Applied Mineral Exploration
Discussion and research relevant to mineral
exploration. Deconvolution of the Brusson mine data
The decrepitation curves of the Brusson samples show complex shapes and difficult to use in interpretation. By performing deconvolution of the curves we can find a subset of population distributions which add together to give the observed curve. It is then possible to compare samples using the parameters of the component sub-populations. A complete discussion of the study at the Brusson mine is here.
All of the Brusson samples were deconvolved using a mathematical software package. Although it is possible to deconvolve the data into gaussian distribution populations a lower residual error (better fit) is obtained by using a skewed gaussian distribution. It is reasonable to expect fluid inclusions populations to be skewed because of the increased likelihood of decrepitation of inclusions near grain surfaces. Skew can also be caused by changes in the gas content of inclusions during entrapment and quartz formation. During curve deconvolution a degree of user input is helpful to constrain the mathematics as the solution is not necessarily unique and some solutions lead to physically unlikely population groupings.
Some of the Brusson deconvolution data is presented here to show how well the method fits the data. At the completion of each fitting procedure, the parameters of the various sub-populations are recorded and used to prepare the tabulation of peak temperatures shown previously.
At the lower adit level a low temperature decrepitation peak is common, but not always present
At the mid-level adit, the quartz varies from complex with multiple populations to simple with only 2 decrepitation peaks.
At the upper adit level, low temperature CO2 caused decrepitation is still prominent.
Above the ore zone in carbonate host rocks, low temperature CO2 caused decrepitation is still present.
Note that the next 2 graphs are 2 fits to the same sample data assuming either 4 or 5 sub-populations. The 5 peak fit is slightly better than the 4 peak fit - but it is difficult to be sure exactly how many populations are really present in the data. The curve fitting does not necessarily lead to a unique result. However, multiple fits of the same data with the same number of assumed populations does lead to identical sub-population parameters.
Note the complexity of the sample with many fluid inclusion populations, indicating that this quartz is strongly zoned.
