Using Second-Harmonic Generation Microscopy Images of Bee Honey Crystals to Detect Fructose Adulteration

Abstract

Second-harmonic generation microscopy is applied to mesquite honey samples with different fructose adulteration concentrations. As a proof of principle, mesquite honey is selected for this test, as it has a monofloral and spreadable-like-butter consistency, besides its economic relevance in the central region of Mexico. Second-harmonic generation microscopy is an optical method that images microstructures, such as sugar crystals in bee honey, without the interference of the liquid phase. Each recorded image is spectrally registered using the photomultiplier detector of the microscope, resulting in several gray-level histograms that are numerically analyzed using signal and image processing techniques. Several samples are prepared, adulterated, and analyzed for this purpose. The inspection requires only a microscopic amount of honey, making it a suitable technique for rare and exotic honey samples that are harvested in limited quantities. The analysis of the experimental results reveals that the second-harmonic generation microscopy signal is sensitive to liquid fructose adulteration in honey, with its signal decreasing as the amount of added fructose increases.

1. Introduction

Bee honey is a natural sweetener made of nectar collected by honeybees from different floral sources. According to the variety of nectar sources the bees visit, it is possible to have monofloral or multifloral honey (the latter with different floral sources) [1]. Honey is also valued for its taste and medicinal properties [2,3]. This sweet food contains carbohydrates, water, proteins, amino acids, enzymes, flavonoids, vitamins, and minerals. Glucose and fructose are the main constituents of honey and could represent up to 95% of the carbohydrates [4]. Fructose, glucose, and water can also reach a high percentage of honey’s composition, leaving the rest to nearly 178 substances that characterize each honey’s geographical and floral origin [5]. The fructose/glucose (F/R) ratio is a standard tool to determine if honey is fast- or slow-crystallizing [6,7]. Crystallization is natural in Apis mellifera honeybees, where sugar crystals create a solid phase that releases water [8]. The crystallization is a complex physical event influenced by several factors such as temperature, storage conditions, and impurities (crystallization seeds) coming from pollen, wax, dust, and other sources [9]; therefore, it could be related to its floral origin.

Honey can be adulterated with cheaper, commercially available syrups that are chemically similar to some honey sugars, such as liquid high-fructose corn syrup. The standard method for inspecting the floral origin of honey is melissopalynology, which microscopically analyzes and identifies honey pollen grains [10,11]. Nevertheless, pollen grains could also be added artificially into filtered honey, rather than adding highly crystallized honey. There are several methods reported in the literature devoted to identifying honey by detecting proteins [12,13], volatile organic compounds [14], and minerals [15], or using chromatography [16,17], near-infrared spectroscopy [18,19,20], hyperspectral imaging [21], and nuclear magnetic resonance [22]. All these methods give relevant information about the samples and adulteration. However, many are indirect measurements, and the few that directly observe the sample and retrieve an image are limited in depth for liquid honey, which can be altered when the honey is adulterated. The possibility of having an alternative inspection method that observes not just the pollen but the microcrystal structure of the honey could add variables and parameters for honey characterization.

Second-harmonic generation microscopy (SHGM), unlike classical microscopy, has the advantage of making the liquid phase of the honey transparent, giving a detailed view of the crystals and particles [23] without the interference of the liquid honey [24]. The latter is a direct measurement of the sample, providing a high-resolution image of the crystallization conditions of the honey. It also enables the analysis of crystals for comparison and numerical evaluation. Currently, SHGM is the only method that directly images sugar crystals for identification purposes in honey. A recently reported monofloral honey identification method uses the second-harmonic signal to directly process the gray-level histogram signal without further numerical analysis [25]. However, adulteration is a different condition in honey, and in this work, mesquite honey samples collected from beekeepers in Aguascalientes, Mexico, are analyzed by means of SHGM. This creamy honey is particularly relevant in the central region of Mexico, where the geographical conditions provide low-moisture and monofloral conditions. Several processing techniques were used to analyze the second-harmonic generation signal (SHGS) profiles (i.e., intensity counts), and each honey sample with different liquid fructose concentrations was inspected. The results show that adulterating the honey sample by adding fructose directly affects the SHGS.

2. Materials and Methods

2.1. Second-Harmonic Generation

SHGM is a non-linear optical process observed in birefringent materials or non-centrosymmetric structures. The SHGS is originated from two excited photons that interact with the material, then reconverted into newly emitted fluorescent photons with half the wavelength [26]. Technological advances in femtosecond illumination sources used to generate second-harmonic generation enable the first observations of biological tissues [27,28]. However, not all materials produce SHGS, such as collagen [29]; for this reason, it is relevant to prove that honey has an SHGS [23]. The presence of SHGS in honey is possible due to the presence of crystals, a feature verified upon inspection.

Each SHGS is recorded using a multichannel photomultiplier that gathers bands of fluorescence light emitted by the sample using spectral separation. Then, these emission bands are registered using a photomultiplier with a linear segmented array detector in a parallel configuration. This method enables the selective selection of spectral regions using a single sample scan. This work utilizes a confocal/multiphotonic microscope scanning system (Carl Zeiss, LSM 710 NLO) equipped with a Quasar multichannel photomultiplier detector on a filtering system and an ultrafast laser (Coherent, Chameleon Vision II) as the illumination source. The emission bands’ detection is possible through wedges as detection borders, while elements acting as light stop sliders block the residual emission.

A scanning routine to image the honey’s crystals is performed to determine if the SHGS behavior is similar at different excitation wavelengths in different samples. Here, excitation wavelengths between 930 and 1060 nm with steps of 10 nm are used, resulting in 14 SHGM images for a single adulteration concentration. Considering 5 adulteration percentages and 1 pure honey sample, a total of 84 samples are processed (6 × 14). As an example, if the sample is illuminated with the femtosecond laser at 930 nm, an SHGS at 465 nm is detected. A similar case occurs with the rest of the available excitation wavelengths, as

Table 1. Excitation wavelength and its corresponding SHGS wavelength.

Excitation Wavelength	SHGS Wavelength
930 nm	465 nm
940 nm	470 nm
950 nm	475 nm
960 nm	480 nm
970 nm	485 nm
980 nm	490 nm
990 nm	495 nm
1000 nm	500 nm
1010 nm	505 nm
1020 nm	510 nm
1030 nm	515 nm
1040 nm	520 nm
1050 nm	525 nm
1060 nm	530 nm

indicates.

Figure 1 illustrates in a schematic form the processing methodology, where an example of a honey sample being illuminated with an excitation wavelength of 930 nm retrieves a second-harmonic signal at 465 nm, which is registered. Here, the result is an SHGM image (indicated in Figure 1), where sugar crystals are visible, and its corresponding histogram is then retrieved (SHGS histogram in Figure 1) using a Matlab (MathWorks, Inc.) algorithm. A second algorithm now analyzes the numerical information of each image’s histogram and the relationship of the gray levels within it; this process helps determine the maximum value of the SHGS from that particular image (orange block in Figure 1). This value will represent the SHGS of the sample according to its adulteration condition.

The microscope’s laser energy (power) is selected to be the same at different excitation wavelengths. A relative power is possible since the microscope system controls the laser’s power using an acoustic–optic modulator (AOM). Then, the gain of the microscope’s photodetector is modulated to avoid signal saturation or a lack of signal emission from the sample. The recording setting of the microscope uses an output power of 1.35 W, an average speed of >40 nm/s, a pulse width of 140 +/− 20 fs, TEM00, M2 < 1.1, a beam diameter of 1.2 +/− 0.2 mm, a beam ellipticity of 0.9 to 1.1, and a nominal repetition rate of 80 MHz, with a frame time of ~968 milliseconds.

Figure 2a shows a basic schematic view of a second-harmonic generation microscope, where the source laser illuminates the sample employing a scanning mirror (after signal conditioning) and a microscope objective. Here, two possible sensors (PMTs) could be used, one in reflection (backward) and a second in transmission (forward). In this study, the transmission path is used. Figure 2b shows an illustrative image of a mesquite honey sample where liquid honey is mixed with sugar crystals (i.e., bright-field microscopy), while Figure 2c shows an SHGM image where the sugar crystals are observed without the noise caused by liquid honey.

2.2. Honey Samples

Professional beekeepers collected the honey from bee apiaries (see Figure 3a), where Apis mellifera honeybees produce the mesquite honey (Prosopis glandulosa, Figure 3b). These honey samples were stored under the same room temperature and relative humidity. The water content was measured using a refractometer (Hanna Instruments, model HI 96801) at 589 nm [30]. Mesquite honey has a moisture content of 15.2% ± 0.2 and a refractive index of 1.4987.

As SHGM requires a microscopic, tiny amount of honey for imaging, it was possible to inspect each honey several times and in different regions. The honey was stirred in the container, and a sample was extracted and fixed in an optically clear microscope glass slide. With the help of a glass coverslip, a uniform honey layer was obtained for each case, keeping the same conditions during the image recording.

Table 2. The ratio of honey/fructose of the six kinds of samples.

Sample Honey/Fructose	Honey (%)	Fructose (%)
75/25	75	25
80/20	80	20
85/15	85	15
90/10	90	10
95/5	95	5
100/0	100	0

indicates the mixing ratios of mesquite honey with liquid high-fructose corn syrup (HFCS), where pure honey is indicated as H/F = 100/0. The advantage of this method is that it can record several images from different regions of the same kind of sample using only a small amount of it.

3. Results and Discussion

Before the experiment with the honey samples, a single HFCS sample was recorded by the microscope to compare its signal with that from the honey. As expected, fructose produces no SHGS whatsoever (see Figure 4), a feature already reported in the literature when it is used as a fixation medium [31,32]. Considering this signal’s absence, it is confirmed that all SHGSs will come only from the honey’s structures.

Now, each honey sample is recorded and processed using the methodology described in Section 2. Figure 5 shows images of the crystal of some selected mesquite honey samples with different fructose adulteration concentrations without the interference of the liquid phase of honey or the liquid fructose added as an adulterant.

It is observed that the SHGM image signal intensities come from the crystals’ surfaces. The crystal size, distribution, and shape make it possible to see differences, but the crystals’ morphology remains pretty much unaltered, similar to what was reported by Flores-Moreno et al. for pure, unadulterated honey [23]. The latter is also in agreement with what was reported by Lopez-Tellez et al., where the crystal clusters of honey produce the highest fluorescence intensity value with respect to the liquid part of the honey [33]. However, as mentioned in Section 2.1, further processing is required, as several samples share similar crystal shapes, and histogram information is needed.

Each honey sample (adulterated or pure) is observed with the second-harmonic microscope in several regions. Then, the SHGM images are processed for each adulteration level and excitation wavelength. This numerical processing results in the values presented in Figure 6, where the average SHGS for each kind of honey sample is observed (orange block in Figure 1; these values consider all the excitation wavelengths). From this figure, it is possible to see that SHGS is highly sensitive to honey adulteration with fructose. The SHGS counts decrease as the level of adulteration increases because liquid fructose represents a higher percentage of the honey’s total volume, despite the excitation wavelength. However, this reduction does not exhibit a linear behavior, as there are two variations from 80% to 75% and from 90% to 85%. This behavior is also observed in the error bars (black lines) of the graph, where variations do not follow a regular response, as they come from full and unfiltered image processing. Nevertheless, the method can detect variations of up to 5% in adulteration; however, further algorithms for processing are currently being developed to reduce the error gap in the results.

Even when the fructose does not add SHGS to the analyzed images, it definitively modifies the fructose/glucose (F/G) ratio, which changes the crystallization speed of the honey. This ratio then modifies the arrangement of the sugar crystal clusters and their distribution within the sample, modifying the SHGS. All the previous tendencies were observed in mesquite honey, and further analysis is required to study the behavior of other monofloral or multifloral honey samples with different crystal structures. Nevertheless, the results obtained with SHGM are in good agreement with those already reported in Apis mellifera honey using different inspection methods, such as LF NMR [34] and UV-HPLC [35]. In these reports, different concentrations were used; however, the differences tended to have a similar response according to the adulteration level to those observed in this work.

4. Conclusions

SHGM is a promising inspection technique that makes liquid honey transparent and retrieves a clear and sharp image of the honey’s sugar crystals. The crystals’ images could help to identify the floral origin, while the post-processing information of the second-harmonic data gives information about purity (adulteration condition). SHGM is a remarkable option for observing the shape of the honey crystals, which are particular to each honey. The histogram distribution enabled the identification of differences in HFCS concentration when the SHGS was analyzed. Despite methods that use pollen to determine honey purity and floral origin, this proposed method does not consider the pollen in the calculations of the adulteration level. The latter is an advancement in the field as pollen could be added artificially to filtered honey, but its sugar crystals’ shape and distribution are not. Even when this method could appear more expensive, it is so far the only one that directly observes sugar crystals with a minimum sample amount, a feature not fulfilled by other methods. Further analysis and tests are required to identify different adulteration substances in more honey samples from different floral origins, which is an ongoing focus.