Natural food additives must be safe for human consumption and packaged safely to enter the European market. This means you must have a food safety management system in place as an exporter to become successful in Europe. Offering traceable and sustainable products is also becoming more and more important in light of the European Green Deal. Expect buyers to demand proof of your product’s traceability, safety and quality before they will buy from you.
Honglai supply professional and honest service.
As an exporter of natural food additives, you must ensure the safety of your products for use in the European market. When exporting to Europe, you must comply with the legally binding requirements of the European Union (EU). The most crucial requirements are related to safe trade in chemicals and food safety.
To enter the European food market, you must comply with several regulations that guarantee food safety. One of these is the General Food Law and its Regulation (EC) No 178/. This law mainly sets requirements for traceability, hygiene, and control. Compliance with this legislation ensures that your natural food additives are safe to eat and that legal limits for food contaminants are not exceeded.
Important for the control of food safety hazards throughout the whole supply chain is the implementation of food safety management based on Hazard Analysis and Critical Control Points (HACCP) principles. HACCP planning consists of consecutive steps to:
As part of its efforts to protect its consumers, the EU carries out regular official controls at the border and at all further stages of marketing. In case of non-compliance with the European food legislation, individual cases are reported through the Rapid Alert System for Food and Feeds (RASFF), which is freely accessible to the general public. In most cases of non-compliance, European importers will not pay for the product or demand their money back. Additionally, a food safety issue will damage your reputation on the market.
Be aware that repeated non-compliance with European food legislation by a particular country might lead to special import conditions or even suspension of imports from that country. Those stricter conditions include laboratory tests for a certain percentage of shipments from specified countries.
For instance, the European Commission has banned titanium dioxide (E 171) as a food additive in the EU as of . This is the result of an updated safety assessment by the European Food Safety Authority (EFSA) as it is not considered safe as a food additive and genotoxicity could not be ruled out. Other examples of banned food additives in the EU are brominated vegetable oil (BVO) and Sudan dyes.
Regulation (EC) / sets rules for the use of food additives, including definitions, conditions of use, labelling, and procedures. In addition to the regulation, two of its annexes require your attention:
Every EU-approved food additive gets an E number. Food additives with an E number have passed safety tests and have been approved for use. The safety evaluations of food additives are done by the expert panel on Food Additives and Flavourings (FAF) of the European Food Safety Authority (EFSA). They examine all relevant scientific data, considering chemical and biological properties, potential toxicity, and estimates of human dietary exposure. Using this information, the panel makes conclusions about the safety of the intended uses of the food additive for consumers.
Regulation (EC) / sets the rules for the use of flavourings in and on foods. Natural ingredients such as oleoresins, essential oils and some extracts are classified as flavourings and must comply with this regulation.
Note that European custom authorities will not allow food products on the market if they contain undeclared substances (i.e., adulterants) or additives or flavourings which are not permitted in the European Union.
Your products must be safe for consumption and must not be contaminated by:
These contaminants could be substances unintentionally added to your product during the stages of production, processing, or transport, or due to environmental contamination.
The European Union has implemented regulations to control and minimise the presence of contaminants in food. Of importance are:
These restrictions on raw materials also extend proportionally to derivative products, such as extracts and food additives. Products containing pesticides or contaminants above the permitted level will be withdrawn from the market.
To ensure safety for consumers, the use of extraction solvents in the production of foods is regulated by the European Union:
European buyers usually verify compliance to the rules by checking information on extraction solvents and residue levels in your Certificate of Analysis (CoA - for an example, see this CoA of Grionia simplicifolia).
Some natural ingredients such as certain essential oils are classified as hazardous. To ensure safety during transport and handling, special packaging must be used and warning labels must be included on the packaging.
Regulation (EC) No / sets out the rules on classification, labelling and packaging (CLP) of substances and mixtures. This CLP regulation applies to food additives and flavours. The aim of this regulation is to identify hazardous chemicals, inform users about their hazards using standard symbols and phrases and provide rules on packaging.
In December , the European Commission reached some preliminary agreements on the revisions of the CLP regulation, including classifying essential oils as ‘potentially harmful chemicals’. The EU Parliament has already proposed an exemption for “renewable substances of botanical origin”, such as essential oils. The European Parliament is now in the process of adopting the new regulation.
If your product is classified as hazardous, your Safety Data Sheet (SDS) as well as your label must include the relevant safety phrases, risk phrases and hazard symbol. Risk phrases indicate the main risks and hazards, while safety phrases indicate the safety measures that need to be taken because of those risks and hazards.
Figure 1: Example of hazard symbol required for clove oil
Source: ECHA,
You are obliged to label your products if you want to export them to Europe. Your product needs to be labelled so European buyers as well as customs authorities, for example, can trace the origins of your product. Another reason is to ensure safety during transport and storage.
In addition to complying with the CLP Regulation, you must apply common export labelling rules. EU’s labelling requirements are outlined in the EU’s food additives and flavourings legislation: Regulation (EC) No /. Whether or not your product is intended for sale to the final consumer determines what labelling requirements you must comply with. This is outlined in chapter IV of the regulation, under Articles 21, 22 and 23.
In general, you must ensure that your labels include the following information in the English language, unless your buyer indicates otherwise:
If you offer Organic certified ingredients, you should also add the name/code of the certifier and the certification number.
If you want to export natural ingredients as food additives, you also need to comply with international treaties. These are especially important if you produce wild-collected ingredients.
You need to comply with theConvention on International Trade in Endangered Species of Wild Fauna and Flora (CITES). CITES aims to protect endangered plants and plant products by regulating their trade. This convention provides a list of plant species that you cannot export/import or where export/import is restricted. If your product is listed in Annex A and Annex B of the EU wildlife legislation, Regulation (EC) No 338/97, you need to get an export permit from your country’s CITES authority. You will also need an import permit from the country you are exporting to. For instance, if you want to export to the Netherlands, you can apply for the CITES import document (in Dutch) at the Netherlands Enterprise Agency, RVO.
European companies adopt diverse strategies when addressing CITES-related concerns. Certain companies avoid utilising raw materials enlisted in CITES, whereas others insist that species listed in CITES should originate from cultivation rather than wild collection. In all cases, suppliers are responsible for complying with current law and international standards, including CITES permits. As a supplier, it is also your responsibility to have the necessary documentation in place. European companies typically collaborate with you to acquire information and ensure adherence to EU regulations.
You also need to determine if and how the Nagoya Protocol of the Convention on Biological Diversity (CBD) applies to your product. This protocol aims to make sure the benefits of genetic resources and traditional knowledge are shared in a fair way. It provides guidelines for accessing and utilising genetic resources and traditional knowledge in Access and Benefit Sharing (ABS) agreements. ABS is especially important for wild-collected ingredients.
The European Union is a signatory of the protocol. Regulation EU 511/ sets the rules for the implementation of the Nagoya Protocol at the EU level. Many other countries have signed this protocol and adopted it into national law. If your home country did as well, you need to comply with these national laws. European companies are legally required to follow those laws that are in force in your country regarding access and benefit sharing. Your buyers will expect you to be aware of and compliant with your country’s regulations on this topic.
Certain European companies require their suppliers to acquire the Internationally Recognized Certificate of Compliance (IRCC) from the Access and Benefit Sharing Clearing House (ABSCH). This certificate verifies that companies have legal access to genetic resources in accordance with the Nagoya Protocol. It is commonly required from suppliers in countries identified with a high risk of non-compliance with ABS regulations.
Figure 2: Video – The Nagoya Protocol and ABS, simply explained
Regulation (EC) No / is concerned with the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH). The REACH system aims to ensure a high level of protection of human health and the environment, including the assessment of hazards in substances. REACH is especially relevant for essential oil exporters.
However, from a legal point of view, essential oils for food are not subject to the same REACH documentation requirements as essential oils for other applications, such as fragrances. The European Union legislation on REACH requires suppliers of essential oils for these alternative applications to provide a lot more documentation. Essential oils for flavourings are exempt from REACH.
Nonetheless, in practice, European buyers of essential oils for flavourings often have the same documentation requirements as buyers of essential oils for other uses. Since they do not know how their customers will use the essential oil, they need to have all the documentation required for different potential applications, including flavourings and fragrances.
Note that exporters that supply less than 1 tonne annually do not need to register their product under REACH. This saves them a considerable administrative burden. The reason that these exporters can still supply to Europe is related to an underlying safety principle of REACH: greater quantities that are manufactured or imported pose greater risks to human health or the environment. A chemical substance that is imported or manufactured by a European company at less than 1 tonne per year falls outside the scope of REACH.
Many buyers have additional requirements which go beyond legislation. These are mainly focused on documentation requirements, sustainability, food safety management, and quality.
European buyers of natural ingredients require exporters to provide them with well-structured and organised product and company documentation. Buyers use it to verify whether you meet their requirements and specifications.
European buyers of natural ingredients usually expect exporters to provide them with a Safety Data Sheet (SDS), Technical Data Sheet (TDS) and Certificate of Analysis (CoA). Table 1 shows what information you need to include in these documents.
Table 1: What is contained in the Safety Data Sheet (SDS), Technical Data Sheet (TDS) and Certificate of Analysis (CoA)
Safety Data Sheet (SDS)Technical Data Sheet (TDS)Certificate of Analysis (CoA)
that matches
Product name, description and classificationProduct name, description and classificationSpecifications mentioned in the TDS Hazard identification Quality that you guarantee to supply Pre-shipment samples approved by buyerInformation on safety measuresInformation on applications Contractual agreements with buyerCertificatesSource: ProFound
Food safety is a top priority in all European food sectors. You can therefore expect European buyers to request extra guarantees in the form of certifications. European buyers of natural food additives often require their suppliers to have the ISO certification on quality management systems and the ISO certification on food safety management.
In addition, for many European buyers the implementation of standards and certifications recognised by the Global Food Safety Initiative (GFSI) is a minimum requirement. The most commonly used food safety management systems in Europe are FSSC , BRCGS and IFS.
For some buyers, especially for large European companies, quality management requirements also apply to the production of the raw materials. In such cases, buyers want to ensure that the raw material used to produce the corresponding natural food additive is cultivated under Good Agricultural Practices (GAP). One of the most common certification schemes used in this regard is Global G.A.P. The Farmer Sustainability Assessment (FSA) of the SAI Platform is another popular tool used by European companies to evaluate quality management in raw materials production.
Figure 3: Examples of food safety certification
Source: standards websites
Table 1: Most important food safety and quality management certifications requested by European buyers
Certification nameType of certification Cost for companiesHow to get certified? ISO : Quality management Certification costs depend on factors such as company profile, sectors, annual turnover, number of sites and staff.You can buy the standard through the ISO website, which lists the requirements for essential features of a quality management system.
If you want to certify your quality management system, look for an accredited certification body in your country that offers ISO certification.
ISO :
Food safety management systemCertification costs depend on factors such as your company’s business activities and location.You can buy the ISO standard through the ISO website.
Always look for an accredited certification body in your country.
FSSC Food safety management systemCertification costs depend on factors such as your company’s business activities and location.Check the FSSC website on how to become certified.
Always look for an accredited certification body in your country.
British Retail Consortium Global Standards (BRCGS)Food safety management systemCertification costs depend on factors such as your company’s business activities, size, and location.Check the website to see how to get BRCGS certified.
Visit the BRCGS partner sections to find a certification body.
International Featured Standards (IFS) FoodFood safety management systemCertification costs depend on factors such as your company’s business activities, size, number of products, and location. See the ‘Roadmap to certification’ on the IFS website to learn how to get certified. Refer to the IFS website to find an accredited certification body in your country.Source: ProFound
Adulteration (the intentional addition of undeclared substances to a product) is a serious issue in the natural food additives trade. Adulteration causes the product to become unusable for the buyer. It loses the necessary properties to fulfil its intended function, and purification is usually either impossible or too costly. Strict controls are in place in Europe to detect potential adulterants. If adulteration is detected, you as a supplier will not be paid for your products and you will lose business with your buyer.
For instance, essential oils for food must be 100% pure (i.e., not mixed with any other essential oils that have similar characteristics) and 100% natural (i.e., not adulterated through the addition of any chemicals). If you dilute an essential oil with solvents, the product must be called a flavour or you must state that the oil is diluted (e.g., 20% in Propylene Glycol). When you have mixed your essential oils, you should call them a blend of essential oils.
European buyers generally send samples to laboratories to analyse them. These laboratories continuously improve their techniques to determine the purity of products. In case there are undeclared substances in your product, it is very likely they will be detected.
In addition, buyers appreciate quality consistency over the year through standardisation. They expect suppliers to standardise products by establishing Standard Operating Procedures (SOP) for collection/harvesting and processing practices (e.g., timing of harvest and blending early and late crops). Often buyers prefer a large volume with a consistent quality level over smaller batches of different qualities.
Packaging requirements can differ from buyer to buyer, and therefore must always be agreed with your buyer. In general, European buyers demand high-quality ingredients. To meet these high requirements, you should preserve the quality of your products by always:
In the specific case of handling essential oil, make sure to always use United Nations approved packaging for dangerous goods.
The EU is committed to environmental sustainability and sustainable growth, something it has made clear in the European Green Deal (EGD). As part of the EGD, new laws are proposed to increase the responsibility of European manufacturers to explain where and how products are produced and what impacts these have on people and the environment.
An example of these legislative changes is the proposal for a Directive on corporate sustainability due diligence, which was adopted by the European Commission in February . The new rules will ensure that businesses address adverse human rights and environmental impacts of their actions, including their value chains inside and outside Europe. This directive proposal is in line with the Farm to Fork strategy, part of the European Green Deal. The Farm to Fork strategy aims for fair, healthy and environmentally friendly food systems across Europe.
A transparent supply chain is key to deal with sustainability concerns. Since , Regulation (EU) / sets the rules on the transparency and sustainability of the EU risk assessment in the food chain. While this regulation brings notable changes to the approval process for items such as food additives and flavourings, its direct impact on your company is improbable.
Nevertheless, European buyers face growing pressure to ensure their supply chains are sustainable. As such, one of the main aspects for European buyers to choose product suppliers is a traceable and transparent supply chain. They want to have guarantees that a product they buy matches product specifications and can be traced back to the source. This might mean that you need to put more rigorous traceability systems in place to be able to deliver the information that your buyers demand of you. It also means you should have information on production and labour practices, as well as environmental issues in your chain.
Buyers expect their suppliers to provide them with all the necessary information. Remember that such information is increasingly digitalised, making it important to follow this trend. Digitalising sourcing information provides improved transparency, increases access to information and statistics, and allows for more efficient purchase and payment processes. At the same time, it puts additional demands on your company to collect and disclose data.
European buyers may also specifically request that you comply with their code of conduct. Most such codes are based on the UN Global Compact principles or Ethical Trading Initiative Base Code and focus on issues such as human rights and fair working conditions and wages. Buyers expect their suppliers to follow these codes of conduct and often require confirmation in writing as part of business contracts.
Companies with global supply chains use EcoVadis and/or the Sedex Members Ethical Trade Audit (SMETA) to assess and audit their suppliers. Both sustainability reporting systems are recognised worldwide. EcoVadis covers four sustainability topics: environment, labour and human rights, ethics and sustainable procurement. As a company, you are either asked by your customers to undergo an EcoVadis assessment or you do it voluntarily. Doing it voluntarily serves as a check and external recognition of your sustainability status as a company. The assessment consists of a questionnaire tailored to the size and products of your company.
SMETA on the other hand is a social auditing methodology. Becoming a Sedex-certified member can give you a big competitive advantage if you wish to supply these large companies. Although SMETA is still a voluntary standard, it is likely to become a basic industry requirement in the coming years.
Verifying and/or certifying sustainable production represents a niche market in the natural food industry. However, it can add value to your product. Organic certification is the most common standard in the EU market for natural food additives. Other social and environmental sustainability standards and requirements include fair trade standards.
Certifying your ingredients may help you to prove the traceability of your products, as this is verified and documented by an independent third-party auditor. See the main certification standards in the table below.
Table 2: Most important product certifications requested by European buyers
Are you interested in learning more about Additives for ETICS? Contact us today to secure an expert consultation!
Certification nameType of certification Cost for companiesHow to get certified? EU Organic OrganicCosts vary and depend on set-up, scale, location and non-conformities.Refer to Regulation (EU) /848 to learn more about the legislative requirements.
Access the list of recognised control bodies and control authorities for EU Organic, issued by the EU.
FairWild Social and environmental sustainability (wild- harvested species)Calculations of the cost of the certification audit are made individually. They depend on the location, size and complexity of operations and include audit, evaluation, certification and office costs. See the approved control bodies and accreditation section on the FairWild website for further information about obtaining certification.Fair For Life Social sustainability for both wild and cultivated species Certification costs vary depending on the size and complexity of operation, location of your operation and of producers.Access the Fair for Life certification process to learn about the steps that must be followed to become certified. Operators usually undergo a full recertification audit process every year.Fairtrade InternationalSocial sustainabilityAccess the cost calculator of FLOCERT to get an estimate of your costs for getting Fairtrade-certified.Consult this link to learn how to become a Fairtrade producer. Operators usually undergo a full recertification audit process every 1-2 years.UEBT certification programmesEthical sourcing and biodiversityCertification fees depend on the type of certification scheme: Ethical Sourcing system certification, ingredient certification and UEBT and Rainforest Alliance Herbs & Spices programme. See the UEBT certification bodies section on the UEBT website for further information about obtaining certification.Rainforest AllianceEnvironmental sustainabilityCertification costs are not fixed. Companies must apply to the authorised certification body for an offer. Total costs include administrative costs, audit fees, a premium price paid to farmers, sustainability investment costs and a volume-based royalty.Contact an authorised certification body in your country to find out about their fees and process to get certified.
Consult the guide for farmers and the guide for companies on how to get certified.
KosherJewish dietary lawsAfter filling out an application form, this will be assessed after which you will receive a quotation. The simplicity or complexity of the product (from a kosher perspective, not necessarily from a manufacturing one) will play a major role in the costs for kosher certification. Read these steps to get kosher-certified.HalalIslamic dietary lawsThe costs of Halal certification depend on factors such as the type and number of products being certified. By filling in an application form you will receive a quotation for Halal certification.Read the Halal certification procedure to learn how to get Halal certified.
Select a certifier accredited by the World Halal Council for recognition of your certificate in Europe.
Source: ProFound
Although the above-mentioned certifications are the most commonly used in the natural food additives industry, requirements may vary by buyer. Therefore, before getting certified you should talk to your buyers and verify their needs.
For certain European companies, organic certification is a prerequisite to enter the market, as they only deal with organic products. Other companies may provide a range of products including both conventional and organic ingredients.
If you want to market organic ingredients to Europe, you need to comply with European requirements on production and labelling of organic products. In January , the new EU organic regulation (EU) /848 entered into force. This regulation adds new checks for imported organic products. Compliance can entail a major shift in your company’s processes. You need to move to permitted pesticides and fertilisers, control weeds naturally, implement a full traceability and internal control system and switch to only use permitted solvents during extraction (water, steam, or organic alcohol).
In the European Union, a product can have the organic production logo if it has been certified, and only if at least 95% of the ingredients from agricultural origin are organically produced . Note that only certain food additives may be used: check the positive list of additives (Part A of Annex V).
Although actual demand for Kosher and Halal products is small, demand for certification of compliance against these religious standards is increasing. Kosher and/or Halal Certification allows food manufacturers to use the ingredients in Kosher and Halal products. European buyers aim to prevent exclusion from these markets. Many European buyers therefore now require Kosher and/or Halal certification.
For instance, Halal certification for gum arabic requires cleaning procedures according to Islamic law. Alcohol cannot be used in the processing. Certified Halal products must also be kept separated from “impure” products, such as products from pigs. A unified standard for Halal does not yet exist.
European consumers and retailers are putting increasing pressure on companies to ensure their products are made according to social and environmental standards. Some European product manufacturers have made meeting environmental and social standards part of their policy and strategy.
Whether or not European buyers of natural food additives are interested in certified ingredients depends on the type of product they produce and how they communicate their sustainability to their customers. If an ingredient makes up only a small share of an end-product, it is difficult to communicate its sustainability to consumers. This means that, in general, standards are used only incidentally for natural food additives.
ProFound – Advisers In Development carried out this study on behalf of CBI.
Food without additives has been favored by consumers. However, there are many foods that contain additives but are marked as additives-free and natural foods. The purpose of this work is to detect several common additives in drinks using three-dimensional fluorescence spectroscopy (TDFS) combined with an independent component analysis (ICA) algorithm. In the experiment, the artificial sample and the real sample are analyzed to determine the components. For artificial samples, acceptable results were able to be obtained even if the components are highly correlated. For real samples, some were shown to contain more than one kind of additive that is not consistent with the label “additive-free.” Three parameters (root mean square error of prediction [RMSEP], a similarity coefficient [p], and an R-squared estimate [R2]) are used to evaluate results. The results indicate that ICA works well in food additive detection. In addition, ICA can analyze the raw spectra without data preprocessing. Therefore, this work is helpful for food safety inspection.
Massive amounts of chemical additives are added to foods to maintain or improve the color, flavor, and sweetness, as well as to extend the expiration date (1). Legislation has indicated the conditions and the maximum permissible quantity of approved food additives based on food safety for consumers (2). In addition, long-term consumption of food containing additives can have adverse effects on health, and food safety issues also arise frequently (3,4). Therefore, additive-free foods are favored by consumers (5).
There are various brand of fresh juice on the market that are labeled “additive-free.”However, some of these juices have been found to contain additives. To avoid use of any illegal additives or excessive quantities of additives, strict quality control, through the identification of food additives and quantification of their levels, is required by regulatory agencies. Therefore, a more accurate method is needed for the detection of additives in fruit juice sold in the market.
There are some effective methods, using Raman spectroscopy (6), near-infrared (NIR) spectroscopy (7,9), ultraviolet–visible (UV–vis) spectroscopy (8), and a hyperspectral imaging (HSI) techniques (10–11), that have been successfully applied to the detection of multicomponent systems. However, these methods are also time-consuming, expensive, and complex compared to fluorescence spectroscopy. Fluorescence spectroscopy, with its high selectivity and sensitivity, has been applied to the analysis of multifluorophoric systems (12). Recently, the most common method for analysis of multifluorophoric systems is second-order correction, such as parallel factor analysis (PARAFAC) (13) or self-weighted alternating trilinear decomposition (SWATLD) (14). Second-order correction, based on three-dimensional fluorescence spectra (TDFS) of multicomponent samples, is a blind separation process by which the source spectra and corresponding concentrations can be inferred from several mixed spectra. But for PARAFAC, for example, its essence is an alternate least squares method, which is easily affected by multicollinearity, resulting in distortion of decomposition results. To address this question, independent component analysis (ICA) has been developed to solve the blind separation problem and was published just before the end of the 20th century. Comon proposed the concept of ICA, and gave the mathematical model of the concept, in (15). Hyv Ärinen and Ojaz proposed the fixed-point iterative algorithm in , which has become a classic ICA algorithm, because of its high convergence rate (16). The ICA is a method based on high-order statistics, which can decompose observation data into statistically independent linear combinations of signal sources to reveal the implicit information inside observation data. Efficient analytical methods based on ICA have also been developed to resolve independent components (ICs) from mixed signals. The approach has been widely applied in signal processing for analytical chemistry, including the extraction of pure-mass spectra from overlapping spectrometric matrices (8), the identification of constituents in commercial gasoline, and in the processing of NIR spectral data (17).
In this work, an ICA method is introduced for the detection of food additive samples by using the TDFS technique. Furthermore, the results of the artificial samples and real samples obtained by the ICA algorithm are discussed in detail. From the results, it can be seen that ICA can be applied to the detection of additives in juices, is helpful in the sampling test for juices, and is potentially of significant importance in food safety inspection.
Methods and Experiments
Experiment
Because the advantages of TDFS include providing the information of both excitation spectra and emission spectra, an FS920 fluorescence spectrometer (Edinburgh Instruments) was used in the fluorescence detection for each sample. To reduce the difference of fluorescence intensity changing over time for the spectrometer and in order to ensure the accuracy of results, the spectrometer was warmed up for 20 min prior to starting the measurements. In addition, the final measurement results were obtained by the averaging of three measurements.
Due to the fact that the juices can be oxidized easily, sample preparation in this study needed to be completed in the shortest time frame. All samples are sealed and stored at
20 °C. Matlab Rb (MathWorks) was used for qualitative and quantitative analysis.
Artificial Samples
The commonly used food additives sodium benzoate (SB) standard substance, potassium sorbate (PS) standard substance, carmine (CA) standard substance, and amaranth (AM)standard substance were selected and blended in varying proportions to 25 sample groups, according to Table I. Figure 1 shows excitation-emission fluorescence matrix (EEM) contour plots for samples 1–5, 10, 15, and 20. For Figures 1e through 1g, it is difficult to identify each component visually. Therefore, chemometric methods are required to resolve this problem.
Real Samples
In order to validate the accuracy of the model based on artificial samples, real samples were selected for follow-up evaluation. Five kinds of grape juices labeled and marketed as “additive-free” were selected as real samples, shown in Figures 4r2 through 4r6, compared with fresh grape juice shown in Figure 4r1.
All samples were prepared using high-precision pipettes. Although errors are inevitable during sample preparation, variation among samples is negligible.
Algorithm
After the detection of each sample, the corresponding TDFS is obtained, and the linear separation model of each sample is determined as follows: Ym = ɑm,1S1 + ∙ ∙ ∙ + ɑm,N SN + Em(m = 1, ∙ ∙ ∙ M) [1]
where Si (i = 1, •••, N) ЄRIxJ are primary spectral signals, am,i (i = 1, •••, N) are the component concentration score, and
Em is the noise.
ICA
For the sake of confirming whether the fruit juices contained food additives, and which additives they contained, independent components analysis (ICA) was combined with TDFS. ICA was applied to identify the additives in the juices. As a blind source separation method, ICA can be used to extract the pure underlying signals from a mixed signals data set (18).
The general model of ICA is (19):
[2]
where X is the matrix of observed spectra, S is the matrix of spectra of single components, and A is the mixed matrix of coefficients which is related to the corresponding concentrations.
Because the main aim of the ICA algorithm is to maximize the non-Gaussianity of the estimated sources, the choice of the number of independence components (ICs) is a key in the ICA process (18).
The Choice of the Number of ICs
It is crucial to choose the optimal number of ICs. Usually, there are three methods that are used, as shown in previous studies. The first method is based on the percentage of variance calculated by principal component analysis (PCA). However, it is hard to determine if a small variance contains useful information (8). The second method is a PCA loadings plot, which was used by Valderrama and associates to choose the chemical rank in molecular fluorescence spectroscopy (20). Based on the Durbin-Watson (DW) criterion, a third method was applied to compute the lower signal-to-noise ratio (S/N) signals (18).
In theory, the number of ICs should be the number of independent variable sources. The ideal case is that the number of ICs is equal to the number of the components. However, it is difficult to determine the number of components in unknown, complex, overlapping signals. Therefore, in this study, the number of ICs was set at 1, and increased by 1 successively, until the evaluation parameters arrived at the optimal value.
Identification
For the purpose of improving the calculated accuracy, the Savitsky-Golay algorithm was used to remove Raman scattering and Rayleigh scattering from the signals. Then, each row or column of the obtained EEMs of these samples was expanded, along the excitation wavelength or emission wavelength, to obtain the extended matrices. The EEMs for EM and EX obtained are shown as:
[3]
[4]
In this work, the results of qualitative analysis were evaluated by two parameters, which are root mean square error of prediction (RMSEP) and similarity coefficient (p), according to the following equations:
[5]
[6]
where s and x are row vectors representing the standard differential spectrum and calculated differential spectrum. Obviously, using equation 6, |p| ≤ 1 can be obtained. The larger the value of p, the more similar the standard differential spectrum is to the calculated differential spectrum. Therefore, the composition of the multi-fluorescence system can be identified according to the p value. In addition, the R-squared estimate (R2) can also be used to indicate the information recovery rate (21) and is given in equation 7:
[7]
Results and Discussion
In order to simplify the interpretation of the data, the 25 different elementary cubes of 25 artificial samples, shown as Table I, were transformed into a three-dimensional matrix. In the end, the matrix obtained was unfolded into a new matrix as the input of the ICA model.
Independent Component Analysis for Artificial Samples
Choice of Number of ICs
The key point of applying the ICA method is confirming a suitable mathematical rank (the number of ICs, k). It has also been considered in previous studies (18,20). For the purpose of finding a suitable mathematical rank (the number of ICs, k) from the spectral data in this study, the values of evaluation parameters (p, R2, and RMSEP) as they changed with the number of ICs (from k = 1 to k = 7) are shown in Table II.
The 25 artificial samples consist of four additives mixed in different proportions. At the beginning, the value of k was taken as 1, but none of the three estimated parameters, p, RMSEP, or R2, met the requirements. As the value of k increased, the performance of the ICA model gradually improved, and the values of the three parameters gradually approached satisfactory results. It is obvious that the average RMSEP is greatly reduced from 31.83% to 13.65% with k = 3 to k = 4; by contrast both the remaining evaluation parameters, p and R2, do not reach satisfying values. Until the k is increased to 6, both parameters p and R2 were significantly increased to an acceptable range and RMSEP was reduced to 8.74%. As the k value continues to increase, there are only slight improvements in these three statistical parameters. The existence of experimental error and data preprocessing are probably the main reason for this phenomenon. Consequently, k = 6 can be considered the optimal value.
Identification and Quantification
Taking artificial sample 20 as shown in Figure 1g as an example, the ICA algorithm is used for qualitative and quantitative analysis. First, the four decomposition results are shown in Figures 2a1 through 2a4, with k = 4. By comparing the contour spectra of amaranth and carmine, shown as (a) and (b) in Figure 1, it can be seen that the contour spectra of the two samples are very similar, but the estimated sources shown in Figures 2a1 and 2a2 can also be seen. This indicates that acceptable qualitative analysis results can be obtained using the ICA method, even when their fluorescence spectra are highly similar and significantly overlapping (22).
Then, the value of k is changed to 6, and sample 20 is used as an input to the ICA model. The obtained six decomposition results are shown in Figures 2b1 through 2b6. It can be seen that Figures 2b1 through 2b4 are consistent with Figures 2a1 through 2a4, but Figures 2b5 and 2b6 represent Raman scattering and Rayleigh scattering, respectively. Therefore, it can be inferred that the appropriate number of factors is necessary to obtain more detailed information. In addition, the existence of Figures 2b5 and 2b6 means that the data preprocessing has not completely removed Rayleigh scattering and Raman scattering. It also indicates that more detailed information can be extracted with a suitable value of k. By comparing the spectra of the four single components in Figures 1a through 1d with Figures 2b1 through 2b4, the results of the decomposition can be well identified. From Table II, the values of RMSEP of CA and AM are less than 10% with k = 4, but the predicted values of SB and PS are 26.24% and 25.46%, respectively. It is then obvious that the latter result is better than the former. The similarity coefficient p is also relatively low, and the original signal cannot be restored well. As the k value increases to 6, the average recovery rate R2 and similarity coefficient p increase from 97.69% to 98.08% and from 0.91 to 0.96, respectively. The error RMSEP significantly decreases from 16.65% to 8.74%. However, due to the high correlation between (a) and (b) shown in Figure 1, and the existence of the experimental error and deviations, the evaluation parameters cannot reach the absolute optimal values of 100%, 1.00, and 0.0, respectively.
Figure 3 shows the error between calculated concentration values and real concentration values for four additives for artificial samples 1–25. It can be clearly seen that in Figures 3a and 3b, the calculated values and the real values of AM and CA can be better fitted, and the errors are relatively small compared to Figures 3c and 3d represented by PS and SB. As can also be seen in Table II, average values of three performance evaluation parameters of the SB and PS are all slightly worse than CA and AM. This indicates that highly similar and overlapping sources will have a significant effect on the results of estimated sources.
Independent Component Analysis for Real Samples
Choice of Number of ICs
Unlike artificial samples, the choice of k has to be estimated due to the unknown number of components. Consequently, k increases from 1 until satisfying results can be obtained.
The fresh grape juice sample is compared with purchased additive-free grape juice shown in Figure 4. It can be observed that the spectra of 4r2 and 4r3 are similar to those of the fresh grape juice 4r1, and the two samples may be considered to be classified as not having any additives. However, it can also be seen that there are more than one fluorescent peak in Figures 4r4 through 4r6. Obviously, in addition to the grape juice, there are other additives in these three samples.
For the real sample (Figure 4r4), R2 significantly changes from 90.8% to 92.1%, and model error RMSEP reduces from 29.1% to 19.2%,when k changes from 3 to 4 (see Table III). However, the RMSEP values of Figures 4r5 and 4r6 are relatively higher as k increases to 5. Also, the R2 values of Figures 4r5 and 4r6 are slightly increased by 7% and 4%, respectively. As expected, the RMSEP values are reduced by 15% and 29%, respectively. The evaluation parameters of Figures 4r5 and 4r6 are satisfactory until k is equal to 6. As k keeps increasing, the result does not improve. Therefore, for the real sample (Figure 4r4), the k is chosen to be 4. Similarly, both k of Figures 4r5 and 4r6 are empirically determined to be 6.
Identification and Quantification
With k = 4, estimated spectra of the sample (Figure 4r4) are presented in Figure 5(a). Figures 5(a)c1, 5(a)c2, and 5(a)c4 were solved for similarity coefficients p, based on spectral data of commonly used additive standards. The average p is 0.92%, as shown in Table III. It can be considered that the components in Figure 4r4 were successfully identified. It can also be seen in Table III that the recovery rate R2 is 92.1%, and increases slightly as k increases. It may be also explained that the experimental error and the data preprocessing mentioned earlier have a significant impact on the results.
Therefore, for the real samples shown in Figure 4r5 and Figure 4r6, ICA decomposition is performed directly with no data preprocessing. As compared with Figures 4r4, Figures 4r5 and 4r6 have higher R2 and lower model error RMSEP. Consequently, ICA decomposition can be performed directly if required.
From Figures 5(b) and 5(c), Raman scattering, Rayleigh scattering, and the main additives of Figures 4r5 and 4r6 are decomposed successfully using k = 6. It can be observed that the decomposition for scattering information effectively removes this signal.
Conclusions
Previous work has documented the effectiveness of ICA in signal processing for analytical chemistry. For example, Alves and associates reported that the combination of UV-vis measurements and ICA makes possible the evaluation of extravirgin olive oil, and can contribute to suggesting that a foodstuff comes from an alleged origin (8). Jorge and colleagues pointed out the advantages of ICA in data analysis by comparing PARAFAC and ICA (21). However, there has been no research to apply ICA to the analysis of food additives.
In this work, we analyzed artificial samples consisting of four additives and real samples purchased from the supermarket. For the artificial samples, a satisfactory result was obtained with the correct number of independent contributions (k), but the excessively correlated original signals had a negative effect on the result. In addition, it is also found that removing the scattering has an adverse influence on the data analysis. Therefore, the ICA algorithm can analyze raw fluorescence spectral data without data preprocessing.
Consequently, ICA may be an alternate tool for resolution of overlapping chemical signals. However, further work is needed to improve the accuracy because of insufficient sample size.
Acknowledgments
This project is supported by the National Natural Science Foundation of China (Nos. ) and Natural Science Foundation of Hebei Province of China (Nos. F).
References
The company is the world’s best Additives for Leveling Compounds supplier. We are your one-stop shop for all needs. Our staff are highly-specialized and will help you find the product you need.