How does UniQuant compare with other programs?
Table of Comparison
Smart-Variable-Step versus Fine-Step scanning
Innovated Concepts

How does UniQuant™ compare with Other Programs?

Introduction

X-Ray Spectrometers are always supplied with what we will call Spectrometer Software.
Examples are WinXRF of ARL and SuperQ of Philips AXR.
Briefly speaking, Spectrometer Software contains of 2 programs, namely:

• The Control Software with which Analytical Programs (AP’s) for Data Collection
  are defined for controlling the spectrometer and sample changer. The AP will control
  the measurement of a sample with a specified set of instrumental parameters like kV,
  mA, collimators, analysing crystals, detectors and measuring times. ·
• The Evaluation Program (EP) which is for conversion of the measured intensities to
  elemental concentrations and possibly mass/area of thin layers.
  UniQuant falls in this category.

Evaluation Programs

There are various type of Evaluation Programs, each one making use of its own typical
way of Data Collection:

• Mathematical Model (MM) program.
  This is still the most widely used. The MM is an equation that approximates the more
  accurate ‘Full Physical Equation’. It comprises calibration coefficients such as for
  slope factor, background, line overlaps and interelement corrections.
  The latter are predicted by theory (theoretical Alpha’s) and/or found by calibration
  (regression Alpha’s). The remaining coefficients must be found by calibration using
  Regression Analysis of several or many standards. Because of the approximations made
  by the MM, a calibration must be made for each ‘family’ of samples, like a family of beads
  and a family of a particular type of alloy. Once all coefficients have been established,
  the MM is called the analytical equation. A set of simultaneous analytical equations is
  used to convert intensities from an unknown sample to concentrations.
  Generally, each family of samples uses its own specific Analytical Program (AP) for
  intensity measurements.
  Regression analysis was first used in XRF analysis in 1967 by the present author
  when with Philips. In the same year, he developed the DJ mathematical model and backed
  it up with ‘theoretical Alpha’s’ and ‘regression Alpha’s’. Publishing was postponed until 1973.
  
  
• UniQuant’s method requires just one Analytical Program (AP), for which UniQuant exactly
  prescribes the instrumental parameters.
  The complete AP comprises 114 fixed spectral positions, which are rather evenly
  distributed over the entire XRF spectrum of interest. Each such position is tuned to the
  wavelength of a potentially present element. One may say that UniQuant makes a smart
  (coarse) step scan of the spectrum. On the other hand its method of measuring is in principle
  the same as with the conventional MM method, see above.
  UniQuant converts intensities to concentrations by using the ‘Full Physical Equations’
  of which most input data are ‘Fundamental Parameters’. As a consequence, calibrations are
  feasible for an extremely wide range of concentrations.
  UniQuant’s method has reduced the ‘calibration curve’ to a simple horizontal line,
  Kappa versus concentration, where Kappa is the instrumental sensitivity for a given XRF line.
  Kappa is independent of samples. The problem of some 1500 potential line overlaps is solved
  in a unique, rigorous and elegant way by using line overlap Kappa’s, which equally are
  instrument sensitivities for interfering XRF lines.
  UniQuant simultaneously solves concentrations and line overlap corrections (in mass%)
  
  
• SQ programs (FSS+FP programs)
  Most spectrometer manufacturers offer a program that use intensity data as obtained
  by Fine Step Scanning (FSS) of the spectrum. The intensities are first converted to net XRF
  intensities which in turn are converted to concentrations by using Fundamental Parameters (FP).
  Examples are SemIQ of PANalytical, QuantAS of Thermo and SSQ of Siemens.
  The names of these programs would suggest that results are semi-quantitative.
  However, for certain types of samples, the results may be accurate, thus quantitative,
  provided that a ‘library of standards’ contains data of a suitable standard.
  Especially if high energy XRF lines are used, the library standard must have about the
  same mass, height and dilution as the unknown sample.
  
  Example 1: An unknown bead requires a ‘library standard’ of a bead with the same dilution
  and total mass as the unknown.
  
  Example 2: an unknown oil sample requires a ‘library standard’ of oil with the same
  total mass (height) as the unknown.
  
  Example 3: The analysis of a dust filter may be accurate only if no high energy XRF lines
  are used, else substantial errors would occur through ‘non-infinite’ thickness of the sample.
  While scanning over XRF lines from trace elements, the effective measuring time is very
  short, for example less then one second. It is then difficult to distinguish between
  background and any fluorescent radiation. As a result, a trace element is easily overlooked.
  Inspired by UniQuant, some SQ programs have therefore been modified so that certain
  elements can be measured with fixed position and adequate time.
  Even so, the burden now is on the analyst who is to specify for which elements this new
  feature is to apply. And, what to do when a sample is totally unknown?
  
  

  Top

  
  

  ---

Table of Comparison

*	Features for SQ programs shown above are to the best of my knowledge.
**	spectral positions are rather evenly distributed over entire XRF spectrum, therefore, one may say that UniQuant makes use of a smart coarse step scan
***	It is common practice to speak about Precision (reproducibility) and Accuracy of an analysis. For programs that can work with totally unknown samples, I here add the characteristic of Reliability. Its meaning is best explained by an example: An SQ program reports 27 %Fe in a sample whilst UniQuant reports 55 ppm Fe. What happened? The sample contained much Pb of which one XRF line interfered the analyte FeKa line . The SQ program intelligently diverted from FeKa to FeKb. But the FeKb line gave the totally wrong result of 27% Fe. Either the library standard used was not an appropriate one or FeKb was itself not free from spectral interference (line overlap). Such problems can hardly occur with UniQuant, which is reliable (rugged) in this respect.

Top

---

Smart-Variable-Step (SVS) versus Fine-Step (FS) scanning

Climbing the highest mountain
In conventional quantitative XRF analysis, intensities are measured at fixed spectral positions.
Almost ten years ago, spectrometer manufacturers started to make software for analysing
totally unknown samples for which no standards are available. They have all chosen the
same method of intensity measurements, namely by Fine-Step (FS) scanning of the X-Ray
spectrum. At the same time, Omega Data Systems started to develop the UniQuant software
which is based on far better concepts. It was like choosing the highest mountain to climb.
Others may reach the top of a mountain only to find out that their mountain is much lower
than Mount UniQuant.
Like conventional methods, UniQuant is based on intensity measurements at fixed
spectral positions. Although UniQuant was originally intended for standardless analysis
of totally unknown samples, it has been so far developed that with the use of standards
and regression analysis, it may give results that are as precise and accurate as with
conventional XRF analysis.

The advocates of Fine-Step scanning
In sales situations and at symposia, sometimes it is emphasised that Fine-Step scanning
is far better than measuring at fixed positions, in particular in view of trace analysis. If this
would be true, why then is conventional quantitative XRF analysis based on fixed
spectral positions and not on Fine-Step scanning? The arguments used are clearly in
defence against UniQuant. This leaflet reviews several of the false arguments in favor
of Fine-Step scanning.

What is Smart-Variable-Step Scanning?
In SemiQuantitative (SQ) programs, intensity measurements are generally done by
Fine-Step (FS) scanning of the wavelength spectrum.
In contrast, the quantitative UniQuant™ (UQ) program prescribes measurements to be made
at about 100 spectral positions. The goniometer scans the entire spectrum in varying
steps from one position to the next, where each position corresponds with an XRF
wavelength of one of 78 elements. We refer to this method as Smart-Variable-Step (SVS)
scanning. The predicate 'Smart' is used here because for a given total time per sample,
all potentially present elements are measured with highest possible precision.

Relative Merits
To appreciate the relative merits of both FS and SVS scanning methods one should be aware
that the SQ programs and UniQuant differ fundamentally in their concepts of finding net
peak intensities, which are the gross peaks corrected for background continuum and
spectrally interfering XRF lines.
SQ: The concept of SQ programs makes the use of FS scanning mandatory. This is
because, for the purpose of line overlap corrections, very many (over a thousand)
spectrally interfering lines must be measured in addition to the about 100 analyte lines.
UQ: Due to its concept, UniQuant can abstain from measuring any of the potentially
3000 interfering lines and devote all its time to the 100 analyte lines.

SQ: When an interference is strong, an SQ program may itself search an alternative free
analyte line. If things get difficult, the help of the analyst is required and this is where the
use of spectrograms comes in.
UQ: In the results obtained from each analyte line, UniQuant gives a full quantitative
account of corrections made for the various line overlaps. Thus it tells exactly which
elements caused an interfering line and to what extent in equivalent mg/kg.
What is the magic here?
The answer is that UniQuant fully exploits the combination of two favorable facts:
   Modern spectrometers can be programmed for spectral positions (wavelengths)
      with avery high precision.
    Both analyte XRF lines and interfering XRF lines are always at their same spectral
      position (apart from chemical shift of soft XRF lines).
Thus, there is no need for scanning and/or searching.

Trace Analysis
SQ: With SQ programs, scanning speed must be relatively fast in order to keep the total
measuring time within 15 minutes. This fast scanning leads to relatively large counting
errors. To partially compensate for this, instrumental parameters are chosen for highest
possible intensity. However, this is at the expense of spectral resolution making the line
overlap corrections even more difficult. The time of scanning across XRF lines of trace
elements is so short that their intensity may not be distinguishable from the stochastic
fluctuations of the background. On the other hand, the same fluctuations are easily
mistaken as peaks from a trace elements. A partial solution of this problem is possible
for samples of which it is known which trace elements can be expected and which not.
The analyst may then enter certain directives to the program for the scanning speed
to slow down where it may be useful or to skip specified ranges of the spectrum
assuming that these do not contain lines from unexpected elements.
It should be clear that such specific approach may not work for less well known samples.

UQ: UniQuant measures trace elements much the same way as in conventional analysis,
that is with fixed time, 10 seconds for example. This is a factor 100 longer than with FS
scanning, where the dwelling time in the vicinity of the peak is as low as 0.1 seconds.
As a consequence, with UniQuant, Detection Limits are better by a factor 10 with respect
to SQ programs if the latter are not tailored to sample specific conditions.

What about background calculations?
The most simple case is that of an XRF peak without any spectral interference in its vicinity.
The background under the peak can easily be calculated from additional measurements at
both sides of the peak. Any program can do that. However, the situation is far less simple
when a lot of interfering lines are cluttering in the vicinity of the analyte line.
SQ: It may be impossible to find a spectral position in the vicinity of the analyte line that is
suitable to measure a background intensity.
UQ: In such cases, the background may be determined by considering a wide spectral
range, like one would do visually. This is exactly what UniQuant does. Fine-step scanning
would not have an advantage. On the contrary, fine-step scanning tempts to look at too
small spectral ranges with the danger of overestimating backgrounds.

What about confirmation by other lines of the same element?
Suppose that Zr is analysed by its ZrLa line. If this line gives a net intensity whilst Zr is
not expected to be present, one may suspect that the ZrLa peak intensity is due to a
spectral interference by a line from another element or by a spectral impurity of the
radiation that is incident to the sample. In order to verify if indeed Zr is present, one
may check the intensities of other XRF lines of Zr, like ZrKa and ZrLb.

SQ: If a fine-step scan was made across the entire spectrum, data for ZrKa and ZrLb
are available. This is sometimes presented as an advantage for FS scanning based
SQ programs. The truth is that such verification may be very much needed by SQ programs.
UQ: By contrast, for UniQuant such checks are not required. This is because, as part
of its results, UniQuant gives a full account of which parts a peak intensity is
composed of, namely:

   Background continuum in equivalent mg/kg.

   Spectral impurity in equivalent mg/kg.

   Spectral line overlaps in equivalent mg/kg and from which elements.

   Counting error in equivalent mg/kg and this is possible for UniQuant because it
     measures with fixed times at the spectral positions of interest.

Top

Innovated Concepts

Any of the calibrated programs can in principle be used for any sample.
The calibration constants differ only slightly. For totally unknown samples,
76 elements are determined.
The AnySample program uses 76+36 XRF lines.

EASY MAINTENANCE
A single "all metal monitors" program for drift corrections supports all calibrated programs.
Each of the 9 monitor samples is a “100“ % element for maximum stability,
easy cleaning and replacement.

UniQuant™5 unites in ONE WORLD:
XRF analysis using standards and standardless XRF analysis, all quantitative
and reports in all cases:

an uncertainty interval for each calculated concentration or sample mass (thickness),
a requirement for programs to be QUANTITATIVE.

Top