Identification of sugars and phenolic compounds in honey powders with the use of GC–MS, FTIR spectroscopy, and X-ray diffraction

This work aimed at the chemical and structural characterization of powders obtained from chestnut flower honey (HFCh) and honey with Inca berry (HBlu). Honey powders were obtained by spray drying technique at low temperature (80/50 °C) with dehumidified air. Maltodextrin (DE 15) was used as a covering agent. The isolation and evaluation of phenolic compounds and sugars were done by gas chromatography–mass spectrometry analysis. Scanning electron microscopy, Fourier-transform infrared (FTIR) spectroscopy, and X-ray diffraction were performed to determine the morphology of the studied honey powders. The obtained results showed that the content of simple sugars amounted to 72.4 and 90.2 g × 100 g−1 in HFCh and HBlu, respectively. Glucose was found to be the dominant sugar with a concentration of 41.3 and 51.6 g × 100 g−1 in HFCh and HBlu, respectively. 3-Phenyllactic acid and ferulic acid were most frequently found in HFCh powder, whereas m-coumaric acid, benzoic acid, and cinnamic acid were the most common in HBlu powder. The largest changes in the FTIR spectra occurred in the following range of wavenumbers: 3335, 1640, and below 930 cm−1. The X-ray diffraction profiles revealed wide peaks, suggesting that both honey powders are amorphous and are characterized by a short-range order only.

Solution preparation and spray drying. Honeys were mixed with water and maltodextrin (MD) DE 15 (PEPEES, Łomża, Poland) to obtain a 60% (w/w) solution in which the ratio of honey solids to MD solids was 75:25. Portions containing 400 g of feed solutions were spray dried in a pilot plant spray drier (Niro Minor, GEA) under with the following conditions: feed ratio speed, 0.25 cm 3 s −1 ; atomization speed, 24,000 rpm; inlet/ outlet air temperature, 80/50 °C. Such low drying temperature was achieved by applying additional force of evaporation by air dehumidification, using a dehumidification system comprising TAEevo TECH020 chilling unit (MTA, Italy) and condensation-adsorption unit ML270 (MUNTERS, Sweden), as described before 22 . During the process of drying, the humidity of the air entering the spray drier was not higher than 1.2 g m −3 . Powders were packed in plastic (BOPA/PE, 55 μm) bags (Pakmar, Garwolin, Poland) and sealed and stored at 25 °C/50% relative humidity.
Chemical properties of powders. Extraction and derivatization of phenolic compounds and sugars. The phenolic compounds were isolated by solid-phase extraction (SPE) 10,26 . For this, 2 g of powdered honey was dissolved in 25 mL acidified water (HCl, pH 2), and the prepared solutions were transferred to the conditioned SPE columns filled with C18 stationary phase (6 mL, 500 mg, Chromabond, Macherey Nagel). Then, the columns were washed with 40 mL deionized water, and sugars and other more polar compounds were eluted. The adsorbed phenolic compounds were eluted with 3 × 5 mL portions of methanol. The collected eluent was dried over Scientific RepoRtS | (2020) 10:16269 | https://doi.org/10.1038/s41598-020-73306-7 www.nature.com/scientificreports/ anhydrous sodium sulfate and then evaporated to dryness on a rotary evaporator under reduced pressure 27,28 . The extracted dry residue was derivatized with 100 µL BSTFA with 1% TMCS for GC derivatization, Supelco) [N,O-bis(trimethylsilyl) trifluoroactamide with 1% trimethylchlorosilane (for GC derivatization, Supelco)] and 200 µL pyridine (anhydrous, 99.8%, Sigma-Aldrich), and the content was heated at 60 °C for 1 h. TMS [Trimethylsilyl] derivatives were subjected to GC-MS analysis. Sugars were isolated by using liquid-solid extraction. Briefly, 0.5 g of the sample was mixed with 20 mL of methanol, and then ultrasound-assisted extraction was performed at 40 °C. The obtained extract was dried over anhydrous sodium sulfate and then evaporated to dryness on a rotary evaporator under reduced pressure. Five milligrams of the obtained dry residue was derivatized and analyzed in the same way as samples were analyzed for phenolic compounds.
GC-MS analysis. The separation and detection of phenolic compounds and sugars were carried out using a 7890B GC System with a 7000C GC/MS Triple Quad mass detector (Agilent Technologies, USA). For the process of separation, the HP-5 ms fused silica capillary column was used (30 m × 0.25 mm × 0.25 μm, Agilent Technologies). Injection temperature was maintained at 260 °C, and the carrier gas flow rate was 1 mL min −1 (helium). Temperatures were programmed from 40 to 300 °C at a rate of 3 °C min −1 (split 1:10) for the separation of compounds. The detection process was performed in the full scan mode from 45 to 600 m/z. Using the same parameters, all compounds were calibrated.
Physical properties of powders. Water content and water activity. Water content was determined by oven method (105 °C/4 h), while water activity was measured using HygroLab C1 (Rotronic, Switzerland) at 25 °C.
Microstructure. The outer morphology of powder particles was observed under a scanning electron tabletop microscope TM3000 (Hitachi, Japan) operating at 15 kV. Before loading into the SEM chamber, the samples were subjected to metallization (sputtering) with a thin layer of gold and were then observed at a magnification of 500×.
FTIR. The infrared spectra of the analyzed samples were measured using 670-IR spectrometer (Agilent, USA). To ensure 20-fold internal reflection of the absorbed beam, Attenuated Total Reflectance (ATR) attachment was used in the form of a ZnSe crystal with adequate geometry (truncated at 45°). Sixteen scans were registered during the measurement, and subsequently, the program averaged the results for all spectra. Prior to the measurement, the ZnSe crystal was cleaned using ultraclear solvents (Sigma-Aldrich). Before (1 h) and during the experiment, the measurement chamber was kept in an inert N 2 atmosphere. Spectral measurements were recorded in the region from 700 to 3800 cm −1 at a resolution of 1 cm −1 . The measurements were taken at the Central Apparatus Laboratory of the University of Life Sciences in Lublin. The spectra were analyzed and processed using Grams/AI software developed by ThermoGalactic Industries (USA). All the spectra were measured at 23 °C.
X-ray diffraction. The structure of the powders was studied using Empyrean X-ray diffractometer (PANalytical) with CuKα radiation (λ = 1.54056 Å) and a generator operated at 40 kV and 30 mA. The radiation was detected with a proportional detector. The source divergence and detector slit were 1/2, and Soller slits were applied. The X-ray diffraction profiles were measured in θ−2θ geometry over a range from 10° to 90° with a step of 0.01° and counting time of 5 s per data point at room temperature. Statistical analysis. All tests were performed in three replicates, and values were expressed as means ± standard deviations (SDs). Statistical analyses were performed using Statistica software (Statsoft Inc.) at a significance level of α = 0.05. The data were analyzed using analysis of variance, and the means were compared using t-test.

Results and discussion
Recently, Samborska et al. 12 and Jedlińska et al. 29 have presented a novel approach by the application of dehumidified air for honey spray drying. The additional force of water evaporation provided by the dehumidified air allows using a lower temperature, which, in turn, reduces the amount of carrier. The honey powders thus obtained have an increased quality with a higher content of honey and reduced degradation of biologically active compounds 22 .
Fructose and glucose were identified and quantified in the honey powders, and the results are provided in Table 1. The sum of glucose and fructose contents exceeded the value of 60 g 100 g −1 which is required for natural honeys 30 . The contents of simple sugars in HFCh and HBlu were 72.41 and 90.19 g × 100 g −1 , respectively. The dominant sugar in the powders was glucose (41.34 g × 100 g −1 in HFCh and 51.61 g × 100 g −1 in HBlu). The total content of disaccharides, considering the share of saccharose, in HFCh and HBlu was 18.03 and 6.34 g × 100 g −1 , respectively. Juszczak et al. 's 31 research on herb honeys showed that the content of fructose, glucose, and sucrose ranged from 25.9 to 36.8 g 100 −1 g, from 23.1 to 33 g × 100 g −1 , and from 0.4 to 24.8 g × 100 g −1 , respectively. Significantly a higher content of trisaccharides was found in the HFCh powders.
Water content and water activity. The obtained honey powders contained, respectively, 3.1 ± 0.1% (HFCh) and 2.8 ± 0.1% (HBlu) of water, while water activity was estimated as 0.196 ± 0.001 (HFCh) and 0.193 ± 0.004 (HBlu). These values are typical for powders obtained by spray drying, both at high temperature applied in a traditional approach 17,35 and with the use of dehumidified air in a novel approach 12,29 . Low water content (below 4%) and water activity (below 0.2) confirmed the proper conditions for water evaporation created in the drying chamber by applying low temperature and low humidity for processing air. Figure 1 shows the external microstructure of the particles of honey powders with a smooth surface and the linkages between the individual particles, which is very common in the case of honey powders containing more than 70% of honey solids. Spherical particles with smooth surface indicate that Table 1. Content of sugars in honey powders (g × 100 g −1 d.m.). Mean values from three repetitions ± SD, Means with different letter in the same row are significantly different (α = 0.05). Linearity range -10 -200 mg/ mL (2 -40 g/kg), limit of quantification -0.03 mg/mL (0.006 g/kg), limit of detection -0.01 mg/mL (0.002 g/ kg).  Table 2. Content of phenolic acids in honey powders (mg kg −1 d.m.). Mean values from three repetitions ± SD; means with different letters in the same row are significantly different (α = 0.05). Linearity range: 10-200 mg mL −1 (2-40 g kg −1 ); limit of quantification: 0.03 mg mL −1 (0.006 g kg −1 ); limit of detection: 0.01 mg mL −1 (0.002 g kg −1 ).  Table 3. Relative composition (%) of the methanol extracts from dry honey. Mean values from three repetitions ± SD, Means with different letter in the same row are significantly different (α = 0.05).  www.nature.com/scientificreports/ the tested honey powders had fully amorphous morphology. Samborska et al. 12 and Jedlińska et al. 29 reported a similar morphology for powders containing 80% of honey solids.

FTIR.
In the next stage of the study, ATR-FTIR spectroscopy was used for analyzing in detail the characteristics of the tested HFCh and HBlu powders. Figure 2 and Table 4 (in the spectral range of 3800-700 cm −1 ) present the spectra of the HFCh and HBlu honey samples, which facilitate the correct interpretation and easier characterization of individual bands. Table 4 also assigns bands to the corresponding vibrations of the functional groups in the identified compounds. According to Anjos et al. 36 and Svečnjak et al. 37 the first spectral area ranging from 3650 to about 3000 cm −1 (for all samples, Table 4 and Fig. 2), characterized by clear bands with a maximum of about 3335 cm −1 , corresponds to the stretching vibration of the -OH group of carbohydrates, water, and organic acids. This area is very often attributed to the stretching vibrations of carboxylic acids and also to the -NH 3 stretching band of free amino acids, which cause a slight strengthening of this area. Other bands, ranging from 3000 to 2800 cm −1 , correspond to the stretching vibrations of the C-H groups (both alkyl and aromatic, which belong to the sugar backbone). These vibrations belong to the functional groups -CH 2 and -CH 3 (alternately symmetrical and asymmetrical). Vibrations with a maximum of ~ 3335 cm −1 may also originate from carboxylic acids, the irregular absorption of which (with a wide band coming from the vibrations of the -OH group) significantly increases the C-H stretching vibrations in the systems of the -CH 2 and -CH 3 groups.
Due to the formation of strong hydrogen bonds, which in this case belong to carboxylic acid dimers 36 , a wide range of vibrations originate from the ν(-OH) groups. A very clear band with a maximum at about ~ 1640 cm −1 (Fig. 2) corresponds, in turn, to the deformation vibrations of the -OH groups. Attention should be paid to a very important area, which is only slightly marked in our spectra (with a maximum at about 1735 cm −1 ) and is due to the stretching vibrations of functional groups such as ketone C = O of fructose and aldehyde CH = O of glucose. It can be seen that it only mildly enhances the vibration with a maximum at 1640 cm −1 .
A very characteristic area of the samples selected for testing is the fingerprint region (extending from ~ 1480 to 700 cm −1 , in our case). This region is rich in bands and provides good information about changes in samples   Table 4). The area from about 1040 to 930 cm −1 and below can be significantly strengthened by the stretching vibrations of C-O in C-OH group and stretching of C-C in the carbohydrate structure 37,39,40 . The area below 930 cm −1 (from 930 to about 700 cm −1 ) is the vibration area, which is very characteristic for vibrations from the anomeric region of carbohydrates or deformation vibrations of C-H and C-C 24,40 . Even small changes in vibrations from this region usually indicate strong modifications/differences in the sugar fraction bonds (glycosidic bonds). In the case of honey varieties selected for testing, the largest changes occurred at the following wave numbers: 3335, 1640, and below 930 cm −1 .
X-ray diffraction. Figure 3 shows the X-ray diffraction profiles of HFCh and HBlu powders. It can be seen that instead of narrow Bragg peaks, the X-ray diffraction profiles reveal wide peaks, suggesting that the tested honey powders are amorphous and are characterized only by a short-range order. It means that the X-ray scattering is coherent for a small volume and is incoherently averaged over the whole sample. The mean response represents the average local order in the sample. The total amorphous X-ray diffraction profile can be treated as a sum of the Gaussian components 41 parameterized by their position, amplitude, and SD. There is no periodic arrangement of atoms and molecules in amorphous materials. However, according to the Gaussian distribution, there are still average characteristic distances between the atoms and molecules located in the material. Because of the normal distribution of these distances, the intensity profiles also have the Gaussian shape. The presence of more than one wide diffraction peak in diffraction profiles indicates that there are a few www.nature.com/scientificreports/ characteristic distances between atoms or molecules. In our case, it was necessary to assume four components to achieve a good fit of the measured profiles. Using four Gaussian components, the experimental profiles for both samples were fitted, and the results are presented together with the profiles in Fig. 3. For the HFCh sample, the Gaussian components were present at the scattering angles 2θ = 18.56°, 22.96°, 34.04°, and 79.41°, which corresponded to the distances 4.78, 3.87, 2.36, and 1.21 Å, respectively. However, for the HBlu sample, the Gaussian components were present at the scattering angles 2θ = 18.39°, 22.92°, 38.04°, and 80.67°, and therefore corresponded to the distances 4.82, 3.88, 2.36, and 1.19 Å, respectively. These distances determine the average distances between atoms in molecules. The amplitudes of the Gaussian components were in the ratio 3.81:1:4.71:2.56 for HFCh sample and 4.24:1:5.08:2.72 for HBlu sample. For individual Gaussian components, SDs of 3.65°, 3.68°, 14.4°, and 21.13° were obtained for both HFCh and the HBlu samples. The results indicate that both powders have a very similar structure characterized by a short-range order only.

Conclusion
The results of the study showed that the content of simple sugar was 72.4 g × 100 g −1 in HFCh and 90.2 g × 100 g −1 in HBlu. Glucose was the dominant sugar at an amount of 41.3 g × 100 g −1 in HFCh and 51.6 g × 100 g −1 in HBlu. The concentrations of total disaccharides were equal to 18 g × 100 g −1 in HFCh and 6.3 g × 100 g −1 in HBlu. 3-Phenyllactic acid and ferulic acid were most frequently found in HFCh powders, whereas m-coumaric acid, benzoic acid, cinnamic acid, and p-coumaric acid were the most common acids in the case of HBlu. The largest changes in the FTIR spectra occurred in the range of the following wavenumbers: 3335, 1640, and below 930 cm −1 . FTIR offers unique advantages, as it reflects the overall vibrations of the components and their interactions within the samples as spectra; it is also shown to be a reliable method to quantify the majority of the sugar content in honey and is easily adapted to the routine analysis of this product. The microstructure analysis and X-ray diffraction profiles revealed wide peaks, suggesting that the honey powders are amorphous and are characterized by a short-range order only. The results of the Gaussian components indicated that both samples have a very similar structure. The use of a novel spray drying method allowed obtaining honey powders that retained their health-promoting properties. These honey powders are an innovative product with potentially wide applications in the food industry.

Data availability
All the data generated or analyzed during this study are included in this published article (and its Supplementary Information files).