoption. Much improvement remains to be realized, but thetechnique is providing information of real value to both academic and commercial users.

2 Kelly · Larson

Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


These performance improvements are crucial, but the introduction of reliable laser pulsingremains the single most important development impacting applications development (22–24).Because APT is based on time-of-flight spectroscopy, the evaporation rate must be pulsed toknow the start time of the evaporated ion. For the first 35 years of atom probe’s history, metalswere studied almost exclusively because the technique required that specimens have high electricalconductivity to accomplish electric field pulsing of the evaporation rate. Laser thermal pulsingof the evaporation rate enabled exploration of atom probe applications from the entire spectrumof condensed matter materials regardless of their electrical behavior. As a result, researchers aredeveloping applications using semiconductor materials of all kinds, ceramic materials, and organicand biological materials. Second-generation laser systems that employ UV photons have now beendeveloped (25), which offer greater performance and specimen yield.

THE STATE OF INSTRUMENTATION

Fundamental Considerations for Design of Instrumentation

Atom probe instrumentation experienced a phase of rapid development in the first decade of the2000s [see Kelly & Larson (26) for a brief history]. These tremendous gains in performance ofAPT instruments (15, 25, 27, 28) give cause to consider the extent to which better performancecan be achieved. Greater MRP is desirable, but not at the cost of field of view: When MRP issufficient for a task, then field of view is more important. The converse is also typically true. Thechallenge for instrument design is to find the optimal trade-off. In Figure 1, a straight-flight-pathatom probe geometry is illustrated. When the flight path length, L, is small, the detector subtendsa large solid angle in the data path, which produces a large field of view, ϕ. The flight times aresmall when the flight path length is small, so high repetition rates may be used. However, thetiming uncertainty will be a finite number, and for small flight times, it can lead to unacceptableuncertainties in the timing resolution or resultant mass resolution for ions. If the energy spread forsimilar ions is negligible, the timing uncertainty does not change during flight, so if the flight pathlength is increased, the relative timing resolution will improve, and greater flight path lengths canproduce higher mass resolution. There is generally thus an inverse relationship between field ofview and MRP. The challenge is to push this envelope of MRP and field of view to greater values.

In laser pulsing of atom probes, the energy spread for many materials is almost negligible,�3 eV, which makes it possible to achieve high MRP with long flight times. However, somematerials may have a large intrinsic energy spread (∼20 eV) in laser pulsing that is large enoughto defeat the goal of achieving high MRP with long flight paths. The source of this large energyspread has not been identified, but fortunately it appears that few materials have a large energyspread in laser pulsing. When long flight times lead to high MRP, optical elements may be usedto bend the ions back toward the central axis to recover field of view for a given L, as shown inFigure 1. This concept was first developed for a three-dimensional atom probe by Bostel et al.(27, 28) and was commercialized by CAMECA.

Reflectron-Based Instruments

Given that energy spread may be greater than a few electron volts even in laser pulsing, theoptions for improving peak discrimination are energy compensation or energy determination.Energy compensation seeks to equalize ion flight time by causing ions of higher energy to takelonger pathways to the detector. When done correctly, all ions of a given mass-to-charge-stateratio have the same flight time. These systems make it possible to achieve high MRP in voltage-pulsing mode as well as in laser-pulsing mode even when large energy spreads exist.

www.annualreviews.org • Atom Probe Tomography 2012 3

Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


Specimen

Ion

Ion optic

R

L

Detector

Counterelectrode

Vext

Vp

Vpa

Vtot

φ

Figure 1Geometry of a typical atom probe. Vtot is the total accelerating voltage of the system. Vext is an extractionvoltage used with both static and dynamic fields. Vpa is a postacceleration voltage that can range in valuefrom zero to magnitudes comparable to Vext . Vp is a voltage pulse that is the time-varying component ofdynamic fields. L is the flight path length to the center of the detector, and R is the radius of the detector.

Mamyrin et al. (29) designed the first reflectrons for general mass spectrometry. Waugh et al.(30) first used reflectrons in APT for one-dimensional atom probe instruments. Cerezo et al. (31)adapted reflectrons to an imaging application in three-dimensional atom probes within a few years.All these reflectrons used planar electric fields, which became a problem in imaging instruments be-cause, although they create an output surface with reduced timing spread, this comes at a cost of in-troducing spatial aberrations for ions of different energy. The field of view was therefore restrictedto small values (∼15 nm diameter) to limit this effect. Panayi (32) solved this problem by designinga curved reflectron in which the output surfaces for time-focused ions and spatial-focused ions aremade to nearly coincide. As a result, the field of view of curved reflectrons can be made an orderof magnitude larger (>150 nm diameter). These curved reflectrons were adapted to a commercialremote-electrode atom probe (22) and are currently used in one model of LEAP instrument.

THE STATE OF SPECIMEN PREPARATION

Automated Electropolishing Instruments

Today, atom probe specimens consist of a sharp needle with an apex radius measuring 20 to200 nm and a modest shank angle (≤10◦ semiangle). The small specimen size enables the high(∼10–50 V nm−1) electric field necessary for field evaporation. Because the needle-shaped spec-imen is the primary optic of the microscope, specimen preparation forms a critical step in asuccessful APT analysis.

4 Kelly · Larson

Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


Historically, needle-shaped specimens formed by electropolishing were the norm for APT.Standard techniques have been applied primarily, although not exclusively, to metals since theinception of high-field work in the 1930s. General information on the techniques can be found inReferences 33 and 34, and a veritable bible of electropolishing and chemical polishing solutionsmay be found in section 9 of Reference 35. In addition to the standard manual techniques, auto-mated apparatuses have emerged commercially (36). Simplex Scientific (Middleton, Wisconsin)offers an instrument called the electropointer that automatically prepares needle-shaped speci-mens. Once the starting wire or piece is inserted into the apparatus, the instrument automaticallycontrols the voltage and current applied to a small wire loop surrounding the specimen normal toits long axis. The loop is moved longitudinally by stepper motor. All motions and electrical signalsare controlled and monitored by a computer. There is also an integrated high-magnification lightoptical microscope that displays the specimen during operation. With this instrument, a blankspecimen is inserted, and a specimen that is ready for atom probe analysis is output without theneed to verify its sharpness.

Focused Ion Beam–Based Preparation

The criteria for most specimens prepared for APT analyses are fundamentally the same. (a) Thespecimen must be sharp enough to produce field evaporation. (b) The specimen must be robustenough mechanically to allow for significant evaporation to occur. (c) The feature of interest tothe user must also be present in the near-apex region to ensure that the feature is included in theevaporated data.

Using ions to assist in the preparation of specimens for field ion microscopy or APT dates backnearly 40 years (37). Several researchers have used broad ion beams (38–43) to assist in specimenpreparation, and early efforts included using focused ion beams (FIB) (44, 45), although the degreeof focusing in these early instruments was not high. The advent of the current generation of FIBinstruments in the 1990s began a revolution in preparation methodology for APT. The abilityto image specimens (using ions in and secondary or backscattered electrons out) coupled withthe ability to remove substantial amounts of materials using ions enabled the ability to see thefeatures of interest in specimens at high magnification while they were being sharpened. Overthe past decade or so, this new capability has totally changed the types of studies to which APTmay be successfully applied.

The earliest FIB-based methods to sharpen specimens for APT relied on attaching a volumeof material to the end of a needle (43, 46) (using non-FIB methods) and shaping the end-formfor APT analysis with FIB (47–50). Optimal annular milling methods that provided for improvedemitter shapes were gradually developed (51–55). These methods allow the user to see the majorityof the specimen apex volume during preparation and are generally applicable to most materials.If care is taken during the final steps of preparation, minimal ion-induced damage will be presentin the specimen region of interest (56). APT lift-out methods (55, 57–60) were adapted fromtransmission electron microscopy (TEM) techniques (61). Using these methods, researchers mayremove a small region of material from nearly any structure while they view it in the FIB. Thissmall region of material eventually forms the apex region of the specimen. Details of these methodsare described below.

The specimen-preparation landscape again changed significantly after Nishikawa developedthe scanning atom probe (62), enabling the local electrode instruments that followed (13, 63, 64).With the electric-field-defining elements of instruments confined locally near the specimen apex(e.g., within less than ∼50 μm), specimen geometries were no longer constrained to needles. Forexample, investigators developed microtip-array carrier tips (60, 65) that can be used in a FIB


Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


to make large numbers of specimens at one sitting with reproducible conditions. This leads toreproducible specimen geometries and greatly improves experimental repeatability.

Today, many applications and variations of the standard lift-out and sharpening methods havebeen reported (55, 66–73). These include variations enabling analysis parallel to the originalspecimen surface (“cross-section” orientation) (74) or inverted relative to the original specimensurface (“backside” orientation) (75, 76).

Site-Specific Specimen Preparation

For many applications, FIB-based site-specific specimen preparation is required. The general stepsused in this preparation applied to the case of a FinFET (77) are illustrated and further describedin Figures 2 and 3. Some preparatory preprocessing or deprocessing of the device structure maybe necessary prior to lift-out (Figure 2) and sharpening (Figure 3). Preprocessing often includesapplying a protective sacrificial cap to the top surface (71).

THE STATE OF APPLICATIONS

Introduction

Applications of APT have expanded rapidly over the past decade. The availability of high-qualitycommercial instruments has pushed the performance and reliability of atom probes, greatly

10 µm 10 µm

30 µm 3 µm 4 µm

d e f 10 µm 10 µm 10 µm

a b c a b c

Figure 2The steps involved in a standard focused ion beam (FIB) lift-out procedure. (a) A FIB-deposited protectivestrip is placed over the region of interest. This protective material is often Pt or W. (b) The material is thenremoved by ion milling around three sides of the region (arrows) as well as underneath to produce a longcantilevered wedge of material. (c) The wedge is removed by using an in situ micromanipulator (attached tothe left end of the wedge) and then cutting the wedge free from the substrate (vertical white line). (d ) Themicromanipulator is used to position the wedge above the carrier microtip (seen in plan view). (e) The wedgeis attached to the carrier-tip surface with FIB-deposited Pt (arrow) and then cut free from the carrier tip(vertical white line) for transfer to additional microtips. Once propagation of the wedge is complete, the FIBstage is rotated 180◦ so that a second Pt braze can be applied to the opposite wedge-post interface of eachmounted post. ( f ) The final mounted wedge section is then ready for sharpening: The line of the targetedFinFET is clearly visible along the line of the arrows.

6 Kelly · Larson

Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


a

d

b c

f e

2 µm

2 µm 2 µm 400 µm

2 µm 2 µm

Figure 3The steps involved in the focused ion beam (FIB)-sharpening process converting the wedge section inFigure 2f into a specimen for APT analysis. (a) The region of interest is shown by the arrow. (b) The firstannular milling pattern shapes the tip into a cylinder, with the FinFET region clearly visible near theupper-center portion of the cylinder. The second and third milling patterns produce (c) a tapered end and(d ) a narrowed end. The final sharpening step is performed at low ion energy (either 2 keV or 5 keV) and issimply a circular pattern that images the specimen end-on (56). During this stage, the specimen is carefullymonitored to ensure that the region of interest is positioned near the apex of the final specimen. (e,f ) Thelow- and high-magnification images of the specimen that is then ready for APT.

impacting the breadth of available applications. However, the most significant factor in expandingthe universe of applications is the renewed utilization of laser pulsing. This is true for materialsthat cannot be field evaporated successfully with voltage pulsing. Yet, laser pulsing is also beneficialfor a range of metals applications for which premature specimen fracture is a problem. Certainmetal specimens, including high-strength steels, for example, will not run with adequate yieldwith voltage pulsing because the stress in a specimen varies with the square of the applied electricfield (78) and voltage pulsing uses higher electric fields than does laser pulsing. These specimenswill often run with much higher success rates in laser pulsing.

Metals

Because the atom probe technique relied almost exclusively upon voltage pulsing in its first threedecades, virtually all the work from the 1970s through 2000 was on metals. As a result, theseapplications tend to be the most mature. However, new and more sophisticated applications areconstantly being pursued and reported in the literature. This section highlights three metalsapplications that illustrate the latest trends and the newest applications.

A trend in microscopy today is the correlative use of TEM and APT for materials studies atthe atomic scale. The complementarity of these two techniques is particularly strong and is likelyto lead to greater integration of the two techniques. Two examples are given here to illustratethe synergistic correlative opportunities that may be exploited with TEM and APT. Srinivasanet al. (79) used high-angle annular dark field STEM (scanning transmission electron microscopy)and APT to study the order-disorder transition between the disordered γ matrix and orderedγ′ precipitates in a nickel-base superalloy (Figure 4). STEM was used to elicit the variation


Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


a

d

b

1.3

1.1

0.9

0.70 2 4

Distance (nm)

Inte

nsit

y ra

tio

for {

002}

6

X1 X2 X3

10

20

30

40

50z

x

60

70

80

90

94 nm

Al = 10 at%Co = 19 at%

cCo

mpo

siti

on (a

t%)

Distance (nm)

35

CoTiCrAlMo

30

25

20

15

10

5

0–3 –2 –1 0

2.0 nm

1 2 3 4

Figure 4Order-disorder transition at a heterointerface. (a) {002} plane intensity in high-angle annular dark fieldSTEM (scanning transmission electron microscopy) at the γ-γ′ interface. (b) Atom probe tomography(APT) image with γ phase (Co atoms shown in blue) and γ′ phase (Al atoms shown in red ). (c) Close-up ofthe lower interface in panel b showing the Al atoms in the {002}, where the interface transition based onatomic ordering is approximately 4–6 atomic layers (0.7–1.0 nm) thick. (d ) A proximity histogram [orproxigram (160)] across the same interface showing a compositional interface width of approximately 2 nm.Reprinted with permission from Reference 79. Copyright 2009, American Physical Society.

8 Kelly · Larson

Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


in order across the coherent {001} heterointerface through intensity ratios of the aluminum-rich/aluminum-poor {002} plane signal. APT was used to measure the aluminum compositionprofile and to estimate the ordering transition width across this same interface type. The twotechniques agree very well. Furthermore, the experimental results matched with high precisionthe atomistic modeling results of Mishin (80) and Ardell & Ozolins (81). This experimental study(79) found that the order-disorder transition width is smaller than the compositional width ofthe interface, which raises fundamental questions concerning the definition of interface width atorder-disorder interfaces. This work illustrates an important trend in APT work today: a strongconvergence between the increasing capabilities of atomistic modeling and the experimental dataavailable from APT (and TEM). This synergistic relationship is expected to continue as a naturalmarriage of atomic-scale predictive and characterization capabilities.

Gault et al. (82) studied an Al-Cu-Li-Mg-Ag model alloy with TEM and APT to elucidate thestructure and composition of precipitates in the material. Excellent correlation was found betweenthe TEM and APT results, as illustrated in Figure 5. Note, for example, that the phase contrast inFigure 5a surrounding the plate-shaped θ′ phase suggests the presence of an ordered L12 phase,which has been suspected to be δ′ (Al3Li). The APT composition profile in Figure 5c shows aclear increase in Li concentration on the outer parts of this precipitate, which is consistent withthe hypothesis of a δ′ wetting layer on the θ′ phase. In this example, each of the two techniques

30

25

20

15

10

5

00 5 10

Conc

entr

atio

n (%

)

Distance (nm)

CuLiMgAg

a bb

c

Figure 5Numerous second-phase features are visible in both the high-resolution transmission electron microscopy image (a) and the atomprobe tomography (APT) image (b), which are taken from different specimens. These include a plate-shaped θ′ phase (Al2Cu) in theorange ellipse that is connected to a spheroidal Guinier-Preston-Bagariastkij (GPB) zone in the light blue circle, an isolated spheroidalGPB zone in the dark blue circle, and a lath of the S phase (Al2CuMg) in the red ellipse. A composition profile from APT across thewhite arrow in panel a is shown in panel c, which reveals the presence of elevated concentrations of Li on the edges of the θ′ phase.Reprinted from Reference 82 with permission from Elsevier.


Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


provides complementary information that is crucial to support the conclusion and that would notbe available from a single instrument.

One of the great unresolved challenges for analytical microscopy has been the ability to maphydrogen quantitatively in materials on a submicrometer scale. APT has been long considered forthis task, but questions concerning the origin of the hydrogen, whether it truly emanates from thespecimen or is adsorbed from the vacuum system, need to be addressed. Furthermore, hydrogen isvery mobile in most materials, which makes it difficult to analyze in thin APT specimens. Showingthat hydrogen concentration varies with position in a microstructure in a way that correlateswith expected behavior would lend credence to qualitative mapping. Furthermore, if a phase withknown hydrogen concentration can be analyzed, then an assessment of the quantitative accuracycan be made.

There have been several efforts in the past three years to demonstrate the capability of APT tomap hydrogen in materials (83–87). Takahashi et al. (86) used an environmental chamber attachedto an atom probe to charge deuterium in steels at room temperature and to quickly get them tothe cryogenic stage of the atom probe. They showed that deuterium is trapped at the interface oftitanium carbides with the matrix. Sepehri-Amin et al. (88) analyzed NdH2 hydrides in an Fe-richmatrix from Nd-Fe-B powder particles (Figure 6) and found a hydride composition of 64 at% Hand 33.1 at% Nd, in good agreement with the expected stoichiometry. Their success is likely due,in part, to the fact that hydrogen is not mobile in this compound.

Cryogenic specimen transport systems for atom probes have become available commerciallyin the past year from a collaboration between CAMECA and Leica. This development shouldenable better studies of materials such as steels in which hydrogen is mobile at room temperatureand that therefore must be quenched and kept frozen after treatment with hydrogen.

Silicon-Based Semiconductor Materials

Characterization of semiconductor device features and defects over the past few decades has beenachieved primarily through use of electron microscopy, ion microscopy, various X-ray technolo-gies, and some light-optical techniques. As the industry has moved further into the nanoelectronicsrealm, the use of these methods has become more challenging owing to the scale and dimension-ality of the devices of interest and limitations of what have become standard techniques within theindustry. One significant goal in the application of APT to silicon-based structures has been theanalysis of functional devices. Downscaling of devices and the reduction of source/drain junctiondepths to avoid short-channel effects, combined with the increased field of view now available inmodern atom probes (13, 64), make analysis of entire device structures a reality. Several researchershave recently used APT in the analysis of portions of partially processed device structures as wellas in analysis of entire devices (71, 72, 89–99); selected examples are reviewed in this section.

Inoue and coauthors (90–92, 99) recently conducted substantial investigations of the distribu-tion of dopants in various parts of MOSFETs. Knowledge of the detailed dopant distributions inmicroelectronic devices is becoming increasingly important as the devices get smaller and moveinto three dimensions. Figure 7 provides details of the n- and p-MOSFET gate regions of a recenttransistor structure.

Various authors have recently investigated FinFET-type devices using APT. FinFETs (77) are“fin”-based field-effect transistors that maximize the area of the gate oxide by wrapping the oxideand gate structure around a feature that resembles a fin (Figure 8a). Intel recently announced aplan to employ this technology in the near future for its 22-mm technology-node products (100).The International Technology Roadmap for Semiconductors refers to these types of structures,which are part of a larger group of devices called multigate transistors, as “advanced nonclassical

10 Kelly · Larson

Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


0

20

40

60NdFeCoH

80

100

0 10 20 30 40 50

0

1

2 10x

20x

3

0 10 20 30 40 50

B

0

1

2

3

Distance (nm)

Distance (nm)

Distance (nm)

Com

posi

tion

(at%

)Co

mpo

siti

on(a

t%)

Com

posi

tion

(at%

) Ga

0 10 20 30 40 50

400 nm

a b

c d

~270

nm

~45 nm

Nd

Nd, B

H

Ga

H

~80 nm ~80 nm

Figure 6Analysis of hydrides in hydrogen-disproportionated Fe-Nd-B powder. (a) TEM image of an atom probe tomography specimen inwhich the brightly imaging phase is NdH2. (b) Atom maps for Nd and H showing the hydrides. The subvolume for atom maps in panelc is shown as the black box. (d ) Composition profiles along the long axis of the subvolume in panel c. Reprinted from Reference 88 withpermission from Elsevier.


Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


n-MOS p-MOS

50 nm

Si (0.1%)OAsBP

Figure 7MOSFET gate regions. (Left) In the n-MOSFET gate region, phosphorus atoms segregate at grainboundaries and polycrystalline silicon/oxide interfaces. (Right) In contrast, no segregation of boron atoms isobserved in the p-MOSFET gate (99). Arsenic and boron atoms were detected in the source/drain regionsand channel regions, respectively, for the p-type FET (and vice versa for the n-type structure). Statisticalevaluation of the data suggested that the fabrication process increases the boron fluctuation but not thearsenic fluctuation, thus producing more variations from device to device in p-type MOSFETs (99) (figurecourtesy K. Inoue, Kyoto University, Japan).

CMOS devices” (77). Gilbert et al. (95) also studied this structure (shown in Figure 8). Theyfound significant segregation of boron at the TiN/Si interface at the top of the FIN, but not atthe same interface on the device sidewalls. They also reported several observations of artifacts inthe nature of the three-dimensional reconstruction of such structures, which highlights the needfor improvements in this area (4, 5, 101).

In FINFETs, as in MOSFETs, device performance may be altered by an inhomogeneous distri-bution of dopants. Recently, Kambham et al. (94) analyzed the conformality of boron implantationas a function of implantation angle in FinFET-type structures. They found that, for the 45◦ im-plantation (measured between the plane of the wafer surface and incoming dopant trajectory),the vertical peak concentration is approximately two times the lateral-profile peak concentration.However, in an implantation done at 10◦, the dopants were more shallow and reduced in con-centration. Observations of high nonconformality of the dopants with the sidewall dose at only10% of the top dose compare favorably with previous modeling results from Vandervorst et al.(102).

Although APT has been applied to transistor and FINFET-type structures as described above,these structures have often been stopped at some point in the fabrication process to accommo-date the analysis. There are many situations in semiconductor processing in which APT anal-ysis of a finished product is desirable; competitive analysis and failure analysis are two goodexamples. Only recently have APT results been obtained from fully processed “off-the-shelf ”transistor structures that are part of a finished product (72, 96–99). Figure 9 presents datashowing the feasibility of APT analysis for fully packaged integrated-circuit microelectronicdevices.

12 Kelly · Larson

Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


µm a b

NiSi

Si

TiNHfSiOSiO2

Figure 8(a) A schematic and (b) an atom probe tomography data reconstruction of an entire high-k metal gateFinFET structure (100 nm in height and 20 nm in width) starting with nickel silicide (near top) and endingnear the bottom of the FIN (bottom) (figure courtesy M. Gilbert, IMEC).

The field of view of the atom probe for this device is adequate to observe portions of the sourceand drain on either side of the channel. In future technology nodes, the transistor characteristiclengths will be smaller, and the present field of view achieved in this work will cover a larger portionof the components outside the channel. Although atom probe analysis of packaged devices is notcurrently routine, we expect that this approach to specimen preparation and analysis will maturequickly.

Compound Semiconductor Materials

Researchers have achieved much success using gallium nitride (GaN) as the basis for a wide range ofoptoelectronic devices. Unintentionally doped GaN emits in the ultraviolet range, and alloying itwith indium nitride (InN) or aluminum nitride (AlN) produces emission with wavelengths rangingfrom green to deep ultraviolet. GaN-based multiple quantum well (QW) structures are used asthe active regions of commercial light-emitting diodes and laser diodes. Such structures oftenexhibit bright-line emission despite very high dislocation densities. Recently, APT has been usedto investigate the chemical and morphological microstructure of GaN- and GaAs-based structures(103–115).

It has been proposed that carrier localization can originate from structural and compositionaleffects such as quantum well width fluctuations and indium clustering (116). APT is ideally suitedto describe compositional fluctuations and to look for nonrandom statistical fluctuations suchas those proposed (18). A light-emitting diode (LED) device structure (50-nm p-GaN/30-nmGaN/3.5-nm In0.15 Ga0.85 N QW/n-GaN) grown by metal-organic chemical vapor deposition ona semipolar (1̄01̄1̄) GaN substrate was recently analyzed with APT (Figure 10) (110). Frequencydistribution analyses were also performed (110) on the indium interatomic distances, and for thisparticular device structure, no evidence was obtained to cause rejection of the null hypothesis atthe 95% confidence level. In other words, the indium composition fluctuations in this device wereconsistent with random fluctuations and could not cause carrier localization.


Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


a b

c

Si O + O2 HfO Ge As

B C

Channel

Source Drain

Gate oxide

Gate

10 nm

10 nm

d

Figure 9Specimens created from a commercially available 32-nm technology node CMOS device that wasdepackaged and then prepared by focused ion beam methods. (a) Atom probe tomography analysis shows aHf-based high-k gate oxide near the top of the data set. (b) Regions of SiGe are visible on either side of adominantly Si region. (b,c) Spatial distributions of the Ge, As, B, and C atoms. Whereas the As is distributedrelatively uniformly (with a concentration of ∼0.016 at%), the B is correlated with the regions of high-Gecontent (96), and the carbon atoms appear to be clustered (72). (d ) A high-angle annular-dark-field STEMimage (97) (courtesy W. Lefebvre, Université de Rouen) of a structure similar to the one analyzed in panelsa–c. An undercut of the gate oxide into the channel is apparent, matching the Ge distribution shown inpanel b.

APT data have also been collected from a CdTe layer within a CdTe solar cell that was preparedwith standard CdCl2 and Cu treatments (117, 118). Figure 11 shows the mass spectrum, whichcontains a number of high-mass molecular ions. This spectrum is one of the most complex APTspectra published to date. Despite this complexity, crucial information concerning segregation ofimpurities at the grain boundaries has been found.

Ceramic and Geological Materials

Although various insulating or ceramic materials were first investigated with the pulsed-laser atomprobe approximately 30 years ago (119), only recently has there been further research in this area.

14 Kelly · Larson

Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


20 nm

a b

0.01

0.1

1

10

–2–1012

In c

once

ntra

tion

(at%

)

Distance (nm) In Ga N2

Figure 10(a) Atom probe tomography reconstruction of a section of a quantum well. (b) Proxigram analysis (160)relative to the 2% indium composition surfaces for the top (blue line) and bottom (magenta line) interfaces.The bottom interface is abrupt—no spectral evidence for indium is present for regions below 0.1 at% indium,whereas the top interface has a ∼2-nm tail of indium into the GaN layer above. Atom colors: In, magenta;Ga, orange; N2, green. Reprinted from Reference 110 with permission from American Institute of Physics.

The reemergence of pulsed lasers in APT analysis has opened the way for the analysis of bulk(or near-bulk) insulating oxides of various metals, including Al (120–122), Si (120, 123–125), Zr(126, 127), Ce (127–129), Li (127), Zn (130–132), In (133), Mg (132, 134), Fe (135), and Ni(E.A. Marquis & R. Reed, unpublished work). Some studies directed more toward the theory offield evaporation of such materials have also been published (136–138). Several examples of theapplication of APT to ceramic materials in materials science are presented in this section.

Carbon at low levels can be difficult to detect by electron microscopy and, as such, has tradition-ally been a very strong area of research for APT (18, 33). SiC additions improve the mechanicalproperties of alumina even at low particle volume fractions. APT was recently used to analyzegrain boundaries in alumina to investigate the position of carbon atoms with respect to grainboundaries (122). A site-specific FIB preparation was used to position a grain boundary in thenear-apex region of a specimen. APT analysis shows a level of carbon segregation of 1.5 ± 0.5atom nm−2 with no measurable carbon found within the alumina grains. The matrix aluminacomposition was measured to be 43.5 ± 1 at% Al–56.5 ± 1 at% O with less than 2 at% H (Al/Oratio is 0.77 ± 0.3 as compared with the expected value of 0.67) (122). One possible explanationfor the missing oxygen is that it may be present as 16O2++ counts at m/n = 16 Da, and thus the16O+ peak is counted as all single oxygen ions when some counts may in fact be double oxygenions.

In addition to bulk alumina, APT has recently been used to investigate metal/oxide interfaces,namely the oxide films formed on a metal surface or along grain boundaries. Elements such asHf, Ta, and Si contribute to improving the oxidation resistance of Ni-base superalloys. WhereasHf may be imaged at Al2O3 grain boundaries by STEM-EDX (139), APT analysis also reveals adistribution of other dopants (Ta and Si) at the oxide/metal interface, at oxide grain boundaries,and within the oxide grains (Figure 12a). Figure 12b shows Zr, Hf, and Y segregation at a grainboundary in the thermally grown Al2O3 scale formed on a Ni-base superalloy (Y. Chen, E.A.Marquis & R.C. Reed, unpublished work).

Segregation at grain boundaries is also of interest in doped Ce-based oxide materials, whichare being investigated as an electrolyte in solid-oxide fuel cells. To explain the decrease in


Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


Cd2Te++

Cd2Te3++

Cd2Te3+ Cd3Te3+Cd2Te2+ CdTe3+

Te2+CdTe+

Cd2+CdTeCO++

CdCO+

CdCl+

CdC2H6O+

TeS+

a

b

215 225 235 245 255 265

105

104

103

102

104

106

102Co

unts

136

800

600

400

200

Coun

ts

1,000

1,200

140 144 148 152 156 160

Mass-to-charge-state ratio (Da)


Coun

ts


0 200 400 600

CdS++

Cd2Te+ CdTe2+

CdTe2++

50 0λ

–50

100

200

300

z40

050

0

CdS CdCl TeS CdS CdCl TeS

Figure 11(a) Atom probe tomography spectrum and (b) three-dimensional atom maps of CdS, CdCl, and TeS molecular ions, with a coloroverlay of CdCl (orange), Cd (black), Te (dark green). These two-dimensional renderings show enrichment of S and Cl at the grainboundary. Quantitative composition profiles that were generated from the dashed-box region show peak Cl concentrations at the grainboundary to be well over 1 at% and S concentrations of approximately 0.5 at%, which are very large amounts for impurities. Reprintedwith permission from Reference 117. Copyright 2012, Cambridge University Press.

16 Kelly · Larson

Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


ONiTaHfSi

Oxide

Metal

80 nm

Oxide

Metal

80 nm

a b

Bondcoat

Top coat Top coat

50 µm

Grain boundarysegregation in Al2O3

Al2O3Al2O3

Al2O3Al2O3

Conc

entr

atio

n (a

t%)

Distance (nm)

0.450.4

0.350.3

0.250.2

0.150.1

0.050

0 20 40 60

Y

Zr

Hf

Figure 12(a) Slice from a three-dimensional atom probe tomography (APT) reconstruction of the oxide/metal regionobtained from an oxidized Ni-base superalloy revealing Hf and Ta segregation at the metal/oxide (alumina)interface and oxide grain boundaries. The distribution of Si in alumina is also shown (courtesy of E.A. Marquis& R.C. Reed). (b) APT analysis of Zr, Hf, and Y segregation at a grain boundary in the thermally grownAl2O3 scale formed on a Ni-base superalloy. The scanning-electron-microscopy image shows the location ofthe analysis within the thermal-barrier-coating structure (courtesy of Y. Chen, E.A. Marquis & R.C. Reed).

ion conductivity in fuel cells, Li et al. (129) recently used APT to investigate segregation tograin boundaries in sintered doped CeO2. They found a depletion of Ce and an increase in Y(Figure 13) or Gd (140) at the grain boundaries in these materials.

Given the success of synthetic ceramic materials analysis with APT, geological materials area natural next step. Analyses of metamorphic magnetite (141), natural olivine (142), calcium car-bonate (143), and ancient zircons (D. Snoeyenbos & D Reinhard, private communication) havebeen conducted, but this field is just getting established. One illustrative example of the value ofAPT to geology comes from an exogeology application: analysis of (Allende) meteoritic nanodi-amond grains (144). In this work, the isotope ratios for carbon have bearing on theories of theircosmic origin, particularly whether they are solar or presolar. Isolated nanodiamond particleswere embedded in Pt and then prepared by FIB lift-out for APT analysis. Figure 14 is an atommap that shows the resulting composite. Standard terrestrial carbon isotope ratios for 12C/13Care 92.4. Stadermann et al. (144) found 12C/13C = 61 ± 4 for C+ and 12C/13C = 54 ± 4 forC++. Although this work is considered preliminary and these isotope ratios must be validated, it


Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


Y

Ce

O

Y

Ce

O

A

B

i

iii

ii

Grain AGrain B

Grain C

A

B

i

iii

ii

Grain AGrain B

Grain C

CeYONominal

70

60

50

40

Conc

entr

atio

n (a

t%)

30

20

10

0 10 20

Distance (nm)30 40

a b

Figure 13(a) Atom map of Y ( green), Ce (red ), and oxygen (blue); the analysis volume includes a triple junction (129). (b) A concentration profileacross a portion of the grain boundary. Volume shown as “A” is also shown in panel b and exhibits clear segregation of Y, which Li et al.(129) determined to be 2–4 at% higher (depending on boundary region i, ii, or iii denoted in panel a) than the Y content within thegrains. The total average concentration of Y, Ce, and O in these data was determined to be very close to the expected nominalcomposition. Reprinted from Reference 129 with permission from Elsevier.

does illustrate the potential impact that APT can have on such studies. Despite the challengesnanoparticles present for specimen preparation for APT, Stadermann et al. (144) have developeda working approach.

Organic and Biological Materials

A recent article (145) recounted the history of attempts to analyze organic materials with APT. Acouple of examples are provided here to illustrate the state of this art. Synthetic polymers, suchas self-assembled monolayer-forming oligomers (146–148), larger-molecular-weight polymers(149), and natural biopolymers (145) made from assemblies of proteins to make subcellular systems,have yielded to APT analysis. The potential to access nanoscale chemical imaging informationon these complexly organized materials promises to open up a new understanding of interchaininteractions and of interfacial interactions that are important in organic device development (150)and biology.

One issue of larger concern for APT analysis of polymeric materials is the potential for largemolecular-ion evaporation. The spatial precision of the three-dimensional reconstruction is par-tially predicated on ion size. For atomic ions, maximum spatial precision is achieved, whereasfor large molecular ions, all the atoms composing the molecule are essentially placed at the sameposition in three dimensions—no molecular orientation information can be collected (146). Thecomplexity of polymer microstructures leads to field-evaporated mass spectra with a wide vari-ety of ion sizes and thus to reduced spatial precision relative to monatomic ions. An example ofthis variety is evident in mass spectra from poly(3-hexylthiophene) (P3HT) films and preparedfor APT by different methods (Figure 15a) (149). For thin films deposited on carrier tips, largemolecular-ion fragments are observed, whereas for FIB-extracted sections, primarily monatomicand diatomic ion species are detected. Ion fragments are observed for nearly every integer massposition, but there are no peaks representing single-carbon ions. For P3HT films doped with

18 Kelly · Larson

Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


PtCN

75 nm

5 nm

54 n

m

Figure 14Atom map of several clusters of meteoritic nanodiamonds (arrows) embedded in Pt. Shown is a 2-nm-thickslice cut through the data set along the plane of the original diamond deposition layer.

C60 molecules, both fragmented and whole C60 ions were observed, with the relative abundancesvarying as a function of analysis conditions (see Figure 15a).

Composite P3HT and [6,6]-phenyl-C61 butyric acid methyl ester (PCBM) films, a popular bulkheterojunction material for organic photovoltaic devices such as solar cells (150), were depositedand prepared via FIB lift-out. Chemically, these are similar in composition and structure to theC60-doped films described above; however, the mass spectra collected from these films are starklydifferent, as shown in Figure 15b. Here, carbon is a dominant monatomic ion, and atomic-massions with m/n > 60 are absent from the mass spectrum. This is consistent with mass spectra ob-served from other FIB-prepared polymer specimens, including cell biopsies where atomic carbon,nitrogen, and oxygen were observed with very few large-mass ion fragments (151).

The most advanced analysis of a biological material comes from the recent work of Gordon &Joester (152) on chiton teeth. Specimens were FIB extracted from the cutting edge of magnetite-based teeth, which contain 5–10-nm-diameter chitin protein fibers. The mass spectra were logicalfor the materials and showed clearly distinct peaks from the organic and mineral components.Atom maps of two analyses are shown in Figure 16. The organic-matter-derived ions are clearlyseen to originate from fibers occluded within the mineral. The total amount of carbon is less thanwhat would be expected from crystalline chitin fibers, which may be a consequence of overlappingpeaks in time-of-flight spectra, preferential evaporation, or biological remodeling of fibers duringmineralization. Nevertheless, the fiber diameter (5–10 nm) closely resembles that of the organicfibers observed in STEM images of the mineralized tooth. Furthermore, Na and Mg clearlycolocalize with organic fibers (Figure 16b,c and Figure 16e, f, respectively).

All the examples of organic materials analysis by APT are from specimens preparedat room temperature. For a general organics capability, it will be important to preparespecimens at cryogenic temperatures and then to transfer them to the atom probe whilekeeping them cold and clean. Cryo-preparation stages for FIBs and cryo-transfer vacuumvessels from FIBs to a CAMECA LEAPTM are now commercially available (for example, see


Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


To 6

,700

To 4

1,90

0

1,400

1,000

600

200

22,000

18,000

14,000

10,000

6,000

2,000

Coun

ts p

er b

inCo

unts

per

bin


800

104

103

102

28

44

18

H+C+

C++

29

18

CH3+

42

120

10570

35

P3HT

PCBM

OCH3

Sn

a

b

0 20 40 60

0 200 400 600

80 100 120

C60++

C60+

O

Figure 15(a) Typical mass spectrum for C60-doped poly(3-hexylthiophene) (P3HT) polymer, highlighting the richdistribution of ion fragments observed for polymer films deposited on sharp carrier tips. (Inset) Expandedmass-to-charge-state-ratio range showing the large number of C60 ions observed. (b) Mass spectrum for[6,6]-phenyl-C61 butyric acid methyl ester (PCBM)-doped P3HT sample prepared by focused ion beamlift-out extraction from a spin-coat-deposited film. The ion fragment distribution has fewer peaks andchemical species such as individual carbon ions that are unambiguously identifiable. The chemical structuresfor PCBM and P3HT are also shown (inset). Reprinted with permission from Reference 160. Copyright2012, Cambridge University Press.

http://www.leica-microsystems.com/products/electron-microscope-sample-preparation/industrial-materials/transfer/details/product/leica-em-vct100/). The EMEZ (Electron Mi-croscopy ETH Zurich) facility at ETH Zurich and the Electron Microbeam Analysis Laboratoryat the University of Michigan are the first laboratories to be equipped for this type of work and areoperational as of early 2012. Although cryogenic micromanipulation is not yet commercially avail-able, commercial suppliers are aware of the need. It is the final hardware item needed for full cryo-genic specimen preparation of specimens. Once available, these hardware advances will enable fullcryo-preparation, which may be the final step in the pursuit of APT analysis of organic specimens.

20 Kelly · Larson

Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.

http://www.leica-microsystems.com/products/electron-microscope-sample-preparation/industrial-materials/transfer/details/product/leica-em-vct100/http://www.leica-microsystems.com/products/electron-microscope-sample-preparation/industrial-materials/transfer/details/product/leica-em-vct100/


Distance from interface (nm)

Distance from interface (nm)

CNa Fe O

C Fe OMg

a b c

d e f

10 nm

10 nm

2 nm

2 nm

–2 0 2 4 6 8 100

0.5

1

1.5

2

2.5

3

3.5

–2 0 2 4 6 8 100

0.5

1

1.5

2

Fiber

Fiber

Mineral

Mineral

CarbonSodium

CarbonMagnesium

Na

Mg

0 2At% C

0 2At% C

Conc

entr

atio

n(a

t%)

Conc

entr

atio

n(a

t%)

Figure 16Atom maps of two analyses using chitin fibers. (a,d ) Three-dimensional reconstructions and proxigrams (160). Two representativesamples containing organic fibers that exclusively bind Na+ (a–c) or Mg++ (d–f ). For clarity, only approximately 5% of the Fe/O ionsare rendered; the edge of the field of view is marked (dashed line in panel b). (b,e) Overlay of Na+ (b, red spheres) and Mg++ (e, magentaspheres) ion positions on carbon concentration maps integrated over the boxed regions indicated in panels a and d. Some regions of thefibers appear devoid of Na or Mg (arrows). (c, f ) Proximity histograms (160) (error bars, ± 1σ) of (c) Na/C and ( f ) Mg/C across theorganic-inorganic interface of fibers indicated by arrows in panels a and d. Interfaces appear graded over 2–4 nm. Reprinted fromReference 152 with permission from Macmillan Publishers Ltd.

CONCLUDING REMARKS

APT has matured over the past decade to become an essential microanalytical technique forresearch on a wide spectrum of materials. It is presently being adopted for, and adapted to,manufacturing needs in the microelectronics industries, which would be the next step in maturationof the technology. As a class of materials, organic and biological materials are the next frontier forapplication of APT technology. If subnanometer compositional mapping of biological materialsin 3D becomes a reality, the impact of such new information on the creation of knowledge couldbe profound.

With all the improvements in APT technology over the past decade, what are the remainingdevelopment needs? Prior to 2000, data-collection times were the bottleneck, as several dayscould be needed to collect an image. Now with improved speed, data collection takes hours, andthe bottleneck has shifted to specimen preparation on the front end and data reconstruction andanalysis on the back end. Automated methods need to be developed both to speed the processand to increase the uniformity of the results. Development work on these topics will likely lead tomuch progress in the next few years. For example, automation of specimen preparation in an FIBis expected, as is use of multivariate statistical analysis methods (153, 154) for data analysis. In theFuture Issues section below, we discuss the two most serious limitations of APT today: specimenyield and data-reconstruction fidelity.


Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


Given today’s state of the art, it is now possible to envision a time when 100% of the atoms ina large volume (�106 nm3) of material will be positioned in three dimensions with high precision(3) that takes APT into the realm of atomic-scale tomography. Although APT offers high-valuemicroanalytical information today, it still has tremendous upside potential. The next decade shouldbe a continuation of exciting times for the field.

FUTURE ISSUES

Specimen-Yield Challenges

Sometimes, when one is working with an unfamiliar material or structure, specimen yield is lowinitially and increases with trial and experience. The challenge is to climb this learning curvequickly by avoiding or overcoming difficulties. In the present context, specimen yield refers to theprobability of obtaining useful amounts of quality data from a specimen before failure. Because thestress in an atom probe specimen varies with the square of the applied electric field and is generallyclose to the cohesive strength of materials, any weakness in a specimen can lead to fracture failure.For specimens with specific regions of interest, any specimen will yield approximately in proportionto the ratio of the strength/adhesion in these regions of interest to the field-induced stress at theoperating field. Fundamentally, the only one way to improve yield is (a) to reduce the stress onthe specimen or (b) to increase strength/adhesion.

There are some general ways to decrease evaporation field, and hence stress, in a specimen:

1. decrease ion detection rate (lower evaporation field),2. increase the background (or other) gas pressure in the vacuum chamber,3. increase specimen base temperature,4. use laser mode rather than voltage pulsing, and5. increase laser energy.

General methods to increase strength/adhesion include the following:

1. change the orientation of the features of interest (e.g., interfaces) in the specimen, such as achange from plan view to cross-section view,

2. remove materials that are not of general interest and that have highly different evaporationcharacteristics,

3. coat (or etch then coat) the specimen with a layer of material, and4. heat the specimen slightly following preparation.

Discontinuities in a specimen, such as an interphase interface, grain boundaries, a notch, or anyhigh-stress distribution, can be initiation points for failure. Avoidance of such weaknesses, wherepossible, is often a successful strategy. For example, layered structures with phases known to havea high evaporation field or a weak interface with other phases can sometimes be eliminated from astructure. Furthermore, the orientation of the structure may be altered to run either bottom-sideup so that the offending phase is encountered at the end of a run or in a cross-sectional orientationso that any weak interface is subject to lower stresses. The cross-section geometry, however, issubject to greater lateral trajectory aberrations that may diminish the spatial resolution across aninterface.

Two further directions for ameliorating these problems that are currently receiving attentionare to coat specimens after FIB preparation and to develop more sophisticated instrument controlprotocols that use real-time detector information to anticipate approaching problems and adaptaccordingly.

22 Kelly · Larson

Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


Reconstruction Challenges

The ability to accurately reconstruct original spatial positions of field-evaporated ions emittedfrom a surface is fundamental to the success of APT. Understanding the evolution of tip shapeand projection during field evaporation plays an important role in improving reconstruction.

Different phases, or even similar phases with different crystal directions, will field evaporatedifferently, leading to topological features on the specimen apex. Because the specimen is theprimary optic in this projection microscope, these features lead to nonuniform magnification inthe image. Several groups are working to improve existing reconstruction algorithms to accountfor these complexities. In extreme cases, ions from the different phases may cross paths, andinformation about their true relative location on the specimen may be corrupted.

State-of-the-art reconstruction algorithms use a simple point-projection method to positiondetected ion events back onto a hemispherical surface. The z increment for each reconstructedion position (x, y, z) is a simple function with respect to the detector impact position (x, y) and ionsequence (z) (4, 155–157). This method works well for limited field-of-view reconstructions ofhomogeneous materials. However, as the field of view is increased, the surface curvature deviatesfurther from that of a hemisphere, and the single point projection poorly approximates a constantangular magnification. In addition, the nonhemispherical nature of the specimen apex shape is fur-ther complicated when materials of differing evaporation fields are simultaneously exposed on a tipsurface. The combination of these effects makes it difficult to achieve an accurate reconstruction ofatomic planes, even for a homogeneous, single evaporation-field material, over the full field of view(4, 5). For nanostructured materials with variations in evaporation field throughout the volume(e.g., multilayer films, precipitates, microelectronic-device structures), the problem is more signif-icant (158). Flat interfaces or spherical precipitates can become highly aberrated and misshapenin the reconstruction. A reconstruction method mitigating these aberrations may require dis-carding simple projection-onto-hemispherical-surface methods in favor of methods that properlytransform detector hit positions onto accurate representations of the evolving specimen surface.

In principle, reconstruction of APT data can be perfected with sufficient knowledge of the spec-imen evaporating-surface shape (specimen apex shape) during an experiment if ion crossing (159)is not present in the data set. To improve reconstruction accuracy, next-generation reconstructionmethods need to make use of increasingly realistic specimen apex shapes and improve the under-standing of how these surfaces project onto the detector and evolve during further evaporation ofthe tip surface. A program has been proposed (1–3) to develop an experimental capability to obtainimages of the specimen apex region with sufficient detail to improve reconstruction. Simulationsof the evolving specimen during field evaporation could also provide the apex shapes needed ifa suitable model of the specimen can be found. The specimen model might be an educated firstguess without experimental input, a result derived from the first reconstructed result of the data,or a model derived from experimental imaging of the specimen. Iteration between reconstructionand model refinement might then yield convergence. In either case, some form of specimen apexshape rendering is likely to play a role in reconstruction in APT going forward.

SUMMARY POINTS

1. Major progress has been made on hardware configurations for APT so that present-daycommercial atom probes have become practical instruments. There have been majorincreases in data-collection rates of four orders of magnitude (to 5 × 106 atoms min−1),in field of view of two orders of magnitude (to 3 × 104 nm2), and in mass-resolvingpower of an order of magnitude (to >1,500) since 2000.


Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


2. Laser pulsing has broadened the application of APT from metals and alloys to semicon-ductors, ceramic and geological materials, and organic and biological materials.

3. Specimen preparation for APT has become straightforward (with a FIB) such that site-specific analyses may be obtained routinely from such material features as grain bound-aries and processed transistors in integrated circuits.

4. APT is often used in correlative microscopy efforts in which a strong synergy with SEM,TEM, SIMS, and EPMA has been demonstrated. On matters for which other analyticaltechniques can provide similar information, very good agreement has been found.

5. APT offers unique information in three dimensions at the atomic scale; such informa-tion often provides valuable insights into materials. In so doing, APT is transformingmicroscopy at the atomic scale.

DISCLOSURE STATEMENT

The authors are employed by CAMECA Instruments, Inc., a manufacturer of atom probe tomog-raphy instrumentation.

ACKNOWLEDGMENTS

The authors would like to thank many individuals at CAMECA Instruments for their help withpreparation of this manuscript, especially T. Prosa, D. Lawrence, R. Ulfig, D. Olson, I. Martin,P. Clifton, D. Reinhard, J. Olson, D. Lenz, J. Bunton, J. Shepard, T. Payne, E. Strennen, B.Geiser, and E. Oltman.

The authors would like to acknowledge all the individuals who contributed to this manuscript,including R. Banerjee (University of North Texas); H. Fraser (The Ohio State University); B. Gaultand S.P. Ringer (University of Sydney); T. Ohkubo and K. Hono (National Institute for MaterialsScience, Japan); K. Inoue (University of Kyoto); M. Gilbert, A.K. Kambham, and W. Vandervorst(IMEC); R. Shivaraman (University of California, Santa Barbara); Vincent Smentkowski (GeneralElectric Global Research Center); Y. Chen and E.A. Marquis (University of Michigan); R.C. Reed(University of Birmingham); D. Isheim and D. Seidman (Northwestern University); C. Floss(Washington University); and L. Gordon and D. Joester (Northwestern University).

LITERATURE CITED

1. Miller MK, Kelly TF. 2010. The Atom TOMography (ATOM) Concept. Microsc. Microanal. 16(Suppl.2):1856–57

2. Kelly TF, Miller MK, Rajan K, Ringer SP, Borisevich AY, et al. 2011. Toward atomic-scale tomography:the ATOM Project. Microsc. Microanal. 17(Suppl. 2):708–9

3. Kelly TF, Miller MK, Rajan K, Ringer SP. 2012. Visions of atomic-scale tomography. Microsc. Today.In press; doi: 10.1017/S1551929512000211

4. Gault B, Haley D, De Geuser F, Moody MP, Marquis EA, et al. 2011. Advances in the reconstructionof atom probe tomography data. Ultramicroscopy 111:448–57

5. Larson DJ, Geiser BP, Prosa TJ, Gerstl SSA, Reinhard DA, Kelly TF. 2011. Improvements in planarfeature reconstructions in atom probe tomography. J. Microsc. 243:15–30

6. Müller EW. 1936. Versuche zur Theorie der Electronenemission unter der Einwirkung hoherFeldstärken. Ztg. Tech. Phys. 11:412–16

24 Kelly · Larson

Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


7. Müller EW. 1951. Das Feldionenmikroskop. Z. Phys. 131:136–428. Müller EW. 1956. Das Auflösungsvermögen des Feldionenmikroskopes. Z. Naturf. 11a:88–949. Müller EW. 1956. Resolution of the atomic structure of a metal surface by the field ion microscope.

J. Appl. Phys. 27:474–7610. Müller EW, Panitz JA, McLane SB. 1968. The atom-probe field ion microscope. Rev. Sci. Instrum.

39:83–8611. Panitz JA. 1973. The 10-cm atom probe. Rev. Sci. Instrum. 44:1034–3812. Cerezo A, Godfrey TJ, Smith GDW. 1988. Application of a position-sensitive detector to atom probe

microanalysis. Rev. Sci. Instrum. 59:862–6613. Kelly TF, Gribb TT, Olson JD, Martens RL, Shepard JD, et al. 2004. First data from a commercial

local electrode atom probe (LEAP). Microsc. Microanal. 10:373–8314. Bunton JH, Olson JD, Lenz DR, Kelly TF. 2007. Advances in pulsed-laser atom probe: instrument and

specimen design for optimum performance. Microsc. Microanal. 13:418–2715. Blavette D, Al Kassab T, Cadel E, Mackel A, Vurpillot F, et al. 2008. Laser-assisted atom probe tomog-

raphy and nanosciences. Int. J. Mater. Res. 99:454–6016. Hono K, Ohkubo T, Chen YM, Kodzuka M, Oh-ishi K, et al. 2011. Broadening the applications of the

atom probe technique by ultraviolet femtosecond laser. Ultramicroscopy 111:576–8317. Schlesiger R, Oberdorfer C, Wurz R, Greiwe G, Stender P, et al. 2010. Design of a laser-assisted

tomographic atom probe at Münster University. Rev. Sci. Instrum. 81:04370318. Miller MK, Cerezo A, Hetherington MG, Smith GDW. 1996. Atom Probe Field Ion Microscopy. Oxford:

Oxford Univ. Press19. Miller MK. 2000. Atom Probe Tomography: Analysis at the Atomic Level. New York: Kluwer Acad./Plenum20. Kelly TF, Miller MK. 2007. Atom probe tomography. Rev. Sci. Instrum. 78:03110121. Kelly TF, Larson DJ. 2012. The second revolution in atom probe tomography. MRS Bull. 37(2):150–5822. Cerezo A, Clifton PH, Lozano-Perez S, Panayi P, Sha G. 2007. Overview: recent progress in three-

dimensional atom probe instruments and applications. Microsc. Microanal. 13:408–1723. Bunton JH, Olson JD, Lenz DR, Kelly TF. 2007. Advances in pulsed-laser atom probe: instrument and

specimen design for optimum performance. Microsc. Microanal. 13:418–2724. Gault B, Vurpillot F, Vella A, Gilbert M, Menand A, et al. 2006. Design of a femtosecond laser assisted

tomographic atom probe. Rev. Sci. Instrum. 77:04370525. Bunton JH, Olson JD, Lenz DR, Larson DJ, Kelly TF. 2010. Optimized laser thermal pulsing of atom

probe tomography: LEAP 4000X. Microsc. Microanal. 16:10–1126. Kelly TF, Larson DJ. 2012. The second revolution in atom probe tomography. MRS Bull. 37:150–5827. Bostel A, Renaud L, Yavor M, Deconihout B. 2007. Sonde tomographique grand angle à haute résolution.

Fr. Patent No. FR292235028. Yavor M, Bostel A. 2007. Sonde atomique à haute acceptance. Fr. Patent No. FR292426929. Mamyrin BA, Karataev VI, Shmikk DV, Zagulin VA. 1973. The mass-reflectron, a new nonmagnetic

time-of-flight mass spectrometer with high resolution. Sov. Phys. JETP 37:4530. Waugh AR, Richardson CH, Jenkins R. 1992. APFIM 200—a reflectron-based atom probe. Surf. Sci.

266:501–531. Cerezo A, Godfrey TJ, Sijbrandij S, Smith GDW, Warren PJ. 1998. Performance of an energy com-

pensated three-dimensional atom probe. Rev. Sci. Instrum. 1:49–5832. Panayi P. 2006. Patents Nos. UK 0509638.3 and US 60/682,86333. Miller MK, Smith GDW. 1989. Atom Probe Microanalysis: Principles and Applications to Materials Problems.

Pittsburgh, PA: Mater. Res. Soc.34. Melmed AJ. 1991. The art and science and other aspects of making sharp tips. J. Vac. Sci. Technol.

B9:601–935. Thompson-Russell KC, Edington JW. 1977. Electron Microscope Specimen Preparation Techniques in Ma-

terials Science. Eindhoven: Philips Tech. Library. 134 pp.36. Kostrna SLP, Peterman JW, Prosa TJ, Wiener SA, Larson DJ, Kelly TF. 2006. An automatic electropol-

ishing system for needle-shaped specimens. Microsc. Microanal. 12:1750–5137. Walls JM, Southworth HN, Rushton GJ. 1974. Preparation of field electron–field ion emitters by ion

etching. Vacuum 24:475–79


Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


38. Hellsing M. 1988. High-resolution microanalysis of binder phase in as sintered Wc-Co cemented car-bides. Mater. Sci. Technol. 4:824–29

39. Henjered A, Norden H. 1983. A controlled specimen preparation technique for interface studies withatom-probe field-ion microscopy. J. Phys. E 16:617–19

40. Kvist A, Andren H-O, Lundin L. 1996. A specimen preparation technique for atom probe analysis ofthe near-surface region of cemented carbides. Appl. Surf. Sci. 94/95:356–61

41. Larson DJ, Camus PP, Vargas JL, Kelly TF, Miller MK. 1996. Specimen preparation and atom probefield ion microscopy of BSCCO-2212 superconductors. J. Phys. 6:271–76

42. Larson DJ, Russel K, Cerezo A. 2000. Sharpening of field-ion specimens and positioning of features ofinterest by ion-beam milling. J. Vac. Sci. Technol. 18:328–33

43. Larson DJ, Petford-Long AK, Cerezo A. 1998. Field ion microscopy of multilayer film devices. Microsc.Microanal. 4:112–13

44. Waugh AR, Payne S, Worrall GM, Smith GDW. 1984. In situ ion milling of field ion specimens usinga liquid metal ion source. J. Phys. 45:207–09

45. Alexander KB, Angelini P, Miller MK. 1989. Precision ion milling of field-ion specimens. J. Phys. Colloq.50:549–54

46. Hasegawa N, Hono K, Okano R, Fujimori H, Sakurai T. 1993. A method for preparing atom probespecimens for nanoscale compositional analysis of metallic thin flims. Appl. Surf. Sci. 67:407–12

47. Larson DJ, Foord DT, Petford-Long AK, Anthony TC, Rozdilsky IM, et al. 1998. Focused ion-beammilling for field-ion specimen preparation: preliminary investigations. Ultramicroscopy 75:147–59

48. Larson DJ, Petford-Long AK, Cerezo A, Smith GDW, Foord DT, Anthony TC. 1998. Three-dimensional atom probe field-ion microscopy observation of Cu/Co multilayer film structures.Appl. Phys. Lett. 73:1125–27

49. Larson DJ, Foord DT, Petford-Long AK, Cerezo A, Smith GDW. 1999. Focused ion-beam speci-men preparation for atom probe field-ion microscopy characterization of multilayer film structures.Nanotechnology 10:45–50

50. Larson DJ, Petford-Long AK, Cerezo A, Smith GDW. 1999. Three-dimensional atom probe studies ofmetallic multilayers. Acta Mater. 47:4019–24

51. Larson DJ, Petford-Long AK, Ma YQ, Cerezo A. 2004. Information storage materials: nanoscale char-acterization by three-dimensional atom probe analysis. Acta Mater. 52:2847–62

52. Larson DJ, Foord DT, Petford-Long AK, Liew H, Blamire MG, et al. 1999. Field-ion specimen prepa-ration using focused ion-beam milling. Ultramicroscopy 79:287–93

53. Larson DJ, Wissman BD, Martens R, Viellieux RJ, Kelly TF, et al. 2001. Advances in atom probespecimen fabrication from planar multilayer thin film structures. Microsc. Microanal. 7:24–31

54. Miller MK. 2005. Sculpting needle-shaped atom probe specimens with a dual beam FIB. Microsc.Microanal. 11:808

55. Miller MK, Russell KF, Thompson GB. 2005. Strategies for fabricating atom probe specimens with adual beam FIB. Ultramicroscopy 102:287–98

56. Thompson K, Gorman BP, Larson DJ, vanLeer B, Hong L. 2006. Minimization of Ga-induced FIBdamage using low energy clean-up. Microsc. Microanal. 12:1736–37

57. Miller MK, Russell KF, Thompson K, Alvis R, Larson DJ. 2007. Review of atom probe FIB-basedspecimen preparation methods. Microsc. Microanal. 13:428–36

58. Thompson K, Lawrence DJ, Larson DJ, Olson JD, Kelly TF, Gorman B. 2007. In-situ site-specificspecimen preparation for atom probe tomography. Ultramicroscopy 107:131–39

59. Miller MK, Russell KF. 2007. Atom probe specimen preparation with a dual beam SEM/FIB miller.Ultramicroscopy 107:761–66

60. Kelly TF, Martens RL, Goodman SL. 2003. U.S. Patent No. 6,576,90061. Giannuzzi LA, Drown JI, Brown SR, Irwin RB, Stevie FA. 1997. Focused ion beam milling and micro-

manipulation lift-out for site specific cross-section TEM specimen preparation. Mater. Res. Soc. Symp.Proc. 480:19–27

62. Nishikawa O, Kimoto M. 1994. Toward a scanning atom probe: computer simulation of electric field.Appl. Surf. Sci. 76/77:424–30

26 Kelly · Larson

Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.ann

ualr

evie

ws.

org

by B

row

n U

nive

rsity

on

07/2

2/12

. For

per

sona

l use

onl

y.


63. Kelly TF, Camus PP, Larson DJ, Holzman LM, Bajikar SS. 1996. On the many advantages of local-electrode atom probes. Ultramicroscopy 62:29–42

64. Kelly T, Larson DJ. 2000. Local electrode atom probes. Mater. Charact. 44:59–8565. Thompson K, Larson DJ, Ulfig R. 2005. Pre-sharpened and flat-top microtip coupons: a quantitative

comparison for atom-probe analysis studies. Microsc. Microanal. 11(Suppl. 2):88266. Cairney JM, Saxey DW, McGrouther D, Ringer SP. 2007. Site-specific specimen preparation for atom

probe tomography of grain boundaries. Phys. B 394:267–6967. Takahashi J, Kawakami K, Yamaguchi Y, Sugiyama M. 2007. Development of atom probe specimen

preparation techniques for specific regions in steel materials. Ultramicroscopy 107:744–4968. Perez-Willard F, Wolde-Giorgis D, Al Kassab T, Lopez GA, Mittemeijer EJ, et al. 2008. Focused ion

beam preparation of atom probe specimens containing a single crystallographically well-defined grainboundary. Micron 39:45–52

69. Saxey DW, Cairney JM, McGrouther D, Honma T, Ringer SP. 2007. Atom probe specimen fabricationmethods using a dual FIB/SEM. Ultramicroscopy 107:756–60

70. Felfer P, Ringer SP, Cairney JM. 2011. Shaping the lens of the atom probe: fabrication of site-specific,oriented specimens and application to grain boundary analysis. Ultramicroscopy 111:435–39

71. Larson DJ, Prosa TJ, Lawrence D, Geiser BP, Jones CM, Kelly TF. 2011. Atom probe tomography formicroelectronics. In Handbook of Instrumentation and Techniques for Semiconductor Nanostructure Charac-terization, ed. R Haight, F Ross, J Hannon, pp. 407–47. London: World Sci./Imperial Coll. Press

72. Larson DJ, Lawrence D, Olson D, Prosa TJ, Ulfig RM, et al. 2011. From the store shelf to device-levelatom probe analysis: an exercise in feasibility. Proc. Int. Symp. Test. Fail. Anal., 36th, pp. 189–97. MaterialsPark, OH: Am. Soc. Mater.

73. Marquis EA, Hu R, Rousseau T. 2011. A systematic approach for the study of radiation-induced segre-gation/depletion at grain boundaries in steels. J. Nucl. Mater. 413:1–4

74. Lawrence D, Alvis R, Olson D. 2008. Specimen preparation for cross section atom probe analysis. Microsc.Microanal. 14(Suppl. 2):1004–5

75. Gorman BP, Diercks D, Salmon N, Stach E, Amador G, Hartfield C. 2008. Hardware and techniquesfor cross-correlative TEM and atom probe analysis. Microsc. Today 16:42–47

76. Prosa TJ, Lawrence D, Olson D, Larson DJ, Marquis EA. 2009. Backside lift-out specimen preparation:reversing the analysis direction in atom probe tomography. Microsc. Microanal. 15:298–99

77. Colinge J-P, ed. 2008. FinFET and Other Multi-Gate Transistors. New York: Springer. 340 pp.78. Smith PJ, Smith DA. 1970. Preliminary calculations of the electric field and the stress on a field-ion

specimen. Philos. Mag. 21:907–1279. Srinivasan R, Banerjee R, Hwang JY, Viswanathan GB, Tiley J, et al. 2009. Atomic-scale structure and

chemical composition across order-disorder interfaces. Phys. Rev. Lett. 102:08610180. Mishin Y. 2004. Atomistic modeling of the γand γ′-phases of the Ni-Al system. Acta Mater. 52:1451–6781. Ardell AJ, Ozolins V. 2005. Trans-interface diffusion-controlled coarsening. Nat. Mater. 4:309–1682. Gault B, De Geuser F, Bourgeois L, Gabble BM, Ringer SP, Muddle BC. 2011. Atom probe tomography

and transmission electron microscopy characterisation of precipitation in an Al-Cu-Li-Mg-Ag alloy.Ultramicroscopy 111:683–89

83. Takahashi J, Kawakami K, Otsuka H, Fujii H. 2009. Atom probe analysis of titanium hydride precipitates.Ultramicroscopy 109:568–73

84. Hudson D, Cerezo A, Smith GDW. 2009. Zirconium oxidation on the atomic scale. Ultramicroscopy109:667–71

85. Gemma R, Al-Kassab T, Kirchheim R, Pundt A. 2009. APT analyses of deuterium-loaded Fe/V multi-layered films. Ultramicroscopy 109:631–36

86. Takahashi J, Kawakami K, Kobayashia Y, Tarui T. 2010. The first direct observation of hydrogentrapping sites in TiC precipitation-hardening steel through atom probe tomography. Scr. Mater. 63:261–64

87. Gemma R, Al-Kassab T, Kirchheim R, Pundt A. 2011. Analysis of deuterium in V-5at.%Fe film by atomprobe tomography (APT). J. Alloys Compd. 509:S872–76

88. Sepehri-Amin H, Ohkubo T, Nishiuchi T, Hirosawa S, Hono K. 2011. Quantitative laser atom probeanalyses of hydrogenation-disproportionated Nd-Fe-B powders. Ultramicroscopy 111:615–18


Ann

u. R

ev. M

ater

. Res

. 201

2.42

:1-3

1. D

ownl

oade

d fr

om w

ww

.a

atom probe tomography 2012 - ایران...

Documents