Guarding food safety with conventional and up-conversion near-infrared fluorescent sensors

Graphical abstract

Highlights

• •Near-infrared (NIR) fluorescent nanomaterials are highly-promising in the quest for a next-generation of food toxicity sensors.
• •Exploiting NIR photoluminescence for food security assessment is effective in detecting inorganic and organic contaminants.
• •Perspectives and challenges concerning the use of NIR photoluminescence for food safety detection are discussed.

Abstract

Background

Acknowledged by the World Health Organisation (WHO), over 200 diseases ranging from mild to fatal are linked to the consumption of food products subjected to physical, chemical, or biological contamination. Nevertheless, conventional methods commonly used for the identification of health hazards in foodstuffs have problems coping with the sensitivity requirements imposed by latest-hour regulations in the field. Additionally, their use and availability is wildly limited by aspects such as instrument dimension, prohibitive costs, detection complexity and required operational knowledge.

Aim of review

This review provides an overview of recent efforts that have focused on the assesment of food contamination based on near infrared (NIR) photoluminescent sensors. Important endeavors that have targeted the precise detection of various inorganic and organic contaminants, including hydrogen sulfide, cyanide anions, mycotoxins, antibiotic residues, etc., are presented and relevant challenges that lie en route as stumbling blocks for such sensors to reach the next level of maturity and to become more available, are systematically discussed and enunciated.

Key scientific concepts of review

Ingenious food contamination sensors that rely on conventional or up-conversion photoluminescence in the NIR region represent an emerging topic. To date, such sensors have been demonstrated as promising detection candidates, possessing important advantages such as: high efficiency, facile implementation, and convenient flexibility, thereby promising significant contributions to expand the current state of the art in food security.

Introduction

Across their journey through the supply chain, from farm to fork via factory and market, food products can be subjected to various health hazards linked to biological, chemical, or physical contamination that carry deep implications for human health. Over 200 diseases ranging from less severe to fatal ones, such as cancers [1], can be linked to the consumption of food containing pathogenic bacteria, viruses, parasites, or chemical substances. A prominent example that we find relevant to provide in the current pandemic context is the transmission via contaminated food products of the SARS-COV-2 virus [2], [3], the pathogen responsible for the unprecedented crisis that is currently affecting the entire planet, accounting for over > 5 million deaths at the time of writing this article, and immeasurable financial losses, in less than 2.5 years since the first reported cases. Also important to mention are the numerous food safety scandals (e.g. prion in beef, clenbuterol in pork and melamine in milk powder) that have continually occurred over the years, which were also linked to severe human health threats accompanied by significant social and economic burden [4], [5], [6]. Thus, the advent of more efficient and more reliable solutions that can probe and certify food safety are of utmost importance for ensuring human physical well-being, life quality and the sustainability of worldwide healthcare and financial systems [7]. As a result of WHO’s continuous emphasis on the importance of this field, the development of novel food safety detection technologies that can identify hazardous substances and pathogenic biological organisms in food, is gaining increasing interest. Such technologies can be complemented by others that can remove the health hazard from the contaminated food products, whose development is equally important.

Recent decades have witnessed the burgeoning of various analytical strategies for food quality and safety control covering sequential injection chromatography (SIC), high-performance liquid chromatography (HPLC), surface-enhanced Raman scattering (SERS) and antigen–antibody specific assay (ELISA) [8], [9], [10], [11]. However, the feasibilities of these strategies are mainly impeded by their insufficient sensitivity and reproducibility, as well as by their associated time-consuming sample handling and detection processes [12], [13]. In addition, due to their high dependence on expensive instruments and highly skilled operators, the availability of these methods in poorly developed economic regions is severely limited.

Among numerous solutions that can overcome the limitations and bottlenecks of conventional approaches for food safety assessment, near infrared (NIR) technologies have always been a central topic. Such solutions have gained lately an important focus of attention, due to the simplicity of implementing them at the end-user level and their ultra-high accuracy and reliability. Covering the wavelength range from 650 nm to 1700 nm, NIR methods for food safety assessment feature inherent optical characteristics including non-invasion, deep penetration and low background noise [14]. The most widely met NIR detection paradigms rely on NIR fluorescent sensors that can signal food contamination, with splendid stability, specificity and sensitivity, circumventing the auto fluorescence interference of test samples [15], [16], [17]. While the employment of NIR fluorescent organic molecules or nanomaterials to enable food safety detection sensors has acquired remarkable progress, this field is far from being exploited at its full potential.

NIR fluorescent organic molecule sensors, organic fluorophores (e.g. squaraines, cyanines, and bodipy derivatives) and optically tunable mechanisms have been jointly combined to date to result in advanced target-sensitive platforms. To this end, a series of schemes such as electron transfer, protection-deprotection, characteristic spirocyclization, and fluorophore integration have been devised to efficaciously endow organic fluorophores with optically available tunable groups, which provide highly significant optical advantages for sensing, including low fluorescence background, long decay length, or structural and functional variability [18], [19], [20], [21]. Meanwhile, driven by the development of nanotechnologies, nanomaterials possessing unique physiochemical properties are substantiated as a welcome component of NIR fluorescent sensors [22], [23]. With prominent temporal and spatial sampling, rapid response and enhanced signal readout, a special type of NIR fluorescent nanoparticles (NPs), coined upconversion nanoparticles (UCNPs) blaze a brand new path for food safety determination. Such nanomaterials upconvert energy, meaning that they emit light whose wavelength is lower compared to that of the excitation source; for example, UCNPs excited in the infrared can emit in the visible domain. As a result of their physico-chemical characteristics suitable for sensing (e.g. high luminescence efficiency, long luminescence lifetime, narrow emission peak, large Stokes-Shift, good chemical stability, low cytotoxicity), NIR UCNPs can overcome the performance limitations of traditional organic fluorophores, such as short fluorescent lifetime, broad characteristic spectrum and proneness to photo-bleaching [24], [25], [26]. This has led to the emergence of a considerable number of UCNP applications focused on food safety.

In this review, we summarize representative NIR fluorescent sensors in light of their potential to detect usual food contamination sources, including inorganic substances (hydrogen sulfide, bisulfite/sulfite anions, cyanide anions, nitrite anions and heavy metal cations) and organic substances (foodborne pathogens, mycotoxins, antibiotic residues, pesticide residues and biological amines). Subsequently, we discuss challenges and future aims of NIR fluorescent sensors for food safety detection. Altogether, we hope that this review not only paints an illustrative picture of the current state of the art, but can also spark fresh ideas leading to the advent of novel techniques for food safety detection based on NIR technologies, and guide the development of novel sensors with superior performance for future food safety applications.

Inorganic substances

Inorganic substances present in numerous food products are highly relevant with respect to the regulation of metabolic processes and of the energy balance. Importantly, the overdose of certain inorganic substances (e.g. toxic metals, metal oxides, salts, sulfides and halides) in foodstuffs may pose danger to the human health, thus identifying such situations is critical. Accordingly, the availability of simple and rapid inorganic substance detection methods as point-of-care devices, or as devices available in relevant health departments is highly desired. In Table 1, we present a number of practical NIR sensors that have been developed so far to identify the presence and dose of inorganic substances in edible substances.

Table 1.

Notable NIR fluorescent sensors for detecting inorganic substances.

Test object	Mechanism	Detection medium	λex/λem (nm)	Linear range	Detection limit	Applications in foodstuffs	Ref
H2S	FRET	PBS solution (pH 7.4, 1x)	720/778 810	0.0367–120 μM	11.0 nM	beer, red wine, stale egg white, riverine water	31.
H2S	ICT	DMF/H2O (pH 7.4, PBS-HCl 10 mM, v/v = 3/7)	543/640 745	12–38 μM	3.09 μM	red wine, real water	32.
HSO3-/SO32-	ICT	PBS solution (pH 7.4)	550/690	3.13–200 μM	0.46 μM	tap water, wine, sugar solution, Chinese liquor	38.
SO32-	ICT	DMSO/HEPES (pH 7.4, 10 mM, v/v = 6/4)	576/675	0–12 μM	31.6 nm	granulated sugar, vermicelli	39.
HSO3-/SO32-	/	PBS solution (pH 7.4, 10 mM)	670/705	0.01–0.15 mM	0.37 μM	crystal sugar, red wine	40.
HSO3-/SO32-	ICT	PBS solution (pH 7.4, 20 mM, with 50% CH3OH, v/v)	450/667 485	0–3 μM	27 nM	wine, soft sugar, crystal sugar, sugar	42.
HSO3-/SO32-	/	PBS solution (pH 7.4, 10 mM, with 10 %DMF, v/v)	500/717 560	0–4 μM	87 nM	crystal sugar, granulated sugar	43.
CN–	ICT	PBS solution/DMF (pH 7.4, v/v = 1:1)	490/519 688	0–80 μM	0.075 μM	sprouting potato, un-sprouted potato, almond, cherry, cherry seed, bitter cassava,	47.
NO2–	ICT	HCl solution (pH 1.0, 0.1 M)	567/656	0–1 μM	6.7 nm	Chinese sauerkraut, river water	51.
Hg2+	IFE	EtOH/H2O (pH 7.2, 10 mM HEPES, v/v = 2:1)	547/756	0.05–10 μM	13.5 nm	tap water, tea	54.

FRET = Fluorescence Resonance Energy Trasnfer; ICT = intramolecular charge transfer; IFE = Inner Filter Effect.

Hydrogen sulfide

Hydrogen sulfide (H₂S) represents a widely met pollutant compromising food safety and endangering human health [27]. For instance, abnormal levels of H₂S result in a high incidence of various severe diseases, such as Alzheimer’s, Down syndrome, liver cirrhosis, or diabetes [28], [29]. Although multiple fluorescence methods featuring easy manipulation and real-time imaging have been engaged to monitor H₂S, a large proportion of probes are distinguished with short emission wavelength which makes them significantly affected by background fluorescence interference and photo damage [30]. As a consequence, the development of NIR fluorescent technologies that can replace, expand or augment the current methods operating in the visible regime is very important for enabling the determination of H₂S contamination with higher specificity and sensitivity.

In recent years, several innovative NIR fluorescent sensors have been synthesized for H₂S detection in food samples. In this respect, a sophisticated NIR fluorescent nanosensor integrating switchable chromic Cy7Cl (a cationic cyanine dye) and NIR775 (a NIR fluorescent dye) into a phospholipid polymer to analyze H₂S in actual food samples such as red wine, beer and stale egg white (Fig. 1a-b) was reported by Xiao et al. [31]. This work showed that Cy7Cl was qualified as a H₂S-reactive chromophore and energy acceptor, and NIR775 efficiently acted as a H₂S-inert fluorophore and energy donor, conspicuously triggering fluorescence resonance energy transfer (FRET), an energy transfer process through dipole–dipole interactions between the donor–acceptor pair, and quenching the fluorescence emission of NIR775 at 778 nm. With the permeation of H₂S, blue-green Cy7Cl was gradually converted to colorless Cy7SH, which caused fluorescence quenching of Cy7Cl at 810 nm and restored fluorescence emission of NIR775. In addition to this, this simple and precise method satisfied a limit-of-detection (LOD) of 11.0 nM, sufficient to specifically detect H₂S among other anionic and sulfur-containing species. Another notable effort is the work of Zhong et al. [32], who reported a hyperchromic NIR fluorescent probe consisting of donor-π-acceptor fluorophore and 2,4-dinitrophenyl ether moiety capable to monitor H₂S in red wine, real water, and living cells. On the one hand, the donor-π-acceptor construction could red-shift the absorption and emission wavelengths to the NIR region ascribing to intramolecular charge transfer (ICT), a fundamental photochemical process relying on charge flow from a donor to an acceptor. On the other hand, the 2,4-dinitrophenyl ether moiety was found to be highly fit in serving as a H₂S recognition probe based on fluorescence quenching. Interestingly, in the presence of HS^-, the 2,4-dinitrophenyl ether moiety was induced to release the fluorescent molecule, thereby offering a significant fluorescence response (reduced intensity at 745 nm and enhanced intensity at 640 nm) and a solution color change from colorless to purple blue.

Fig. 1.

NIR fluorescent sensors for H₂S and HSO₃^-/SO₃^2- detection. (a) Scheme of the preparation process and H₂S detection of a NIR775/Cy7Cl sensor. (b) Absorption spectra of Cy7Cl with and without H₂S. Adapted with permission [32]. Copyright 2020, Elsevier. (c) Scheme of mitochondria-target NIR fluorescent probe mediated HSO₃^-/SO₃^2- detection. (d) UV–Vis spectra changes with SO₃^2-. (e) Fluorescence spectra changes with SO₃^2-. Adapted with permission [39]. Copyright 2020, Elsevier. (f) Scheme of coumarin-indolium based NIR fluorescent probe mediated HSO₃^-/SO₃^2- detection. (g) Color changes with HSO₃^-. (h) UV–Vis spectra changes with HSO₃^-. Adapted with permission [43]. Copyright 2015, Elsevier.

In brief, in these past works, the fluorescence intensity of fluorophores designed to perform in the NIR region had been successfully elevated or decreased, at significant levels, via H₂S mediated thiolysis and thiolation. The advantages of NIR fluorescence technologies in terms of Stokes-Shift and signal-to-noise ratio (SNR), enabled thus valuable practical applications for H₂S identification.

Bisulfite/sulfite anions

Serving as food preservatives, bisulfite/sulfite (HSO₃^-/SO₃^2-) anions validly suppress browning, oxidation, and microbial reactions during the period of storage [33]. Nevertheless, ingestion of these anions can lead to tissue and cell injury, thus inducing asthma and allergies in individuals [34], [35]. Among the current methods of HSO₃^-/SO₃^2- detection, several fluorescent probes tend to stand out due to splendid on-site and real-time testing properties [36], [37]. Moreover, fluorescent probes distinguished with significant NIR fluorescence response and colorimetric effect in HSO₃^-/SO₃^2- measurement procedures are capable even to optimize the background and enhance the resolution.

The number of NIR fluorescent probes possessing high sensitivity in the detection of HSO₃^-/SO₃² molecules, augmented by minimum autofluorescence interference, achieved a sustained and substantial growth. For example, an excellent NIR probe suitable for different types of analyte solutions was constructed to rely on the ICT mechanism (Fig. 1c-e) [38]. In this ingenious chemical sensor implement by Zeng et al., (E)-3-(4-(dimethylamino)phenyl)acrylaldehyde, a common fluorescent sensor with high fluorescence yield and chemical stability, was utilized as the acceptor, and positively charged 1-Benzylquinolin-1-ium as the donor and mitochondria-targeting component. As the SO₃^2- level improved, a linearly correlated reduction in the fluorescence signal at 690 nm was observed, which simultaneously faded the solution color from bluish violet to colorless. The applicability of this method was demonstrated in various sample types such as sugar, wine, HepG2 cells and even zebrafish, demonstrating its versatility. Another effort important to mention is the work of Duan et al. [39], where a novel NIR probe based on dicyanomethylene-benzopyran and quinolinium was reported to exhibit an extremely fast response (<50 s) in the detection of SO₃^2- molecules, with a LOD of 31.6 nM. To achieve this, dicyanomethylene-benzopyran was chosen as a photostable fluorophore precursor, and quinolinium, possessing an electron-withdrawing ability, was used as a structure block to quench the fluorescence of the obtained probe. Once quinolinium was interrupted by SO₃^2- anions through 1, 4-Michael addition reaction (an organic conjugate addition reaction), a notable chromogenic reaction (from yellow to purple) and an easily measurable NIR fluorescence response (from off to on) could be observed. Importantly, this method was successfully employed to visually detect SO₃^2- anions in very common food samples such as granulated sugar and vermicelli, upon some mild conditions (pH 7.4, 37 °C). In another study that we find important, stabilized hemicyanine skeletons on the basis of IR-780 were designed as a fluorescent probe for HSO₃^- determination in crystal sugar and red wine samples [40]. As HSO₃^- anions increased gradually, HSO₃^- mediated nucleophilic addition reaction towards the carbon atom in the probe was shown to trigger conspicuous fluorescence intensity attenuation at 705 nm with the fading of the blue reaction solution. In addition, this probe, featuring prominent biocompatibility and cell-membrane permeability, was applied for HSO₃^- monitoring in HeLa cells and BALB/c mice, thereby exhibiting great potential for biological research such as in vitro and in vivo imaging.

Different approaches rely on ratiometric NIR fluorescent sensors, which not only exhibit notable SNR advantages, but are also enriched with the attributes of ratiometric fluorescent sensors which measure emission intensities at two disparate wavelengths to offer a self-calibration correction [41]. Such approaches had represented a welcome addition to the family of NIR sensors developed for HSO₃^-/SO₃^2- determination. In an interesting study, a super-assembled probe conjugating electron-donating 7-diethylamino coumarin fluorophore and an electron-withdrawing indolium derivative was found to test HSO₃^-/SO₃^2- anions with evident dual ratiometric and colorimetric signal alterations (Fig. 1f-h) [42]. More specifically, the nucleophilic attack of HSO₃^-/SO₃^2- anions could disrupt the conjugated bond of coumarin-indolium structures and discourage the ICT process, extinguishing this way the NIR fluorescence of the coumarin-indolium hybrid at 667 nm but recovering that of the coumarin moiety at 485 nm. On account of sufficiently low LOD (27 nM), the applications of this probe covered HSO₃^- identification in food and serum samples, as well as bioimaging of HSO₃^- anions in living cells. Furthermore, a straightforward detection dipstick system relying on this probe was also successfully prepared to enable rapid-response and high-sensitivity HSO₃^- analysis. In a later effort, the same research group devised an aza-coumarin-indolium conjugated probe with longer absorption and emission wavelengths via condensation reaction [43]. Upon addition of HSO₃^- anions, the optimized probe was qualified to provide significant ratiometric and colorimetric fluorescence effects at 560 nm and 717 nm, respectively, indicating a typical ratiometric fluorescence pattern. On the strength of the superior response (down to 30 s) and LOD (estimated to be 87 nM), this probe effectively enabled HSO₃^- detection in various food samples and exogenous/endogenous bioimaging in living cells.

In summary, the efforts discussed in this section have not only led to the development of NIR fluorescent sensors based on the nucleophilicity of HSO₃^-/SO₃^2- anions for HSO₃^-/SO₃^2- measurement in foodstuffs, but also introduced a new perspective to design sensors that can be used to address other anion species in cells and even animal models. Hence, we find these investigations to represent an important progress of NIR fluorescent techniques in food chemistry.

Cyanide anions

Cyanide (CN^–) anions, which are generally acknowledged as deadly anions, are known to induce cellular respiratory paralysis once they become tightly attached to the cytochrome oxidase, even at a low dosage [44]. Despite the fact that they represent a serious health hazard, extensive applications of CN^– anions in various industrial processes still exist, generating contaminated industrial wastewater, which can pollute foodstuffs [45]. Furthermore, cyanogenic glycosides diffusely distributed in food crop species (sorghum, bamboo shoot, almonds, and cassava) can also release CN^– anions in the case of plant cell rupture caused by improper processing [46]. To address the issue of food products contamination, an ingenious ratiometric NIR fluorescent probe was established to sense CN^– anions in numerous food samples (e.g. bamboo shoots, almonds and sprouting potatoes) without elaborate instrumentation (Fig. 2a-e) [47]. In this work the authors exploited NIR fluorescence in a method that differed in terms of working principles from ICT based ones; it relied on increasing the conjugation length of the probe to reduce the energy gap and realize red-shifted emission centered at 688 nm. This CN^– detection probe was based on the unique nucleophilic function of the CN^– anions to electron deficient double bonds such as carbon–carbon double bond, carbon–oxygen double bond, and carbon–nitrogen double bond. Upon treatment with CN^– anions, a noticeable ratiometric fluorescence response appeared with the fluorescent intensities respectively at 519 and 688 nm. The ratios of the fluorescent intensities at these two wavelengths were found to be closely correlated to the concentration of CN^– anions, endowing this distinguished probe with the ability to accurately measure CN^– levels.

Fig. 2.

NIR fluorescent sensors for CN^–, NO₂^– and Hg²⁺ detection. (a) Design of a NIR fluorescent probe for CN^– detection. (b) The sensing reaction of a NIR probe for CN^– detection (c) The fluorescence colors of the probe paper strips with CN^–. (d) The fluorescence color of the probe paper strip in a sprouting potato extract sample. (e) The fluorescence color of the probe paper strip in an un-sprouted potato extract sample. Adapted with permission [48]. Copyright 2020, Elsevier. (f) Fluorescence spectra changes with NO₂^–. (g) The “covalent assembly” mechanism with NO₂^–. Adapted with permission (Yu et al., 2020). Copyright 2020, Elsevier. (h) Scheme of RhDCP-UCNPs based NIR fluorescent probe mediated Hg²⁺ detection. Adapted with permission [55]. Copyright 2019, Elsevier.

Despite promising efforts such as those discussed in this section, the number of reported fluorescence based sensors developed for CN^– detection and analysis in food samples is low, and such sensors operating in the NIR domain are even scarcer. However, given the commendable results obtained in past related works, and their intrinsic advantages, we argue that NIR based methods aimed at detecting CN^– anions qualitatively and quantitatively are in urgent need to enable food safety applications addressing this critical problem. The introduction of NIR ratiometric approaches, could, to some extent, overcome or alleviate artifacts deriving from photo bleaching and probe distribution, which would significantly enhance the accuracy of CN^– contamination assessment.

Nitrite anions

As a vital intermediate of the nitrogen cycle, nitrite (NO₂^–) anions are actively under development in food products as additive and preservative agents [48]. To some degree, NO₂^– is a form of biological nitric oxide (NO, a toxic volatile compound) storage rather than a metastable final product [49]. Ascribing to this, excessive intake of NO₂^– anions induces adverse effects on the population, especially in children and pregnant women, being responsible for the production of carcinogenic N-nitrosamines, which can be linked to blue baby syndrome and methemoglobinemia [50]. Therefore, extending the effectiveness of NO₂^– quantification methodologies, and developing novel methods that are more sensitive compared to current ones, and which can be more conveniently implemented, is of extreme value. In this quest, Yu et al. [51] introduced a coumarin framework based NIR fluorescent sensor possessing emission capabilities at 656 nm that was established via the covalent assembly principle (Fig. 2f-g). Owing to an intermolecular azo-coupling mechanism, the prepared sensor could react with NO₂^– anions through diazotization in a strong acidic environment and afterwards with the newly formed aryldiazonium through an electrophilic aromatic substitution reaction to obtain the azo-dye. More interestingly, in this identification process of NO₂^–, orange-colored azo-dye was further protonated into purple-colored form possessing NIR fluorescence. Additionally, the sensor in its initial acidulated state was completely dark, offering zero background, which significantly augmented the results of the performed NO₂^– analysis in river water, Chinese sauerkraut and Escherichia coli samples. The results were found to be consistent with measurements performed with the conventional NO₂^– identification method, the Griess assay.

Regarding the final step in the Griess assay, azo-coupling with activated aromatic rings is undoubtedly a reliable fluorescent sensor strategy for NO₂^– determination. The combination of this strategy and covalent-assembly principle has significant potential to accomplish zero background, effective spectral red-shift towards NIR region and optimal detection sensitivity, holding great promises in food security and basic scientific research.

Heavy metal cations

The overwhelming exploitation and usage of metals or ores have deep implications for environment pollution and food contamination. This topic has hence attracted the attention and criticisms of the society. Posing a huge threat to human health, heavy metal cations should only be present in trace amounts in foodstuffs, however levels that exceed the acceptable ones are frequently observed. Such hazard levels can be linked to severe health issues such as neural disorders and cognitive deficit which have been found to occur after long term ingestion [52], [53]. Although many traditional methods for heavy metal cations detection exist, such as atomic absorption/emission spectroscopy (AAS/AES) and inductively coupled plasma mass spectroscopy (ICP-MS), these solutions are highly dependent on either complicated pre-preparation or the availability of expensive and bulky instrumentation. NIR-based biosensors conceived for testing heavy metal cations in a highly rapid and low-cost fashion could play a huge role in the field. A prominent example of the significant value of NIR materials/technologies to address the heavy metal cations detection problem is the work of Annavaram et al. [54]. In this effort, spirolactam appended rhodamine-B (a fluorescent dye) based organic complex-RhDCP and UCNPs (type NaYF₄: Yb, Er) were combined to constitute a sensing probe for Hg²⁺ analysis in tap water and black tea (Fig. 2h). Once the complexation between Hg²⁺ cations and RhDCP occurred, an internal filter effect (IFE), a radiative energy transfer phenomenon involving the absorption of excitation or emission from fluorophores, would be elicited due to the evident spectral overlap between the absorption and emission bands of UCNPs and RhDCP-Hg²⁺. By means of an IFE effect, the emission quenching of UCNPs at 547 and 756 nm was triggered. The efficiency of this process was proportional to the concentration of Hg²⁺ cations.

Although a wide variety of fluorescent sensors have been reported and applied to date for Hg²⁺, Fe³⁺, Al³⁺, Zn²⁺, Cu²⁺, and Pb²⁺ detection [55], [56], to ensure food safety. The number of NIR sensors for metal cation detections is very limited. Given the success of the methodology based on the IFE effect reported by Annavaram et al. [54], we hope that not far from now novel NIR based sensors will be developed to address this pressuring problem.

Organic substances

Organic substances are, in general, adverse for physical health when potentially ingested by humans. Unfortunately, the contamination of foodstuffs with organic compounds is frequent [57], [58], [59]. For example, benzoyl peroxide exists in flour, formaldehyde in seafood, and organic benzenethiol in drinking water, etc [60], [61]. In this context, a wide palette of fast and effective organic matter detection methodologies based on NIR fluorescence have been developed to date [62], [63]. Furthermore, the introduction of antibodies and aptamers (short single-stranded DNA, RNA or peptide molecules selected from combinatorial libraries) as solutions to this problem enabled a new class of measurement techniques exhibiting highly specific targeting performance. Compared with traditional antibodies, aptamers are easily chemically synthesized, modified and stored for a large variety of biomolecules, chemical entities or cell targets [64]. In Table 2, we provide an overview of prominent NIR fluorescent sensors that have been developed for the detections of organics, categorized according to the targeted hazardous substances.

Table 2.

Notable NIR fluorescent sensors for detecting organic substances.

Test object	Detection medium	λex/λem (nm)	Linear range	Detection limit	Applications in foodstuffs	Ref
S. aureus, S. typhimurium	PBS solution	325/527 (SA), 980/806 (ST)	50-106cfu/mL	16 cfu/mL (SA), 28 cfu/mL (ST)	/	67.
Ochratoxin	borate buffer (pH 8.5)	980/660	0.1–1000 ng/mL (standard solution)	0.098 ng/mL (standard solution), 0.449 ng/mL (wine), 0.108 ng/mL (grape juice), 0.208 ng/mL (beer)	red wine, beer, grape juice,	73.
Patulin	PBS solution (10 mM)	980/543	0.01–100 ng/mL	3 pg/mL	apple juice	74.
Aflatoxin B1	PBS solution (pH 7.4, 10 mM)	980/550	3.13–125 ng/mL	0.17 ng/mL	peanut	75.
Aflatoxin B1, Ochratoxin A	PBS solution	980/452 (AFB1), 980/660 (OTA)	0.01–10 ng/mL	0.01 ng/mL	maize	77.
Zearalenone	assay buffer (with BGG and BSA based blocking buffer)	980/800	50–500 pg/mL	20 pg/mL	maize	78.
Sulfaquinoxaline	PBST solution	370/552 (QDs), 980/542 (UCNPs)	0–20 ng/mL	1 ng/mL (standard solution), 8 μg/kg (sample)	chicken, shrimp	83.
Enrofloxacin	nuclease-free water	980/543	0.976–62.5 ng/mL	0.47 ng/mL (standard solution), 1.59 ng/mL (sample)	milk powder	84.
Sulfaquinoxaline	PBS solution	980/474	0.1–100 μg/L (standard solution), 0.5–500 μg/kg (sample)	0.1 μg/L (standard solution), 0.5 μg/kg (sample)	shrimp, milk, sea bass, beef, pork, chicken	85.
Enrofloxacin	PBS solution	980/544	1–10 ng/mL	0.06 ng/mL	fish	86.
β-lactams, Tetracyclines, Quinolones, Sulfonamides,	PBST solution (pH 7.4, 0.02 M, 0.05% Tween 20)	774/789	0.26–3.56 ng/mL (β), 0.04–0.98 ng/mL (T), 0.08–2.0 ng/mL (Q), 0.1–3.98 ng/mL (S)	8 ng/mL (β), 2 ng/mL (T), 4 ng/mL (Q), 8 ng/mL (S)	milk	87.
Diazinon	cyclohexane	980/800	0.1–50 ng/mL	0.05 ng/mL	tap water, apple, lake water, pear, green tea powder	94.
Diazinon	water	980/547	0.05–500 ng/mL	0.023 ng/mL	tea, apple	95.
Atrazine	PBS solution	980/552	0.005–10 μg/mL	0.002 μg/mL (PBS), 2ng/mL (river water/ fresh cane juice), 20 ng/kg (corn/rice)	river water, fresh cane juice, corn, rice	96.
Tyramine, Histamine	PBS solution (pH 7.4, 0.01 M)	980/483 (tyramine), 980/550 (histamine)	0.5–100 μg/mL (tyramine), 0.1–100 μg/mL (histamine)	0.1 μg/mL (tyramine), 0.01 μg/mL (histamine)	pork, bacon, trachitus ovatus, carassius auratus, turbot, cheese, soy sauce, rice vinegar, rice wine	100.

Foodborne pathogens

Although largely available and well-established food packaging techniques (e.g. pasteurization and vacuum packaging) offer foodstuffs relatively long preservation time, the contamination of the so packaged food products with foodborne pathogens still occurs, accounting for problematic health issues such as acute emesis and acute abdominalgia [65]. Over the past decade, food infections mediated by various foodborne pathogens such as Salmonella, Listeria, and Escherichia coli have been occurring all over the world [66]. On account of the warning of these events, preventive measures need to be taken, especially in the form of improving the efficiency of the pathogen detection strategies in food samples, and developing new monitoring strategies based on latest generation advanced materials. As one of the advanced identification strategies, NIR fluorescent technologies with almost perfect specificity and sensitivity have been proposed to detect pathogens. For example, a proeminent aptasensor method uniting the anti-Stokes type emission of UCNPs (type NaYF₄: Yb, Tm) with the Stokes type emission of CdTe quantum dots (QDs) was conducted to simultaneously monitor multiple food pathogens (Fig. 3a-d) [67]. Owing to the upconversion property of UCNPs excitated by NIR irradiation (980 nm) and down-conversion property of QDs by ultraviolet (UV) irradiation (325 nm) respectively, the detection system effectively avoided typical issues in conventional methods such as spectral overlap and signal crosstalk. Additionally, the DNA aptamers included in this sensor provided admirable sensitivity and specificity for the enabled bioassay, yielding LOD values of 16 and 28 cfu mL⁻¹ for Staphylococcus aureus and Salmonella typhimurium sensing, respectively.

Fig. 3.

NIR fluorescent sensors for foodborne pathogen detection. (a) Design of a dual-excitation NIR fluorescent probe for foodborne pathogen detection. (b) Fluorescence spectra changes with Staphylococcus aureus (c) Fluorescence spectra changes with Salmonella typhimurium. (d) Specificity evaluation of a dual-excitation NIR fluorescent probe. Adapted with permission [67]. Copyright 2016, Elsevier.

In brief, NIR-activated foodborne pathogen recognition biosensors proffer an alternative scheme to address the deficiencies of the traditional models such as moderate sensitivity and slow response. Moreover, different combinations of UCNPs and other luminescent units possessing non-overlapping fluorescence emission have broad research prospects and great research value for simultaneous detections of several analytes.

Mycotoxins

At present, mycotoxins are widely spread throughout the entire world, representing a significant concern given their strong toxicity even in trace amounts [68], [69], which has been the main cause of dramatic outbreaks of various foodborne diseases. A variety of mycotoxins (e.g. ochratoxin, zearalenone, aflatoxin, patulin, and deoxynivalenol) coexisting in foodstuffs trigger a potential threat to biological health including skin irritation, immunosuppression, neurotoxicity, and even death [70], [71]. Due to the expensive and bulky required equipment along with low screening efficacy, traditional mycotoxin detection methods are inadequate to meet the current requirements imposed by this critical problem at the present stage [72]. NIR fluorescent detection technologies represent a great promise to overcome the limitations of traditional methodologies, by enabling novel mycotoxin measurement tools and instruments that are fast, effective and simple to operate. In the following paragraphs, we discuss several experiments conducted to date that nicely reflect this situation.

On the basis of a nonradiative energy transfer process that occurs between lanthanide cation donor fluorophores and acceptors, NIR excited luminescence resonance energy transfer (LRET) techniques have been developed for intelligent mycotoxin determination. Besides additional quenchers, LRET techniques usually need aptamers to realize specific recognition with the corresponding complementary targets in various mycotoxin detection scenarios. In this respect, a single-step LRET aptasensor comprising Mn²⁺-doped UCNPs (type NaYF₄: Yb, Er) and black hole quencher 3 (BHQ3, a quencher dye) was reported to determine ochratoxin A without complex procedures (Fig. 4a-b) [73]. Interestingly, Mn²⁺ doping hindered the transition of Er³⁺ ions to generate an appropriate red emission which overlapped with the absorption of BHQ3 for LRET activation. As the samples containing ochratoxin A were added, the formation of an aptamer-target complex was observed to diminish the linker length between the UCNPs and BHQ3, significantly improving the LRET effect. To some extent, this aptasensor was endowed with increased quenching efficiency in synchronism with the quantity of ochratoxin A, thereby showing satisfactory capacity for quantitative detection in colored food samples (e.g. wine, beer, and grape juice). Moreover, an intriguing LRET assay based on UCNPs (type NaYF₄: Yb, Er) donors and gold NPs acceptors was designed for exonuclease-catalyzed target recycling patulin measurement [74]. In this research, complementary single-stranded DNA strands, availably connected UCNPs and gold NPs were combined to exploit the controlled quenching of the upconversion luminescence. With the presence of patulin, the quenching effect could be effectively attenuated as patulin was bound to the corresponding aptamer for the construction of the stem-loop structure and for the liberation of UCNPs. Most importantly, additional exonuclease owning specificity to single-stranded DNA was exploited to digest the patulin aptamer selectively, thus releasing patulin for subsequent recycle. By virtue of the remarkable linear range (0.01–100.00 ng mL⁻¹) of the method, accompanied by efficient recoveries (93.33–105.21%), this ingenious LRET assay was successfully employed to sense patulin in apple juice samples. In another notable effort exploiting LRET, Wang et al. developed a method based on aptamer-modified UCNPs (type NaYF₄: Yb, Er) and gold NPs to detect aflatoxin B1 in peanut samples [75]. In the presence of aflatoxin B1, gold NPs attached to the UCNPs were gradually replaced by the aflatoxin B1, thus leading to dose-interrelated fluorescence recovery. By counting the number of luminescent particles on the glass slide surface, this single-particle measurement method successfully led to the readout of the concentration of aflatoxin B1.

Fig. 4.

NIR fluorescent sensors for mycotoxin detection. (a) Scheme of single-step LRET aptasensor mediated ochratoxin A detection. (b) Fluorescence spectra changes with ochratoxin A. Adapted with permission [74]. Copyright 2017, American Chemical Society. (c) Scheme of the preparation process and aflatoxin B1/ochratoxin A detection for a NIR-activated immunosorbent sensor. Adapted with permission [78]. Copyright 2011, Elsevier. (d) Design of simultaneous mycotoxin detection based on a small portable device. Adapted with permission [80]. Copyright 2018, The Royal Society of Chemistry.

In competitive immunoassays, an analyte and a labeled antigen compete for a limited number of antibody binding sites. This allows a precise quantification of the antibody amount that becomes bound with the attached analog antigens [76]. As a handy and sensitive strategy, competitive upconversion-linked immunoassay have made a profound impact on mycotoxin detection. For instance, a NIR-activated immunosorbent method was exploited to simultaneously analyze aflatoxin B1 and ochratoxin A in contaminated maize samples (Fig. 4c) [77]. This ingenious method adopted artificial antigen-modified magnetic NPs as immunosensing probes and antibody functionalized UCNPs (type NaYF₄: Yb, Tm; NaYF₄: Yb, Er) as multicolor signal probes. The antibody-antigen affinity was exploited as the link between the two types of NPs. Additional magnetic NPs were found adequate to optimize the overall assay efficacy by separating and purifying the immunecomplexes. Upon laser irradiation at 980 nm, a robust fluorescence signal was generated by the UCNPs, which was very useful for the efficient sensing of small toxin molecules. This approach based on energy upconversion helped overcome the problem of target toxins autofluorescence. Farther, amount of immunocomplexes was reduced as mycotoxin concentrations increased, resulting in lower fluorescent signal recorded from the emitting UCNPs. In a different study, a zearalenone mimetic peptide was chemically synthesized and extended via a biotin linker, as an alternative to a conventional protein conjugate, to enable an efficient immunoassay [78]. For mycotoxin measurement in spiked and naturally contaminated maize samples, the obtained peptide was applied with streptavidin (a biotin-binding protein)-conjugated UCNPs (type NaYF₄: Yb, Tm). The mimetic peptide contended with zearalenone in the samples for a limited number of antibody binding sites, while UCNPs acted as a background-free optical label and were bound towards the biotin molecules that were conjugated to the mimetic peptide. More sensitive than commercial assays, this valuable method met the rigorous demands of European legislation, enabling a LOD of 20 pg mL⁻¹, and high specificity for zearalenone and its metabolites (α- and β-zearalenol). In a different work, Yang et al. [79] introduced a small portable device as an innovative quantitative analysis platform for real-time and off-site mycotoxin determination (Fig. 4d). This method relied on multicolor UCNPs (type NaYF₄: Yb, Tm@NaYF₄, Er@NaYF₄: Yb, Tm; NaYF₄: Yb, Tm; NaYF₄: Yb, Er) barcodes and fluorescence image processing algorithms. More concretely, encoded signals from microspheres doped with red/blue/green-emitting UCNPs were identified, which effectually mediated indirect competitive immunoassays that were demonstrated to be efficient for the simultaneous detections of diverse mycotoxins including aflatoxin B1, ochratoxin A, and zearalenone. A proprietary algorithm was used to process the images captured by the portable device, which offered a reliable result about the type and concentration of mycotoxins within 1 min. Such handheld devices are likely to play a huge impact in coming years in ensuring food safety in geographic regions with less developed economies, where the availability of complex specialized equipment and trained operators is limited.

Given that mycotoxin pollution yields huge economic losses to food enterprises, livestock and poultry farms, and food processing industries, finding the best detection strategies is of ultra-high importance. Ingenious methods that rely on NIR fluorescent sensors for mycotoxin detection, such as those discussed above represent a great promise in the quest of enhancing the current state of the art. They were shown to provide very high accuracies and sensitivities, features that directly intertwine with the NIR emission/excitation properties responsible for reduced background. Additionally, NIR based multidisciplinary detection assays can be designed in a myriad of ways, fostering and stimulating further developments in connected technologies such as aptamer probes, target recycling, digital enumeration, peptide design, detection equipment, etc.

Antibiotic residues

In animal husbandry, the widespread use of antibiotics including tetracyclines, quinolones, sulfonamides and β-lactams has become a standard as a result of their roles in growth promotion, disease prevention and cure [80]. However, such practices are not always implemented carefully. Using uncontrolled antibiotic dosing levels can lead to significant issues such as increasing the emergence of antibiotic-resistant bacteria or the presence of drug residues in the final product that is marketed, which can cause allergies, and various other health problems, including severe pathologies such as cancers [81], [82]. In case antibiotic residues are found above critical limits, the product is regarded as a health hazard and needs to be discarded. Hence, there is a great need for cutting-edge, low-cost, and super-sensitive approaches for testing antibiotics’ signatures in various types of foodstuffs, in different stages along their route from farm to market.

Over recent years, the LRET technology has attracted significant attention, in the context of antibiotics detection given its promising performance in connected applications. Among the relevant efforts conducted on this topic, in a work performed by Hu et al., an UCNPs (type NaYF₄: Yb, Er)-based luminescence quenching immune chromatographic strip was confirmed to be efficient in monitoring sulfaquinoxaline (a sulfonamide antibiotic) in foods of animal origin (Fig. 5a-b) [83]. Upon adding sulfaquinoxaline samples, a competition between the free sulfaquinoxaline and immobilized sulfaquinoxaline-ovalbumin took place in a race to combine with a colloidal gold labeled antibody. In this case, less colloidal gold labeled antibody was captured whereby colloidal gold acceptors were inadequate to quench the luminescence of UCNPs donors completely. Overcoming typical problems associated to traditional colloidal gold-based strips, this strip featured low cost, one-step operation and visual result judgment, and was shown to generate an optical signal positively correlated with the targets even at weak intensity. Impressively, the obtained strip revealed detection results highly consistent with those yielded by commercial kits in chicken and shrimp samples, with significantly higher performance. With approximately 13-fold lower LOD (1.59 ng mL⁻¹) than a commercial kit, another LRET based aptasensor consisting of core–shell UCNPs (type NaYF₄: Yb, Er, Gd) donors and graphene oxide (GO) acceptors was found to be a highly efficient solution for enrofloxacin (a quinolone antibiotic) analysis in milk powder samples [84]. The core–shell structure and Gd³⁺ doping provided UCNPs with enhanced fluorescence intensity and improved the efficiency of the LRET process. This sensitive and cost-effective aptasensor was found to exhibit splendid specificity, being capable to discard signatures associated to other antibiotic residues that could have resulted in a high false-positive rate.

Fig. 5.

NIR fluorescent sensors for antibiotic detection. (a) Scheme of an immune chromatographic strip exploiting UCNPs luminescence quenching for sulfaquinoxaline detection. (b) Fluorescence spectra of UCNPs and QDs alongside UV–Vis spectra of colloidal gold. Adapted with permission [84]. Copyright 2017, Elsevier. (c) Schematic representation of a recent NIR fluorescent hybrid probe developed for enrofloxacin detection. Adapted with permission [87]. Copyright 2016, Elsevier. (d) Design of an efficient multiplex lateral flow immunoassay method for antibiotic detection. Adapted with permission [88]. Copyright 2015, Elsevier.

Immunoassays represent as well very promising antibiotic detection candidates. In competitive immunoassay format, UCNPs (type NaYF₄: Yb, Tm) and magnetic polystyrene microspheres were integrated in a system addressing sulfaquinoxaline measurement [85]. Building on the advantages of the simple and fast extraction procedure, not involving organic solvents, recoveries of sulfaquinoxaline could be significantly measured in the range of 69.80–133.00% in animal-derived food samples. Particularly, this immunoassay showed equal performance in the analysis of sulfaquinoxaline in diluted milk samples with a reported LOD of 0.5 μg kg⁻¹. In another notable work, Fe₃O₄ NPs were ingeniously used to immobilize aptamers as magnetic substrates in a new-style NIR fluorescent hybrid probe, with UCNPs (type NaYF₄: Yb, Er) as signal sources (Fig. 5c) [86]. On account of the aptamer recognition process, enrofloxacin quantification was simply realized via reading signal alterations from UCNPs. Showing satisfactory values in accuracy, sensitivity, selectivity, linearity, and precision, the proposed hybrid probe was capable of detecting enrofloxacin in various fish samples, including spanish mackerel, perch, catfish, and snakehead. As regards multiplex lateral flow immunoassays, a novel NIR label was combined with a monoclonal antibody of broad-specificity for screening different antibiotic residues in milk samples (Fig. 5d) [87]. In this interesting approach, diverse antigens were located in separate test zones of a nitrocellulose membrane as capture agents, completing simultaneous analyses of four antibiotic samples including tetracyclines, quinolones, sulfonamides and β-lactams. By means of analyzing the fluorescence intensity of the NIR label, qualitative and quantitative assessment of antibiotic residues was successfully implemented, with LODs and cut-off values meeting the requirements imposed by EU legislation.

Serious health hazards arise from antibiotics being widely used as food additives in livestock and poultry industries given that significant amounts are not properly absorbed nor excreted in the faeces, causing contamination of the subsequent foodstuffs. Recent efforts such as those discussed above suggest that NIR hybrid sensors integrating multiple functions such as specific recognition, stable fluorescence, and reliable magnetic properties can intelligently transform and significantly improve the antibiotic identification process yielding enhanced contamination control.

Pesticide residues

In modern agriculture, pesticides equipped with prominent insect killing effectiveness are widely applied to protect crops from pests and so ensuring high yields [88]. However, the overuse of pesticides, which is very common [89], [90], represents an extremely serious food safety issue as it results in the accumulation of indigestible, and highly harmful, pesticides in the body for long periods of time [91]. Furthermore, inappropriately disposed pesticides might be ingested by humans via different contamination routes including atmosphere, water, and agricultural products, resulting in enormous threats to human health [92], given that even low-doses can result in the installment of various conditions and pathologies [93]. Thus, the advent of novel strategies that overcome the bottlenecks of pesticide detection (e.g. limited storage time, complicated pretreatment and expensive instruments) is highly important, and very urgent. Considering recent results, methods based on NIR fluorescence detection technologies are likely to play a very important role in the next years in the context of pesticide residues monitoring, acting as reliable guardians of food safety.

LRET and immunoassays are probably the most commonly employed NIR fluorescence detection schemes for recognizing pesticide residues. With respect to the former, we find important to mention the effort of Wang et al. [94], where an acetylcholinesterase modulated biosensor featuring an UCNPs (type NaGdF₄: Yb, Tm)-Cu²⁺ mixture was manufactured for Diazinon (an organ phosphorus pesticide) determination (Fig. 6a-b). In this platform, thiocholine (considered as an enzymatic hydrolysate of acetylthiocholine) was able to seize Cu²⁺ from UCNPs-Cu²⁺, which was accountable for the luminescence quenching of the UCNPs. Since Diazinon irreversibly impaired the enzymatic activity of acetylcholinesterase, the production of thiocholine would decrease greatly in samples containing Diazinon, thus reducing the luminescence recovery. Based on the above mechanisms, this biosensor, featuring a reliable linear detection (0.1–50 ng mL⁻¹) was demonstrated in contaminated environmental and agricultural samples. Analogously, in a different effort, UCNPs (type NaYF₄: Yb, Er) and GO were selected as foundations to construct a practical fluorescence sensor for Diazinon analysis in real food samples [95]. Once adsorbed onto the GO surface via a π-π interaction, the aptamer-modified UCNPs donors would quench their own luminescence. With the addition of Diazinon, this phenomenon was terminated, leading to a linear recovery in the fluorescence intensity of the UCNPs. In a different type of approach, a competitive upconversion-linked immunoassay technology was harnessed to result in a highly sensitive system for Atrazine monitoring, a pesticide commonly found in sugar cane juice, rice, corn, and river water samples [96]. In this immunoassay system, anti-Atrazine antibody conjugated UCNPs (type NaYF₄: Yb, Er) and antigen conjugated polystyrene magnetic microspheres were utilized as the signal and capture probes, respectively. Interestingly, the antigen was entitled to compete with Atrazine for antibody binding and immunocomplex formation. Upon excitation at 980 nm, the intensity of green fluorescence yielded by the magnetic separated complexes was measured to reflect the count of pesticide residues.

Fig. 6.

NIR fluorescent sensors for pesticide and biological amine detection. (a) Scheme of an UCNPs-Cu²⁺ NIR fluorescent sensor for Diazinon detection. (b) Plausible energy transfer mechanism of an UCNPs-Cu²⁺ NIR fluorescent sensor. Adapted with permission [95]. Copyright 2019, American Chemical Society. (c) Scheme of an analytical sensor based on multi-color UCNPs labels’ fluorescence for tyramine and histamine detection. (d) Fluorescence spectra changes upon tyramine and histamine contamination. (e) Specificity analysis of the proposed fluorescent analytical sensor. Adapted with permission [102]. Copyright 2020, Elsevier.

Current forecasts on agricultural economy suggest that the use of pesticides will continue to increase, globally. This will result in higher needs for detection tools that are not only better compared to the current solutions but also more affordable and more easy to use. Detection methods based on NIR fluorescence are expected to be of great help in this quest, as efforts such as those discussed above demonstrated that they could enable specific target analyses in cooperation with various technologies, e.g. enzyme inhibition technology, aptamer technology and immune technology.

Biogenic amines

Known as bioactive nitrogen-containing compounds with low molecular weight, biogenic amines (BAs) including histamine, tyramine, cadaverine, spermine, spermidie, phenylethylamine and putrescine are transformed from protein in foods through microbial amino acid decarboxylase [97]. Adequate ingestion of BAs from various foods (e.g. fruits, vegetables, dairies, seafoods, fermented foods, meat products and beverages) can expedite physiological metabolism, enforce immunity, and improve body constitution. However, improper intake of BAs may bring about diverse reactions, such as abdominal cramps, tachycardia, vomiting, and migraine [98] and can even result in various diseases [99]. Thus, efficient detection approaches are urgently needed for the screening of BAs in daily protein-rich foods.

Recent efforts suggest the potential of NIR based technologies to significantly contribute to BA detection. For example, Wu et al. [100] reported a method for histamine sensing based on a material that was synthesized by coating a layer of molecularly imprinted polymers doped with silver nanoparticles on the surface of UCNPs. In the presence of histamine, the fluorescence intensity of the proposed material was quenched gradually while, conversely, the intensity of Surface Enhanced Raman Scattering (SERS) signals, yielded by the same material, increased gradually. This dual effect led to achieving a histamine LOD of 0.009 mg L⁻¹ for the fluorescence mode, and 0.04 mg L⁻¹ for the SERS mode, showing the huge potential held by dual responsive materials for BA detection. In a more recent effort, an analytical strategy based on the NIR fluorescence of multi-color UCNPs (type NaYF₄: Yb, Er, NaYF₄: Yb, Tm) labels was developed to detect tyramine and histamine concurrently in meat, fermented products and aquatic origin foodstuffs (Fig. 6c-e) [101]. Upon laser excitation at 980 nm, two types of UCNPs displayed different single emission peaks respectively at 483 nm and 550 nm, laying the foundation for multiple BAs identification scenarios. In addition, magnetic microspheres were linked to the tyramine or histamine coating antigens as capture probes for the competition with analytes. By reason of the simple competitive immune process that was ingeniously exploited, this strategy reduced the total test time, saving approximately 10–20 min of the color development time, and promoted the efficient detection of the targeted BAs.

At present, the detection of BAs in food samples generally refers only to the analysis of histamine. However, many foodstuffs contain other BAs, some of which may possibly induce a synergistic effect to heighten the toxicity of histamine. Therefore, simultaneous detection of multi-component BAs is of great significance for ensuring food safety. Given the flexibility of detection technologies that rely on NIR fluorescence discussed in this review, we are confident that novel, highly efficient, BA detection schemes will soon be reported.

Conclusion and perspectives

In conclusion, all NIR fluorescent sensors discussed herein demonstrate the tremendous potential of such technologies to ensure food safety. Works reported in recent years show that various materials capable of NIR fluorescence can be functionalized for many applications targeting different contaminants and different foodstuffs. Nevertheless, it should be noted that so far most efforts addressing the development of NIR fluorescent sensors had been focused on biological imaging, whereas food safety methods based on NIR fluorescent sensors are still in their infancy. Even though such methods have many unique requirements, some of them not easy to address, the connected body of work performed to date suggests that NIR fluorescent sensors will assume in the coming years a significant role in food safety, once several barriers to further progress are addressed. Some of the most representative issues are outlined below:

First, a wide palette of sophisticated NIR fluorescent sensors have been reported [102], [103], and although they present important advantages in terms of SNR, the entire elimination of the background signals in food samples is still a challenge. Therefore, additional efforts need to be placed on the following designs: (1) time-gated luminescence performance, (2) unique fluorescence signal peaks; (3) stability in weak acid or alkali; (4) excellent sensitivity for the target. In this context we find noteworthy to mention that the introduction of functional ligands and polymers is promising to foster the development of novel NIR fluorescent sensors that will yield important breakthroughs in terms of detection sensitivity and specificity [104].

Secondly, most NIR fluorescent sensors introduced to date were reported to enable single-species detection, whereas different types of food contaminant sources may co-exist in foodstuffs such as foodborne pathogens and mycotoxins. Obviously, single-species detection is inadequate to match all contaminants in foodstuffs. Multispecies detection would allow to completely rule out the possibility of food spoilage, thus such methods are desperately demanded in the food industry, to effectively safeguard the quality of foodstuffs.

Third, NIR fluorescent sensors should be more practical (e.g. more easy to create, more simple to operate, available in miniaturized, portable and affordable devices) to satisfy the customers and market requirements. For this purpose, continually in situ techniques are well-suited to help individuals obtain the real-time situation of food samples. In parallel, developing fluorescent sensors endowed with color variability upon disparate target concentrations would contribute to naked eye detection or to the implementation of accessible read-out strategies, e.g. test paper designs.

We also find important to mention that an increasing number of multi-modal sensors tactfully integrate multiple detection strategies into a single platform, exploiting in tandem their superiorities, and thus yielding higher detection accuracy, sensitivity and stability. However, such efforts are still tentative when it comes to using NIR based sensors for food safety applications. We anticipate that the development of such combined approaches could play an important role in further consolidating the role of NIR technologies in this field, given that many available tools and methods can significantly augment their use. For instance, nuclear magnetic resonance (NMR), identified as another fast, precise and non-invasive technology can cooperate with NIR fluorescent sensors for improved food safety detection. Such combined approaches would provide more comprehensive food quality information for vegetables, fruits, meat, and aquatic products. In this respect, NMR parameters are engaged to reflect microbiological growth and effectually verify the reliability of foodborne pathogen detection by NIR fluorescent sensors.

Other NIR fluorescence applications that are highly expected are those referring to the in vivo detection of ingested substances regarded as health hazards. With regard to the timely detection of contaminants, in vivo, in the human body, NIR fluorescent sensors are equipped with unrivaled advantages since UV and visible fluorophores hold limited penetration depth. On the grounds of qualitative and quantitative analysis of data linked to the outputs of NIR fluorescent sensors, physicians could detect the nature of the hazardous substances that was ingested, its amount, and consequently could arrange the corresponding treatment plan to relieve suffering and restore health of patients. Similar approaches based on in vivo probing of NIR fluorescent materials could be of help for researchers in efforts to understand nosogenesis aspects linked to various food contamination sources. In addition, the advent of such methods and applications would play a tremendous role in medicine, which could be extended to the early diagnosis of other non-foodborne diseases such as metabolic syndrome and cancer [105].

We strongly hope that this review will facilitate the future development of highly efficient and versatile NIR fluorescent sensors, by providing an overview of notable recent efforts that have been reported in the field of food safety detection.

Compliance with ethics requirements

This review does not contain any studies with human or animal subjects.

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.

Biographies

Fang Yang is currently acting as an Associate Professor at Cixi Institute of Biomedical Engineering, Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences (CIBE-NIMTE) and is involved in multiple research projects related to nanobiomaterials and their biological effect. His Bachelor in Physics (2010), Masters of Physics (2012) and PhD in Chemistry (2016) titles were awarded by the Phillips University of Marburg (PUM) in Germany. Throughout his studies Fang was also deeply involved in research: 2008-2012 Research Assistant at the Biophotonics Department of PUM; 2010-2012 Researcher at the Biophysical Chemistry Department of PUM; 2012-2016 full time Researcher and PhD student at the Biophysical Chemistry department of PUM. He has authored more than 30 scientific publications with an H-index 10.

Junlie Yao obtained his Bachelor degree in Biological Engineering from China University of Mining and Technology in June 2018, and recived his Master degree in Materials Physics and Chemistry at Cixi Institute of Biomedical Engineering, Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences (CIBE-NIMTE) in June 2021. His research project focuses on the functional nanomaterials for cancer diagnosis and treatment.

Fang Zheng obtained her Bachelor degree in Materials Chemistry from Anhui Normal University in June 2018, and recived her Master degree in Materials Engineering at University of Science and Technology of China in June 2021. Her research project focuses on the magnetic nanomaterials for cancer diagnosis and treatment.

Hao Peng (male), is acting as PhD student at Cixi Institute of Biomedical Engineering, Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences (CIBE-NIMTE). He has obtained his bachelor degree in Materials Science and Engineering from University of Chinese Academy and Science in 2018 and master degree in Materials Science and Engineering from National University of Singapore in 2020. At present, he is studying at Cixi Institute of Biomedical Engineering, Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences (CIBE-NIMTE) for his PhD degree in the areas of Optical upconversion nanoparticles and their bioimaging application.

Shaohua Jiang received his PhD from University of Bayreuth, Germany in 2014. Then he worked as Scientific Coworker during 2014-2017 at Neue Materialien Bayreuth GmbH, Germany and TransMIT GmbH, Germany, respectively. In 2017, he was rewarded “Jiangsu Distinguished Professor” and worked at the College of Materials Science and Engineering in Nanjing Forestry University. His current research interests include but not limit to the following topics: biobased materials, stimuli materials, functional wood materials, high performance fibrous materials, energy storage materials. He has published more than 120 papers with a citation more than 5000 and H-index of 45.

Chenyang Yao obtained his Bachelor degree in Material Chemistry from Central South University in June 2017, and recived his Master degree in Materials Engineering from Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences in June 2020. Now he is a PhD student at Cixi Institute of Biomedical Engineering, Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences (CIBE-NIMTE). His research project focuses on the magnetomechanical nanomaterials and their biological effect.

Hui Du obtained her Bachelor degree in Chemical Engineering and Technology from Changan University in June 2019. Now she is a postgraduate student at Cixi Institute of Biomedical Engineering, Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences (CIBE-NIMTE). Her research project focuses on the preparation of zinc-doped ferrite nanomaterials and their bio-applications.

Bo Jiang obtained his Bachelor degree in Electronic Information Science and Technology from Binzhou University in June 2013, and recived his Master degree in Communication and Information Systems from Ningbo University in June 2016. Now he is acting as an engineer at Cixi Institute of Biomedical Engineering, Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences (CIBE-NIMTE, CAS). With a background in Electronic Information and Optical Communication, he has worked for many years in interdisciplinary fields such as Electronics, Optical, Materials, and Biology. Currently, His main research interest focuses on the preparation of nanomaterials by laser ablation in liquid.

Stefan G. Stanciu received the Ph.D. degree in electronics and telecommunications from University Politehnica of Bucharest (UPB), Bucharest, Romania, in 2011. He was a Postdoctoral Researcher with UPB and ETH Zurich, Zürich, Switzerland. He is currently a Principal Investigator with Center for Microscopy-Microanalysis and Information Processing, UPB. He has coauthored >60 Web of Science journal articles, with >30% as main author, and several book chapters. His research focuses on high- and super-resolution imaging by scanning laser and scanning probe microscopies. He acted as a Management Committee Member in three EU COST Actions dealing with bioimage analysis (NEUBIAS), emerging microscopy techniques (BioBrillouin), and nano sensing (EsSENce), in the latter serving also as Short-Term Scientific Mission Coordinator. Stefan currently coordinates various research projects that focus on super-resolved imaging of cells, tissues, and advanced materials, and on the development of related image analysis and processing methods, with focus also on artificial intelligence.

Aiguo Wu received his PhD from Changchun Institute of Applied Chemistry, Chinese Academy of Sciences (CAS), China, in 2004. He started his independent career working for NIMTE after taking up his research associate appointment at Northwestern University, USA and his postdoctoral positions at Caltech, USA and University of Marburg, Germany. His research is focused on the synthesis of nanomaterials and their biomedical applications in biosensors, bioimaging, and drug delivery. He has authored over 180 scientific publications and has received some scientific awards and honors. The published papers have been cited by others more than 7000 times with an H-index of 43.