Characterisation of coffee by ToF-SIMS
•
1 like•108 views
This document summarizes a study that used Time-of-Flight Secondary Ion Mass Spectrometry (ToF-SIMS) and multivariate analysis to characterize coffee samples from four farms in Sao Paulo, Brazil based on their brewing methods. Coffee samples were prepared through dilution and thin film deposition or compressed into granules. ToF-SIMS analysis was performed on the samples. Multivariate analysis of the data, including PCA, NMF, and k-means clustering, found separation between hot and cold brewed samples as well as granule versus film samples. The hot brewing preparation method showed the clearest separation between farms. The study demonstrated the potential of using ToF-SIMS and multivariate analysis for coffee
![Classification by origin
ToF-SIMSSample preparation
Samples of ground roast coffee from four different farms in Sao Paulo
region, Brazil were prepared following systematic procedures:
Dilution followed by thin film deposition: Granules were dissolved
in deionized water with 200 mg of coffee for 8 mL of water. The water
temperature varied in order to mimic the solution of different
brewing methods. After brewing, two drops of the solution were
dropped onto clean pieces of clean silicon wafer and dried in an oven
Compact granules: Ground coffee granules were compressed into a
small hole of a stainless steel stub using a spatula.
Gustavo F. Trindade1*, Lucio L.F.S2. Rosa, Elis M. Stori2, Carla E.L. Dos Santos3, and John F. Watts1
1The Surface Analysis Laboratory, Department of Mechanical Engineering Sciences, University of Surrey, Guildford, Surrey, GU2 7XH, UK
2 Ion Beam Centre, Advanced Technology Institute, University of Surrey, Guildford, Surrey. GU2 7XH UK
3 IMEF, Universidade Federal do Rio Grande Rio Grande Do Sul (Brazil). *g.ferraztrindade@surrey.ac.uk
Introduction
Characterisation of coffee by ToF-SIMS
Data analysis | The simsMVA app
In a first step of the analysis, peak areas
of ion fragments related to scent
molecules reported in [1] were imported
into the simsMVA app for multivariate
analysis (PCA, NMF, K-means clustering).
The main data pre-processing steps
employed were normalisation by total
counts, Poisson scaling and mean
centring
Hot brew: water at
92o C and brewing
for 10 minutes
Wrong brew: water
at room temperature
and brewing for 10
minutes
Cold brew: water at
room temperature and
brewing for 24 hours
ToF-SIMS analysis was carried out using a TOF.SIMS 5 system from IONTOF
GmbH (Münster, Germany) using 25 keV Bi3
+ion beam operated in the high
current bunched mode delivering 0.3 pA, over areas of 50 x 50 µm2. Several
different regions were analysed for each sample. Positive secondary ions
spectra were acquired for 30 seconds each, keeping the total PI dose below
the static limit of 1013 ions/cm2/analysis. For the granules samples, an
electron flood gun was used to eliminate charge building up on the samples
surfaces during the analyses.
Differentiation of brewing methods
Find out more at www.mvatools.com
Description of results
Fig. 2 shows a scatter plot of the scores of principal components (PC) 1
and 2 of the dataset containing all samples. The symbols are categorised
by sample preparation method. There are two categories for the room
temperature brewed samples because those were prepared and measured
on different days. Nevertheless, the fact that both showed very similar
scores certifies that the method is reproducible. K-means clustering was
applied to the first 3 principal components and shown as the background
colours of the plot. PC 1 separates the granules samples from the film
samples while PC 2 separate the hot and cold brewed samples from
the room temperature brewed and granules samples.
Fig. 3 shows overview spectra for each of the four kinds of coffee
analysed and each sample preparation method. All spectra are
extremely similar at first glance, yielding always the same peaks regardless
of the kind or brewing method. The differences among the spectra rely on
relative intensities of their peaks, which can be better assessed if
multivariate analysis is employed. Figs 4 & 5 present PCA results for
each preparation method. The Hot brewing preparation method
showed the clearest separation amongst farms
Coffee is the world's most widely traded tropical agricultural commodity with global production for 2016 of over 151 million 60 kg bags, according to the International Coffee Organization. The
geographic origin of coffee influences its quality and therefore it is important to develop methods that are able to identify and authenticate coffee samples. This work proposes a methodology that
combines ToF-SIMS and multivariate data analysis for the characterisation of coffee. The straight forward sample preparation and the wide range of masses that can be analysed provided information
of both volatile and non-volatile compounds before and after brewing.
Conclusion & Future work
Successfully developed sample preparation methodology that enables
now more experiments to be carried out.
Value of combination of ToF-SIMS and MVA for a simple and effective
characterisation which carries rich chemical information.
Future work: Collect more data from a wider range of farms in Sao
Paulo and create, test and validate an accurate classification model
based on ToF-SIMS data (initial tests were done using K-nearest
neighbour and random forest methods).
Fig. 2
Fig. 5
Fig. 4
Fig. 3
Fig. 1: Brewed coffee is dropped onto pieces of silicon wafer on top-mounted sample holder
Acknowledgments
The authors wish to thank the funding agency Comissao
de Aperfeicoamento de Pessoal de Nivel Superior (CAPES
project: 11995-13-0), the Worshipful Company of
Armourers and Brasiers and The UK Surface Analysis
Forum for enabling this work to be presented.
[1] R.A. Buffo, C. Cardelli-Freire, Flavour Fragr. J. 19 (2004) 99–104.](https://image.slidesharecdn.com/posterecasia2017gft-171026101428/85/Characterisation-of-coffee-by-ToF-SIMS-1-320.jpg)
Characterisation of coffee by ToF-SIMS
- 1. Classification by origin ToF-SIMSSample preparation Samples of ground roast coffee from four different farms in Sao Paulo region, Brazil were prepared following systematic procedures: Dilution followed by thin film deposition: Granules were dissolved in deionized water with 200 mg of coffee for 8 mL of water. The water temperature varied in order to mimic the solution of different brewing methods. After brewing, two drops of the solution were dropped onto clean pieces of clean silicon wafer and dried in an oven Compact granules: Ground coffee granules were compressed into a small hole of a stainless steel stub using a spatula. Gustavo F. Trindade1*, Lucio L.F.S2. Rosa, Elis M. Stori2, Carla E.L. Dos Santos3, and John F. Watts1 1The Surface Analysis Laboratory, Department of Mechanical Engineering Sciences, University of Surrey, Guildford, Surrey, GU2 7XH, UK 2 Ion Beam Centre, Advanced Technology Institute, University of Surrey, Guildford, Surrey. GU2 7XH UK 3 IMEF, Universidade Federal do Rio Grande Rio Grande Do Sul (Brazil). *g.ferraztrindade@surrey.ac.uk Introduction Characterisation of coffee by ToF-SIMS Data analysis | The simsMVA app In a first step of the analysis, peak areas of ion fragments related to scent molecules reported in [1] were imported into the simsMVA app for multivariate analysis (PCA, NMF, K-means clustering). The main data pre-processing steps employed were normalisation by total counts, Poisson scaling and mean centring Hot brew: water at 92o C and brewing for 10 minutes Wrong brew: water at room temperature and brewing for 10 minutes Cold brew: water at room temperature and brewing for 24 hours ToF-SIMS analysis was carried out using a TOF.SIMS 5 system from IONTOF GmbH (Münster, Germany) using 25 keV Bi3 +ion beam operated in the high current bunched mode delivering 0.3 pA, over areas of 50 x 50 µm2. Several different regions were analysed for each sample. Positive secondary ions spectra were acquired for 30 seconds each, keeping the total PI dose below the static limit of 1013 ions/cm2/analysis. For the granules samples, an electron flood gun was used to eliminate charge building up on the samples surfaces during the analyses. Differentiation of brewing methods Find out more at www.mvatools.com Description of results Fig. 2 shows a scatter plot of the scores of principal components (PC) 1 and 2 of the dataset containing all samples. The symbols are categorised by sample preparation method. There are two categories for the room temperature brewed samples because those were prepared and measured on different days. Nevertheless, the fact that both showed very similar scores certifies that the method is reproducible. K-means clustering was applied to the first 3 principal components and shown as the background colours of the plot. PC 1 separates the granules samples from the film samples while PC 2 separate the hot and cold brewed samples from the room temperature brewed and granules samples. Fig. 3 shows overview spectra for each of the four kinds of coffee analysed and each sample preparation method. All spectra are extremely similar at first glance, yielding always the same peaks regardless of the kind or brewing method. The differences among the spectra rely on relative intensities of their peaks, which can be better assessed if multivariate analysis is employed. Figs 4 & 5 present PCA results for each preparation method. The Hot brewing preparation method showed the clearest separation amongst farms Coffee is the world's most widely traded tropical agricultural commodity with global production for 2016 of over 151 million 60 kg bags, according to the International Coffee Organization. The geographic origin of coffee influences its quality and therefore it is important to develop methods that are able to identify and authenticate coffee samples. This work proposes a methodology that combines ToF-SIMS and multivariate data analysis for the characterisation of coffee. The straight forward sample preparation and the wide range of masses that can be analysed provided information of both volatile and non-volatile compounds before and after brewing. Conclusion & Future work Successfully developed sample preparation methodology that enables now more experiments to be carried out. Value of combination of ToF-SIMS and MVA for a simple and effective characterisation which carries rich chemical information. Future work: Collect more data from a wider range of farms in Sao Paulo and create, test and validate an accurate classification model based on ToF-SIMS data (initial tests were done using K-nearest neighbour and random forest methods). Fig. 2 Fig. 5 Fig. 4 Fig. 3 Fig. 1: Brewed coffee is dropped onto pieces of silicon wafer on top-mounted sample holder Acknowledgments The authors wish to thank the funding agency Comissao de Aperfeicoamento de Pessoal de Nivel Superior (CAPES project: 11995-13-0), the Worshipful Company of Armourers and Brasiers and The UK Surface Analysis Forum for enabling this work to be presented. [1] R.A. Buffo, C. Cardelli-Freire, Flavour Fragr. J. 19 (2004) 99–104.