Scanning photoelectron microscope for nanoscale three-dimensionalspatial-resolved electron spectroscopy for chemical analysisK. Horiba, Y. Nakamura, N. Nagamura, S. Toyoda, H. Kumigashira et al. Citation: Rev. Sci. Instrum. 82, 113701 (2011); doi: 10.1063/1.3657156 View online: http://dx.doi.org/10.1063/1.3657156 View Table of Contents: http://rsi.aip.org/resource/1/RSINAK/v82/i11 Published by the American Institute of Physics. Related ArticlesIdentification of boron clusters in silicon crystal by B1s core-level X-ray photoelectron spectroscopy: A first-principles study Appl. Phys. Lett. 99, 191901 (2011) Effect of residual gases in high vacuum on the energy-level alignment at noble metal/organic interfaces APL: Org. Electron. Photonics 4, 237 (2011) Effect of residual gases in high vacuum on the energy-level alignment at noble metal/organic interfaces Appl. Phys. Lett. 99, 183302 (2011) Note: Heated sample platform for in situ temperature-programmed XPS Rev. Sci. Instrum. 82, 076106 (2011) A flexible apparatus for attosecond photoelectron spectroscopy of solids and surfaces Rev. Sci. Instrum. 82, 063104 (2011) Additional information on Rev. Sci. Instrum.Journal Homepage: http://rsi.aip.org Journal Information: http://rsi.aip.org/about/about_the_journal Top downloads: http://rsi.aip.org/features/most_downloaded Information for Authors: http://rsi.aip.org/authors

Downloaded 13 Nov 2011 to 133.11.175.250. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions

REVIEW OF SCIENTIFIC INSTRUMENTS 82, 113701 (2011)

Scanning photoelectron microscope for nanoscale three-dimensionalspatial-resolved electron spectroscopy for chemical analysis

K. Horiba,1,2,3,a) Y. Nakamura,1 N. Nagamura,1 S. Toyoda,1 H. Kumigashira,1,2,4

M. Oshima,1,2,3 K. Amemiya,3,5 Y. Senba,6 and H. Ohashi61Department of Applied Chemistry, The University of Tokyo, Tokyo 113-8656, Japan2Synchrotron Radiation Research Organization, The University of Tokyo, Tokyo 113-8656, Japan3Core Research for Evolutional Science and Technology (CREST), Japan Science and Technology Agency(JST), Tokyo 102-0075, Japan4Precursory Research for Embryonic Science and Technology (PRESTO), Japan Science and TechnologyAgency (JST), Saitama 332-0012, Japan5Photon Factory, Institute of Materials Structure Science, High Energy Accelerator Research Organization(KEK), Tsukuba 305-0801, Japan6JASRI/SPring-8, Hyogo 679-5198, Japan

(Received 11 May 2011; accepted 8 October 2011; published online 3 November 2011)

In order to achieve nondestructive observation of the three-dimensional spatially resolved electronicstructure of solids, we have developed a scanning photoelectron microscope system with the capabil-ity of depth profiling in electron spectroscopy for chemical analysis (ESCA). We call this system 3Dnano-ESCA. For focusing the x-ray, a Fresnel zone plate with a diameter of 200 μm and an outermostzone width of 35 nm is used. In order to obtain the angular dependence of the photoelectron spectrafor the depth-profile analysis without rotating the sample, we adopted a modified VG Scienta R3000analyzer with an acceptance angle of 60◦ as a high-resolution angle-resolved electron spectrometer.The system has been installed at the University-of-Tokyo Materials Science Outstation beamline,BL07LSU, at SPring-8. From the results of the line-scan profiles of the poly-Si/high-k gate patterns,we achieved a total spatial resolution better than 70 nm. The capability of our system for pinpointdepth-profile analysis and high-resolution chemical state analysis is demonstrated. © 2011 AmericanInstitute of Physics. [doi:10.1063/1.3657156]

I. INTRODUCTION

With the evolution of nanotechnology, there is a criticalrequirement for the nanoscale analyses of device structuresand self-assembled structures. In particular, it is important toinvestigate the nanoscale distribution of electronic structuresand chemical states along depth as well as lateral directionsin order to understand the mechanisms and characteristics ofcomplicated nanostructure species such as stacking structuresin semiconductor devices and the surface/interface reactionsof catalytic materials.

Photoelectron spectroscopy or electron spectroscopy forchemical analysis (ESCA) is one of the most powerfultechniques for investigating the electronic characteristics ofsolids, such as chemical bonding states, chemical-potentialshifts, the work function at the surface, and electronic struc-tures. ESCA equipped with a focused light probe has beenused to realize spatial distribution analysis of an electronicstructure while providing the advantages of ESCA. This com-bined system is called a scanning photoelectron microscope(SPEM). SPEM using synchrotron radiation (SR) has beenrealized at Elettra,1, 2 ALS,3 NSRRC,4 PLS,5, 6 and manyother synchrotron facilities. The advantage of SPEM overother microscopic techniques like photoemission electronmicroscopy7 is that an ESCA analysis using SPEM provides

a)Electronic mail: horiba@sr.t.u-tokyo.ac.jp.

detailed electronic and chemical information. However, a nor-mal SPEM system, and other microscopic techniques, enablesus to obtain the information only for lateral distributions ofthe electronic structures in two dimensions. One can obtainthe depth information by removing gradually materials fromthe sample surface by ion sputtering. However, such a destruc-tive analysis involves risk of misleading due to degradation ofthe sample during the destructive process.

In this paper, we report a newly developed SPEM sys-tem with a nondestructive depth profiling analysis capa-bility for three-dimensional (3D) spatially resolved ESCAanalysis of solids. We call this system 3D nano-ESCA.8

The concept of 3D nano-ESCA is illustrated in Fig. 1. ASR x-ray beam is focused to nanometer size on the sam-ples. We can obtain the lateral (x and y directions) dis-tribution of photoelectron spectra by scanning the samplesalong the lateral directions and acquiring the photoelectronspectrum at each point. This is the typical scheme withSPEM. In addition to SPEM measurements, we simultane-ously detect the angular distribution of emitted photoelectronsusing an angle-resolved photoelectron spectrometer. The an-gular distribution corresponds to the probing-depth depen-dence of the photoelectron spectra and can be converted intothe depth profiling information (z direction) using maximum-entropy methods (MEM).9 Thus, we can obtain the three-dimensional x, y, and z distribution of the electronic structureand chemical bonding states of the samples on a nanometerscale.

0034-6748/2011/82(11)/113701/6/$30.00 © 2011 American Institute of Physics82, 113701-1

Downloaded 13 Nov 2011 to 133.11.175.250. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions

113701-2 Horiba et al. Rev. Sci. Instrum. 82, 113701 (2011)

x

θ

SR

FresnelZone Plate

Angle-ResolvedPhotoelectron Spectrometer

Two-AxisSample Scanning

y Two-DimensionalDetector

ΔE

Δθ

zKinetic Energy

Em

ission Angle

Angular Distribution ofPhotoelectron Spectra

Con

cent

ratio

n

Depth Profile

SiO

N F

ilms

Si S

ub

stra

teMEM

analysis

FIG. 1. (Color online) Conceptual scheme of 3D nano-ESCA.

II. SYSTEM DESCRIPTION

A. Hardware overview

Figure 2 shows the schematic illustrations of the 3Dnano-ESCA system. This system has been installed at theUniversity-of-Tokyo Outstation beamline, BL07LSU,10 atSPring-8. As shown in the figure, the SR x-ray beam entersthe ultra high vacuum (UHV) chamber of the 3D nano-ESCAsystem. In the UHV chamber, a Fresnel zone plate (FZP) andan order-sorting aperture (OSA) are mounted on three-axisstages (Kleindiek Nanotechnik GmbH, LT6820XEYEZE-UHV) for precise alignment of optical components.

For focusing, we use a 200-μm-diameter FZP (NTT-AT)patterned on a 700 μm × 700 μm SiC membrane fabri-cated on a Si substrate; the FZP has a 100 μm center stopand zone patterns with an outermost zone width (�r) of35 nm. The OSA is an 80 μm pinhole placed between theFZP and the sample for extracting only the focused beam fromthe first-order diffraction. The OSA is designed to shield thespectrometer from secondary electrons produced at the FZPand OSA.

The theoretical spot size of an FZP can be expressed byEq. (1) (Ref. 11) as the square root of the sum of the squares of(1) the diffraction-limited resolution of the FZP, δi (obtainedby applying the Rayleigh criterion), (2) the geometrical reduc-tion ratio of the FZP, δg, which depends on the source size σ ,the source-optic distance p, and the optic-specimen distanceq, and (3) the chromatic aberration, δc, which depends on theFZP diameter, 2r, and the energy resolution of the SR x-ray(�E/E),

δm =√

δi2 + δg

2 + δc2

(a) Top View

(b) Side View

SR

Ion PumpAngle-Resolve

d

Photoelectron

Spectrometer

MCP DetectorSample

Extremely WideAcceptance AngleElectrical Lens

Coarse ScanSample Manipulator

Fine ScanPiezo Stage

FZP StageOSA Stage

SR

SR

FZP OSA Sample

30°

90°

SR

e-

θe

Configurationof Optics

(c) Components in UHV Chamber

Sample

MCP

FZP

OSA

Lens

FIG. 2. (Color online) Schematic diagram of 3D nano-ESCA system. (a) Topview of the 3D nano-ESCA chamber. The inset shows the configuration of thefocusing optics. (b) Side view of the chamber. (c) Photograph and 3D viewof the components in the UHV chamber.

=√(

1.22�r

m

)2

+(

σq

p

)2

+(

2r�E

E

)2

. (1)

In order to eliminate the effect of the geometric reductionratio, the 3D nano-ESCA system is set downstream from theexit slit of the beamline (imaginary source point of the x-ray)at a distance of 16 m.10 Since the focal length of our 200-μm-diameter FZP is calculated to be 5.6 mm at a photon energy of1000 eV, the reduction ratio q/p is reduced to ∼1/3000. There-fore, for a suitable energy resolution of the SR, the spatialresolution is limited by a function of the diffraction-limitedresolution δi = 1.22�r/m = 42.7 nm (where m is the diffrac-tion order).

Downloaded 13 Nov 2011 to 133.11.175.250. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions

113701-3 Horiba et al. Rev. Sci. Instrum. 82, 113701 (2011)

The FZP focuses the SR at a nanometer scale on the sam-ples mounted on a sample stage. The scanning sample stageconsists of a combination of a three-axis high-stability ma-nipulator (TOYAMA Co., Ltd.), a differentially pumped ro-tary feedthrough (DPRF) stage, and a two-axis piezo scanner(Physik Instrumente GmbH, PI-734). The three-axis manipu-lator is used for coarse sample movement perpendicular to theSR beam axes in the horizontal (x) and vertical (y) directions,as well as along the incident SR beam axis (z). The DPRFstage is used for adjusting the incident beam angle and thephotoelectron emission angle detected by the photoelectronspectrometer. The manipulator and DPRF stage are driven us-ing stepping motors; these stepping motors are placed out-side the UHV chamber to avoid leakage magnetic fields fromthe motors, which may affect the emitted photoelectrons. Thetwo-axis piezo scanner is used for precise sample scanningalong the x and y directions for SPEM measurements with amaximum scan range of 100 μm × 100 μm and an accuracyof 2 nm. Behind the scanning sample stage, a multichannelplate (MCP) detector is placed to detect the shape and flux ofthe transmitted photon for optical alignment or measurementsin the transmission mode.

An angle-resolved photoelectron spectrometer is at-tached to the UHV chamber at an angle of 60◦ from the inci-dent SR beam. In order to achieve high energy resolution anda wide acceptance angle, a modified angle-resolved photo-electron spectrometer (VG scienta R3000) with an extremelywide angle lens (EWAL) and a two-dimensional (energy andangular distributions) detector are utilized. The acceptanceangle of the EWAL is 60◦; the energy resolution and angu-lar resolution of the spectrometer is better than 2.5 meV and0.9◦, respectively. From the measurement configuration of oursystem, shown in the inset of Fig. 2(a), we can simultane-ously detect the angular distribution of photoelectron spectrawithin a range of 30◦–90◦ from the direction perpendicularto the sample surface. The two-dimensional detection systemis based on an MCP detector monitored by a FireWire CCDcamera. The angular distribution of photoelectron spectra canbe recorded in a flash by taking a snap shot of the detectorimage, covering a kinetic-energy range of 12% of the passenergy and an angular range of 60◦. By combining snap-shotdetection and high-speed piezo scannning, we can obtain aphotoelectron-intensity mapping within 50 ms per pixel.

B. Data analysis for depth profiling

For the characterization of three-dimensional distribu-tions of the electronic structures, depth profiling analysis isone of the most critical techniques. In fact, angle-resolvedphotoemission spectroscopy is a powerful, non-destructiveanalytical method to determine the atomic concentration andthe chemical bonding states as a function of depth in solids.The variation in photoemission spectra by changing the pho-toelectron emission angle arises because the photoelectrons ata given depth must travel through more material as the emis-sion angle is increased, so that the probability of inelastic pho-toelectron scattering increases. As a result, the surface sensi-tivity of grazing-emission measurements is enhanced relative

to that of normal-emission along the perpendicular directionto the sample surface.

Based on simple exponential attenuation for the photo-electrons traveling in solids, the intensity contribution froma given depth, z, at the emission angle, θ e, is expressedby Eq. (2) as

I = I0 exp

(− z

λ cos θe

), (2)

where λ is the inelastic mean free paths of the solids. Then,for example, in the case of a uniform film with thickness d ona substrate, the total photoelectron intensity from the film IF

and the substrate IS are given by Eq. (3).

IF =∫ d

0I0F exp

(− z

λF cos θe

)dz

IS =∫ ∞

dI0S exp

(− z − d

λS cos θe

)dz

× exp

(− d

λF cos θe

). (3)

Therefore, the thickness of the film, d, can be calculated fromthe photoelectron intensity ratio as

d = λF cos θe ln

(1 + 1

R0

IF

IS

) (R0 = I0F λF

I0S λS

). (4)

In the case of more complicated depth profile structures,some difficulties arise for the reconstruction of the depth pro-file from angle-resolved photoemission data because the prob-lem is fundamentally under-determined, such that many dif-ferent reconstructions (in principle, an infinite number) canappear to satisfy the constraints imposed by the set of angle-resolved measurements. Furthermore, the data are relativelynoisy, particularly for elements present in low concentrationsor buried deeply in the film, causing the reconstructed profilesto be extremely sensitive to small, not necessarily real varia-tions in the data. MEM has proven to be a powerful tool forconstructing concentration versus depth profiles from angle-resolved photoemission measurements.9, 12 The key conceptof this method is that it minimizes artificial correlations withinthe data by proposing a reconstruction that agrees with thedata but has the minimum amount of information. In order tofind this reconstruction, the problem can be described math-ematically as one in which a regularization function must bemaximized. It can be shown that the appropriate regulariza-tion function has the same form as the well-known expressionfor entropy. A useful expression for the entropy, S, is given by

S =∑

i

∑j

{ni, j − mi, j − ni, j log(ni, j/mi, j )}, (5)

where ni, j is the proportion of element i at depth j and mi, j

is the initial estimate for the proportion of element i at depthj in the absence of any data to the contrary. The logarithmicterm in the equation forces the proportions of the elements inthe reconstruction to be positive. The MEM solution for theangular distribution of photoelectron spectra is obtained bysearching for a maximum S for all possible depth profiles ni, j,subject to the condition that the calculated data agree withthat measured, within the noise. This condition is imposed by

Downloaded 13 Nov 2011 to 133.11.175.250. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions

113701-4 Horiba et al. Rev. Sci. Instrum. 82, 113701 (2011)

(a) (b)

765760

30˚40˚

50˚60˚

70˚80˚

Kinetic Energy (eV)

Em

ission Angle θ

e

4

3

2

1

0

ln(1

+I S

iON/R

oISi

)

432101/cosθe

30˚ 50˚ 60˚ 70˚65˚ 75˚Emission Angle θe

Si 2p

SiONSi sub.

d/λ = 1.33

SiONSi sub.

e-θe

d

FIG. 3. (Color online) (a) Angle-resolved Si 2p core-level photoelectron spectra of 2.5-nm-thick SiON thin films fabricated on Si substrates. All spectra aremeasured simultaneously using the two-dimensional detector of the photoelectron spectrometer without energy scanning and sample rotation. (b) Plot of intensityratio of photoelectrons ln(1 + ISiON/R0ISi) versus emission angle 1/cos θ . The slope of the linear fit (dashed line) represents the thickness d/λ. The inset showsthe measurement configuration.

ensuring that the weighted sum of squares error, χ2, is con-sistent with the uncertainty in the data. Thus, the MEM anal-ysis involves the process of maximizing the joint function, Q,combining χ2 and S with the regularizing coefficient, α, as

Q = αS − χ2

2. (6)

Sometimes, ambiguity in the parameter of α brings a differentoutput at particular cases in MEM analysis. Without properbounds on the magnitude of α, one could conclude incorrectresults of thickness and composition in the film. Therefore, itis necessary to optimize α elaborately by performing an em-pirical simulation.9

III. EXPERIMENTAL RESULTS

A. The depth profiling capability of the system usingangular distribution of photoelectron spectra

In order to demonstrate the depth profiling capabilityof the system using the angular distribution of photoelec-tron spectra, simultaneously detected angle-resolved Si 2pphotoelectron spectra of 2.5-nm-thick SiON thin films on Sisubstrates were obtained at an excitation photon energy of870 eV, as shown in Fig. 3(a). Sharp two-peak structures ata kinetic energy of around 765 eV and broad peak struc-tures at around 762 eV are derived from the Si substrates andSiON films, respectively. By increasing the emission angle,the intensity of the peak from the SiON films increases, andthe peak from the Si substrates disappears, since the grazing-emission condition is more sensitive to the surface.

The thickness of the SiON films, d, is represented byEq. (4) as

d

λSiON cos θe= ln

(1 + ISiON

R0 ISi

), (7)

where λSiON is the inelastic mean free paths of the SiONfilms; ISiON, the experimental photoelectron intensity fromthe SiON; and ISi, the photoelectron intensity from the Sisubstrates. R0 for SiON films on Si substrates is taken as

0.9329.13 Our experimental data fit very well with this the-oretical line for emission angles below 65◦, as shown inFig. 3(b). The discrepancy between the experimental data andthe theoretical line above 65◦ is because of the influence ofthe elastic scattering of photoelectrons in solids.14 The thick-ness of the SiON is determined from this analysis to be 1.33λ,which is a reasonable value since the thickness is 2.5 nm in thecase of λ = 1.88 nm.15, 16 Therefore, we confirm that precisedepth information on solids can be obtained from the angulardistribution of the photoelectron spectra using this system.

To determine depth profiles of atomic concentration andchemical states using MEM analysis for a more complicatedcase, we have also analyzed the angular distribution of core-level photoemission spectra for HfO2/SiO2 stack films fabri-cated on Si substrates. Figure 4(a) shows the core-level in-tensity ratio as a function of emission angle in HfO2/SiO2

stack films on Si substrates. Experimental data are simulatedby MEM depth profiling using prior model depth profilesas shown in Fig. 4(b) and a regularizing coefficient, α, of 5× 10−4.9 A depth resolution of 1 nm is applied to depth pro-files by convolution of Gaussian functions.17 Optimized depthprofiles explicitly exhibit double layer structures of HfO2

and SiO2 components, which is in good agreement with thecross-sectional transmission electron microscope (TEM) im-age shown in the inset of Fig. 4(b), indicating that we havesucceeded in obtaining precise depth profiles for such com-plicated stacking structures using our MEM analysis.

B. Photoelectron imaging and spot-size estimation

For testing our system, we carried out SPEM measure-ments on nano-patterned poly-Si gate electrodes fabricatedon high-k gate-stacking structures of HfO2/SiO2/Si, as shownin Fig. 5(a). Figure 5(b) shows the Hf 4f photoelectron in-tensity mapping of the poly-Si/HfO2 gate electrode patternswith a gate width wg of 1000 nm. The dark positions of theimage correspond to the poly-Si gate electrodes fabricatedon HfO2, reproducing the shape of the pattern well for thetop view of the structure, shown in Fig. 5(a). The spatial

Downloaded 13 Nov 2011 to 133.11.175.250. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions

113701-5 Horiba et al. Rev. Sci. Instrum. 82, 113701 (2011)

Con

cent

ratio

n (%

)

100

80

60

40

20

0

O

HfSi4+

Si0(b) Model

706050403020100100

80

60

40

20

0706050403020100

Depth (Å)

(c) MEM

O

Si4+

Si0

Hf

1.0

0.8

0.6

0.4

0.2

0.0

Inte

nsity

Rat

io

Emission Angle θe

(a)

Si 2p (Si0)

Si 2p (Si4+)

Hf 4f

30˚ 40˚ 50˚ 60˚20˚ 70˚ 80˚ 90˚

HfO2

Si sub.

e-θe

SiO2HfO2

Si sub.

SiO2

2 nm

FIG. 4. (Color online) (a) Core-level intensity ratio as a function of emissionangle in HfO2/SiO2 stack films fabricated on Si substrates. Open symbols in-dicate experimental data obtained from Hf 4f, oxidized (Si4 +), and substrate(Si0) Si 2p core-level spectra. The solid curves are angle profiles of intensityratios estimated by Eq. (2) after calculating depth profiles based on the MEMalgorithm. (b) A prior model for the initial guess for the MEM analysis, and(c) optimized depth profiles of the HfO2/SiO2 stack film on Si substrates. Theinset of (b) shows a cross-sectional TEM image of the measured sample.

Si sub.

Poly-SiHfO2

Top View Cross-SectionalView

(a)

SiO2

wg

0 1 2 3 4 50

1

2

3

x Position (μm)

y Po

sitio

n (μ

m)

(b) Hf 4f

A

wg = 1000 nm

Inte

nsity

(ar

b. u

nits

)

-400 -200 0 200 400y Position (nm)

Line Profile along A

Hf 4f Intensity Fitting

Δy = 70 nm

(c)

HfO2(on Si sub.)

HfO2(on SiO2)

FIG. 5. (Color online) (a) Schematic view of nano-patterned poly-Si/HfO2/SiO2/Si high-k gate-stacking structure for SPEM measurements. (b)Hf 4f photoelectron-intensity mapping of the gate-stacking structure with agate width of 1000 nm, obtained at an excitation photon energy of 870 eV.(c) Estimation of the spatial resolution from the line profile at the edge of thegate pattern along line A in (b).

resolution is estimated from a line profile at the edge of thegate pattern along line A in Fig. 5(b). The Hf 4f intensity pro-file in Fig. 5(c) is fitted using a step function convoluted witha Gaussian function, and the full width at the half maximumof the Gaussian function under the best-fit condition is de-termined to be 70 nm. Therefore, we have confirmed that thespatial resolution better than 70 nm is achieved. This is almostequal to the theoretical value of 66.1 nm. From our presentexperimental conditions with a source spot size σ of 100 μmand energy resolution of the SR x-ray E/�E of ∼ 5000, thetotal spot size (the spatial resolution) is calculated by Eq. (1)to be

√(42.7 nm)2 + (30.7 nm)2 + (40 nm)2 = 66.1 nm.

C. Pinpoint chemical state analysis

Figures 6(a) and 6(b) show the drain-current map andHf 4f photoelectron-intensity map, respectively, of the gate-stacking structure illustrated in Fig. 5(a), with a gate widthof 200 nm. Since the drain current originates from the com-pensation of the charge loss due to the emission of secondaryelectrons, the drain-current map corresponds to the image ob-tained using a conventional scanning electron microscope. Inboth maps, the shape of a gate line with a width of 200 nm isclearly observed. On the drain-current map, we can see somecontrast between the Si substrates and the SiO2 shallow trenchisolation (STI) layers under the HfO2 films, probably due tothe difference in the electrical conductivity of the underlay-ers. On the other hand, on the Hf 4f photoelectron-intensitymap, the contrast between the Si substrates and the SiO2 STIlayers disappears because the top HfO2 layers have the samecomposition and thickness. Nevertheless, different levels ofcontrast are observed between the edge and inner sides of thegate-stacking structure on the Si substrates, suggesting thatsome variations occur at the inner side of the HfO2 layers.

In order to clarify the origin of the contrasts between theedge and inner sides of the HfO2 layers on the Si substrates,we carried out a detailed analysis of the high-resolution pho-toemission spectra for different sample positions, at the points*a and *b in Figs. 6(a) and 6(b). The Hf 4f and O 1 s core-levelphotoemission spectra along with the results of the curve fit-ting analysis are shown in Figs. 6(c) and 6(d). In the Hf 4fspectra at *b, the intensity of the Hf-silicate component18, 19

increases and a Hf-silicide component20–22 appears, comparedwith the spectra at *a. Furthermore, in the O 1 s spectra, theintensity of the Hf-oxide component in the spectra at *b sig-nificantly decreases. Note that we can detect the very minorHf-silicide component, the spectral intensity of which is notmore than 0.2% of that of the main peak. From these results,we have found that a reduction in the HfO2 thickness andsome intermixing reactions between the HfO2 and SiO2 in-terfacial layers occur at the inner sides of the HfO2 layers onthe Si substrates. The origin of this inhomogeneity can be ex-plained by a problem during the Ar+ ion etching process forforming poly-Si gate patterns. Since the poly-Si/HfO2 stack-ing layers on the STI region are positively charged during ionetching, the etching rates can change between the peripheralregion and the inner region of the gate patterns depending onthe distance from the STI region. Thus, we have demonstrated

Downloaded 13 Nov 2011 to 133.11.175.250. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions

113701-6 Horiba et al. Rev. Sci. Instrum. 82, 113701 (2011)

0 20 40 60

0

10

20

x Position (μm)

y Po

sitio

n (μ

m)

(a)

(b)

0

10

20

Inte

nsity

(ar

b. u

nits

)

20 15

*a

*b Hf-O

Hf-O

Hf-Si-O

Hf-Si-O

Hf-Si

×30

535 530Binding Energy (eV)

*a

*bO-Si

O-Si

O-Hf

O-Hf

(c) (d)Hf 4f O 1s

*a *b

*a *b

hv = 870 eV hv = 870 eV

wg = 200 nm

×30

FIG. 6. (Color online) (a) Drain-current map and (b) Hf 4f photoelectron-intensity map of the gate-stacking structure with a gate width of 200 nmobtained at an excitation photon energy of 870 eV. (c) Hf 4f and (d) O 1score-level photoemission spectra at points *a and *b in (a) and (b).

the capabilities of our system and the importance of a pinpointchemical state analysis with high energy and spatial resolu-tions for developments in nanotechnologies.

IV. SUMMARY

We have developed a scanning photoelectron microscopesystem 3D nano-ESCA with depth profiling ESCA capa-bility. For focusing the x-ray, a FZP with a diameter of200 μm and an outermost zone width of 35 nm is used.In order to obtain the angular dependence of the photo-electron spectra for the depth profiling analysis without ro-tating the sample, we used a modified VG Scienta R3000analyzer with an acceptance angle of 60◦ as a high-resolutionangle-resolved electron spectrometer. The system has been in-stalled at the University-of-Tokyo Materials Science Outsta-tion beamline, BL07LSU, at SPring-8. From the results of theline-scan profiles of the poly-Si/high-k gate patterns, we haveachieved a total spatial resolution better than 70 nm. The ca-pability of our system for pinpoint depth-profile analysis andhigh-resolution chemical state analysis is demonstrated. Webelieve that this system is a powerful tool for investigatingnanoscale depth profiles and interface electronic structureson artificial and also self-assembled nanostrucutres, suchas semiconductor devices, nano-catalyst, nanoscale phase-separation phenomena, and so on.

ACKNOWLEDGMENTS

This research is supported by the Core Research forEvolutional Science and Technology (CREST) of the JapanScience and Technology Agency (JST) and the Japan Societyfor the Promotion of Science (JSPS) through its FundingProgram for World-Leading Innovative R&D on Scienceand Technology (FIRST Program). The authors would liketo thank Semiconductor Leading Edge Technologies, Inc.,(Selete) and Semiconductor Technology Academic ResearchCenter (STARC) for providing the semiconductor samples.The synchrotron-radiation photoemission measurementswere performed partly under project 08U004 at the Instituteof Materials Structure Science at KEK.

1L. Casalis, W. Jark, M. Kiskinova, D. Lonza, P. Melpignano, D. Morris,R. Rosei, A. Savoia, A. Abrami, C. Fava, P. Furlan, R. Pugliese, D. Vivoda,G. Sandrin, F. Q. Wei, S. Contarini, L. DeAngelis, C. Gariazzo, P. Nataletti,and G. R. Morrison, Rev. Sci. Instrum. 66, 4875 (1995).

2M. Marsi, L. Casalis, L. Gregoratti, S. Güenther, A. Kolmakov, J. Kovac,D. Lonza, and M. Kiskinova, J. Electron Spectrosc. Relat. Phenom. 84, 73(1997).

3T. Warwick, H. Ade, S. Cerasari, J. Delinger, K. Franck, A. Garcia,S. Hayakawa, A. Hitchcock, J. Kikuma, S. Klingler, J. Kortright, G. Moris-son, M. Moronne, E. Rightor, E. Rotenberg, S. Seal, H. J. Shin, W. F. Steele,and B. P. Tonner, J. Synchrotron Radiat. 5, 1090 (1998).

4I.-H. Hong, T.-H. Lee, G.-C. Yin, D.-H. Wei, J.-M. Juang, T.-E. Dann,R. Klauser, T. J. Chuang, C. T. Chen, and K.-L. Tsang, Nucl. Instrum.Methods Phys. Res. A 467, 905 (2001).

5M. K. Lee and H. J. Shin, Rev. Sci. Instrum. 72, 2605 (2001).6H. J. Shin and M. K. Lee, Nucl. Instrum. Methods Phys. Res. A 467, 909(2001).

7B. P. Tonner, G. R. Harp, S. F. Koranda, and J. Zhang, Rev. Sci. Instrum.63, 564 (1992).

8M. Oshima, S. Toyoda, H. Kamada, T. Tanimura, Y. Nakamura, K. Horiba,and H. Kumigashira, ECS Trans. 33, 231 (2010).

9S. Toyoda, J. Okabayashi, M. Oshima, G. L. Liu, Z. Liu, K. Ikeda, andK. Usuda, Surf. Interface Anal. 40, 1619 (2008).

10Y. Senba, S. Yamamoto, H. Ohashi, I. Matsuda, M. Fujisawa, A. Hara-sawa, T. Okuda, S. Takahashi, N. Nariyama, T. Matsushita, T. Ohata, Y. Fu-rukawa, T. Tanaka, K. Takeshita, S. Goto, H. Kitamura, A. Kakizaki, andM. Oshima, Nucl. Instrum. Methods Phys. Res. A 649, 58 (2011).

11S. Güenther, B. Kaulich, L. Gregoratti, and M. Kiskinova, Prog. Surf. Sci.70, 187 (2002).

12A. K. Livesey and G. C. Smith, J. Electron Spectrosc. Relat. Phenom. 67,439 (1994).

13M. P. Seah, Surf. Interface Anal. 37, 300 (2005).14K. Kimura, K. Nakajima, T. Conard, and W. Vandervorst, Appl. Phys. Lett.

91, 104106 (2007).15C. J. Powell and A. Jablonski, NIST Electron Effective Attenuation Length

Database, Version 1.0, National Institute of Standards and Technology,Gaithersburg, MD, 2001.

16S. Tanuma, C. J. Powell, and D. R. Penn, Surf. Interface Anal. 21, 165(1994).

17P. J. Cumpson, J. Electron Spectrosc. Relat. Phenom. 73, 25 (1995).18M. Oshima, S. Toyoda, T. Okumura, J. Okabayashi, H. Kumigashira,

K. Ono, M. Niwa, K. Usuda, and N. Hirashita, Appl. Phys. Lett. 83, 2172(2003).

19M. Helot, T. Chevolleau, L. Vallier, O. Joubert, E. Blanquet, A. Pisch,P. Mangiagalli, and T. Lill, J. Vac. Sci. Technol. A 24, 30(2006).

20L. Sha, R. Puthenkovilakam, Y.-S. Lin, and J. P. Chang, J. Vac. Sci. Tech-nol. B 21 2420 (2003).

21S. Toyoda, J. Okabayashi, H. Kumigashira, M. Oshima, K. Ono, M. Niwa,K. Usuda, and G. L. Liu, Appl. Phys. Lett. 84, 2328 (2004).

22H. Takahashi, S. Toyoda, J. Okabayashi, H. Kumigashira, M. Oshima,Y. Sugita, G. L. Liu, Z. Liu, and K. Usuda, Appl. Phys. Lett. 87, 012903(2005).

Downloaded 13 Nov 2011 to 133.11.175.250. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions