ABSTRACT
Honey adulteration by cheaper sweeteners such as sugar syrups, synthetic honey, molasses, and sugar beet has become a common vice thus negatively affecting the quality of honey production, and diminishing its market value. Lack of label–free, easy to use and rapid quality assessment honey adulteration detection techniques in the market has encouraged honey producers and processors to cheat on its quality. Furthermore, the current honey adulteration detection techniques such as, Stable Carbon Isotope Ratio Analysis (SCIRA), Liquid Chromatography (LC), Gas Chromatography (GC), and High Performance Liquid Chromatography (HPLC) suffer from the disadvantages that include being less rapid and expensive to use. Hence the need for rapid and affordable honey adulteration detection techniques. In this research, laser Raman spectroscopy robustness as an emergent technique for definitive molecular fingerprint analysis was explored to study honey adulteration. Authentic honey was intentionally adulterated by molasses in varying concentration ranges. Raman spectra was collected separately with each done under 60 seconds from small quantities of 1 g of authentic honey, molasses and molasses - adulterated honey samples. PCA was employed to perform exploratory analysis of the combined authentic and adulterated Raman spectral data sets, while machine learning techniques namely, random forest (RF), and support vector machine (SVM), and artificial neural networks (ANN) were used to create multivariate classification and regression models for forecasting authentic honey and molasses - adulterated honey samples. The most variant bands between authentic honey and molasses that were confirmed using ANOVA and PCA showed characteristic bands centered around: 690 cm-1 (stretching of CO and CCO, and bending of OCO); 732 cm-1 (glucose ν(C-C) vibrations); 754cm-
1 (weak (C = O) bond vibrations); 845 cm-1 (glucose spectrum); 970 cm-1 (glucose, ν (C- O) vibrations). Furthermore, high classification accuracies ranging from 86 – 100 % were achieved using RF and SVM classification models. Artificial Neural Networks (ANN) was built as a regression model using the concentration ranges of 0 – 10%. The coefficient of determination (R–squared) was R2 = 0.5786 and the mean absolute error (MAE) was 1.51. In order to calculate the limit of detection (LOD), the training data set obtained from the ANN regression model were used to determine the LOD. The median and mean absolute deviation values of the samples with known concertation versus samples whose concertation were predicted were used to calculate the LOD because they were found to be statistically stable, thus they yielded minimized error bars. Using the ANN model an LOD value lower than 1% was obtained. Thus, the results discussed in this research demonstrate the capability of Raman spectroscopy coupled with PCA, RF, and SVM, and ANN for molecular distinction of authentic and molasses - adulterated honey using the Raman spectral data.
TABLE OF CONTENTS
DECLARATION i
DEDICATION ii
ACKNOWLEDGEMENT iii
ABSTRACT iv
LIST OF TABLES ix
LIST OF FIGURES x
LIST OF ABBREVIATIONS xii
CHAPTER 1: INTRODUCTION
1.1 : Background to the Study 1
1.2 : Honey Adulteration 2
1.3 : Conventional Methods of Honey Adulteration Sensing 3
1.4 : Statement of the problem 4
1.5 : Research objectives 5
1.5.1 : Main objective 5
1.5.2 : Specific Objectives 5
1.6 : Significance and Justification of the study 5
1.7 : Scope and limitation of the study 6
1.8 : Hypothesis of the Study 6
CHAPTER 2: LITERATURE REVIEW
2.1 : Chapter Overview 7
2.2 : Honey Adulteration Detection Using Raman Spectroscopy 7
2.3 : Detection of Honey Adulteration by HFCS and Maltose Syrup Using Raman Spectroscopy 8
2.4 : Detection of Honey Adulteration Using Chemometrics – Integrated Raman Spectroscopy 8
2.5 : Other Spectroscopic Techniques Used in Honey Adulteration Detection 9
2.5.1 : Chemometrics-Integrated Infrared Spectroscopy 9
2.5.2 : Nuclear Magnetic Resonance Spectroscopy (NMRS) 9
2.5.3 : Chemometrics-Integrated High-performance Anion-Exchange Chromatography Coupled with Pulsed Amperometric Detection (HPAEC –PAD) 10
2.6 : Tools Used to Extract Raman Spectral Data for Analysis 10
2.7 : A Summary of Honey Adulteration Detection Techniques Studies 12
2.8 : Summary of Literature Review 14
CHAPTER THREE 3: THEORETICAL FRAMEWORK
3.1 : Chapter Overview 15
3.2 : Theory of Laser Raman Spectroscopy 15
3.3 : Sample and Raman Intensity 17
3.4 : Variants of Raman Spectroscopy and Their Advantages 17
3.5 : Utility of Raman Spectroscopy in Molecular Analysis 18
3.6 : Surface Enhanced Raman Scattering (SERS) 19
3.7 : Tip –enhanced Raman spectroscopy (TERS) 19
3.8 : Machine Learning Techniques Applicable for Raman Spectroscopy Analysis 20
3.8.1 Principal Component Analysis 20
3.8.2 : Random Forest (RF) 21
3.8.3 : Support Vector Machine 23
3.8.4 : Artificial Neural Network 25
CHAPTER 4: MATERIALS AND METHODS
4.1 : Chapter Overview 28
4.2 : Collection of Honey and Molasses Samples 28
4.3 : Raman Sample Substrates Used in the Experiment 33
4.4 : Sample Preparation 34
4.5 : Instrumental Optimization and Sample Analysis 37
4.6 : Raman Spectral Data Acquisition and Analysis 38
4.6.1 : Analysis of Collected Raman Spectra 39
CHAPTER 5: RESULTS AND DISCUSSIONS
5.1 : Chapter Overview 41
5.2 : Characteristic Raman Spectra of Authentic Honey and Molasses 41
5.3 : Identification of Raman Marker Bands for Molasses Adulteration in Honey 46
5.4 : Results of Classification Analysis Using RF and SVM 48
5.5 : Results of Prediction Model Using ANN 51
5.5.1 : Calculation of Limit of Detection (LOD): 51
CHAPTER 6: CONCLUSIONS AND RECOMMENDATIONS
6.1 : Conclusions 54
6.2 : Recommendations 55
REFERENCES 56
APPENDICES 68
Appendix 1: ANN Model Script in R 71
Appendix 2: PCA Script in R 73
Appendix 3: Random Forest Script in R 76
Appendix 4: Support Vector Machine Script in R 79
LIST OF TABLES
Table 2.1: Comparative study of distinctive methods of honey adulteration detection 13
Table 4.2: Mass of molasses expressed as a percentage of the mixture of honey and molasses 35
Table 5.3 (a): Component and Vibrational assignments of Raman bands/peaks identified in authentic honey, molasses and honey adulterated by molasses… 44
Table 5.3 (b): Component and Vibrational assignments of Raman bands/peaks identified in authentic honey, molasses and honey adulterated by molasses… 45
Table 5.4: Classification Models of Set.1, Set.2, and Set.3 using RF and SVM… 49
Table 5.5: Classification models of Set. 4 and Set. 5 using RF and SVM… 50
Table 5.6: Details of variables from the RF Classification Analysis 50
Table 5.7: Training Phase of ANN Model 52
Table A: Carbohydrate content comparison between molasses and honey for 100 g sample 68
Table B: Mineral content comparison between molasses and honey for 100 g sample 68
Table C: Water soluble Vitamin comparison between molasses and honey for 100 g sample 69
Table D: Honey composition in g/100g 69
LIST OF FIGURES
Figure 3.1: Diagrammatic representation of energy transitions between ground vibrations sates and the virtual states of Rayleigh scattering, Stokes Raman and Anti –stokes Raman scattering (Source: Krishnan, 2019) 16
Figure 3.2: Random Forest schematic… 23
Figure 3.3: A multi – layered neuron network architecture with an input layer (layer 1), hidden layer (layer 2), and an output – layer (layer 3) with the back propagation algorithm (Source: (Kadhm et al., 2021) 26
Figure 4.4: Authentic honey sample obtained from the bee department of ICIPE-Kenya (a), and Molasses, a honey adulterant (b) 28
Figure 4.5: Conductive silver paint (a), and three cleaned glass slides (b) 33
Figure 4.6: Electronic balance used to measure, the weight of an empty sample bottle (a), and the weight of a sample bottle with honey and molasses in it (b) 34
Figure 4.7: Prepared mixtures of honey – molasses (HMM) samples at different percentage concentrations of molasses (a), the prepared mixtures stored in the open on a laboratory bench (b). 36
Figure 4.8: Glass slides with smears of silver paint left to dry in the open (a), honey—molasses smears (HMM) over dry silver paint (b) 37
Figure 4.9: Laser Raman Spectrometer equipment in the department of Physics at the University of Nairobi. 38
Figure 5.10: Displaying Characteristic Raman spectral profiles of authentic and molasses (average of – each) in the range of 300 – 1800 cm-1. The variance plot is also plotted with significantly variant bands indicated with dotted lines as an eye guide 42
Figure 5.11 (a): PCA score plot of authentic honey and molasses samples shows close clustering of honey samples and molasses samples, indicating that each sample has the its own unique characteristics. 44
Figure 5.12 (b): Plot of PCA loadings for authentic honey and molasses samples. Unique bands responsible for honey and molasses' segregation have been identified, these bands are the ones with the highest loading value. 44
Figure 5.13: ANOVA on averaged Raman spectra of authentic honey and honey adulterated by molasses at various concentrations. The variance plot has been plotted showing significantly variant bands indicated with dotted lines as an eye guide 46
Figure 5.14: SVM plot utilizing linear kernel function 50
Figure 5.15 (a): Linear fit plot using the median and mean absolute deviation from the training data set of the ANN model. 52
Figure 5.16 (b): Linear fit plot using the median and mean absolute deviation from the test data set of the ANN model. 53
Figure 17: PCA score plot of authentic honey, molasses and all samples adulterated by molasses. LL – Low concentrations, MM – Middle concentrations, HH – High concentrations. 70
LIST OF ABBREVIATIONS
ANN Artificial Neural Network
CARS Coherent Anti – stokes Raman Spectroscopy
GC Gas Chromatography
HMF Hydroxymethylfurfural
HMM Honey Molasses Mixture
HFCS High Fructose Corn Syrup
HPLC High Performance Liquid Chromatography
IUPAC International Union of Pure and Applied Chemistry
LC Liquid Chromatography
LOD Limit of Detection
MAE Mean Absolute Error
MIR Mid Infrared
MSS Maltose Sugar Syrup
NIR Near Infrared
NMR Nuclear Magnetic Resonance
PCA Principal Components Analysis
RF Random Forest
RRS Resonance Raman Spectroscopy
RMSEP Root Mean Square Error of Prediction
SCIRA Stable Carbon Isotope Ratio Analysis
SERS Surface Enhanced Raman Spectroscopy SVM Support Vector Machines
TERS Tip Enhanced Raman Spectroscopy UV Ultraviolet
CHAPTER 1
INTRODUCTION
1.1 : Background to the Study
Laser Raman spectroscopy is a spectroscopic technique typically used to determine vibrational modes of molecules when such molecules interact with laser source of light (Yang and Yi, 2011). Raman scattering of light arises when incident radiation is scattered by molecules resulting in a frequency shift that is either the same as incident radiation or a shifted frequency (Smith and Dent, 2019). The frequency shifts in the molecular transitions of scattered radiation lies between rotational, electronic and vibrational level, and they constitute about 0.0001 % of the incident radiation (Wiley, 2006). The frequency shifts results into Raman effects corresponding to wavelength shifts of a vibrating molecular bond (Long, 2005). Hence, unique molecular structures that form the molecular fingerprints of the sample under study are revealed by the shifts in wavelength (Py et al., 2015).
As a non- destructive technique, Raman spectroscopy involves simple sample preparation procedures that do not need dissolving of samples, or pressing of pellets, or alter the physical or chemical structure of a sample (Zeitler et al., 2007). This limits the possibility of sample cross – contamination (Korth and Ralston, 2002) and reduces clean ups. Moreover, the rich and informative Raman spectra has instrumentally aced Raman spectroscopy capabilities over other analytical methods (Alula et al., 2018) in definitive molecular analyses. In addition, Raman spectroscopy technique is advantageous since, it operates at wavelengths that are independent of vibration modes being studied, with tunable wavelength ranges spanning from UV to NIR being possible (Gaft and Nagli, 2008) even the far – infrared which are very difficult to access (Petry et al., 2003). Furthermore, Raman spectroscopy combined analytical techniques have incredibly facilitated the progressive qualitative and quantitative assessment of food components, thus the quality of food (Yang et al., 2005; Se et al., 2019).
In assessing adulteration, Laser Raman spectroscopy is employed to predict levels of adulteration in virgin oil using soya bean, corn and olive residue (Baeten and Meurens, 1996). Moreover, the technique has also been used in the discrimination of olive oil from various vegetable oils as well as detecting adulteration (Donfack and Materny, 2009). The robustness of laser Raman spectroscopy has been extensively harnessed in detecting urea adulteration in milk (Khan and Krishna, 2015), analysis of chemical profiling of medical counterfeits (Dégardin et al., 2011), and the detection of butter adulteration by margarine (Selin et al., 2013).
1.2 : Honey Adulteration
Honey adulteration is the process of lowering the quality of honey by combining it with low-cost sweeteners that have similar properties but are of lower quality than honey (Tura and Seboka, 2020). The use of cheap sweeteners to adulterate honey is credited with the ability to artificially manufacture cheap sweeteners with profiles that resemble authentic honey (Yaacob, et al., 2019). (Se et al., 2019) specifically mentioned starch syrups, inverted syrups, jaggery syrup, molasses, and date syrup as common cheap sweeteners used to adulterate honey. For example, the distinct dark brown color of jaggery syrup has made it difficult to distinguish it from natural honey on many occasions (Mishra et al., 2010), thus its wide use as a honey adulterant. For instance, molasses, a product of sugarcane refining is commonly used to adulterate honey. The presence of such low-quality honey on the market may have lowered the market price of genuine honey (Zábrodská et al., 2014), thus, limiting the country's economic growth.
Honey adulteration crept into honey production in recent decades as a fraudulent act to meet the insecure global honey supply (Zábrodská et al., 2014). Honey adulteration, according to (Ertelli et al.,2010), has become an appealing vice that helps honey fraudsters meet the ever-increasing market demand for honey and honey products. In a case study of the Czech market, (Zábrodská et al., 2014) found that consumers frequently encounter bogus honey as well as honeys that are occasionally tainted in some way by deadly chemicals such as antibiotics, colorings, and Hydroxymethylfurfural. Excessive consumption of such phony honey is harmful to people's health, particularly diabetics (Fakhlaei et al., 2020). Adulterants like molasses can also cause digestive issues like loose stools, according to the researchers (Gupta, 2020). Ill-health, such as obesity and high blood sugar levels (Ismail et al., 2018), are some of the consequences of consuming tainted honey.
1.3 : Conventional Methods of Honey Adulteration Sensing.
Adulteration sensing in honey includes not only the detection of adulterants directly added to honey, but also the detection of adulterants through the indirect feeding of sugars to honey colonies (Guler et al., 2014). In honey adulteration sensing, conventional methods such as thin layer chromatography (TLC), stable carbon isotope ratio analysis (SCIRA), gas chromatography (GC), liquid chromatography (LC), and high performance liquid chromatography (HPLC) are popular. TLC has been used to detect HFCS (Yaacob, et al., 2019) and to investigate the authenticity of honey by looking into the ratio of fructose to glucose (Cimpoiu et al., 2013). This method, however, is unreliable because more extensive works needs to be done to assess its reliability in detecting (Se et al., 2019a). SCIRA, on the other hand, has proven useful in distinguishing honeys from various botanical sources (Bontempo et al., 2015). GC has proven to be accurate in detecting sugar adulterants and the aroma of honey (Yaacob, et al., 2019). However, using this technique necessitates time-consuming sample preparation (Matute et al., 2007). While LC has been found to be effective in sensing C3 and C4 adulterants (Yaacob, et al., 2019), HPLC has been found to be reliable for routine monitoring of large numbers of samples, despite being labor intensive and requiring an expert (Roussel et al., 2003).
In this research, robustness of Laser Raman spectroscopy as an emergent technique for definitive molecular fingerprint analysis was explored in the detection of the molecular distinction of honey and molasses (honey adulterant) which are two compounds with almost similar molecular composition. The Raman spectra obtained from both authentic honey and honey samples adulterated with molasses were subjected to exploratory analysis to ascertain unique properties of authentic samples and adulterated samples. Similarly, classification analysis was done to check on the ability of laser Raman spectroscopy to in molecular distinction of similar compound. Furthermore, regression analysis to help in building a model that could be used confirming the authenticity of honey samples. Using principal component analysis (PCA) loadings plot, definitive spectral bands characterizing authentic honey, molasses and honey adulterated by molasses at different concentrations were identified. Moreover, for the purposes of building rapid models, principle components (PCs) were used as inputs for both classification and regression models. This greatly reduced the architecture of the models, thus making the modes to be rapid. Classification analysis were done using support vector machine (SVM) and random forest (RF) models, high classification accuracies ranging from 86 – 100 % were achieved. In addition, regression analysis was done using artificial neural network (ANN) model in the concentration range of 0 – 10 %. This yielded a coefficient of determination R2 = 0.5786 and mean absolute error (MAE) of 1.51.
This research is ordered into six chapters: the introduction entails background information and captures a statement of the problem, the objectives, significance and justification of the study and the hypothesis of this study. The chapter on literature review presents a review of literature on honey adulteration detection by Raman spectroscopy and other spectroscopic techniques, a review of machine learning tools applied in extraction of Raman data in honey adulteration studies as well as the discussion of honey adulteration studies. The chapter on theoretical framework discusses Raman spectroscopy and its utility in various applications by considering emissions that are relevant in the interpretation of molecular vibrations that are Raman active. In addition, discussions of Chemometric techniques, namely PCA, RF, SVM, and ANN are also outlined. In the materials and methods chapter, sample preparation approaches are described in detail as well as how Raman spectral data was acquired from each sample set and preprocessed before analysis. In addition, data analysis approaches utilized in interpreting the spectral data are described. Results and discussion chapter entails a description of the collected results, analyses using various techniques and the discussion of the findings. Conclusion and recommendations chapter contains a summary of the main findings and directions for future research.
1.4 : Statement of the problem
The current techniques used in assessing quality of honey are cumbersome and expensive, this makes it hard to frequently carry out quality assessments of honey at every stage of processing, marketing and selling (Kružík, 2017). Thus there is a compelling demand to study and create techniques that are rapid, non-destructive and portable that can easily be used for routine assessments of the quality of honey. If more rapid techniques are developed, it will significantly ensure quality food products are sold in the market, thus the economic viability of each food product such as honey will be realized (Zábrodská et al.,2014).
1.5 : Research objectives
1.5.1 : Main objective
To create an effective quantification and prediction model for assessing honey adulteration through analysis of Raman spectra using PCA, RF, SVM, and ANN
1.5.2 : Specific Objectives
i. To obtain characteristic Raman profiles and identify unique Raman bands of molasses, authentic honey and molasses – adulterated honey.
ii. To perform exploratory data analysis using multivariate calibration on the collected Raman spectra of authentic honey and adulterated honey using selected Chemometrics techniques
iii. Build a prediction model from the collected Raman spectra of authentic and adulterated honey and employ it in determining limit of detection.
1.6 : Significance and Justification of the study
The consequences of honey adulteration as a vice range from health risks such as high blood sugar levels and obesity to economic risks which include and not limited to brake – down of the economy due to low prices of adulterated honey and loss of consumer confidence on the quality of honey (Ismail et al., 2018). The shortcomings due to ineffective honey adulteration detection techniques in the markets today has also greatly contributed to the increase in honey adulteration fraud (Jaafar et al., 2020). The current techniques are unable to cope with the newest and sophisticated adulteration methods, and in so doing low quality honey gets its way into the market thus diminishing the market price of authentic honey (Zábrodská et al., 2014). In addition, the current techniques of honey adulteration detection are cumbersome, use expensive chemicals with some being able to detect only one adulterant (Cotte et al., 2007; Bougrini et al., 2016). These pertinent issues call for a solution. It is for such reasons that this study explores the adequacy of laser Raman spectroscopy coupled with machine learning techniques as a robust definitive molecular analysis method in the assessment of honey adulteration.
The scientific significance of this work is geared towards improving and adding to the existing spectral libraries on honey adulteration using laser Raman spectroscopy. By identifying unique and prominent spectral bands that significantly contribute to honey adulterated by molasses, this research has proved the possibility of laser Raman to be used as a hands on technique for routine monitoring and assessment of honey quality.
1.7 : Scope and limitation of the study
Production of natural honey under varying climatic conditions has made it an easy target of adulteration by low – cost sweeteners. In the market today, honey adulterants range from industrially manufactured syrups as well as indirect feeding of honey bees on sugar syrups. In this work, molasses, a highly viscous by – product from sugar refining is studied as honey adulterant. On its own, the molecular composition of molasses is majorly dominated by fructose, glucose, and sucrose which are also the basic compounds characterizing the honey profile.
1.8 : Hypothesis of the Study
It is possible to create a quantification scheme for honey adulteration detection lower than 1%. Using the distinct vibrational bands present in honey together with the target adulterant, the prediction of trace adulteration levels is achievable with prediction models being developed from a range of low adulteration levels of honey.
Click “DOWNLOAD NOW” below to get the complete Projects
FOR QUICK HELP CHAT WITH US NOW!
+(234) 0814 780 1594
Login To Comment