Characterization of Synthetic Hydroxyapatite Fibers Using High-Resolution, Polarized Raman Spectroscopy

In the Raman spectrum of B-type carbonated apatites, the ν1 CO32– mode (at ∼1070 cm–1) overlaps the ν3 PO43– band. The latter is readily observed where the CO32– content is low (up to ∼3 wt%). The CO32– content of bone is considerably higher (∼7–9 wt%). As a result, the ν3 PO43– band becomes completely obscured. The 1000–1100 cm–1 spectral range of carbonated apatite is frequently considered a combined ν3 PO43– and ν1 CO32– region. Here, high-resolution polarized Raman spectroscopy (step size of 0.74 ± 0.04 cm–1) provides new insights into synthetic hydroxyapatite (HAp) obtained as micrometer-sized fibers. Compared to bone mineral (deproteinized bovine bone), spectral features of HAp fibers are highly resolved. In particular, the ν3 PO43– band resolves into nine distinct sub-components: 1028, 1032, 1040, 1043, 1047, 1053, 1055, 1062, and 1076 cm–1. Parameters including full width half-maximum, intensity, area fraction, intensity ratio, and area fraction ratio vary between parallel and perpendicular polarized configurations. It is likely that the ν1 CO32– band of B-type carbonated apatites may contain a small but not insignificant contribution from the 1076 cm–1 sub-component of the ν3 PO43– band. Furthermore, the 1076 cm–1/1047 cm–1 ratio changes between parallel and perpendicular scattering configurations, suggesting that the contribution of the 1076 cm–1 sub-component may vary as a function of local orientation of bone mineral, thus skewing the ν1 CO32– band and compromising accurate estimation of carbonate-to-phosphate ratios in B-type CO32– substituted apatite.


Introduction
Synthetic-and natural-derived calcium phosphates, e.g., hydroxy-or hydroxy(l)apatite (HAp), are by far the most frequently used biomaterials for bone repair. 1 Naturally occurring (geological and biological) apatites can incorporate a plethora of anionic, cationic, and anionic complex substitutions. 2 Likewise, ion substitutions can be easily achieved in synthetic apatites in order to tailor chemical stability or degradation behavior, and bone/biological response. 3 Raman spectroscopy is frequently employed for nondestructive assessment of bone quality, 4,5 and has been used extensively to study carbonate substitution in bone mineral, 6 which is considered an important marker of bone turnover. 7 In a typical Raman spectrum of B-type carbonated apatites (where CO 3 2substitutes for PO 4 3-), the symmetric stretching n 1 CO 3 2-(b 2g ) mode overlaps the antisymmetric stretching n 3 PO 4 3-(a 1g ) band. The latter is observed up to $3 wt% CO 3 2but tends to be completely enveloped by the n 1 CO 3 2peak in bone, 8 where the CO 3 2content is significantly higher ($7-9 wt%). 9 CO 3 2substitution for PO 4 3-influences physical properties including crystallite size, solubility, and thermal stability of biological apatites, 10 in effect restricting mineral crystallinity to below that observed for carbonate-free apatites. Polarized Raman spectroscopy has been used previously to investigate orientation dependence of bone mineral. [11][12][13] Attributed to A-and B-type n 1 CO 3 2bands, previous investigations of scattering configuration dependence of the n 3 PO 4 3mode in carbonated fluorapatite have been unsuccessful in precisely locating the positions of individual subcomponents. 14 Using high-resolution polarized Raman spectroscopy, this work investigates micrometer-sized fibers of carbonate-free synthetic HAp and carbonated apatite of bovine cortical bone to better understand the n 3 PO 4 3and n 1 CO 3 2overlap in B-type CO 3 2substituted apatite such as bone mineral in order to enable more accurate estimation and interpretation of carbonate-to-phosphate ratios.

Synthetic HAp Fibers and Bovine Bone Mineral
The HAp fibers were obtained by dispersing 5-80 nm apatite particles in an aqueous solution of 200 kDa pullulan (Sigma-Aldrich). This solution was extruded at 1.4 kg/cm 2 pressure while simultaneously discharging air at 250 m/s to form a stream of fibers. Using a far-infrared heater under the extrusion nozzles, the fiber stream was heated at 400 C and blown onto a screen conveyor belt to produce a non-woven fabric that was further heated at a rate of 50 C per hour and calcined at 1100 C for one hour. 15 Commercially sourced (https://boneslices.com) bovine cortical bone stored in 96% ethanol was deproteinized using 5% sodium hypochlorite (Honeywell Fluka) for 16 h at 4 C in order to isolate the inorganic/mineral phase.

Scanning Electron Microscopy and Micro-Raman Spectroscopy
The HAp fibers were visualized using scanning electron microscopy (Ultra 55 FEG SEM, Leo Electron Microscopy Ltd, UK) in the secondary electron mode. Micro-Raman spectroscopy was performed using a confocal Raman microscope (Renishaw inVia Qontor) equipped with a 633 nm laser and LiveTrack focus-tracking technology. The laser was focused down on to the sample surface using a Â 100 (0.9 NA) objective. In the 390-1100 cm -1 spectral range (step size of 0.81 cm -1 at 390 cm -1 and 0.68 cm -1 at 1100 cm -1 ), the Raman scattered light was collected using a Peltier-cooled charge-coupled device deep depletion nearinfrared-enhanced detector behind a 2400 g mm -1 grating, 10 s integration time, and 10 accumulations. Using a halfwave plate and polarization analyzer, spectra were collected in parallel zðxxÞz and perpendicular zðxyÞz polarized configurations from 3 Â 3 point grid-matrices (n ¼ 8) for HAp fibers and from 9 Â 8 point grid-matrices (n ¼ 1) for deproteinized bovine bone. The laser power at the sample was $15 mW. Background subtraction and cosmic ray removal were performed in Renishaw WiRE 5 software. Spectra from each grid-matrix were averaged and normalized to the 428 cm -1 n 2 PO 4 3sub-component. For HAp fibers, one Lorentzian curve was fitted to the n 1 PO 4 3peak ($962 AE 10 cm -1 ). Centered at inflection points in calculated second-derivative spectra, n 3 PO 4 3band (1020-1090 cm -1 ) sub-components were resolved by fitting multiple Lorentzian curves (r 2 > 0.99). For each sub-component, the full width half-maximum (FWHM), intensity, and area fraction of the fitted range were obtained. Polarization configuration dependent variation in intensity ratios and area fraction ratios between n 3 PO 4 3sub-components were expressed as percentage difference values.

Statistical Analysis
For statistical analysis, the non-parametric Wilcoxon signed-rank test was used (SPSS Statistics, v.25, IBM Corporation) and p values < 0.05 were considered statistically significant. Mean values AE standard deviations are presented.
Between scattering configurations, n 3 PO 4 3sub-component positions remained consistent, while FWHM, intensities, and area fractions of several sub-components vary ( Fig. 3), which alters the intensity ratios and area fraction ratios between individual sub-components. It is, therefore, demonstrated that n 3 PO 4 3band profile/shape is not conserved between polarization configurations and that the individual sub-components exhibit dissimilar levels of polarization dependency. Of particular significance is the relationship between 1076 cm -1 and 1047 cm -1 sub-components, which are two of the strongest features of the n 3 PO 4 3band. Attributable to increases in 1076 cm -1 subcomponent intensity and area fraction, the 1076 cm -1 / 1047 cm -1 ratios differ by $68-69% (p ¼ 0.012) between the polarization configurations (Fig. 4).
Biological apatites contain varying amounts of CO 3 2-. The CO 3 2content of bone is taken as an indicator of bone maturation and turnover, 7,17 and considerable attention is given to the carbonate-to-phosphate ratios of bone affected by compromised systemic conditions [18][19][20] and bone surrounding implant biomaterials. 21,22 It is likely that the 1076 cm -1 sub-component of the n 3 PO 4 3band contributes to the n 1 CO 3 2band, but remains a challenge to  deconvolute. The scattering configuration dependency of the 1076 cm -1 /1047 cm -1 ratio is also noteworthy. The contribution/intensity of the 1076 cm -1 sub-component may vary as a function of local, sub-micrometer/nanoscale orientation of bone mineral, thereby skewing the shape, intensity, and/or integral area of the n 1 CO 3 2band, thereby compromising the accuracy of carbonate-to-phosphate ratios in Btype CO 3 2substituted apatite.

Conclusion
The n 3 PO 4 3band in carbonate-free HAp fibers resolves into at least nine sub-components in the 1020-1090 cm -1 spectral range. Sub-component FWHM, intensities, area fractions, and intensity and area fraction ratios vary between zðxxÞz and zðxyÞz configurations. The sub-components at 1028, 1032, 1047, 1055, and 1076 cm -1 exhibit pronounced sensitivity to the polarization configuration. A noteworthy change is in the 1076 cm -1 /1047 cm -1 ratio between parallel and perpendicular polarized configurations. It is likely that the 1000-1100 cm -1 range of Btype carbonated apatites, containing the n 1 CO 3 2band, also includes a small contribution of the 1076 cm -1 subcomponent of the n 3 PO 4 3band, which may vary as a function of local, sub-micrometer/nanoscale orientation of bone mineral, thereby skewing the shape, intensity, and/or integral area of the n 1 CO 3 2band, thereby compromising the accuracy of carbonate-to-phosphate ratios in B-type CO 3 2-  substituted apatite. For this reason, carbonate-to-phosphate ratios determined using Raman spectroscopy must be interpreted with caution.

Acknowledgments
HAp fiber samples were kindly provided by Dr. Takayuki Miyahara.

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
This work was supported by the Svenska Sällskapet fö r Medicinsk Forskning (SSMF) postdoctoral scholarship, the Adlerbertska Foundation, the IngaBritt and Arne Lundberg Foundation, and the Hjalmar Svensson Foundation.