For the Windows Version: soon
We are studying structures of thin metal films deposited on silicon by XPS. Would
the QUASES software be suited for our analysis?
Concerning the morphology of the deposits I would appreciate to see a comparison with STM
and/or AFM results.
As you may know, we have previously used the method to determine the structure of
several metal/silicon systems.
The structures are, for those systems where STM measurements were available, in agreement
with these.
We have also recently completed a study of Ge-growth on Si where we have compared the
structures determined by the peakshape method and AFM.
The morphology we have found for Ge growth on Si agrees very well with AFM measurements we
took on the same samples before we did the XPS analysis.
I am sure that QUASES will be helpful for your studies of nano-structures of metal
- silicon structures.
I enclose copies of the papers on metals/Si and Ge/Si.
I am currently running QUASES on my notebook at home. The reason is that I cannot
run it on my PC at work.
When I start the Analyze program, I get the error message: program too large to fit in
memory.
QUASES requires about 570 kB conventional RAM memory.
To see how much free conventional memory you have, type "mem" at the DOS prompt.
If you run it under Windows 95 this is not a problem.
If you run it under DOS you may have a memory problem if you are running other resident
programs. However such problems can usually be fixed simply by running the
"memmaker" program which is part of the DOS6.0 or higher. Memmaker frees
conventional memory by loading some of the device drivers and programs into upper memory
area. To run memmaker, type "memmaker" at the DOS prompt and follow the
instructions on the screen. I have used this many times on several PC's in the past and
always found it reliable and effective. For details, consult your DOS manual.
When I make a print of the results with PrintScreen, the concentration profile
diagramme at the bottom of the page turns completely black, how can I change this?
For best black and white print-results, set colors by clicking the Setup menu in QUASES
and then click Color and then click Color2. You can get better contrast if you then do the
following: In the Main menu, select "Analyze". Then select "6. Screen
setup" . Move to line "3) 4 Fig. color" . Hit return and type a new value
for the color. If you choose 11 or 12 you will get a better contrast in the printout
(although it is reversed i.e. black on the printout is white on the screen).
Occasionally, I have had the following problem. After analyzing a peak from a file,
and then changing to another peak, sometimes the screen does not show the spectrum when I
use Analyze. The problem disappears if I exit to the main menu and then go back to
Analyze.
To fix the problem: In the first line in the menu under "Maximum Energy" put in
any high value, for example 10000.
The phenomenon you describe will occur if you after having analyzed one peak and then
change to a new peak with higher energy. The reason is that the first line in the menu:
"Maximum Energy " is automatically set to the highest energy in the spectrum. In
the case just described, it will then not recognize any data points below that maximum
energy. The default value is 10000 eV and therefore the problem disappears when you exit
and restart Analyze.
Two practical questions on the software :
How much conventional memory does it need to run ?
Is it possible to print a graph from the screen on a postscript printer ?
QUASES needs approximately 570 kB memory.
If you have problems, run "memmaker" from the dos prompt. This usually will free
enough memory.
I have no experience with printing the graphics on a postscript printer. But the printer
might accept for example a laserjet format.
If you run QUASES under Windows 95 the easiest way to print is to open a document
(for instance Word) at the same time as QUASES. In QUASES press
Alt+PrintScreen and then paste the captured image into the document. And finally print the
document as you usually do. For best black and white print-results, set colors by clicking
the Setup menu in QUASES and then Color and then Color2.
The first tries with QUASES seem great. Yet I have some questions on your model, it
seems that there is some subtlety that I did not quite understand : When I compared the
results of QUASES to those I obtain with a home made programm (to check if my calculations
where correct !!) on a homogeneous sample (gold for instance) I observed a constant factor
between the two. I quickly found out the reason : in the ANALYZE programm the first thing
QUASES does is to divide the input file by the IMFP*cos(emission angle), which I do not do
in my calculation, and indeed the constant factor is equal to this. Does this means that
using QUASES, if I want to estimate the relative concentration of two elements in a
homogeneous material by comparing the peak areas obtained with ANALYZE, I just have to
make a correction for the photoionisation cross section and for the angular asymetry
factor, but that no correction for the IMFP is needed ?
You are right about the peak areas and IMFP factors.
The reason is the following:
What you determine with QUASES is the atomic spectrum F(E). This means the spectrum from a
single atom. When you do the correction you should therefore compare directly F(E)
determined from a given sample with unknown depth profile with F(E) from a pure elemental
sample. When the area (and shape) of the two agree then you have determined the
concentration depth profile.
So all factors are taken care of. When you interpret the peak areas done in Analyze, you
should always remember that the corrections carried out by QUASES brings you all the way
back to the spectrum from a single atom.
Do I have to beware of a similar problem when calculating a home determined
x-section from REELS, if I want to use it with QUASES ? (From equations 7 and 12 in Phys.
Rev. B 35, 6570 (1987) it seems to me that there is no particular problem).
With the REELS determined cross section you should have no particular problems.
I have done recently background subtraction using your model, either with the
universal inelastic scattering cross section one or the REELS one. This last one gives
very interesting results with semiconductor and dielectric materials. My problem is the
following : I have been using the Auger electron gun to achieve the REELS spectrum, and
the elastic peak is quite large (1.7 eV or so at half maximum). In none of your papers
that I read you say how to deal with the elastic peak, do you ignore it and cut the
experimental spectrum quite arbitrarily before the elastic peak and lose the information
in the first eVs, do you subtract it from the experimental data before the calculation of
the cross section, or do you deconvolve the elastic peak from the experimental data in the
hope of getting a Dirac in the place ? I have tried these three and I am not very
satisfied with any of them ! Could you help me on this point ?
For the cross sections from REELS, we normally draw a straight line from the minimum point
in the spectrum on the low kinetic energy side of the elastic peak (1-3 eV below the
elastic peak) to zero intensity at the energy of the elastic peak maximum. We have also
tried more sophisticated procedures, but the point is that
1) whatever you do, it will be an approximation so why not choose the simplest.
2) the cross section must go to zero at zero energy loss and this procedure ensures that.
3) the resulting background subtraction will be practically independent of the procedure.
The procedure is described in PRB, Vol.43, p.1652 (1991) which also gives a description of
the importance of and how to check for the linearity of the spectrometer
How can I interpret the intensities determined by Analyze? Does the software require
reference spectra? How do you analyze spectra if you do not have reference spectra?
The reference spectra are mainly just used to determine the total amount of material and
not for the determination of the depth profile. The reference spectra are therefore
helpful but not necessary.
The intensity presented by QUASES is very easy to interpret.
After ANALYZE, what you have determined is the peak shape and intensity F(E) corresponding
to the emission from a single atom. This means that you get the same peak area as from a
reference of a pure homogeneous sample of the element. The information on the
quantification lies in the structure that you specify as you apply ANALYZE. The structure
is varied until you get the same peak area as from the reference sample.
If you have a reference spectrum, then you vary the structure until the peak area (and
shape) of F(E) is as close as possible to that of the reference. Then you have determined
the quantitative nano-structure of the surface.
If you do not have a reference spectrum, then vary the structure until the shape of F(E)
looks reasonable (the background meets the spectrum about 30 eV below the peak and follows
the spectrum over an energy range of 20 or more eV. This determines the in-depth profile
for the element but not the absolute concentration. To get concentration information you
can then take the ratio of peak areas determined from peaks from the other elements found
in the spectrum. Because the relative concentrations of the different peaks can be
directly compared also without a reference spectrum.
Is it possible to analyze spectra where two peaks overlap in energy? In your
procedure of shape analysis of wide scans, can you introduce several elastic core level
contributions (for instance C1s + Cu 3p + Cu3s) and if so, how many are allowed?
Yes, in principle any number. But if the atoms have different in-depth distributions their
contribution must be treated separately and then added. This is only of concern when peaks
from different elements are overlapping in energy.
In general the procedure of analysis is simplest if you don't have overlapping peaks, but
I think this is not the case with your systems. If you have overlapping peaks, the
software requires that you supply the reference spectra determined from using Analyze on
the spectrum from a sample with known in-depth composition (for example a pure elemental
sample). Then you use Generate for each of the elements that overlap in energy. Finally
you add these generated spectra and compare to the measured spectrum. When you get a good
match with respect to both intensity and shape then you have determined the in-depth
distribution of the involved elements. The procedure is straight forward but it is more
time consuming since you have to fit parameters for two or more elements at the same time.
I have been able to use the Correct option successfully, but I have been getting
some funny-looking results with the Analyze option. I have been using some Ni 2p data
covering a 70 eV region from about 583 eV to 653 eV. The background found by QUASES is
almost constant. If I artificially stretch the energy scale so that it covers more than
about 100 eV, the Analyze option appears to work. Does successful use of the QUASES
Analyze option require that the energy range of the input spectrum be greater than some
amount (e.g., 100 eV)?
Concerning Ni2p data:
Look e.g. at fig. 6 page 1018 in SIA vol.20.
From this we see that the background meets the spectrum at ~570 eV and that the peak ends
at 650eV. Then we need some energy region below 570eV to have a unique determination with
ANALYZE of the profiles (550 eV as in this figure is a reasonable value although we would
always measure a bit further down say to ~520eV) but 583 eV is certainly not enough since
the peak has still not even reached zero intensity at this energy). The second thing we
see is that since the peak starts at 650eV, we need an energy range of at least 15-20eV
(i.e. measure the spectrum at least to 665 but preferably a bit higher) to be able to
determine the straight line in CORRECT, and certainly 653 is not sufficient to give a
unique determination of the slope of this line. There is however no minimum requirement as
far as the program itself is concerned as to the length of the files. You should not get
any "strange" looking results by applying a short energy range (but the physical
significance of the analysis depends on the range as explained above).
Did you compare the structures determined by QUASES to other techniques like STM
and/or AFM.
We have recently completed a comparison between the peakshape method and AFM. (the paper
is submitted to JVST). The morphology we have found for Ge growth on Si agrees very well
with AFM measurements we took on the same samples before we did the XPS analysis. The
series of morphology determinations we have determined for several metal/silicon systems
(2-3 papers published in 1995/96) are also in agreement with results found by other groups
(by STM, AFM and other measurements) for the same systems.
Is the software capable of dealing with balls for example, that has an inner content
different from the shell ? Is it possible to generalize it also to a similar structure
with lower symmetry, like polyhedron ?
In principle any structure can be modelled. What this means is that one can assume an
in-depth profile of any sort. But whether or not it is possible to distinguish the spectra
from different distributions of atoms depends on how different the structures are. In
general, structures that differ significantly on a depth scale of about 0.5nm can be
determined by the method.
I read your page in internet and I am interested in information about your software.
I use AES and XPS in corrosion studies, specially in analysis of oxide films. Is the
QUASES program adequate for this kind of studies? Our objective is to know the composition
and thickness of thin surface films.
We have experience with studies of iron oxides and in particular copper oxide thicknesses
grown at elevated temperatures. There should be no problem in quantifying the oxide films.
Thicknesses up to ~ 100 Å can be determined by the method. (The maximum depth probed is
approximately 5-10 times the inelastic mean free path of the transition studied). Please
do not hesitate to call back if you have further questions.
I wonder whether your QUASES program can accept the data file format the sample of
which I am sending to you (first column is the binding energy, the second is the pulse
count).
90.000 1194.289
90.293 1192.464
90.586 1188.144
90.879 1157.832
91.172 1183.447
91.465 1238.155
91.758 1224.389
Yes the data file format is acceptable for QUASES. Your data is binding energies in order
of increasing energy. QUASES needs kinetic energies in increasing order. QUASES
provides the necessary program to convert from binding energy to kinetic energy. First
apply "B_toKinE" from Tools on the menu bar. This converts your data to kinetic
energies in decreasing order. Then apply "Reverse" from Tools on the menu bar to
convert data to kinetic energies in increasing order.
What is the maximum number of spectral points which can be used in each spectrum?
The maximum number of datapoints is 2000.
Thank you for sending QUASES which I have received this afternoon. I was very
anxious to use it but encountered some problems from the very beginings. After selecting
Analyze (p.15 in the user's guide) the message "cannot execute
c:\quases\analyze.exe" appeared. The analogous message can be seen after selecting
Generate.I would appreciate your help.
First please answer the following three questions:
1. did you install quases under the default directory
c:\quases
2. Are you running it under DOS, Win95 or Win3.11
3. How much free memory do you have under DOS. Type "mem" at the dos prompt.
Please call back
Answer:
1. Yes
2. DOS
3. the message reads: 504880 largest executable program.
It is probably a memory problem. QUASES needs about 570 kB conventional RAM memory.
Try the following:
1. Can you run Correct from the main menu?
If you can, it is probably a RAM memory problem since Correct takes considerably less
memory than Analyze and Generate.
2. If it is a memory problem, you can probably free enough memory by typing
"memmaker" at the dos prompt and follow the instructions on the screen.
3. Or you can run it under Win95 which does not have the 640k memory limit.
Let me know what you get.
Answer:
Thanks for the message. I fixed the problem now by making some changes in config.sys
I am preparing to install the QUASES software. As I am not an expert in computers I
would like to ask you some information: -My computer is a Pentium 486 with 8MB RAM 75Mhz
800MB disk and Windows 3.11. Can I install and run QUASES under this Windows version?
-What is the advantage of running under Windows in comparison with DOS? -My ESCA software
is a VSW E-C V7.02 under DOS
QUASES should run well on your PC. QUASES runs fine under both Windows 95,
and DOS but I do not recommend to run it under Win3.11. The advantage of running under
Windows95 is that as it runs under a window, you can have other applications running
simultaneously and you do not have memory restrictions when you run the program. QUASES
requires about 570 kB conventional RAM memory. To see how much free conventional memory
you have, type "mem" at the DOS prompt.
If you run it under DOS you may have a memory problem if you are running other resident
programs. Then you may get a message like "Out of memory problems" when you
execute the Analyze and Generate programs. However such problems can usually be fixed
simply by running the "memmaker" program which is part of the DOS6.0 or higher.
Memmaker frees conventional memory by loading some of the device drivers and programs into
upper memory area. To run memmaker, type "memmaker" at the DOS prompt and follow
the instructions on the screen. I have used this many times in the past and always found
it very reliable and effective. For details, you may consult your DOS manual but it should
not give any problems, just follow the instructions on the screen.
If you run it under Windows 95, it is unlikely that you will have memory problems.
Do you account for the plasmon (surface + bulk) contributions?
Yes, the plasmon excitations are accounted for.
Is it possible to implement specific model of growth, not only the three classical
ones (VW, FvdM, SK)?
Yes in principle any profile can be modelled.
I suggest you have a look at the paper:
D. Fujita, M. Schleberger and S. Tougaard, "XPS Study of the Surface Enrichment
Process of Carbon on C-doped Ni(111) Using Inelastic Background Analysis. Surface Science.
331-333, 343-348 (1995). where we have studied a system similar to yours (growth of C on
Ni). You can find some information on the software and you can also download a
demo-version of the software package from the QUASES web site at http://www.spo.dk/quases
The demo shows some of the capababilities of the software package.
Thanks for the information on QUASES and for your advice on REELS. I have one more
question about the software before I send you a purchase order: for the cross section, is
there the choice between a REELS-calculated and the universal one ?
You may put in and apply any shape of cross section that you wish. This can be one
determined experimentally from REELS, or one of the several that are supplied with the
software or you can simply put in any set of numbers that has the shape of any energy loss
feature that you wish. So any loss function may be applied.
Is there any possibility to test in some way this piece of software using our
experimental data?
Yes, you may send me a file with a sample spectrum from your instrument and I will make
and send you the result of an analysis with the software.
I am interested in your QUASES software package and would appreciate very much to
have more detailed information. As you know there are many matrix dependent processes
leading to changes of the XPS peak shapes and of the background. With regard to this fact
I wonder how accurate your method is. Does it take for instance into account all energy
loss processes like shake-off etc.? How are the results influenced by surface roughness of
the sample? How the results of the depth profiling and of determination of the surface
morphology of the deposits depend on the angle at which the spectra were measured? What
about the influence of the surface contamination ? It is not clear from the demo version
available on the www what the procedure called Correction for electron cascade and
analyzer transmission actually means. The same holds true for File management facilities.
Intrinsic processes are what is left behind after the background correction procedure,
since those are the processes that are inherent to the peak and usually one does not want
to remove these.
Essentially all geometries are covered by the present software like e.g. layered
structures with any number of layers and spacings etc. But there is ofcourse a limit as to
how detailed information can actually be distinguished. This question is adressed in
detail in a recent paper which will be submitted shortly to Surface Interface
Analysis(1996/97).
Thank you very much by your mail message about the QUASES program. I would like very
much to have more information about it, namely about the mathematical procedures and basic
assumptions and if possible some examples of applications. Is that possible to send me
some technical information about it?
Have a look at this website under Publications. There you will find different sections
with papers. One section contains papers which describes the theory behind the method,
other papers describe practical applications and another group contains papers with tests
on the absolute validity of the formalism. Indicate in an e-mail to me the code number of
the papers that have your interest and include also your address, and I will be happy to
mail you the copies.