Unveiling non-linear water effects in near infrared spectroscopy: A study on organic wastes during drying using chemometrics
Graphical abstract
Summary of methodology: near infrared spectral variations related to moisture content variations are obtained for a variety of substrates, and application of principal components analysis is used to analyze the effects of water. The biochemical characteristics of substrates are obtained to investigate water effects’ dependency to chemical types.
Introduction
A growing number of solid organic waste treatment processes such as anaerobic digestion, composting or pyrogaseification are currently being developed and industrialized. Usually, organic wastes cover a wide range of physical characteristics and bio-chemical compositions, making substrate characterization a key issue in optimizing any of these processes. Recently, near infrared spectroscopy (NIRS) has been used to offer a fast, non-destructive, and cost-effective waste characterization system in the anaerobic digestion context (Charnier et al., 2016, Fitamo et al., 2017, Godin et al., 2015, Lesteur et al., 2011, Mayer et al., 2013, Mortreuil et al., 2018) and composting context (Albrecht et al., 2008, Galvez-Sola et al., 2010, Vergnoux et al., 2009). However, a freeze-drying step is always required, due to strong interferences in the near infrared region related to the presence of water in the substrates (Lobell and Asner, 2002, Williams, 2009). Not only is this drying step cumbersome and impedes any online application, but the volatilization process that takes place during drying makes some characteristics (volatile fatty acids) impossible to predict directly. Though some applications have been developed for the characterization of liquid samples with the presence of water, these are usually restricted to a limited moisture content range, as well as one substrate type (Jacobi et al., 2009, Stockl and Lichti, 2018). In fact, near infrared spectroscopy is sensitive to numerous factors including the spectrometer lamp temperature (Sánchez et al., 2003), sample presentation (Sørensen et al., 2014), light penetration depth (Padalkar and Pleshko, 2015), sample particle size distribution (Igne et al., 2014), sample temperature (Sánchez et al., 2003), and moisture content (Lobell and Asner, 2002). All these interfering factors need to be accounted for in order to build robust quantitative calibrations (Acharya et al., 2014, Zeaiter et al., 2004). Furthermore, these factors may interact together, leading to more complexity for their correction. Indeed, for example, a close relationship between moisture effects and temperature has been outlined (Renati et al., 2019, Wenz, 2018), leading to account for both factors in conjunction (Hans et al., 2019).
The effect of moisture content on near infrared spectra has been described for a wide variety of different matter types including soil (Bogrekci and Lee, 2006, Bowers and Hanks, 1965, Chang et al., 2005, Knadel et al., 2014, Lobell and Asner, 2002, Sudduth and Hummel, 1993, Wu et al., 2009), crops (Gaines and Windham, 1998, Gergely and Salgó, 2003, Peiris et al., 2016, Popineau et al., 2005, Williams, 2009), food (Büning-Pfaue, 2003), plants (Carter, 1991), wood (Giordanengo et al., 2008), pharmaceuticals (Igne et al., 2014), object models (Reeves, 1995, Reeves, 1994, Wenz, 2018), and water-dominant systems (Muncan and Tsenkova, 2019). In addition, though not focused on the analysis of moisture content effects in NIRS, some studies use NIRS to monitor drying or hydration processes where moisture content varies (Caponigro et al., 2018, Raponi et al., 2017). However, no study has yet analyzed and compared moisture content effects in one comprehensive experiment with a wide variety of biochemical and physical types. Better understanding water effects and how they relate to the substrate properties appears as key for the development of robust calibrations models on wet substrates. Indeed, groups could then be used for building local models, an approach which has been shown to be successful for biochemical methane potential (BMP) prediction on plant biomasses (Godin et al., 2015).
The main effect of moisture content variations on NIR spectra usually put forward in studies relates to the apparition of three broad OH absorbance bands (detailed further on); but one major effect of water relates to physical effects (ie. changes in scattering). This is why, when speaking about water effects, an important aspect to have in mind concerns the measurement mode. For transparent liquid samples such as pure water or clear suspensions, transmission or transflexion mode is usually preferred (Pasquini, 2003), while for solid samples like powders, diffuse reflection appears most suitable. When studying large moisture content variations, one substrate may cover various states from a clear suspension, to a sludge-type material, to a powder when fully dried.
Because near infrared spectra contain both physical information (such as granulometry) and chemical information (compound concentration of interest), a pre-processing step is commonly used to maximize the chemical information in the spectra. This is done by getting rid of baseline effects due to scattering (referred to additive and multiplicative effects), as well as using spectral derivation to deconvolve the peaks. A wide variety of pre-processing techniques are used (Rinnan et al., 2009, Zeaiter and Rutledge, 2009), sometimes even in combination (Roger et al., 2020). However, these pre-processing steps may bring important artefacts (Rabatel et al., 2019) in the spectra when applied inappropriately. As well, some pre-processing steps such as derivation may deport the chemical information on shifted peak positions which can make the assignment of bands more complicated (Oliveri et al., 2019). Nevertheless, such pre-processing steps will most likely remain necessary when building quantitative models.
In the context of highly diverse matter types, water effects are expected to vary at least according to the biochemical characteristics. Exploring such differences in effects is the aim of this article. A customized air-drying system was built, allowing the simultaneous monitoring of samples’ moisture content and acquisition of near infrared spectra during drying. Using this system, spectral variations related to moisture content variations were obtained for a large variety of substrates. A principal component analysis was used to explore the various effects. The aim of this global PCA was to identify major groups of substrates in regards to water effects. This was done by analyzing the scores’ kinetics of each substrate during drying in relation with the interpretation of each component loadings using band assignments (Williams and Antoniszyn, 2019, Workman and Weyer, 2012). Because the aim of the study was to explore the water effects, including baseline modifications related to scattering effects, data analysis was done on the raw spectra, without any prior pre-processing steps.
Section snippets
Sample preparation
The study was conducted on substrates chosen to represent a wide range of organic wastes with different chemical compositions: fruits (banana, apple), vegetables (carrots, onions, salads, potato), farm wastes (manure, silage, soya meal, grass), dairy products (cream, yoghurt, butter), meat products (beef, grilled/fresh meat, fish), as well as food industry materials (sugar, sauces, fried potatoes, wheat flour). In order to provide control samples with simplified water effects due to
Biochemical characteristics
Fig. 2 presents the predicted characteristics obtained using the near infrared spectroscopy calibrated model for freeze-dried and ground samples. Samples (detailed in Section 2.1) cover a very wide variety of biochemical types which is representative of the variety of inputs possibly used in the anaerobic digestion process, in particular in co-digestion plants. All biochemical characteristics show non-Gaussian distributions, which will impact the structure of the data. Some extreme samples will
Conclusion
The present study investigated the complexity of water effects in near infrared spectroscopy and highlighted the close dependency with the biochemical and physical characteristics of samples.
A customized acquisition system allowed to obtain a unique dataset comprising NIR spectral variations related to water content modifications in standard conditions (ambient temperature/humidity) with no heating nor chemical altering (oxidation, Maillard reactions). Such water spectral variations were
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
This work was supported by the French Agency of National Research and Technology (ANRT) [grant number 2018/0461].
Authors would like to thank Guillaume Guizard and Philippe Sousbie for their technical support during the setting-up of the experiment; as well as the ChemHouse group for limitless discussions on chemometrics.
References (77)
- et al.
Efficiency of near-infrared reflectance spectroscopy to assess and predict the stage of transformation of organic matter in the composting process
Bioresour. Technol.
(2008) Analysis of water in food by near infrared spectroscopy
Food Chem.
(2003)- et al.
Assessing the influence of temperature on NIRS prediction models for the determination of sodium content in dry-cured ham slices
Food Chem.
(2018) - et al.
Effect of temperature variation on the visible and near infrared spectra of wine and the consequences on the partial least square calibrations developed to measure chemical composition
Anal. Chim. Acta
(2007) Temperature dependence of light absorption by water
Nucl. Instruments Methods Phys. Res. Sect. A Accel. Spectrometers Detect. Assoc. Equip.
(2013)- et al.
Influence of drying on functional properties of food biopolymers: From traditional to novel dehydration techniques
Trends Food Sci. Technol.
(2016) - et al.
Rapid biochemical methane potential prediction of urban organic waste with near-infrared reflectance spectroscopy
Water Res.
(2017) - et al.
The potential of near infrared reflectance spectroscopy (NIRS) for the estimation of agroindustrial compost quality
Sci. Total Environ.
(2010) - et al.
Biochemical methane potential prediction of plant biomasses: Comparing chemical composition versus near infrared methods and linear versus non-linear models
Bioresour. Technol.
(2015) - et al.
Robustness of calibration models based on near infrared spectroscopy for the in-line grading of stonefruit for total soluble solids content
Anal. Chim. Acta
(2006)
First step towards a fast analytical method for the determination of Biochemical Methane Potential of solid wastes by near infrared spectroscopy
Bioresour. Technol.
The impact of signal pre-processing on the final interpretation of analytical outcomes – A tutorial
Anal. Chim. Acta
Free/bound water absorption in an epoxy adhesive
Polymer (Guildf).
Temperature dependence analysis of the NIR spectra of liquid water confirms the existence of two phases, one of which is in a coherent state
J. Mol. Liq.
Review of the most common pre-processing techniques for near-infrared spectra
TrAC - Trends Anal. Chem.
Sequential preprocessing through ORThogonalization (SPORT) and its application to near infrared spectroscopy
Chemom. Intell. Lab. Syst.
EPO-PLS external parameter orthogonalisation of PLS application to temperature-independent measurement of sugar content of intact fruits
Chemom. Intell. Lab. Syst.
Discriminating from highly multivariate data by Focal Eigen Function discriminant analysis; application to NIR spectra
Chemom. Intell. Lab. Syst.
Near-infrared spectroscopy (NIRS) for a real time monitoring of the biogas process
Bioresour. Technol.
Achieving robustness to temperature change of a NIRS-PLSR model for intact mango fruit dry matter content
Postharvest Biol. Technol.
Monitoring of the evolution of an industrial compost and prediction of some compost properties by NIR spectroscopy
Sci. Total Environ.
Examining water in model membranes by near infrared spectroscopy and multivariate analysis
Biochim. Biophys. Acta - Biomembr.
Robustness of models developed by multivariate calibration. Part I: The assessment of robustness
TrAC - Trends Anal. Chem.
Robustness of partial least-squares models to change in sample temperature: II. Application to fruit attributes
J. Near Infrared Spectrosc.
Accuracy improvement of in-line near-infrared spectroscopic moisture monitoring in a fluidized bed drying process
Front. Chem.
Effects of soil moisture content on absorbance spectra of sandy soils in sensing phosphorus concentrations using UV-VIS-NIR spectroscopy
Trans. ASABE
Reflection of radiant energy from soils
Soil Sci.
Hydration of hydrogels studied by near-infrared hyperspectral imaging
J. Chemom.
Primary and secondary effects of water content on the spectral reflectance of leaves
Am. J. Bot.
Influence of soil moisture on near-infrared reflectance spectroscopic measurement of soil properties
Soil Sci.
Fast characterization of solid organic waste content with near infrared spectroscopy in anaerobic digestion
Waste Manage.
Influence of sample temperature for measurement accuracy with FT-NIR spectroscopy
J. AOAC Int.
Effect of wheat moisture content on meal apparent particle size and hardness scores determined by near-infrared reflectance spectroscopy
Cereal Chem.
Changes in moisture content during wheat maturation - What is measured by near infrared spectroscopy?
J. Near Infrared Spectrosc.
Humidity Fixed Points of Binary Saturated Aqueous Solutions
J. Res. Natl. Bur. Stand. Phys. Chem.
Temperature and Moisture Insensitive Prediction of Biomass Calorific Value from Near-Infrared Spectra Using External Parameter Orthogonalization
J. Near Infrared Spectrosc.
Cited by (8)
Fast and robust NIRS-based characterization of raw organic waste: Using non-linear methods to handle water effects
2022, Water ResearchCitation Excerpt :The histograms of obtained reference values are presented in Fig. 1. Furthermore, in order to evaluate the robustness of developed NIRS models towards moisture content effects, a dataset of NIRS measurements acquired during N2-drying experiments of various organic substrates was used as described in (Mallet et al., 2021a). The oven-drying was not used because of possible chemical modifications at high temperatures (Maillard reactions), and freeze-drying was not used because it requires to freeze the sample and temperature strongly modifies the measured near infrared spectra.
Agronomic characterization of anaerobic digestates with near-infrared spectroscopy
2022, Journal of Environmental ManagementCitation Excerpt :This characterization method can still be optimized by finding a way to remove water-related disturbances and, thus, performing the NIR measurements directly on fresh samples. This improvement path is a current research subject and requires first identifying the effects of water (Mallet et al., 2021) on the NIR signal in order to subtract them afterwards from the NIR signal. This could also allow the calibration of an ammonium prediction model in addition to organic nitrogen, as ammonium is at the very heart of spreading strategies reducing the risk of acidification and eutrophication of natural environments.
Applications of near infrared spectroscopy and hyperspectral imaging techniques in anaerobic digestion of bio-wastes: A review
2022, Renewable and Sustainable Energy ReviewsCitation Excerpt :The moisture content and particle size distribution of a sample can affect the precision and robustness of a NIRS prediction model because high moisture content and large particle size will mask some NIRS spectral information [39]. Prior to collecting NIRS spectra, the organic waste samples usually need to be dried and ground to avoid the effect of water content and sample heterogenicity [40]. The water removal from organic waste samples can be achieved via air drying [41], oven-drying [39,42–46] or freeze-drying [1,20,47,48] as summarized in Table 1.
Drying and Valorisation of Food Processing Waste
2023, Drying and Valorisation of Food Processing Waste