Sei sulla pagina 1di 15

Measuring the Thickness of Aluminium Alloy Thin

Foils Using Electron Energy Loss Spectroscopy


A. Bardal* and K. Lie†
*SINTEF Materials Technology, N-7465 Trondheim, Norway and †Department of Physics,
Norwegian University of Science and Technology, N-7491 Trondheim, Norway

Combining electron energy loss spectroscopy and convergent beam electron diffraction
measurements, we have determined the mean-free-path for inelastic electron scattering for
four different aluminium alloys. Electron energy loss spectroscopy spectra were acquired
with the transmission electron microscope in image mode, without any objective aperture
inserted. The value for pure aluminium was determined as 119 ⫾ 5nm at an incident elec-
tron energy of 150keV. For the most common alloys with moderate amounts of alloying ele-
ments, no significant changes in the mean-free-path values are measured within the 5% ac-
curacy of the method. The-mean-free path values are required for accurate determination of
the thickness of thin foils for transmission electron microscopy using electron energy loss
spectroscopy. On the basis of our experimental results, we estimate a single measurement
of foil thickness to have an accuracy of better than 8% if the mean-free-path is known
within 5%. The outlined procedure for thickness measurements is very robust, with thick-
ness results being insensitive to experimental details such as inclination of the specimen or
diffracting conditions. © Elsevier Science Inc., 2000. All rights reseved.

INTRODUCTION chemical composition of the sample using


energy-dispersive X-ray spectroscopy. A
When using transmission electron micros- number of methods have been used in the
copy for measuring the densities of micro- past for measuring the thickness of thin
structural features such as dislocations, foils for transmission electron microscopy.
precipitates, or dispersoids in metallurgical These include the contamination spot
samples, information on the thin-foil thick- method, the use of thickness fringes at an
ness is needed. It is important to obtain as inclined boundary, and the use of the X-ray
accurate measurements of the thickness as yield’s dependence on thickness (as re-
possible. At the same time, to obtain statis- viewed, for example, in [1, 2]). These meth-
tically reliable data, a large number of den- ods suffer from being time consuming, of
sity measurements is often required. If a poor accuracy, or both. Convergent-beam
thickness measurement is required from electron diffraction (CBED) yields good ac-
each sample used for the density measure- curacy, better than 2–5% relative error in
ments, the time consumption involved the simplest version where the periodicity
with the thickness measurement becomes a of extinction fringes under two-beam con-
very important issue. Apart from the im- ditions is utilized [3–5]. Even better accu-
portance for density measurements, an ac- racy can be obtained by quantitative many-
curate knowledge of sample thickness is parameter fits to the intensity profiles of
also in some cases important for absorption the CBED disks [6]. The CBED methods
corrections when determining the local are, however, quite time consuming (sev-
329
MATERIALS CHARACTERIZATION 44:329–343 (2000)
© Elsevier Science Inc., 2000. All rights reserved. 1044-5803/00/$–see front matter
655 Avenue of the Americas, New York, NY 10010 PII S1044-5803(99)00072-8
330 A. Bardal and K. Lie

eral minutes per measurement at best), in- deformed over a wide range in the Zener-
volving controlled tilting of the sample and Hollomon parameter (a dimensionless para-
fairly elaborate data evaluation. meter for deformation kinetics that includes
The electron energy loss spectroscopy strain rate, temperature, and activation en-
(EELS) method for measuring sample ergy for dislocation movement). Here, more
thickness is based on a simple relationship than 600 individual samples of ␳ were ob-
between the intensity distribution in the tained, each sample involving one thick-
EELS spectrum and the ratio between the ness measurement. Lodgaard and Ryum
sample thickness and a mean free path for (Norwegian University of Science and
inelastic scattering [7]. With use of a paral- Technology, Private Communication, 1998)
lel EELS spectrometer, the method is very carried out a comprehensive study of how
fast, with a recording time of fractions of a the densities of ␣-Al(Fe,Mn)Si dispersoids
second per spectrum, and a data evaluation in a 6082 alloy varied with annealing time
time per spectrum of on the order of 20 s. in the temperature range of 520–550–580⬚C,
The method is, therefore, many times faster which involved approximately 400 thick-
than the alternatives. The real benefit from ness measurements.
this is gained when many thickness mea- The EELS method has potential for being
surements need to be made. For instance, very accurate, although not reaching the
500 thickness measurements using EELS accuracy of 2% or better obtained using
would take about 3 h, whereas the same CBED. The accuracy of the EELS method
number of measurements using CBED or depends on the accuracy with which the
the contamination spot method would take mean free path value is known for the mate-
50 h or more. EELS, therefore, opens up the rial in question, and on the accuracy with
possibility of sampling real distributions of which the intensity of the zero loss peak can
sizes or densities, where the alternative be extracted from the experimental spec-
would have been to obtain only mean trum. Malis et al. [10] established a parame-
values based on a limited number of mea- terized expression for the mean-free-path
surements. Quite large matrices of materi- value based on experimental measure-
als conditions can be studied, without ments on 11 different materials. Using this
thickness measurements being a severe approach, thickness measurements are
time-limiting factor. expected to be accurate within ⫾20% for
Some examples from recent use of the most inorganic specimens, without any ex-
method in projects at our laboratories may plicit calibration of the mean-free-path
illustrate the benefits of having a really fast value being necessary. Considerably better
measurement technique at hand. In [8], we accuracy is within reach, however, if it is
studied deformation structures in a plane- possible, and one takes the trouble, to ex-
strain compressed aluminium alloy (3004) plicitly calibrate the mean-free-path value
to relate these to recrystallization behavior. for the material in question. There are few
The distributions of dislocation density (␳), reports in the literature on the attainable
subgrain boundary misorientation (␪), and accuracy in the latter case. Working on an
subgrain size (␦) were measured within the AA3004 aluminium alloy, Botton et al. [11]
different texture components of the mate- found a mean-free-path value of ␭ ⫽ 175 ⫾
rial. The work involved obtaining approxi- 5nm at an accelerating voltage of 300kV
mately 200 individual samples of disloca- and a spectrometer collection angle of
tion densities from different regions and 9.5mrad. That is, ␭ was determined with an
texture components of the transmission accuracy of 3% by means of combined
electron microscopy specimens, each sam- CBED and EELS measurements.
ple involving one thickness measurement. The motivation for the work reported in
In [9], the distributions of the same ␳, ␦, and this paper was to determine mean-free-
␪ microstructural parameters were mea- path values and accuracies for a selection
sured for a series of materials that had been of aluminium alloys, and to establish a
Measuring Thickness of Aluminum Alloy Thin Foils with EELS 331

measurement procedure that takes full measurements, we determined the mean-


advantage of the possible speed of EELS free-path value ␭ appropriate for the cho-
thickness measurements, while at the same sen procedure and investigated how ␭ var-
time maintaining a highest possible accu- ied with alloy composition and temper,
racy. The major concern for such a proce- and with experimental condition such as
dure is that one should avoid any time-con- diffracting condition and specimen tilt. On
suming controlled tilting of the transmission the basis of this work, mean-free-path
electron microscopy specimen. This means values and accuracies can also be estimated
that the crystal orientation and surface in- for samples in which information on alloy
clination might vary, and might be un- composition and temper is incomplete, and
known for the recorded EELS spectra. If the for which no explicit calibration of ␭ has
specimen is in a strongly diffracting condi- been made.
tion, the Bragg diffraction may cause an an-
gular redistribution of the ratio between in-
elastically and elastically scattered electrons.
If the EELS spectrum is recorded with an THEORY
angle-limiting aperture, the relative inten-
sity of the zero-loss electrons will be af- The total intensity of the electrons that are
fected by such elastic-inelastic coupling. inelastically scattered while being transmit-
The coupled plural inelastic scattering and ted through a thin foil of thickness t can be
elastic Bragg diffraction is not easily treated expressed as:
theoretically, but experimental observations
i
show that the effect may be significant. Bot- I tot ( t ) = I surf + Σ n ,i P n ( t ) (1)
ton et al. [11] measured variations in elastic/
inelastic ratios of 10% across bend contours, where Isurf is a thickness-independent sur-
and attributed this to Bragg diffraction. face contribution, and Pni are probabilities
Our approach was to work with the for the different inelastic scattering pro-
transmission electron micoscope (TEM) in cesses at thickness t. These probabilities can
image mode while doing the EELS mea- be expressed by a multiparameter Poisson
surements, as well as the imaging of the distribution [12]:
microstructural feature of interest. The i n
EELS spectrometer can then collect all elec- P n ( t ) = 1 ⁄ n! ⋅ ( t ⁄ λ i ) ⋅ exp ( – t ⁄ λ ) (2)
trons scattered up to large angles, and there
where the index i denotes various types of
is no need to worry about angular redistri-
scattering processes, and the index n de-
bution of the scattering, or to tilt the speci-
notes the order of plural scattering. The
men to nondiffracting or well-defined dif-
total mean-free-path value ␭, and the mean-
fracting orientations. The specimen tilt and
free-paths for the individual scattering pro-
crystallographic orientation can still, in
cesses ␭i, are related by:
principle, affect the result by modifying the
probability for some of the inelastic scatter- 1 ⁄ λ = Σi 1 ⁄ λi (3)
ing processes. The excitation of surface plas-
mons is enhanced if the sample surface is as derived from the scattering cross-section:
inclined to the incident electron beam,
σ = Σσ i (4)
whereas excitation of inner-shell electrons
is affected by localization of the electron because
Bloch waves under strongly diffracting
conditions. These effects were expected to σi = 1 ⁄ na ⋅ λi (5)
be small and probably negligible, but an where na is the number of atoms per vol-
experimental verification of this view is in ume. Setting n ⫽ 0 in Eq. (2), yields:
order and is made here.
Using a combination of EELS and CBED t ⁄ λ = ln ( I tot ⁄ I 0 ) (6)
332 A. Bardal and K. Lie

where I0 is the intensity of the zero-loss The dependence of ␴p and ␭p on the energy
peak, and Itot is the total intensity in the of the incident electrons and on the plasmon
spectrum. energy Ep is given by Eqs. (8) and (5) [7], and
The most important inelastic scattering the dependence of Ep on the volume density
process is the excitation of volume plas- of conduction electrons (n) by Eq. (9).
mons, i.e., collective oscillations of conduc- 2
σ p ( β ) ≈ Ep ⋅ ln ( β ⁄ θ E ) ⁄ n a a 0 m 0 v (8)
tion electrons [13]. With n ⫽ 1 and i ⫽ plas-
mon scattering in Eq. (2): 2 2 1⁄2
E p = [ ne h ⁄ ε 0 m ] (9)
t ⁄ λ plasmon = I 1 ⁄ I 0 (7)
Here, ␤ is the collection angle, ␪E is the
where I1 is the intensity contained in the characteristic scattering angle, a0 is the Bohr
first plasmon peak. ␭ and ␭plasmon have been radius, m0 and m are the rest mass and rela-
shown experimentally to differ by approxi- tivistic mass of the electron, respectively,
mately 30% (see, e.g., [14]), illustrating that and n is the volume density of conduction
other processes than plasmon scattering electrons. Alloying elements in solution
also contribute significantly. The other in- give rise to a changed volume-density of
elastic scattering processes that need to be conduction electrons, Eq. (9), causing shifted
considered are coherent double plasmon Ep, ␭p and ␴p values. The Ep value shows a
excitation, excitation of single conduction linear dependence on solute concentration
electrons, excitation of single core elec- in certain cases, as reported for a number of
trons, and excitation of surface plasmons. alloy systems such as Al–Mg, Al–Mg–Zn,
In addition, the incident electrons undergo Al–Zn, Al–Ag, and Al–Li [15]. As an exam-
phonon scattering. The latter process is ple, we would expect 5% magnesium in
quasielastic (losses less than 100meV) and solid solution to yield a decrease in Ep of
the thermal-diffuse scattered electrons con- approximately 0.25eV, corresponding to a
tribute to the zero-loss peak under the ex- decrease in ␴p of about 2%, Eq. (8).
perimental conditions given by a transmis- Apart from excitation of ordinary volume
sion electron microscope. plasmons, there is also the possibility for co-
It is appropriate here to briefly discuss herent double plasmon excitation, i.e., the
the different scattering processes, the theo- excitation of two plasmons by one electron
retically expected effect of microstructural in a single scattering event [12]. The most
features of the aluminium alloy on the recent literature on this subject [16, 17] agrees
cross sections, and the effects of experimen- that the probability of this process is very
tal parameters such as sample inclination small, with P2/P1 being less than 1–3% (P2
and diffracting conditions. Because the and P1 are the probabilities for coherent
plasmon excitations are collective excita- double plasmon excitation and ordinary
tions of a medium, the cross-sections of plasmon excitation, respectively). In [16], a
these processes depend on the microstruc- review is also given of the wide scatter in
tural state of the alloy. In principle, it reported experimental values for this ratio
makes a difference whether the medium as well as a discussion of possible sources
consists of a single phase with solute at- of error in the different experiments. In the
oms, or if it is a composite medium with present context, this process can be com-
several phases in the form of fine precipi- pletely ignored.
tates, dispersoids, or primary particles in a In addition to the volume plasmons that
matrix. It is convenient to discuss and com- propagate inside the solid, longitudinal
pare the contributions from different scatter- waves of charge density that travel along
ing processes in terms of their cross sections, the surface, i.e., surface plasmons, are ex-
because these are additive [see Eq. (4)]. cited by the incident electron. In the practi-
Volume plasmons in aluminium can be cal case when the metal is covered by a thin
well described by the free-electron theory. layer of surface oxide, the characteristic en-
Measuring Thickness of Aluminum Alloy Thin Foils with EELS 333

ergy of the surface plasmon is Es ⫽ Ep/(1 ⫹ number roughly as ␴ ∝ Z1/3, which can be
ε1)1/2, where ε1 is the real part of the dielec- used to estimate how the e–e cross-section
tric permitivity of the oxide. Calculations varies with alloy content.
and experiment confirm that a typical na- For excitation of core electrons (ioniza-
tive oxide thickness of 4nm is sufficient to tion), the scattering cross-sections can be
yield a surface plasmon energy of alumin- treated within an atomic framework to a
ium of Es ⬇ 7eV [18, 19]. For an Al speci- good approximation (although the fine
men with native oxide at normal incidence, structure of ionization edges, of course, can
the probability for surface plasmon excita- depend strongly on bonding and the chemi-
tion is estimated as 2% [7]. The surface cal state of the compound). Hence, ioniza-
plasmon peak is easily visible in spectra tion cross-sections for the alloy can be
from thin specimens, and will be stronger if found by simple averaging of cross-sections
the specimen is tilted (i.e., the incident elec- for the different species, according to the
tron arrives at a non-normal incidence an- overall composition of the probed volume,
gle). A general expression for the effect of and are essentially independent of the
incidence angle on the probability for sur- alloy’s microstructural state and temper.
face plasmon excitation is given in [20]. The Scattering cross-sections for the excitation
total surface plasmon intensity for a given of core electrons can be calculated to good
angle of incidence, integrated over all scat- accuracy using, for example, the hydro-
tering angles, can be derived from this ex- genic model [7] or the Hartree-Slater model
pression, but cumbersome integration is re- [24]. Both approaches are implemented in
quired. the commercial EELS software [25]. The
A competing mechanism to excitation of core-loss scattering process of major rele-
the conduction electrons as plasmons, is the vance in this context is the ionization of
excitation of single conduction electrons aluminium L electrons with an ionization
[7]. These single-electron (e–e) excitations edge onset of approximately 73eV. For al-
involve larger momentum transfer than loys, the core-loss contribution to the total
plasmon excitations and scattering of the scattering cross-section will be an average
transmitted electrons to higher angles than based on the fractions (atomic percentage)
the cutoff angle for plasmon excitation. In a of the alloying elements.
theoretical treatment, Ritchie and Howie Because core electrons are localized close
[20] link the plasmon and high-angle sin- to the atomic nuclei, the excitation of these
gle-electron regimes. Their scattering cross- exhibit channeling effects. That is, an ion-
section reduces to that of plasmon scatter- ization edge will become more or less
ing in the limit of small angles and to that prominent depending on the location in the
of Rutherford scattering from a free elec- unit cell of the atoms being ionized, relative
tron in the limit of large angles. Their to those that lie on the Bragg reflecting
mean-free-paths would suggest that about planes [26]. Hence, the t/␭ ratio measured
10% of all electrons suffer e–e scattering to by EELS will, in principle, be affected by
a high angle in a thin sample. Eaglesham the diffracting conditions in the probed
and Berger [21] give experimental evidence region of the sample, as given by its crystal-
for single-electron excitations being a sig- lographic orientation with respect to the
nificant contribution to the energy loss electron beam. Calculations of the localiza-
spectrum. For aluminium they find the e–e tion of inelastic scattering by Kohl and
contribution to be comparable to that of Rose [27] show that ionization of alumin-
phonon scattering at room temperature. ium L electrons (onset at ⌬E ⫽ 73eV) is de-
Calculations by Egerton and Wong [22] are localized over more than the unit-cell
in good accordance with this. Their calcula- dimensions. Hence, the channeling effects
tions of e–e cross-sections, using the Lenz on measured t/␭ ratios are believed to be
model [23], show an increase with atomic small and possibly negligible.
334 A. Bardal and K. Lie

EXPERIMENTAL METHODS 220 (or 311) rocking curve intensity profiles


were plotted in a si2/(ni,xk)2 vs. 1/(ni,xk)2
The transmission electron microscopy work plot. Here, si is the deviation parameter at
was carried out using a Philips CM30 the i-th minimum or maximum, ni are inte-
TEM/STEM microscope, equipped with a gers corresponding to the order of minima,
Gatan model 666 PEELS system. A liquid– and xk are numbers corresponding to the
nitrogen-cooled sample holder was used, order of maxima, as described in more de-
to reduce contamination and contribution tail in [3–5]. The t- and ␰g-values were ex-
to the CBED patterns from thermal diffuse tracted by fitting a straight line to the data,
scattering. The measurements were made determining the line’s slope (b) and its in-
at an accelerating voltage of 150kV. For tercept with the ordinate axis (a), and using
measuring the densities of dislocations, the relationships: t ⫽ a⫺1/2 and ␰g ⫽
precipitates, or dispersoids in aluminium (⫺b)⫺1/2. As long as the extinction distance
alloys it is advisable to work at an acceler- of the operating Bragg reflection is known
ating voltage below 170kV, which is the approximately, there is no need to resort to
threshold voltage for radiation damage by dynamical calculations to extract thickness
the knock-on process [28, 29]. The small values, an approach taken in [30]. Instead,
dislocation loops resulting from this pro- the starting index n1 can be unambiguously
cess, particularly when working close to chosen as the one that yields a constant ex-
low-index zone axes, can greatly obscure tinction distance for all patterns analysed
the image contrast and lead to erroneous under the same diffracting conditions [5].
results. Our measured extinction distances were
For each of the four alloys we investi- compared to the calculated values of ␰220 ⫽
gated, between 6 and 14 pairs of CBED/ 130nm and ␰311 ⫽ 164nm (at E ⫽ 150kV) ob-
EELS measurements were made from posi- tained using two-beam theory and Doyle-
tions on the samples having thickness Turner atomic scattering factors [31]. For
values in the range 80–400nm. the CBED measurements, the standard er-
The purpose of the CBED measurements rors of the least-squares fit were taken as
was to obtain reasonably accurate measure- the uncertainities of the single measure-
ments of sample thickness, which could be ments of t and ␰g.
use as reference values when extracting EELS measurements were made with the
mean-free-path values ␭ from the less accu- microscope in the image mode, and with
rate t/␭ values obtained through the EELS the objective aperture removed. Hence,
measurements. CBED patterns were re- electrons that had been scattered through
corded with the sample oriented in the large angles were admitted into the spec-
two-beam diffracting condition with a trometer entrance aperture. The collection
(220)- or (311)-reflection excited. Care was angle in this condition is larger than
taken to avoid the well-known sources of 100mrad, meaning that essentially all trans-
error that can affect thickness measure- mitted electrons are collected. A spectrom-
ments, as summarized by Ecob [5]. That is, eter entrance aperture having a diameter of
both intensity minima and maxima of the 3mm or less, and an image magnification of
extinction fringes were included in the 50,000⫻ was used. The probed areas thus
analysis, only g-vectors of sufficient magni- had diameters less than 60nm, and thick-
tude were used, and measurement errors ness variations in the probed areas (which
were minimized by scanning the negatives cause non-Poissonian intensity distribu-
from the transmission electron microscope tions in the spectrum) were negligible.
and using intensity line profiles for deter- The electronics of our spectrometer (pur-
mining positions of minima and maxima. chased 1989) originally had a defect caus-
Data evaluation followed the well-estab- ing a nonlinear detector response on the in-
lished procedure as described, for example, cident signal as described by Egerton et al.
in [3]. That is, maxima and minima of the [32]. The problem was cured by making a
Measuring Thickness of Aluminum Alloy Thin Foils with EELS 335

modification to the electronics as suggested be regarded as sufficiently correct to yield


in Egerton’s paper, and a linear response thickness values with an accuracy of ap-
was verified by plotting zero-loss peak proximately 8%, as witnessed by our re-
(ZLP) intensity (with the electron beam sults and as discussed later. The procedure
through a hole in the sample) vs. incident yields a thickness independent ␭ value
beam current. The background level of the over the thickness range 80–400nm at least,
spectra caused by detector dark current, as required.
and readout noise was subtracted using We also tested inferior procedures for ex-
standard procedures of the commercially tracting Io (as implemented, for example, in
available software [25]. To correct for gain earlier versions of commercial software) in
variations across the photodiode array of which no check of the background level is
the EELS detector, we divided the spectra made, and in which the ZLP is extrapolated
by a gain correction curve. This was ob- using an exponential. These procedures
tained through the procedure described by yielded apparent ␭ values that varied with
Boothroyd et al. [33], in which several spec- thickness when t/␭ was calibrated against
tra are acquired at different positions on CBED thickness values. We observed a sys-
the photodiode array and averaged to pro- tematic apparent increase in ␭ with thick-
duce a gain-corrected spectrum. After this ness of 5–10% relative over the investigated
correction, we found that sequential EELS thickness range, when using the inferior
recordings at the same position of the sam- procedures.
ple, but onto different parts of the EELS
detector array, yielded identical intensity
distributions in the recorded spectrum. On MATERIALS
our spectrometer setup, this is not entirely
the case when using only the default, stan- Four different aluminium alloys were in-
dard procedure [25] for correcting gain vestigated. A commercially pure alloy
variations across the detector array. (sample A) was included, along with alloys
The ZLP and full spectrum intensities, I0 that have a reasonably high density of fine
and Itot in Eq. (2) were determined using a precipitates (sample B), or that have a rea-
standard procedure of the commercially sonably high concentration of solute atoms
available software [25]. Using this proce- (samples C and D). The effect of solute at-
dure, the background level of the recorded oms on the plasmon mean-free-path de-
spectra (caused by detector dark current pends on the solute species, as different
and readout noise) was checked on the far species can modify the density of free (con-
low-energy side of the ZLP and corrections duction) electrons (n) to a varying degree.
were made to the background if necessary. For all samples, it was verified with energy
The tails of the ZLP were modeled using a dispersive X-ray spectrometer (EDS) in the
Lorentzian extrapolated under the loss TEM that the composition of the probed
data. Provided that proper care has been volumes in the thin foil was close to the
taken to ensure a linear response of the de- overall composition of the bulk material.
tector on the incident signal, the main No elaborate EDS studies were carried out,
source of error in determining the t/␭ though, to determine very accurately the
value lies in the extraction of the ZLP inten- exact composition; that is, only theoretical
sity from the total intensity of the spec- k-factors were used. Also, absorption ef-
trum. The details of the procedures for ex- fects were ignored, because EDS recordings
tracting I0 affects the resulting t/␭ value were made in very thin parts of the TEM
significantly. It is difficult to assess theoret- specimen. The quoted compositions should,
ically what the accuracy of the generic ap- therefore, only be regarded as approximate
proach of fitting a Lorentzian is, but the values, possibly with up to about 20% rela-
empirical results of the present study give a tive error in the different elements. This ac-
good indication. The chosen procedure can curacy is sufficient for the present context.
336 A. Bardal and K. Lie

Specimen A was aluminium of 99.5% 164nm. For all points, the measured ␰g
commercial purity, with Fe and Si present value lies within 2␴ of the mean value,
only in significant amounts as large inter- showing that the CBED results are very re-
metallic particles, and with no significant liable. The standard errors of the single ␰g
amount of alloying elements in solid solu- values range from 23% relative to 3% rela-
tion or as fine scale precipitates. Specimen tive, as indicated with vertical error bars.
B was a 7030 type alloy containing 5 wt.% As expected, the measured ␰g values do not
Zn and 1.2 wt.% Mg. The alloy was given a vary with thickness. The standard errors of
solution heat treatment at 480⬚C for 30 min the single thickness values range from 0.8%
followed by quenching, and aging at 150⬚C realtive to 5% relative, as indicated with
for 2 h. In this peak-aged condition, the ma- horizontal error bars. The average standard
jor alloying elements are present as approx- error of the thickness values was 1.8%.
imately 1 vol.% ZnMg2 precipitates. The All the mean-free-path (␭) values, as ob-
reason for including this alloy was its fairly tained by comparing pairs of CBED thick-
high density of fine precipitates. Although ness and EELS t/␭ results, are plotted vs.
a volume percentage of around 1% may sample thickness in Fig. 2. There is no vari-
sound quite modest, this is about as high a ation of the ␭ values with thickness, show-
value as one may find in the most impor- ing that that the procedure used for extract-
tant 2xxx, 3xxx, 5xxx, 6xxx, and 7xxx alloy ing the values is working. In the plot, the
series. Specimen C was a model alloy of mean-free-path values for specimens A–C
99.99% purity aluminium alloyed to con- appear to be similar, whereas the values for
tain 1.6 wt.% of manganese, and given a so- specimen D, with 5% magnesium in solid
lution heat treatment at 640⬚C for 1 h, fol- solution, appear to be slightly higher. Sta-
lowed by quenching, to bring all the tistical treatment of the data, as summa-
manganese into solid solution. Specimen D rized in Table 1, shows that there are no
was a model alloy of 99.99% purity alumin- significant differences in the ␭ values for
ium alloyed to contain 5 wt.% magnesium, specimens A, B, and C. The average ␭ value
and given a solution heat treatment at for specimen D is 5% higher compared to
525⬚C for 3 h, followed by quenching, to pure Al (specimen A), but this difference still
bring all the magnesium into solid solution. cannot be termed significant at the ␴ level,
All specimens for transmission electron given the accuracy of the measurements. The
microscopy were prepared by mechanical accuracy with which the ␭ values are deter-
grinding and polishing, and standard elec- mined are 2–5%.
tropolishing procedures using a Struers Apart from the measured ␭ values, we
Tenupol electropolishing unit, a methanol– have also included in Table 1 the measured
nitric acid electrolyte at approximately plasmon energies Ep, the total scattering
⫺30⬚C, medium flow rate, and a potential cross-section ␴total calculated from ␭ using
of 20V producing a current near 200mA Eq. (5), and the scattering cross-section for
during thinning. plasmon excitation [calculated using Eqs.
(4) and (7), and the measured Ep value].
Cross-section values were included for the
RESULTS AND DISCUSSION purpose of comparing the relative impor-
tance of different scattering processes. For
Figure 1 shows results of the CBED mea- calculation of the plasmon mean-free-path,
surements, with extinction distance plot- the cutoff angle ␪c was set equal to ␤ in Eq.
ted vs. thickness. The arithmetic mean of (7), and a value of ␪c ⫽ 5mrad was used, on
the measured ␰220 values is 130 ⫾ 6nm, the basis of the discussion in [7]. ␴p values
which is equal to the theoretical two-beam are found to be typically 9.3 ⫻ 10⫺5nm2,
value of 130nm. The arithmetic mean of the which amounts to approximately 66% of
measured ␰311 values is 170 ⫾ 10nm, which the total cross section ␴total (typically 14.0 ·
is close to the theoretical two-beam value of 10⫺5nm2). The statistical relative errors in
Measuring Thickness of Aluminum Alloy Thin Foils with EELS 337

FIG. 1. Extinction distances as measured using convergent beam electron diffraction (CBED) plotted vs. thin-foil
thickness.

the calculated ␴p values are equal to those mately 1.5% in ␴p is in good agreement
of the corresponding Ep values, i.e., near with theory, Eqs. (8) and (9).
0.2%. The total uncertainty in the calculated The cross-sections for core-loss excitation
␴p value (related to the theoretical model) is were calculated using the Hartree-Slater
not known, but any error would yield the model implemented in [25], and the average
same systematic shift for the different al- core-loss cross section for each alloy was
loys. Hence, ␴p for specimen D is signifi- found from the alloy composition. The
cantly lower. It is probably the main contri- edges which were considered were Al K,
bution to the lowered value of ␴total for this Mg K, Mn L, and M, Zn L, and M. The ap-
alloy, but more precise statements cannot proach yields a cross-section of 2.4 ⋅ 10⫺5nm2
be made with the given measurement accu- for all alloys, and no significant differences
racy of ␭ (and ␴total). The decrease of ap- between the alloys. This process is seen to
proximately 0.20eV in Ep and of approxi- contribute with approximately 17% to ␴total.
338 A. Bardal and K. Lie

FIG. 2. Experimentally determined mean-free-path values plotted vs. thin-foil thickness.

The remaining approximately 17% of the to- traction of I0 by fitting of a Lorentzian to


tal scattering cross-section must be due to the ZLP tail, there is no general answer to
other scattering processes, essentially sin- what the magnitude of ␦Lj/Lj is. An esti-
gle-electron scattering of conduction elec- mate of the typical, approximate ␦Lj/Lj ra-
trons. The apparent relative contribution of tio can, however, be made on the basis of
the single-electron process is in good accor- our experimental measurements.
dance with earlier work [20–22]. For this, we again look at the pairs of
It is now interesting to discuss the accu- EELS and CBED measurements. In Fig. 3,
racy, ␦tj/tj, of a single EELS measurement for each pair (j) of measurements, we have
(j) of thin foil thickness, once the ␭ value plotted the relative deviation between the
has been calibrated, and has a known accu- determined ␭j and the average value ⬍␭j⬎
racy ␦␭. From Eq. (6) we obtain: found for the alloy in question. This de-
viation is caused by the uncertainty in
δt j ⁄ t j = δλ ⁄ λ + δL j ⁄ L j (10)
extraction of I0 from the EELS spectrum
where Lj ⫽ ln (Itot/I0)j. Apart from the sys- and by uncertainties in the CBED thick-
tematic uncertainty in the calibrated value ness determination (0.8–5%, illustrated
of the mean-free-path , ␦tj/tj only depends by the horizontal error bars). In all
on ␦Lj/Lj, i.e., how accurately the ZLP in- cases, [␭j ⫺ ⬍␭j⬎]/⬍␭j⬎ is found to be
tensity can be extracted from the EELS less than 10% relative. The larger devia-
spectrum. With the generic approach to ex- tions tend to be found for thickness values,
Measuring Thickness of Aluminum Alloy Thin Foils with EELS 339

Table 1 Experimentally Measured Mean Free Paths (␭) and Plasmon Energies (Ep) and
Derived Scattering Cross-Sections
Specimen A B C D

Alloy composition Al (comm.pure) Al–5 Zn–2 Mg Al–1.6 Mn Al–5 Mg


and temper Peak aged— Sol. treated— Sol. treated—
1 vol.% ZnMg2 1.6 wt.% Mn 5 wt.% Mg
precipitates in solid sol. in solid sol.
␭ (nm) Experimental 119 ⫾ 5 121 ⫾ 2 118 ⫾ 5 125 ⫾ 4
result
Ep (eV) Experimental 15.18 ⫾ 0.02 15.18 ⫾ 0.02 15.24 ⫾ 0.03 14.98 ⫾ 0.01
result
␴total (10⫺6 nm2) 140 ⫾ 6 138 ⫾ 3 140 ⫾ 6 133 ⫾ 4
Calculated from exp. ␭
␴p (10⫺6 nm2) 95.0 95.0 95.4 93.7
Calculated from exp. Ep

t/␭ ⬍ 1.5, but do not correlate particularly One advantage of working with the TEM
with the CBED errors. The standard devia- in image mode during the EELS recordings
tion of all the [␭j ⫺ ⬍␭j⬎]/⬍␭j⬎ values is is that one should presumably not have to
3.4%. It should then be a good approxima- worry about how the crystallographic ori-
tion to state that the accuracy in ␦Lj/Lj is entation of the probed region of the foil
better than 3%. This is then the precision would affect the measured thickness value,
when making relative comparison between as discussed in the introductory part of this
single thickness measurements from the paper. This presumption was checked ex-
same sample. From (10), the accuracy in a perimentally by making EELS recordings at
single thickness measurement by EELS is some well-defined crystallographic orien-
better than 8% if ␭ is known within 5%. tations of the thin-foil specimen. A pair of

FIG. 3. Relative error in single ␭j values, as found from CBED and EELS results. The errors have contributions
from CBED thickness determination (horizontal error bars) and from extraction of I0 from the EELS spectra.
340 A. Bardal and K. Lie

EELS spectra were recorded from a fixed was gradually tilted more and more, and
position on the sample, first with the sam- EELS recordings were made at a fixed posi-
ple oriented at the diffracting condition tion, while ensuring that no strong beams
and then oriented at a nondiffracting condi- were operating in the diffraction pattern.
tion 1–2 degrees off. The conditions checked After correcting for the influence of speci-
were the strongly diffracting (111) and men tilt (␥) on the effective foil thickness
(220) two-beam conditions. Results are dis- (t⬘ ⫽ t/cos␥), variations in measured t/␭
played in Fig. 4, which shows the ratios be- values could be studied. Results are shown
tween the measured t/␭ value at the dif- in Fig. 5, for a specimen thickness of ap-
fracting condition and the measured t/␭ proximately 0.7␭. No significant variations
value at the nondiffracting condition, plot- with surface inclination are found, i.e., any
ted vs. specimen thickness. All measured possibly increased probability for surface
ratios were found to deviate with less than plasmon excitation does not significantly
approximately 6% from unity, and there affect the results.
are no systematic shifts to above or below When working with aluminium alloys for
unity. There are, furthermore, no systemtic which no explicit calibration of the mean-
variations with specimen thickness. free-path value has been made, fairly accu-
Another measurement series was made rate estimates of ␭ can still usually be made
to investigate the effect of the surface orien- based on the present investigation. For most
tation on the mean-free-path. The sample 1xxx, 2xxx, 3xxx, 5xxx, 6xxx, and 7xxx series

FIG. 4. Ratio between t/␭ with the specimen in a strongly diffracting condition and t/␭ with the specimen in a
nondiffracting condition.
Measuring Thickness of Aluminum Alloy Thin Foils with EELS 341

FIG. 5. Thickness/mean-free-path ratio (t/␭) plotted vs. tilt of thin-foil specimen.

alloys, a value of ␭ ⫽ 120 ⫾ 10nm can be CONCLUSIONS


used (at 150kV), irrespective of alloy tem-
per, as long as larger particles are avoided Pairs of EELS and CBED measurements
in the probed volume. In the mentioned al- were compared to determine the mean-
loy series, only magnesium can (in special free-paths ␭ for inelastic electron scattering.
alloys occasions) be suspected to be present EELS spectra were acquired with the TEM
at sufficiently high solute concentrations to in image mode and without any objective
motivate the use of a slightly higher ␭ aperture inserted. With the procedures
value. Other solute atoms will have a more used, ␭ values could be determined with an
negligible effect at the concentrations that accuracy of better than 5% for four differ-
are possible in the mentioned alloys series. ent aluminium alloys. The value for pure
Fine precipitates and dispersoids in the aluminium was determined as 119 ⫾ 5nm
mentioned series will be present at volume at an incident electron energy of 150keV.
densities that are comparable to or smaller For the most common alloys with moderate
than in the Al–Mg–Zn sample studied in amounts of alloying elements, no significant
this work. Accuracies and appropriate ␭ changes in the mean-free-path values are
values for more strongly alloyed systems, measured within the attainable accuracy of
such as, for example, Al–Li cannot easily be the method. An alloy with 5% Mg in solid
predicted by the present approach. solution exhibited a ␭ value of 125 ⫾ 4nm.
342 A. Bardal and K. Lie

With the given accuracy, the measured crystallographic aspects of recrystallization, N.


shift cannot be termed significant at the ␴ Hansen, D. Juul Jensen, Y. L. Liu, and B. Ralph,
eds., Risø national laboratory, Roskilde, pp. 261–
level. Still, one expects in this alloy an in- 266 (1995).
crease in ␭ of 2–3% compared to the pure
9. T. Pettersen, A. Bardal, I. Lindseth, and E. Nes:
alloy, owing to a lower density of conduc- Characterization of Deformation Microstructure in
tion electrons and a corresponding lower Hot Plane Strain Compressed Al–1Mg–1Mn. Proc.
probability for plasmon excitation. On the ReX 96 - 3rd Intern. Conf. on Recrystallization and
basis of our experimental results, we esti- Related Phenomena, T. R. McNelley, ed., ReX’96
International Advisory Board and Organizing
mate a single EELS measurement of thin-foil
Committee, pp. 495–501 (1997).
thickness to have an absolute accuracy of
10. T. Malis, S. C. Cheng, and R. F. Egerton: EELS log-
better than 8% if ␭ is known within 5%, the ratio technique for specimen-thickness measure-
difference coming from inaccuracy in extrac- ment in the TEM. J. Electron Microsc. Tech. 8:193–
tion of the zero-loss intensity from the EELS 200 (1988).
spectrum. Using the outlined procedure, the 11. G. A. Botton, G. L’Esperance, C. E. Gallerneault,
EELS method for thickness measurements is and M. D. Ball: Volume fraction measurement of
very robust, with thickness results being in- dispersoids in a thin foil by parallel energy loss
spectroscopy. Development and assessment of the
sensitive to experimental details such as in-
technique. J. Microsc. 180:217–229 (1995).
clination of the thin-foil specimen or crystal-
12. J. C. Ashley and R. H. Ritchie: Double-plasmon ex-
lographic orientation/diffracting conditions. citation in a free-electron gas. Phys. Stat. Sol. 38:
In cases where no explicit calibration of ␭ has 425–434 (1970).
been made, for most 1xxx, 2xxx, 3xxx, 5xxx, 13. H. Raether: Excitation of plasmons and interband
6xxx, and 7xxx series alloys a value of ␭ ⫽ transitions by electrons. In: Springer Tracts in Mod-
120 ⫾ 10nm can be used (at 150kV), irrespec- ern Physics, vol. 88. Springer, Berlin (1980).
tive of alloy temper. 14. P. E. Batson and J. Silcox: Experimental energy-
loss function, Im[⫺1/ε(q,␻)], for aluminum. Phys.
Rev. B 27:5224–5239 (1983).
15. D. B. Williams and J. W. Edington: High resolu-
References tion microanalysis in materials science using elec-
tron energy-loss measurements. J. Microsc. 108:
1. D. B. Williams: Practical Analytical Electron Micros- 113–145 (1976).
copy in Materials Science. Philips /Verlag Chemie 16. P. Schattschneider and P. Pongratz: Coherence in
International, Weinheim, Basel (1984). energy loss spectra of plasmons. Scanning Microsc.
2. D. B. Williams and C. B. Carter: Transmission Elec- 2:1971–1978 (1988).
tron Microscopy. Plenum Press, New York (1996).
17. R. F. Egerton and Z. L. Wang: Plural-scattering de-
3. P. M. Kelly, A. Jostsons, R. G. Blake, and J. G. convolution of electron energy-loss spectra re-
Napier: The determination of foil thickness by corded with an angle-limiting aperture. Ultrami-
scanning transmission electron microscopy. Phys. croscopy 32:137–148 (1990).
Stat. Sol. (a) 31:771–780 (1975).
18. E. A. Stern and R. A. Ferrell: Surface plasma oscil-
4. S. M. Allen and E. L. Hall: Foil thickness measure- lations of a degenerate electron gas. Phys. Rev. 120:
ment from convergent-beam diffraction patterns. 130–136 (1960).
An experimental assessment of errors. Philos. Mag.
A46:243–253 (1982). 19. C. J. Powell and J. B. Swan: Effect of oxidation on
the characteristic loss spectra of aluminum and
5. R. C. Ecob: Comments on the measurement of foil
magnesium. Phys. Rev. 118:640–643 (1960).
thickness by convergent beam electron diffraction.
Scripta Met. 20:1001–1006 (1986). 20. R. H. Ritchie and A. Howie: Electron excitation
and the optical potential in electron microscopy.
6. J. C. H. Spence and J. M. Zuo: Electron Microdiffrac-
Philos. Mag. 36:463–481 (1977).
tion. Plenum Press, New York (1992).
7. R. F. Egerton: Electron Energy Loss Spectroscopy in 21. D. J. Eaglesham and S. D. Berger: Energy filtering
the Transmission Electron Microscope. Plenum Press, the “thermal diffuse” background in electron dif-
New York (1986). fraction. Ultramicroscopy 53:319–324 (1994).
8. A. Bardal, I. Lindseth, H. E. Vatne, and E. Nes: 22. R. F. Egerton and K. Wong: Some practical conse-
Dislocation Densities, Subgrain Sizes and Aub-bound- quences of the Lorentzian angular distribution of in-
ary Misorientations Within the Different Texture elastic scattering Ultramicroscopy 59:169–180 (1995).
Components of Hot-Deformed AlMgMn. Proc. 16th 23. F. Lenz: Zur streuung mittelschneller Electronen in
Risø Int. Symp. on Mater. Sci.: Microstructural and kleinste Winkel. Z. Naturforsch. 9a:185–204 (1954).
Measuring Thickness of Aluminum Alloy Thin Foils with EELS 343

24. R. D. Leapman, P. Rez, and D. F. Meyers: K, L, and 30. F. R. Castro-Fernandez, C. M. Sellars, and J. A.
M shell generalized oscillator strengths and ion- Whiteman: Measurement of foil thickness and ex-
ization cross sections for fast electron collisions. J. tinction distance by convergent beam transmis-
Chem. Phys. 72:1232–1243 (1980). sion electron microscopy. Philos. Mag. A 52:289–
25. Gatan EL/P Software, version 3.0. Gatan Inc., 303 (1985).
Pleasanton, CA (1995). 31. P. A. Doyle and P. S. Turner: Relativistic Hartree-
26. J. Taftø and O. L. Krivanek: Characteristic energy- Fock x-ray and electron scattering factors. Acta
loss from channeled 100 keV electrons. Nucl. In- Crystallogr. A24:390–397 (1968).
strum. Methods 194:153–158 (1982). 32. R. F. Egerton, Y.-Y. Yang, and S. C. Cheng: Char-
27. H. Kohl and H. Rose: Theory of image formation acterization and use of the Gatan 666 parallel-
by inelastically scattered electrons in the electron recording electron energy-loss specrometer. Ultra-
microscope. Adv. Electron. Electron Phys. 65:175– microscopy 48:239–250 (1993).
200 (1985). 33. C. B. Boothroyd, K. Sato, and K. Yamada: The De-
tection of 0.5 at % Boron in Ni3Al Using Parallel
28. H. M. Simpson and R. L. Chaplin: Damage and re-
Energy Loss Spectroscopy. Proc. XIIth International
covery of aluminum for low-energy electron irra-
Congress for Electron Microscopy, L. D. Peachey
diations. Phys. Rev. 185:958–961 (1969).
and D. B. Williams, eds., San Francisco Press, San
29. A. Wolfenden: Electron radiation damage near the Francisco, pp. 80–81 (1990).
threshold energy in aluminum. Rad. Effects 14:225–
229 (1972). Received December 1998; accepted November 1999.

Potrebbero piacerti anche