Applications Overview

| Oxides
| Alloys
| Glass
| Oil
| Polymers
| Catalyst Samples
| Small Samples
| Coal
| Carbonblack

Last update: 2010-02-01

XRF Analysis of Glass 
Overview of Samples

XRF analysis of Green and Brown Glass
Same Glass sample in 3 different forms
Boron in Glass or Enamel
Special Glass with High Boron

XRF analysis of Green and Brown Glass

The analytical results shown here are by courtesy of Mr. Theo W. Verkroost
of the Technical University, TU-Delft, The Netherlands.

On a regular basis, TU-Delft analyses a great variety of glass samples including glass
with high Lead and/or Barium content. To enhance accuracy, use is made of UniQuant’s
ability to firm up the calibration by one or more international standards.
The procedure is to first copy UniQuant’s ‘universal’ calibration data to a new folder
(sub-directory), for example named C:\UQ4\Glass. In these data, the Kappa’s are firmed
up by one or more standards. In the example below, just one standard is used, namely
European Standard Glass EC1.1, see last column.

Working with pressed powders has an advantage here over working with flat glass discs.
A disc is cut for example from the bottom of a bottle assuming that its composition is
representative for that one bottle as well as for an entire batch of bottles (no segregation’s in
the process of making bottles). Powder on the other hand can be made from pieces of
glass of one or several bottles which, when ground together, ensures ‘representativity’.

Columns ‘XRF+UQ3’ show concentrations found by a XRF + UniQuant 3. Glass No. 7
is treated as an unknown sample to compare results with certified concentrations.

Columns ‘XRF’ show concentrations found by a different lab using the conventional
calibration with several or many standards, where standards and unknown samples
were presented as flat glass discs.

       Clear glass        Green glass       Brown glass       EC1.1
        Glass No. 7 of                                         used to refine
        Society of Glass                                       calibration
                 XRF+UQ3   XRF     XRF+UQ3   XRF     XRF+UQ3   XRF+UQ3
        Cert.    pressed   glass   pressed   glass   pressed   pressed
        Conc.    powder    disc    powder    disc    powder    powder
        ----------------   ---------------   ---------------   -------
SiO2    72.74    72.66     72.41   72.55     71.54   71.74     71.97
CaO     11.03    11.36     10.67   10.58     11.69   11.51      8.63
Na2O    13.90    13.46     12.40   12.29     12.42   12.38     13.41
Al2O3    1.50     1.50      1.63    1.70      1.57    1.65      1.08
MgO      0.14     0.14      1.45    1.47      1.63    1.65      3.78

Fe2O3    0.044    0.061     0.298   0.30      0.245   0.25      0.10
K2O      0.43     0.45      0.60    0.61      0.56    0.57      0.59
TiO2     0.042    0.039     0.047   0.052     0.062   0.069     0.04
Cr2O3                       0.26    0.27      0.042   0.042   
MnO                         0.020   0.013     0.013   0.011   

BaO                         0.041   0.040     0.037   0.030   
PbO                         0.015   0.014     0.013   0.010   
SO3      0.19     0.20      0.054   0.036     0.078             0.24
S                                                     0.033   
LOI(500°C) 0.07                                         

UniQuant finds and automatically reports Sulphur as %SO3 in Green glass and as %S in
Brown glass. The colour of brown glass is caused by Ferro-sulphide. In production,
FeS2 may be added to the mixture. Carbon is used as a reduction agent whilst Na2SO4 is
used a an oxidation agent. In the analysis shown above, UniQuant has been set to treat
all S as SO3 or all S as S, depending on which of the two has the largest concentration.

In principle, UniQuant can determine SO3 and S simultaneously. In fact, this feature is in
routine use at a cement plant, where Sulphur concentrations are in the order of 1 mass%.
At Sulphur concentrations of 0.03% as in glass, simultaneous determination is more difficult
and it is expected that a super-fine collimator is required to improve resolution of SKa
from S and ‘SxKa’ from SO3. We intend to make a simple feasibility study shortly and report
the results at this place. The topic is perhaps interesting because technology of making
brown glass requires that reduction is not overdone. Presumably, the ratio (S as SO3) / (S as S)
is about 80/20. There seems to be no easy and quick way to determine this ratio.
Comments by E-mail are welcome


Same Glass sample in 3 different forms

        UniQuant1 (May 1989)
Original 670mg broken loose 1+33LiT Chemical solid piece powder Bead spec. ------------ ------ ------- -------- MgO 1.6 ± 0.1 1.3 1.4 1.4
Al2O3   12.8 ± 0.3     11.7      12.6       14
SiO2    42.9 ± 0.5     39.9      43.2       43
CaO     34.1 ± 0.06    37.2      34.8       33
Fe2O3    1.8 ± 0.1      1.7       1.6        1.7
ZnO      0.4 ± 0.06     0.5       0.3        0.4 
ZrO2     4.3 ± 0.2      4.6       4.5        4.5
BaO      0.3 ± 0.05     0.3       0.3        0.3 


Boron in Glass or Enamel

Most spectrometers can measure the intensity of B Ka. The question is if it
is possible to determine the concentration of B2O3 on the basis of the
B Ka intensity. The answer is not a simple YES or NO.

The problems are:
- the B Ka line is interfered by XRF lines from many elements (30 or more).
- many of the interfering lines almost coincide with the B Ka line.
- there is no spectral position at which the background of B Ka can be
 reliably measured because all positions in the vicinity of B Ka are suspect
 of being spectrally interfered.
-both background and line overlap corrections (expressed as equivalent %B2O3)
 are relatively high.

If a glass sample does not contain interfering elements at levels higher than
say 0.1%, it may be quite feasible to determine B2O3 directly by the B Ka intensity,
However, in case one or more interfering elements are present at higher levels,
corrections of B Ka for spectral line overlaps becomes doubtful. This is
because the equivalent correction values (expressed as %B2O3) are relatively
high. As a consequence, the B2O3 content would have to be calculated from
the difference of large values. For example:
30% - (28% ± 3%) = 2%B2O3 ± 3%B2O3,
where the 28% ± 3% represents the total line overlap correction.

Interfering elements are: Mo, Nb, Zr, Cl, Sb, Sn, Te, Hf, Ru, Ag, Cd, In,
all Lanthanide elements (La...Lu), and many more.

There are 2 general solutions:

Solution 1
Determine B2O3 by a technique other than XRF and enter its value
as known concentration in UniQuant.

Solution 2
Determine B2O3 as the non-measured rest from the difference of the
sum of concentrations with 100%. This method works very well with
UniQuant if 2 conditions are satisfied, namely:
- The Kappa's (Kii's) of the major elements must be very accurate.
 When analysed with UniQuant, glass without B2O3 must give a sum
 of concentrations of 100 ± 0.5%, or better 100 ± 0.3%.
- Samples must be measured as large flat glass specimen or as beads.
 This is in order to avoid 'shadow' effects as with powder samples, which
 would decrease the sum of concentrations.
Depending on the quality of calibration of the Kappa's, the B2O3 content
calculated (as difference) may then be accurate to 0.3 to 0.5 weight%. 
Below is an early example of 'Special Glass with High Boron', for which no
special attention was given to the first condition mentioned above.


Special Glass with High Boron

            Other           (1991)   
            method          UniQuant1 
           --------         ---------
La2O3       40.7             40.1
Y2O3        10.57            10.20
ZrO2         6.20             6.01
ZnO          4.53             4.70

SiO2         2.07             2.20
Gd2O3        2.0              2.7
Nb2O5                         0.70
TiO2                          0.11
Al2O3        0.04             0.04
P2O5         0.01             0.03
Fe2O3        0.01             0.01
B2O3        32.3             33.2*
           ------           ------
            98.43           100.00
* Boron was not actually measured. Instead, it was specified in UniQuant
that the non-measured rest is B2O3. The %B2O3 was then determined by
difference, taking the absorption by B2O3 into account.


UniQuant® 1989...2010 is a registered trademark of Thermo Fisher Scientific Inc.
Copyright © 2010 Thermo Fisher Scientific Inc. All rights reserved.