1. Introduction

Starch granules have an interesting and somewhat tunable ultrastructure, allowing them to be used in different industrial applications such as a binder in pharmaceuticals to hold medication together, a disintegrant to break the medication up into smaller particles, an absorbent to help with stability, and a lubricant [1]. However, native starch can be limited by its inability to withstand chemical conditions used during manufacturing such as high temperatures and large changes in pH. For that reason, modified starch granules and alternative polysaccharides are often sought for pharmaceutical formulations.

Starch granules are primarily composed of the polysaccharides, amylose and amylopectin. Amylose is a mostly unbranched α-glucan and accounts for a small percentage (10%-35%) of the total starch dry weight in wild-type starches. Amylopectin is the major constituent of wild-type starches. Amylopectin is a branched α-glucan capable of forming highly ordered double helices. From X-ray diffraction (XRD) and nuclear magnetic resonance analyses of starch, two crystallographic starch structures, that differ in their arrangement and ultrastructure of the double helices, have been previously identified as A- and B-type allomorphs. The A-type allomorph contains maltodextrin A-type crystals that exhibit monoclinic B2 type symmetry [2] while the B-type allomorph contains B-type crystals which exhibit hexagonal C₆ type symmetry [3]. Examples of A-type crystallinity include starch from cereals such as maize, rice and wheat while starch from tubers such as potatoes, yams, squash, and carrots have B-type crystallinity. The ultrastructure of the starch allomorphs is influenced by various chemical parameters including the distribution of the chain length, the frequency of branching points and physical factors including temperature and hydration conditions [4]. An alternative polysaccharide to starch used in pharmaceutical formulations is maltodextrin. Recently, crystallized maltodextrin particles (CMPs) have been prepared by two different procedures and XRD confirmed that the two different preparations were able to create A- and B-type allomorphs consisting of the same α-glucan [4].

The content of starch and alternative starch-like polysaccharides like maltodextrin polymers are well understood however, the heterogeneity of the ultrastructure of starch and CMPs within and between individual granules, and its effect on bulk material properties needed for applications in different industrial processes is obscured by bulk measurements. Furthermore, attempts at analysis of individual granules by standard high-resolution techniques including electron and atomic force microscopies do not supply ultrastructure information directly and they are fraught with complications such as destructive sample preparation which alters their natural structure. For example, to examine the ultrastructure of starch granules’ interior with either technique, the microscopic granules must be freeze-fractured, complicating in vivo analysis. However, since starch gives rise to strong second harmonic generation (SHG) signals [5–7], the unique internal crystalline ultrastructure of individual granules of starch and starch-like polysaccharides can be probed in 3D using polarization-sensitive SHG (PSHG) microscopy [8]. Therefore, PSHG microscopy can be used to investigate the heterogeneity of the ultrastructure of starch and starch-like polysaccharides within and between individual granules. Furthermore, since CMPs have structural features that mimic native starch granules, CMPs can be used as model starch granules for PSHG microscopy to improve the understanding of SHG from starch granules and the ultrastructure of the two allomorphs.

The focus of this study is to use PSHG microscopy to determine the ultrastructural organization of CMPs for furthering their use as starch-model systems, which could aid in the increased understanding of the ultrastructural organization of starch granules and further the potential use of CMPs for different applications such as in the pharmaceutical industry.

2. Materials and methods

2.1 A- and B-type CMPs preparation

CMPs were synthesized from commercial maltodextrin with a dextrose equivalent of 4.0-7.0 (Sigma-Aldrich, CAS 9050-26-6) following two procedures described previously with several modifications [4,9]. 75 g of maltodextrin was first dissolved in 130 mL of deionized water at 95 ± 4°C in a beaker. The solution was then allowed to cool to ∼50°C before it was quantitatively transferred into a 250 mL volumetric flask and diluted with deionized water at ∼50°C. The solution in the flask was then left to cool to 4°C overnight. 45% of the maltodextrin was recovered while the remainder stayed suspended in solution. 3% of the total maltodextrin initially used was recovered after being fractionated using 95% ethanol to obtain a degree of polymerization range between 5-40, as described in the procedure by Hejazi et al., except to dissolve the recovered maltodextrin, a two-neck round bottom flask attached to a condenser was used and in place of freeze drying, high vacuum was used until completely dry due to equipment availability [4]. Samples were left on high vacuum for 1-8 h depending on their amount. The fractionated maltodextrin powder was dissolved in deionized water using a microwave reactor (Discover 2, CEM Corp.) for 15 minutes. For A-type CMPs, 50% w/v fractionated maltodextrin in deionized water was heated in a sealed microwave reactor vessel at 65 W to reach ∼120°C while for B-type CMPs, 30% w/v fractionated maltodextrin was heated in a sealed microwave reactor vessel at 75 W to reach ∼120°C due to the increased amount of water. The sealed microwave reactor vessels were removed when they cooled to 60°C due to microwave reactor limitations (i.e. not being able to open the microwave reactor door until the solution was cooled to 60°C). The sealed microwave reactor vessels were then placed into two beakers containing 60°C water to prevent the sample from cooling rapidly. For A-type CMPs, the appropriate microwave reactor vessel immersed in the beaker was placed directly into a CO₂ incubator (MCO-5AC-PA, Panasonic Corp.) and left at 30°C overnight, while for B-type CMPs, the microwave reactor vessel immersed in the beaker was left to cool for 20 min and then placed in the refrigerator at 4°C overnight. The CMPs were then centrifuged at 800 g (accuSPIN 8C, Fisher Scientific International, Inc.) for 10 min, washed 5 times in deionized water, collected using vacuum filtration, and the powder was stored at -20°C.

Initial attempts to synthesize the A- and B-type CMPs created CMPs of lower SHG intensity than maize starch however, improvements were made to the procedure which resulted in CMPs that generated more intense SHG. These improvements (already incorporated above) included using deionized water, ensuring that little water evaporates by use of a condenser, using high vacuum to dry the maltodextrin, and improved dissolution of the intermediate crystalline maltodextrins using a microwave reactor.

2.2 CMPs preparation for imaging

Hydrated CMPs samples were prepared for SHG imaging by resuspending ∼0.010 g of the CMPs in 1 mL of deionized water and then sonicating for 180 s. Subsequently, the hydrated CMPs were placed in a hydrogel to prevent movement during imaging. The hydrogel contained 0.68 M acrylamide (Sigma Aldrich), 8.3 mM bis(acrylamide) (Sigma-Aldrich), 0.18 mM ammonium persulfate (Sigma-Aldrich) and 26 mM tetramethylethylenediamine (TEMED; Sigma-Aldrich) [10]. Within a few seconds after adding TEMED, 30 µL of the solution was pipetted onto a microscope slide containing a parafilm spacer. A No. 1.5 microscope glass coverslip was immediately placed over the sample allowing the polymerization of the film to a thickness of ∼20 µm. The sample was subsequently sealed with nail polish to avoid drying. Sample slides of corn starch, from the local grocery store, were prepared in a similar manner except that the corn starch solution was not sonicated.

To image dry CMPs, A- and B-type CMPs were dried by being placed into 95% ethanol and centrifuged at 800 g for 10 minutes and washed 3 times. The pellet was collected using vacuum filtration and transferred into a glass vial where they were left to dry for 4 days at ambient temperatures (21 ± 1°C). Microscope slides for imaging the dried CMPs were prepared by resuspending the dried CMPs in 1.5 mL of 95% ethanol and sonicating the solution for 180 s. Then 30 µL of solution was placed onto a microscope slide which had a parafilm spacer on top. The solution was left for a minimum of 1 h at ambient temperatures to allow for the ethanol to evaporate. After 1 h, a glass cover slip was placed over the sample and then sealed with nail polish.

2.3 SHG microscopy

SHG microscopy was performed on the A- and B-type CMPs using a microscope previously described [11]. Briefly, a custom-built microscope was coupled with an ultrafast femtosecond duration pulsed laser (FemtoLux 3, EKSPLA) which operated with 5 MHz repetition rate and 290 fs duration pulses at a wavelength of 1030 nm and delivered ≤0.095 nJ of pulse energy at the sample when the SHG intensities of the CMPs and maize were compared. The laser beam was raster scanned across the sample using a pair of galvanometric scanning mirrors (ScannerMAX, Pangolin Laser Systems), and polarization control of the laser beam was achieved using a stationary linear polarizer (LPVIS100, Thorlabs, Inc.) followed by a half-wave plate (MWPUM2-25-1030-V, Karl Lambrecht Corp.) in a motorized rotation mount, both located just before the microscope objective. The laser was focused on the CMPs using a 0.8 numerical aperture (NA) air immersion microscope objective (Plan-Apochromat 20x, Carl Zeiss AG). The SHG signal was collected in transmission mode by a custom polarization independent 0.85 NA collection objective (Omex Technologies, Inc.), and the SHG signal was collected using an interference filter centered at 515 nm with a 10 nm bandwidth (65-153, Edmund Optics Inc.). The SHG signal was measured using a single-photon counting detector (H10682-210, Hamamatsu Photonics K.K) and acquired using a data acquisition card (PCIe-6363, National Instruments). To measure the variation in SHG intensity between hydrated and dry CMPs, a quarter-wave plate (WPMP4-22-HEAR-1030, Karl Lambrecht Corp.) was added just before the microscope objective to obtain circularly polarized light.

2.4 Polarization-sensitive SHG image analysis

The polarization-in polarization-out (PIPO) SHG technique was used for polarization-dependent measurements of the SHG signal as previously described [12]. Polarization-sensitive detection was achieved using the SHG microscope setup described in section 2.3 along with the addition of a motorized rotation stage containing an SHG polarizer (LPVISA100, Thorlabs, Inc.) in the forward detection path. The PIPO SHG technique consisted of acquiring 100 individual frames with 6 µs pixel dwell time and 100 × 100 pixels for each image. A total of 65 images were obtained at each combination of 8 evenly spaced analyzer angles (0 to 157.5$^\circ $) for each of 8 evenly spaced half-wave plate angles (0 to 78.75$^\circ $), followed by one control image at the initial polarization parameters to verify that the samples had not moved or burned. PIPO SHG imaging of hydrated CMPs required 0.38 mW or 0.076 nJ of pulse energy while PIPO SHG imaging of dry CMPs required 0.75 mW or 0.15 nJ of pulse energy.

For the analysis of the PIPO SHG data performed here, a laboratory Cartesian coordinate system, XYZ, was defined with respect to the principal propagation direction of the laser, Y, where XZ is the imaging plane (Fig. 1 in Ref. [12]). Analysis of the PIPO SHG data was performed under several assumptions to obtain the second-order nonlinear optical susceptibility tensor component ratio, $\rho = \chi _{ZZZ}^{(2 )}/\chi _{ZXX}^{(2 )}$, including cylindrical symmetry [13,14]. It was also assumed that the crystal axis of A- and B-type CMPs is oriented along the focal plane when an equatorial optical section is imaged however, due to possible out of plane tilting of domains, this is likely untrue in heterogeneous CMP regions. Furthermore, Kleinman symmetry was assumed. These assumptions are similar to previous studies on the PSHG of starch granules [15], allowing comparison of the results of the ratio of two unique nonzero tensor elements, $\chi _{ZZZ}^{(2 )}$ and $\chi _{ZXX}^{(2 )}$, which permits the use of Eq. (1).

 

Fig. 1. Characterization of CMPs using powder X-ray diffraction (PXRD), a differential scanning calorimeter (DSC) and SHG microscopy. The PXRD data for A-type CMPs (a) and B-type CMPs (b) over the range 3–30° in a 2θ scale. The DSC endotherms for 40% w/v A-type (black squares) and B-type (grey circles) CMPs dissolved in deionized water (c). A bar graph representing the average SHG intensities for hydrated and dry maize and A- and B-type CMPs where the SHG intensity of dry maize is where the SHG intensity of dry maize was obtained from Ref. [15] (d). SHG intensity images of hydrated A-type CMPs in a hydrogel (e), hydrated B-type CMPs in a hydrogel (f), dry A-type CMPs (g), and dry B-type CMPs (h). The scale bar in (e) represents 25 µm. The asterisks *, **, *** and **** in (d) indicate p < 0.05, p < 0.02, p < 0.01, and p < 0.001 significance, respectively.

Download Full Size | PDF

The SHG intensity $({{I_{2\omega }}} )$ is defined in Eq. (1) in terms of the orientation of the laser electric field polarization ($\theta $), the analyzer orientation (φ), the in-plane angle between the laboratory Z-axis and the crystal axis ($\delta $) as well as $\rho $, the molecular second-order nonlinear optical susceptibility tensor component ratio projected onto the image plane (XZ).

The value of $\rho $ has been previously related to the helical tilt of SHG emitters in starch [16] as well as the orientation of structured water within starch [15]. $\rho $ also represents disorder of SHG emitters within the focal plane, with theoretically $\rho \to 3$ as disorder increases [17].

Additional quantification was performed with the $\rho $ values by taking the ratio of either the number of red pixels with fitted $\rho $ values between 5 and 6 per image (‘Ratio of Ordered Pixels for $\rho $’ in Table 1), or the blue pixels with fitted $\rho $ values between 2 and 3.5 per image (‘Ratio of Disordered Pixels for $\rho $’ in Table 1), to the total number of fitted pixels per image with a goodness of fit parameter $\ge $0.8.

Table 1. SHG and PSHG analysis summary for hydrated and dry A- and B-type CMPs. The average SHG intensity, the average $\rho $ values, the distribution width of the $\rho $ values, the average ratio of ordered pixels for $\rho $, the average ratio of disordered pixels for $\rho $, the average DOLP values and the distribution width of the DOLP values are shown for at least 8 CMPs of hydrated and dry A- and B-type CMPs.

View Table

In addition to obtaining the value of $\rho $, another parameter known as the degree of linear polarization (DOLP) was determined using the following equation.

The SHG is described by the Stokes parameters, ${s_0}$, ${s_1}$ and ${s_2}$ defined as: ${s_0} = {I_{2\omega ,0}} + {I_{2\omega ,90}}$, ${s_1} = {I_{2\omega ,0}} - {I_{2\omega ,90}}$ and ${s_2} = {I_{2\omega ,45}} - {I_{2\omega , - 45}}$ where ${I_{2\omega ,a}}$ is the SHG intensity at the analyzer angle a. An average DOLP is obtained by using Eq. (2) to calculate the DOLP at 8 incident laser polarizations (0°, 22.5°, 45°, 67.5°, 90°, 112.5°, 135°, and 157.5°), and for each one, two DOLP additional calculations were averaged, one using measurements at analyzer angles 0°, 45°, 90° and 135° and another at 22.5°, 67.5°, 112.5° and 157.5°.

2.5 Powder X-ray diffraction

Powder X-ray diffraction (PXRD) spectra were collected for the A- and B-type CMPs at the Department of Physics and Atmospheric Science, Dalhousie University using a Siemens D500 Powder X-ray Diffractometer with Cu K_α1 radiation (λ=1.5418 Å). The scans were performed with measurements between 3°-30° (2θ), with increments of 0.04° and 4 s integration. The degree of crystallinity of CMPs, which has also been referred to in literature as the crystallinity index, was calculated from the PXRD data by integrating the peaks and taking the ratio of the area under the crystalline peaks to the total area under the PXRD curve.

2.6 Differential scanning calorimetry

The onset (T_o), melting peak (T_m) and end (T_c) temperatures were measured using a differential scanning calorimeter (DSC) located at the Department of Chemistry, Saint Mary’s University (Q100 DSC, TA Instruments). The DSC calibration was verified using an indium standard. For DSC measurements, 3.5 mg of A- and B-type CMPs was placed in aluminum pans, followed by 5.25 µL of deionized water (40% w/v), and gently stirred with the tip of a micropipette. The sample pans were then hermetically sealed using a Tzero Sample Press (Tzero press, TA Instruments). After sealing, the solutions were allowed to equilibrate at ambient temperatures for an hour before scanning. The DSC was set to heat at a rate of 10°C/min over the temperature range of 30 to 120°C, and all measurements were performed in triplicate.

3. Results

3.1 Crystallographic and thermal characterization of CMPs

Synthesis of A- and B-type CMPs was confirmed using PXRD and a DSC. The A-type CMPs’ PXRD data is consistent with what has been previously observed in the literature for starches with A-type morphology, with a major peak present at 18° (Fig. 1(a)) [18], while PXRD data for the B-type CMPs also had peaks which are consistent with what has been previously observed for starches with B-type morphology, with peaks at 5° or 22°, but not 18° (Fig. 1(b)) [18]. The DSC endotherms were also consistent with what was previously observed in the literature for starches with A-type and B-type morphology [19], namely, A-type CMPs have a broader peak and higher melting temperature than B-type CMPs (Fig. 1(c)). The B-type CMPs have a T_o of 70.8 ± 0.9°C, a T_m of 91.6 ± 0.9°C, and a T_c of 97.8 ± 0.8, whereas the A-type CMPs have a T_o of 53 ± 2°C, a T_m of 106.22 ± 0.01°C, and a T_c of 113.0 ± 0.2 °C. Further, the A-type CMPs have a temperature range (T_c-T_o) of 60 ± 2°C which is greater than the temperature range of the B-type CMPs (33 ± 1°C).

3.2 SHG intensity of A- and B-type CMPs

After confirmation with PXRD and the DSC, the A- and B-type CMPs were compared at dry and hydrated conditions via SHG intensity, using circularly polarized excitation at 1030 nm, and no polarization components in the forward detection path. The size of the analyzed particles was kept below 10 μm to avoid birefringence effects. The SHG intensity of the hydrated A- and B-type CMPs were found to be 3-4 times greater than the SHG intensity of hydrated commercial maize starch (Fig. 1(d) and Supplement 1 Fig. S1). In comparison to one another, the SHG intensity of the dry A-type CMPs (Fig. 1(g)) was significantly higher than the SHG intensities of the hydrated A-type (Fig. 1(e), p < 0.001) and hydrated B-type CMPs (Fig. 1(f), p < 0.001), as well as the dry B-type CMPs (Fig. 1(h), p < 0.001). The SHG intensity of the dry B-type CMPs was also significantly higher than the SHG intensities of the hydrated B-type (p < 0.01) and the hydrated A-type CMPs (p < 0.001). A larger difference in SHG intensity is also apparent in hydrated versus dry A-type CMPs than hydrated versus dry B-type CMPs (Fig. 1(d) and Table 1).

3.3 PSHG of A- and B-type CMPs

To gain insight into the ultrastructural differences of the CMPs, PIPO SHG microscopy was also used to study dry and hydrated A- and B-type CMPs. A summation of PSHG intensity images of similar sized CMPs are shown in Fig. 2(a1-a4) while color-coded maps of the fitted $\rho $ values are shown in Fig. 2(b1-b4). The corresponding occurrence frequency histograms of the fitted $\rho $ values are shown in Fig. 2(c1-c4) and the fitted $\delta $ parameters are displayed as vector images in Fig. 2(d1-d4).

 

Fig. 2. PIPO SHG analysis of A- and B-type CMPs under different treatment conditions. Hydrated A-type CMPs (a1 – d1), hydrated B-type CMPs (a2 – d2), dry A-type CMPs (a3 – d3), and dry B-type CMPs (a4 – d4). Sum of PIPO SHG intensity images (a1 – a4) where the scale bar in (a1) represents 2 µm. Color-coded maps of the fitted $\rho $ values where a $\rho $ value of 2 is represented by blue and a $\rho $ value of 6 is represented by red (b1 – b4). The corresponding occurrence frequency histograms for the $\rho $ values along with Gaussian fits of the histograms (c1 – c4). Vector diagrams of the fitted orientation of the crystal axis $\delta $ (d1 – d4).

Download Full Size | PDF

The color-coded maps of the fitted $\rho $ values represent the level of organization of a given region within the CMPs. Regions with $\rho $ values ≤3.5, which are colored blue, represent low levels of organization while regions with $\rho $ values ≥5, which are colored red, represent high levels of organization, although this level is modified by the divergence of crystallites near nucleation regions due to the radial CMP ultrastructure. The occurrence frequency histograms of the fitted $\rho $ values for each CMP were fitted with a Gaussian function to obtain the mean $\rho $, and this value was averaged among several CMPs to obtain an overall average $\rho $ value for the dry and hydrated CMPs. The width of the fitted occurrence frequency histogram obtained via full-width at half-maximum of the Gaussian fit, was also averaged among several samples to obtain an average $\rho $ distribution width. Analysis was performed on at least 8 CMPs at both hydrations for A- and B-type CMPs, whose diameter ranged between 3 and 8 μm.

The average $\rho $ values, and the average $\rho $ distribution widths of the occurrence frequency histograms are shown in Table 1. The average $\rho $ values were found to be 3.8 ± 0.4 and 3.7 ± 0.2 for the hydrated A- and B-type CMPs, respectively, while the average $\rho $ distribution width for the two types was 1.3 ± 0.3 for both. The average $\rho $ values for the dry A- and B-type CMPs were 3.8 ± 0.4 and 3.9 ± 0.4, respectively, and the average $\rho $ distribution width for the two types were 1.6 ± 0.3 and 1.9 ± 0.5, respectively. A significant difference in the $\rho $ values (Fig. 3(a)) between hydrated and dry A-type CMPs as well as hydrated and dry B-type CMPs was not observed using a two-tailed t test however, a significant difference in the $\rho $ distribution width (Fig. 3(b)) was found between hydrated and dry A-type CMPs (p < 0.05) as well as hydrated and dry B-type CMPs (p < 0.01).

 

Fig. 3. Bar graphs of the $\rho $ values (a), the distribution width of the $\rho $ values (b), the ratio of ordered pixels (ROP) for the $\rho $ values (c), the ratio of disordered pixels (RDP) for the $\rho $ values (d), the degree of linear polarization (DOLP) (e) and the distribution width of the DOLP values (f). The asterisks *, **, *** and **** indicate p < 0.05, p < 0.01, p < 0.002 and p < 0.001 significance, respectively.

Download Full Size | PDF

The plots of the fitted $\delta $ parameters displayed as vector images in Fig. 2(d) reveal that both the hydrated and dry CMPs are radially arranged with respect to their nucleation regions, similar to starch granules.

For the $\rho $ values of the CMPs, additional analysis was performed by finding the fraction of red (ordered) and blue (disordered) pixels to the total number of fitted pixels per image, ROP and RDP, respectively (see section 2.4). Statistically significant differences were observed between the ROP for $\rho {\boldsymbol \; }$(Fig. 3(c)) of dry and hydrated A-type CMPs (p < 0.01), dry and hydrated B-type CMPs (p < 0.001) and dry A-type and dry B-type CMPs (p < 0.001). However, statistical significance between the RDP for $\rho {\boldsymbol \; }$(Fig. 3(d)) of the groups described above was not found.

The DOLP was also investigated for hydrated and dry A- and B-type CMPs (Fig. 3(e)). Statistically significant differences were found between the mean DOLP values of hydrated and dry A-type CMPs (p < 0.001) as well as for the DOLP distribution width (Fig. 3(f)) between these two groups (p < 0.001). A significant difference was also observed for the mean DOLP and DOLP distribution width between hydrated and dry B-type CMPs (p < 0.001 and p < 0.002, respectively). A significant difference was not observed between the DOLP values of the hydrated A-type and hydrated B-type CMPs, but a significant difference was found between the DOLP values of the dry A-type and the dry B-type CMPs (p < 0.001).

4. Discussion

4.1 Macroscopic similarities between CMPs and starch

The synthesized CMPs were found to share many important physical and nonlinear optical parameters with starch granules making them useful model systems of starch for nonlinear optics. The CMPs are composed of crystallized maltodextrin domains in similar arrangements as the crystalline regions inside starch granules, in A- or B-type crystal packing, supported by PXRD data (Fig. 1(a),(b)), and with appropriate melting temperatures as supported by DSC data (Fig. 1(c)). Furthermore, the particles share a 3D macroscopic arrangement of crystallites; they are spherulites, having a radial symmetry with respect to nucleation regions, as shown by the vector diagrams which were calculated from fits of PSHG data at each pixel (Fig. 2(d1-d4)). The diameters of synthesized particles reach as high as small maize granules (∼10 μm), an ideal size for SHG imaging since it is significantly larger than even a moderately high NA focal volume (∼0.5 μm lateral and ∼2 μm axial with a 0.8 NA objective), while minimizing birefringence effects which are prominent at higher granule diameters [20]. Importantly, an SHG intensity of CMPs higher than maize starch was achieved (Fig. 1(d) and Supplement 1 Fig. S1), allowing PSHG properties to be studied at lower laser powers than starch granules. Further analysis of PSHG data reveals that while some properties of CMPs are similar to starch granules, several others are not, highlighting important ultrastructural differences between these synthetic starch models and natural starches.

4.2 SHG intensity of CMPs and starch

The increased SHG intensity of CMPs as compared with maize starch granules may be explained by variations in the degree of crystallinity between the two structures. SHG in starch has been attributed to semi-crystalline amylopectin bands in starch [5,21], while the rest of the granule is filled with alternating bands of mostly amylose which is thought to possess an amorphous structure and hence likely does not emit SHG. The molecular origin of the SHG was further postulated to be aligned intra and interhelical hydroxide and hydrogen bonds mediated by structured water inside the semi-crystalline domains [15,18,22]. The degree of crystallinity measured by PXRD is related to the amount of ordered material in the sample, and hence its sensibly related to SHG intensity. The degree of crystallinity, also referred to as the crystallinity index, was calculated from dry CMPs PXRD data (see section 2.5), revealing values of 0.93 ± 0.01 and 0.89 ± 0.05 for the A- and B- type CMPs, respectively, similar to what has been reported previously for A- and B-type CMPs (1 and 1, respectively) [4]. In comparison, the degree of crystallinity or the crystallinity index of maize starch was previously reported as 0.52 [4], nearly half the value of A-type CMPs, which partially explains the significantly lower SHG intensity of dry maize starch as compared with dry CMPs. The lower degree of crystallinity of starch is likely due to the presence of amylose dominated amorphous regions located as layers between semi-crystalline shells and observed as growth rings (see Fig. 5 in [23]). The self-assembly process utilized to synthesize CMPs likely results in particles without amorphous layers, and hence CMPs are composed of a higher concentration of crystalline material explaining their higher SHG intensity. With twice the crystalline material according to the degree of crystallinity, we expect 4× the SHG signal in dry CMPs as compared to granules, however the variation is closer to 10× (see Fig. 1(d)), and hence additional differences likely occur. It is sensible that the 9-10 nm periodicity in starch, composed of alternating crystalline and branching regions of amylopectin, and observed with small angle XRD scattering [24–26], restricts the size of individual crystalline domains to ∼10 nm long radial structures, resulting in a more uniform radial structure. Furthermore, it is unclear if the branching regions of semi-crystalline amylopectin bands contribute to starch SHG. In CMPs crystallites are not restricted by rings, and hence they can grow larger, explaining the additional signal variation. Furthermore, the observation that the dry A-type CMPs have a higher SHG signal than the dry B-type CMPs may be explained by the slightly higher degree of crystallinity for the A-type CMPs.

It is interesting that both A- and B-type CMPs had higher SHG intensities when dried to ambient humidity conditions as compared to hydrated (Fig. 1(d-h) and Table 1), since previous reports showed that maize, potato and barley starch all have higher SHG signal intensities when hydrated as compared to dried [12,21]. Higher SHG intensities from dried CMPs was statistically significant and is here attributed to the aforementioned lack of semi-crystalline rings which limit crystalline domain size. In starches, the connected domain of crystallites in a ring allows even swelling of each layer when hydrated, strengthening the alignment of domains by increasing the content of bound water stabilizing intra- and interhelical hydrogen bonds, and hence it increases the SHG intensity. In CMPs, the crystalline domains are likely isolated and not in rings due to their self-assembly synthesis, and hence domains are not adjacent to neighboring domains, thus crystalline domains swell at different rates and orientations, losing their ability to contain increased structured water between them, preventing SHG intensity from increasing, and reducing alignment with each other. The presence of water may also disrupt larger crystallites in CMPs. This is supported by PSHG data where the average $\rho $ does not change much between the wet and dry CMPs and explains the different behavior of CMPs compared to starches.

4.3 PSHG parameters of CMPs are biased near nucleation regions

The ultrastructural organization of the CMPs was also investigated with PSHG microscopy, which provides an SHG intensity independent parameter set. Two PSHG parameters were obtained for each pixel by fitting using Eq. (1), resulting in an effective susceptibility ratio, $\rho $, and the effective in-plane orientation of the dominant crystal axis, $\delta $. In CMPs, the typical $\rho $ values are lower near the supposed nucleation centers (∼2), and higher near the periphery (∼6) (see Fig. 2(b1-b4)), similar to what is observed in starch granules from maize, potato and barley [15,22]. The $\delta $ parameter shows the effective crystal orientation at each pixel and indicates that the particles have a macroscopic radial structure with respect to supposed nucleation regions (Fig. 2(d1-d4)). Nucleation regions have very low SHG intensity (compare to Fig. 2(a1-a4)), often 100× less than the remaining CMPs material, similar to starch granules [17], and is here attributed to the same reason, due to a radially organized and hence centrosymmetric nucleation region which cancels SHG signal. At points surrounding the nucleation regions, intense SHG is emitted, however the measured $\rho $ parameter is altered from its true value because the laser focal volume encompasses SHG emitters with a cone of angles due to the radial macrostructure. The SHG signal from a variation of emitter angles inside a focal volume has been previously investigated, and it was found that as the cone angle increases, the effective measured $\rho $ value reduces to 3 [17], explaining the low $\rho $ values near nucleation regions inside CMPs. This indicates that the $\rho $ parameter is therefore biased in these regions and cannot be used to indicate microscale structure. Conversely, domains with high $\rho $ values of CMPs, which mostly occur away from nucleation regions, are attributed to highly aligned regions of crystallized maltodextrin, which mimics the crystal structure of crystalline material in starch, and hence results in similarly high $\rho $ values intrinsic to starch granules, similar to previous measurements in barley, maize and potato starch granules [15,22]. This is supported by the vector diagrams (Fig. 2(d1-d4)) of those regions which appear aligned.

4.4 Effect of hydration on the PSHG parameters of CMPs

The average $\rho $ values of hydrated and dry CMPs did not exhibit a statistically significant variation (Fig. 3(a)), which is different from two starch granule studies, where higher $\rho $ values were observed in dry maize and barley starch granules. The difference is attributed to averaging the $\rho $ parameter across spherulitic particles, which biases the result. In one study, the average difference in $\rho $ between hydrated and dry maize starch was statistically significant (4.7 versus 5.7) [11,20], and in another publication on barley starch granules, higher $\rho $ values could also be observed in dry versus hydrated conditions (compare Fig. 2 b2 and b5 in [22]) even though the average $\rho $ values were similar [15,22]. The higher average $\rho $ variation between hydration conditions in maize as compared to CMPs and barley starch likely occur because larger granules with a diameter of 20 μm were investigated, and hence they have more periphery regions that tend to have a higher $\rho $ value, as compared to 10-15 μm barley granules or 3-8 μm CMPs, where more area is encompassed by nucleation regions which reduce $\rho $ values within their radius. Hence increased particle diameter can have the effect of artificially increasing the average value of $\rho $, and this creates a bias when comparing the average $\rho $ values in particles of different sizes.

A significant difference was also observed in the width of $\rho $ histograms (Fig. 3(b)) between hydrated and dry CMPs. Larger widths of $\rho $ were previously also observed in dry maize and barley starch granules [15,22], similarly implying that dry CMPs have more domains of different hydrations. However, due to the aforementioned bias on the low $\rho $ values around nucleation regions, the $\rho $ distribution width is biased on one side, hence two new metrics were devised to quantify granule structure, termed the ratio of ordered pixels (ROP) and the ratio of disordered pixels (RDP) (see section 2.4 for the definition). Analysis revealed that dry CMPs had a higher ROP than hydrated CMPs (Fig. 3(c)), with higher significance than $\rho $ width, while the RDP were similar (Fig. 3(d)), suggesting ROP is a more useful metric for discriminating hydrated CMPs, and indicates that dry CMPs do have more aligned domains, similar to starch granules.

4.5 Degree of the linear polarization of CMPs

The average DOLP of the emitted SHG was also determined from the same PIPO SHG data sets. The DOLP parameter indicates that the SHG is increasingly linearly polarized as the DOLP goes to 1 and becomes increasingly depolarized or circularly polarized as the DOLP approaches zero. Further, by assuming that birefringence is negligible, and that chirality does not play a role in starch SHG, the DOLP indicates the relative amount of coherent signal and incoherent scattering [27–29]. Unlike in some other studies where the DOLP was measured at a single input polarization, which induces an orientation bias, here the DOLP was calculated as the average of DOLPs for 8 equally spaced input polarizations. The average DOLP values for all CMPs were found to be high (>0.8), similar to previously observed hydrated potato starch granules [30–32]. The average DOLP values of dry CMPs were found to be significantly lower than the average DOLP values of hydrated CMPs (Fig. 3(e)), and this change could be attributed to increased domains with variations in hydration in dry CMPs, which induces differences in refractive index and results in increased scattering. However, due to the small size of the CMPs, scattering may not play such a significant role, and alternatively, the reduced DOLP indicates that a laser focal volume in dry CMPs has more smaller SHG emitting domains oriented in different directions within a single focal volume.

4.6 Effect of CMPs crystal type on SHG

Several SHG parameters differed between A- and B-type CMPs. For instance, hydrated B-type CMPs emitted more SHG signal than hydrated A-type CMPs and is here attributed to B-type CMPs having increased structured water content as compared to the A-type. It does not correspond with earlier results that hydrated potato starches (B-type) have lower SHG signal than maize (A-type) [15], but perhaps that difference is more due to species than crystal type. On the other hand, dry A-type CMPs emitted more SHG signal than B-type. This interesting observation is attributed to A-type having a higher density, and mostly glucan-water bonds holding its water tighter than B-type which uses water bridges to hold its network of structured water together inside helical cavities absent in A-type. The width of the $\rho $ parameter was also greater in B- than A-type in dry CMPs, supporting the idea of a higher heterogeneity of hydration domains in B-type, and sensible since the structure can hold more ordered water. This is supported by the higher ROP for dry B-type CMPs as compared to dry A-type CMPs. Additionally, the DOLP was higher in hydrated B-type CMPs. This can be attributed to A-type CMPs having more smaller domains than B-type which is not predicted by the higher $\rho $ distribution of the B-type, or perhaps the higher water content in B-type reduces Fresnel scattering to account for this difference.

5. Conclusions

Synthesized CMPs have many PSHG properties in common with starch granules making them useful model systems for PSHG microscopy. CMPs had intense SHG emission properties, owing to well-ordered crystalline domains, and they had a spherulitic 3D macrostructure similar to natural starches, with radially ordered SHG emitters with respect to nucleation regions. CMPs had an artificial reduction in $\rho $ around nucleation regions also similar to starch granules owing to the radial ultrastructure. The PSHG properties of CMPs were investigated as a function of hydration and crystal type, with significant effects similar to starch granules, including a decrease in $\rho $ heterogeneity, observed via reduced $\rho $ width during hydration. Interestingly, unlike starch granules, which have higher SHG intensities when hydrated, CMPs have higher SHG intensities when dry, indicating that water might not be well-arranged or not as much water can be absorbed in the CMPs. The reduction in $\rho $ near nucleation regions biases the average $\rho $ parameter, which is typically reported, and as a result, the ROP parameter was developed and revealed higher order statistically significant differentiation between wet and dry states can be observed as compared with the $\rho $ average or distribution width parameters. Lower DOLP values were observed with decreased hydration, attributed to increased scattering or higher variation of emitter orientations within focal volumes. Some variations in PSHG parameters between A- and B-type CMPs could be explained by B-type having more water content, as in higher SHG intensity when hydrated, higher $\rho $ width and higher ROP when dried. The DOLP had a higher magnitude in hydrated B-type CMPs than A-type due to reduced in-pixel heterogeneity of B-type and possibly reduced Fresnel scattering due to higher water content as well. Therefore, the utilization of CMPs as model systems for starch granules offers additional understanding of the ultrastructure of CMPs, starch granules, and possibly other hybrid spherulitic structures which may be of potential interest to the pharmaceutical industry.