Hoffman D.M., Singh B., Thomas J.H. (Eds). Handbook of Vacuum Science and Technology

Подождите немного. Документ загружается.

734 Chapter 5.6: Common Analytical Methods for Surface and Thin Film

Electron spectroscopy provides qualitative analysis and quantitative analysis of

material surfaces, thin-film elemental depth profiling, chemical surface analysis,

and elemental mapping. These features and the instrumentation required to pro-

duce these effects are discussed in the following sections.

5.6.2.2 Qualitative Analysis

Qualitative analysis is performed through the analysis of the electron cloud emit-

ted from a material surface excited by an electron beam or a low-energy X-ray

source such as Mg K^ or Al K^ X-rays (1253 or 1468 eV, respectively). Materials

that can be analyzed include metals (conductors), insulators (polymers, inorganic

insulators, etc.), and semiconductors. Some organic materials can be damaged by

electron or X-ray irradiation, but in general, most materials can be reliably quali-

tatively analyzed. As described next, analysis is straightforward, and spectral in-

formation can be obtained from well-documented handbooks

[1-9}.

The physical excitation process in Auger electron spectroscopy is the Auger

transition, shown in Figure 2, where a bombarding electron inelastically scatters,

for example, from a core K-level electron producing a K-level hole. The scattered

and primary electron are free to leave the sample or to produce a cascade of sec-

ondary electrons through electron-electron interactions. To maintain charge equi-

Rg.2.

0) The photoemisslon process to produce a photoelectron;

(2) Auger emission process—an X-ray excitation of the atom is shown.

Atomic Energy Band Diagram

•^ (1)E,^ = hv-8i

(1)EK = hv-8B

(2)EK«A8B(K-L2)

In an

electron-excited

process,

primary

electrons

would

scatter

electrons

from

core

levels

and

Auger transitions

shown

would

occur.

5.6.2 The Electron Spectroscopies 735

librium in the atom, one of two relaxation processes may occur. An electron from

a shallower core level, Li, will lose its energy by filling the deep-level hole in the

K core level (generated by the scattering event) emitting an X-ray of energy E^ ~

EL-

Another process, known as the Auger process, can occur where the K hole is

filled by the electron from the L2 level by losing energy. The energy loss is syn-

chronously coupled to an electron in a nearby level,

L3.

The

electron is excited

to a kinetic energy approximately equal to the energy loss,

and may leave

the atom. The excited Auger electron is designated by the three levels involved in

the process, KLL. The energy associated with the Auger transition is unique to the

atom of origin and is used for chemical analysis. Typically, the core level is des-

ignated by standard spectroscopic notation (K, L12, L23,

Note that Auger processes can also be excited by X-rays.

Since a large number of secondary and primary electrons are observed in the

Auger spectrum, the signal is generally differentiated with respect to the kinetic

energy. This removes a large portion of the background and produces a very sen-

sitive method of observing the Auger transitions.

If the material is excited by nearly monochromatic X-rays, core-level electrons

can be excited. The excitation process must conserve energy, and hence the ki-

netic energy of the excited electron

Ey^

— hv

—

E^ where E^ is the binding en-

ergy of the core-level electron designated by spectroscopic notation, K, Li, L2,

M3,

M4, etc. The core-level excited electron has a unique binding energy, de-

pending on the atom of origin, and is used for chemical identification.

5.6.2.3 Surface Sensitivity Mechanism

When an electron leaves an excited atomic level in a material, it loses its origi-

nal kinetic energy primarily through electron-electron interactions and thermal

losses (phonons). The original electron kinetic energy is indicative of the atom of

origin (qualitative analysis). After the electron loses energy, it is no longer readily

identified with the original excitation and is called a loss electron. The range the

electron traverses before losing its first quanta of energy through some process in

the material, is known as the "electron inelastic mean free path" and is generally

of the order of a few atomic layers. This range depends on material and kinetic

energy. Electrons that escape from the surface can have energies equal to or less

than its kinetic energy {hv - E^). To identify surface species, the excited elec-

trons that leave the surface must not lose kinetic energy. Only "elastically" scat-

tered electrons leaving the surface add to the detected atomic core-level transition

(either electron- or photon-excited spectra). Therefore, both Auger electron and

photoelectron spectroscopy are surface sensitive for the same reason.

The electron "escape depth" is a function of electron kinetic energy, the mate-

736 Chapter

5.6:

Common Analytical Methods

for

Surface

and

Thin Film

Fig.

Electron escape depths taken from literature references.

Electron Escape Depth

100

100 1000

Kinetic Energy (eV)

The straight line fit shows that escape depth depends on kinetic energy of

the

emitted electron.

The usual power law dependence is

E^-^-.

rial parameters and the physical geometry of the detection system relative to the

sample surface. Figure 3 shows representative electron escape depths (at a 90-

degree takeoff angle relative to the sample surface) as a function of the emitted

electron kinetic energy.

5,6.2.4

Chemical Information (chemical shift)

In addition to providing near-surface region elemental analysis, electron spec-

troscopy can also provide limited chemical information. When an electronegative

atom and an electropositive atom bond to form a compound, the electron charge

distribution around the atoms redistributes to maintain electrical neutrality. The

redistribution of charge produces a change in the core-level binding energies. Us-

ing photon-stimulated or electron-stimulated spectroscopy, the perturbed kinetic

energy of a core level can be probed and appears as a shift in energy and/or a

change in line shape. The perturbation is known as "chemical shift," and its de-

velopment for use in ESCA (electron spectroscopy for chemical analysis) or X-

ray photoelectron spectroscopy [1] resulted in Kai Seigbaghn receiving the Nobel

prize in chemistry.

5.6.2 The Electron Spectroscopies

737

Fig.

The spectrum of the Si 2p region was obtained using 130 eV photons

from a synchrotron source.

Classical Q v AEg

S 5 r-

Silicon

J-

5 4 3 2 10

Change in Binding Energy (eV)

1 2 3

Chemical Shift (eV)

The sample is a thin silicon dioxide layer on silicon. Interfacial states at the Si02/Si interface

are believed to be due to silicon bonded in SiOx where

A-

= 0, 1, 2, 3, 4. The chemical shift is

plotted as a function of classical charge

and is nearly a straight line [11].

A simple example is shown in Figure 4 where the oxide states of silicon are

compared. The binding energy of the Si Ip core level shifts ~5 eV from its ele-

mental position. Using classical theory to compute the amount of charge pres-

ent on silicon (electronegativity calculation), and plotting this as a function of

the shift in binding energy (chemical shift) results in a nearly linear dependence.

Therefore, the peak position along with quantitative chemicaly analysis can be

used to identify functionality at the surface. This is mostly used in polymer chem-

istry, where the shifts and the shape of the C Is peak are most useful in functional

identification.

5.6.2.5 Quantitative Analysis

Both AES and XPS are reasonably quantitative techniques. Historically, elemen-

tal standards have been employed to calibrate an experiment. This requires the

use of in situ cleaning of the standard to produce a clean surface for this purpose.

Both methods obtain spectra in the N(E) mode, that is, direct electron counting is

employed. Elemental sensitivity factors have been derived empirically and theo-

retically

[2-6].

Over the periodic table of elements, the sensitivity factors may

change by as much as 20. By the proper choice of the transitions used in the

analysis, the effect of sensitivity variations can be minimized.

With elemental standards available, a good first approximation to quantitative

analysis is possible by obtaining the elemental signal amplitudes from the un-

738 Chapter 5.6: Common Analytical Methods for Surface and Thin Film

known and standard spectrum (usually by integrating the spectral region digi-

tally).

The unknown concentration is then obtained from the ratio of the unknown

to the standard.

When standards are not available, quantitative analysis is still possible. By

measuring all the elemental peak areas of the unknown, the normalized atomic

concentration is given by

X?,

= (/«>«)-Y/y^M (1)

where X is the normalized atomic concentration of element a in the matrix m, / is

the amplitude of the /-th element in the matrix m, and a is the elemental sensitiv-

ity factor. The sensitivity factors can be improved by including the element in a

material similar to what is being analyzed. Generally, this expression provides a

reasonably good normalized concentration. For further information, the reader is

referred to References

[2-7].

As an example of the application of this methodology. Figure 5 shows a typical

XPS spectrum of an unknown surface. The elements are identified from peak bind-

ing energies. With the aid of software, the peak areas are integrated. The elemen-

tal peak areas are placed in a table and the normalization procedure, as shown in

Equation (1), is performed. After the computation is performed, an elemental sur-

face compositional table is available, as shown next for the unknown.

Compare this with what is expected from the chemical formula

CH3

-esio-^„

CH3

where the O to Si to C ratio is expected to be

1:1:2.

The experimental data are

found to be in reasonable agreement. Therefore, the unknown is identified as sili-

cone (polydimethylsiloxane).

AES quantitation is very similar. The exception is when the data are obtained

in the derivative mode, that is, the spectral data are A^(^) • dNIdE versus E. In-

stead of measuring the peak area, the peak-to-peak signal amplitude is measured,

and the appropriate elemental sensitivity factors for derivative data are applied.

5.6.2.6 Elemental Depth Profiling

and Cliemicai Depth Profiling

Both AES and XPS are typically equipped with ion milling or sputtering using

rare gases (He, Ne, Ar, Xe). Ion gun designs permit a focused beam of rare gas

5.6.2 The Electron Spectroscopies

739

Fig.

Polydimethylsiloxane spectrum obtained using an aluminum K^

monocliromatic X-ray source.

Element

Si2p

C \s

0 Is

350

Binding Energy (eV)

Elemental Composition Table

Sensitivity

Factor [12]

0.800

1.000

2.950

Peak

Area

3000

6550

8400

Atomic %

The peak amplitudes are determined by integrating the area under the peaks.

ions to be produced with kinetic energy from 500 eV through 10 keV. The

beams can be focused to spot sizes on the order of 10 /im but are generally used

at larger spot sizes (depending on the ion energy). To produce an elemental depth

profile that is accurate, the ion beam is rastered over an area larger than the ana-

lyzed area [3].

Auger elemental depth profiling is typically performed by obtaining the N(E)

spectrum, taking the derivative, and subsequently measuring the peak-to-peak

amplitude of the measured peaks as a function of sputter time. Data may also be

obtained from the peak amplitude minus a point near the peak to give a peak

height. This information is used to obtain a normalized atomic concentration. Fig-

ure 6 shows a typical Auger elemental depth profile where the sputter time was

740 Chapter 5.6: Common Analytical Methods for Surface and Thin Film

Fig.

/""]. Titani

Titanium

0 1 2

Sputter Depth based on Si02

(jmm)

Elemental depth profile of a soldered connection to metalized silicon where the metalization

was peeled off the silicon prior to analysis.

converted to depth in

/mm

based on the sputter rate for Si02. The Auger peak-to-

peak amplitude was not converted to elemental concentration.

It is possible to estimate the film thickness from the sputter time. However, this

can vary widely depending on the matrix and the difference in sputter yields. The

time scale can be converted to depth using thickness standards of

known material.

When X-rays are the excitation source, the photoelectron peaks are tracked as

a function of depth. The peak areas are obtained and normalized using elemental

sensitivity factors to provide a plot of atomic concentration as a function of sput-

ter time. When the ion bombardment process does not degrade the surface, this

technique can be used to obtain chemical information from chemical shift data as

a function of sputter time.

5.6.2 The Electron Spectroscopies 741

5.6.2.7 Auger Elemental Mapping

One of the main features of using Auger electron spectroscopy as a surface ana-

lytical method is the use of the very finely focused electron beam obtainable with

modern electron columns (similar to scanning electron microscopes). The elec-

tron sources are typically LaB^ (single-crystal) or field emission sources. These

sources provide a high brightness beam at very small spot sizes (<100 A diame-

ter beams).

Elemental mapping is accomplished by rastering the electron beam across the

surface where at each point (pixel), the elemental signal amplitude is measured

relative to the background. The background amplitude is used to correct for some

of the topographic effects encountered in the analysis. These data are stored in a

data file on a computer where it can later be manipulated using software. The pic-

ture can also be obtained in an analog fashion by photographing the Auger signal

amplitude displayed as Z-axis modulation on an oscilloscope.

An Auger elemental map of silicon and carbon in silicon carbide fibers in a ti-

tanium matrix is shown in Figure 7. The fibers consist of a graphite core sur-

rounded by SiC and a graphite coating encapsulating the SiC.

5.6.2.8 Instrumentation

Electrostatic analysis of the emitted electrons is performed using a number of

conmion analyzers. In the following, three of the most common analyzers are

briefly described.

The cylindrical mirror analyzer (CMA) is used mainly in AES systems. This

instrument consists of concentric cylinders. The size of the outer cylinder is fixed

depending on the inner cylinder diameter. Electrons are accepted into the struc-

ture where the outside cylinder is biased negative, and this voltage is scanned.

The geometrical ratio of the cylinders fixes the energy resolution to —0.2-0.5%

of the kinetic energy.

The excitation source, generally an electron gun, is mounted coaxially in the

inner cylinder of the analyzer but can be mounted externally to the CMA. Both

assemblies are at ground potential. The electron gun can be a high-quality mag-

netically focused gun or an electrostatically focused electron gun. Both instru-

ments are available on the commercial market. The coaxial geometry provides a

small-size structure where the source is axially aligned with the analyzer. The

sample for analysis must be at the proper location relative to the analyzer to angu-

lar restrictions. A typical simple AES system based on this analyzer is shown in

Figure 8. Electrons are collected from a 360-degree cone at an angle of —42 de-

742 Chapter 5.6: Common Analytical Methods for Surface and Thin Film

Fig.

AES map of

carbon

and silicon in a multilayer

fiber

in a titanium matrix.

grees relative to the cylindrical axis. Other geometries degrade the amount of sig-

nal obtainable from the analyzer. Analyzed electrons are collected using a multi-

channel plate or simple multiplate electron multiplier. This information is used to

produce spectra as shown.

The second most popular analyzer is the hemispherical sector analyzer (HSA).

Historically, this instrument has been used with XPS systems due to its high-

energy resolution. The HSA consists of two hemispherical sectors placed one

within the other (Figure 9). The spacing of the sectors (the difference in the radius

of the spheres) determines the basic energy resolution. As with the CMA, the

outer sphere is biased negative, relative to the inner sphere. The HSA is equipped

with entrance and exit slits, which may be either physical or electronic. The phys-

ical size of the slits determine the analyzed area.

Fig.

A simple

AES

single-pass

CMA,

detector, and

e-gun

electronics.

Outer Cylinder (-)

Inner

A Cylinder

Sample

Detector

jComputer Control

Cylinder Supply

Detector

Data Acquisition

The detector can be an electron multiplier or multichannel plate type detector.

Fig.

Hemispherical Sector Analyzer

Channel Plate

\ ' I '^ Multiplier

Position Sensitive

Detector

Computer:

Lens,

cylinders,

multiplier,

detector control

& data acquisition!

Source (x-ray or

e-beam)

' "" ' SAMPLE

A hemispherical-analyzer-based system used either for XPS or AES.