| |

Received: 1 March 2021    Revised: 7 April 2021    Accepted: 7 April 2021

DOI: 10.1111/jace.17853

ORIGINAL ARTICLE

Development of microfibers for bone regeneration based on

alkali-­free bioactive glasses doped with boron oxide

Anuraag Gaddam1   | Przemysław Gołębiewski2,3   | Hugo R. Fernandes1   |

Dariusz Pysz2  | Ana S. Neto1   | Ryszard Diduszko2   | Agnieszka Malinowska2   |
Ryszard Stępień2   | Jarosław Cimek2   | Ryszard Buczyński2,3   | José M. F. Ferreira1

1
CICECO –­Aveiro Institute of Materials,
Department of Materials and Ceramic Abstract
Engineering, University of Aveiro, In this paper, we present a new series of alkali-­free bioactive glasses (BG) based
Aveiro, Portugal
2
on FastOs® composition (38.49 SiO2 –­36.07 CaO –­19.24 MgO –­5.61 P2O5 –­0.59
Łukasiewicz Research Network
–­Institute of Microelectronics and
CaF2, expressed in mol %), which was modified by partially replacing silicon diox-
Photonics, Warsaw, Poland ide network-­former with boron trioxide network-­former, utilizing calcium oxide as a
3
Faculty of Physics, University of charge compensator. The main objective of this study was to obtain a new family of
Warsaw, Warsaw, Poland
bioactive glasses suitable for the fabrication of glass fibers. The BGs were prepared
Correspondence by melt quenching technique and their structural and thermal properties were deter-
Anuraag Gaddam and José M. F. Ferreira, mined. Glass rods were used to obtain fibers by the classic drawing technique. The
CICECO –­Aveiro Institute of Materials,
bioactivity of the fibers was subsequently assessed through immersion tests in simu-
Department of Materials and Ceramic
Engineering, University of Aveiro, lated body fluid (SBF) to establish their ability to form hydroxyl carbonated (HCA)
Santiago University Campus, 3810-­193 apatite onto their surfaces. Glasses with moderate substitution of SiO2 with B2O3
Aveiro, Portugal.
Email: anuraagg@ua.pt (A. G.); jmf@
exhibited enhanced thermal properties, allowing to significantly suppress the crystal-
ua.pt (J. M. F. F.) lization trend, and favoring to draw the fibers. The structure of the studied glasses was
obtained by NMR spectroscopy. The structure-­property correlations were established
Funding information
Fundação para a Ciência e a by their relationship to the configurational entropy. Smaller amounts of substitution
Tecnologia, Grant/Award Number: resulted in larger entropy of the glasses. Moreover the SBF tests revealed an extensive
6818 -­311, UIDB/50011/2020
formation of HCA, comparable to the parent FastOs®BG composition, which assures
and UIDP/50011/2020; Centro de
Investigação em Materiais Cerâmicos fast bonding to the bone. Thus, presented glass fibers may be considered as promis-
e Compósitos, Grant/Award Number: ing materials for wool-­like bone implants or as reinforcing constituent of biopolymer
022161; Narodowa Agencja Wymiany
matrix composites.
Akademickiej, Grant/Award Number: PN/
BIL/2018/1/00253
KEYWORDS
bioactive glass, fibers, structure, structure-­properties

1  |   IN T RO D U C T IO N soft tissues.1,2 For over 50 years, researchers explored variety

Since 1969, when Larry L. Hench discovered 45S5 glass
(also known as Bioglass®) with nominal composition of: 46.1
SiO2, 2.6 P2O5, 24.4 Na2O, 26.9 CaO (in mol %), glass mate-
rials have been considered very promising for biomedical ap-
plications with their outstanding ability to bond to bone and
of glass compositions, not only silicate3,4 but also borate-­5,6
and phosphate7-­based glasses. However, only two composi-
tions: 45S5 (NovaMin®, Biogran®, Novabone®, Perioglas®)8
and S53P4 (53% SiO2, 4% P2O5, 23% Na2O, 20% CaO (in
wt. %), also named BonAlive®)9,10 succeeded commercially.
Nevertheless, both above-­mentioned materials exhibit several

|
4492    ©
wileyonlinelibrary.com/journal/jace
2021 The American Ceramic Society J Am Ceram Soc. 2021;104:4492–4504.
GADDAM et al.
drawbacks. 45S5 glass may have a negative effect on remod-
eling of bone as its dissolution and resorption rates are rea-
sonably fast, thus leading to the formation of a gap between
the implant and the bone.11 Moreover its high sodium oxide
content and the easiness of sodium ions’ leaching leads to sig-
nificant pH increase, causing cytotoxicity.12,13 Finally, 45S5
exhibits poor sintering ability due to its very small difference
between Tg and the onset of crystallization temperature (Tc),
also called thermal stability (ΔT = Tc − Tg), 132°C.14 As a re-
sult, thermal processing of Bioglass® is difficult and usually
leads to poor densification of, for example, 3D printed porous
scaffold causing poor mechanical properties. These scaffolds
are also partially crystallized which tends to hinder glass
bioactivity.15 In turn, S53P4 exhibits slower dissolution rate
and better thermal stability as a result of its higher network
connectivity (NC). Nevertheless, clinical test proved that its
dissolution rate is too slow in the case of large defect sizes,
and remnants were observed even 14  years post-­surgery.16
Attempting to overcome these drawbacks, a different con-
cept was recently proposed by Ferreira et al.,17 aiming at
developing glasses with network connectivity similar to that
of 45S5, while avoiding alkali oxides in glass structure that
are responsible for cytotoxicity. Accordingly, a new series of
alkali-­free bioactive glasses based on ternary diopside (Di)
–­fluorapatite (FA) –­tricalcium phosphate (TCP) system
were developed.17 In this series, TCP-­20 glass (70 Di –­10
FA –­20 TCP, also known as FastOs®BG) showed outstand-
ing properties. Its thermal stability allowed in obtaining fully
dense and amorphous compacts after sintering at 800°C.
Furthermore, the formation of hydroxyl carbonated apatite
(HCA) was detected just after 1 h of immersion in simulated
body fluid (SBF), anticipating a fast-­bonding ability to hard
and soft tissues in comparison to other previously reported
bioactive glasses. Moreover amorphous sintered compacts
of FastOs®BG powder provided proper conditions for adhe-
sion and proliferation of mesenchymal stem cells (MSCs).17
Further studies revealed superior stimulation for differentia-
tion of human MSCs when compared to Bioglass®.18 Finally,
in vivo tests in a sheep animal model showed slower resorp-
tion and higher osteointegration and osteoconduction than
observed for 45S5 used as control.19 In conclusion, it can be
stated that in the field of implants for large bone defects, the
potential application of FastOs®BG is even greater than that
of 45S5 Bioglass®.
Bioactive glasses may be incorporated into human skel-
eton in different ways. Bone defects can be filled with glass
graft (single piece of glass), powder as a part of cement,20
cotton-­wool-­like form21 or as composites with bioresorbable
polymers.22,23 The composites may incorporate glasses either
as powders or in the fibers’ form. The fibers’ drawing re-
quires glasses with an extended working range to prevent the
occurrence of crystallization. Achieving this target implies
compositional changes towards increasing glass network
  
|
   4493
connectivity, partially compromising their bioactivity.24 One
of the first bioactive glasses successfully used for fiber draw-
ing was the composition labeled as 13–­93 (53 SiO2, 6 Na2O,
12 K2O, 5 MgO, 20 CaO, 4 P2O5, in wt. %).24,25 The partial
replacement of SiO2 in 13–­93 bioactive glass with B2O3 en-
hanced the degradation rate and the surface mineralization of
trabecular scaffolds immersed in simulated body fluid (SBF).
However, high B2O3 contents of the glass severely reduced
the ability of the scaffolds to support proliferation and func-
tion of osteogenic MLO-­A5 cells during conventional in vitro
cell culture studies.26
In this study, we investigated the possibility of drawing
FastOs®BG-­based bioactive glass fibers from glass rods by
the same classic method used for drawing glass fibers for
photonic applications. After setting proper parameters, this
process allows us to draw a large quantity of fibers with
very precise control of their diameters. However, this pro-
cess demands high thermal stability of the glasses to avoid
crystallization. As mentioned above, early crystallization
leads to poor mechanical properties of obtained fibers, and
in case of bioactive glasses it also results in decrease of bio-
activity.15 Aiming at further enhancing ΔT, a new series of
alkali-­free bioactive glasses based on the FastOs®BG com-
position was prepared, in which silica was partially replaced
by boron oxide and calcium oxide. Additional calcium oxide
is needed to maintain NC at the same level, as boron oxide
will react with alkaline earth oxides to form fourfold coor-
dinated (BIV) units. Boron oxide will increase the entropy of
the investigated bioactive glasses and is expected to improve
their thermal properties.27,28 Furthermore, boron oxide con-
centration may also help adjusting the glass dissolution
rate and tune the glass bioactivity.29,30 Addition of boron to
the glass may also induce positive effects on angiogenesis
and osteogenesis as suggested elsewhere.31–­33 The struc-
ture of glasses was assessed by Fourier Transform Infrared
Spectroscopy (FTIR) and Nuclear Magnetic Resonance (11B
NMR, 29Si NMR, and 31P NMR). The thermal properties
of obtained glasses were assessed by differential scanning
calorimetry (DSC) and dilatometry (DIL) measurements.
The ability of BGs fibers to form hydroxyl carbonated apa-
tite on their surfaces in vitro was investigated by immersing
them in SBF. The surface changes occurred upon immersion
of the fibers in SBF were assessed using X-­ray diffraction
(XRD), FTIR spectra, and Scanning Electron Microscopy
(SEM).
2  |  EXPERIM ENTAL
A series of alkali-­free, silica-­based bioactive glasses with a
general formula: (38.49 − 2x) SiO2 · (x) B2O3 · (36.07 + x)
CaO · 19.24 MgO · 5.61 P2O5 · 0.59 CaF2 (where x = 0.25,
0.50, 1.00, 3.00, and 5.00 mol.%) was prepared by the melt
 |
4494       GADDAM et al.

quenching technique. Glass samples were accordingly named
B2x, where x corresponds to the amount of B2O3 in the glass
composition. FastOs®BG glass was the base glass compo-
sition with label B0.0. All glass compositions are presented
in Table 1. Starting powder reagents of high purity were
used, which are: SiO2 (>99.5%), CaCO3 (>99.5%), MgCO3
(BDH Chemicals Ltd, >99.0%), NH4H2PO4 (Sigma–­Aldrich,
>99.0%), and CaF2 (Sigma Aldrich, 325 mesh, >99.9%).
Weighed powders were mixed carefully using mortar and
pestle and placed in Pt-­Rh crucibles and calcined at 900°C
for 1  h (heating rate 3°C  min−1) in the air. The calcined
batches were reground and melted at 1570°C for 1 h (heating
rate 10°C min−1). Finally, the melts were cast into a copper
mold to form bars and transferred to an annealing furnace
that was preheated to 800°C, followed by 1 h of dwelling at
this temperature, and then cooled down slowly.
DIL measurements were performed using bulk cuboid-­
shaped glass samples (10 mm in length, cross section of c.
3  ×  3  mm), using a DIL 801 Bähr Thermoanalyse GmbH
apparatus. All the measurements were carried out in air from
room temperature to 800°C with a heating rate of 5°C min−1.
DSC measurements were performed from 25°C to 1300°C
at a heating rate of 10°C  min−1 using Netzsch STA 449F1
calorimeter. All measurements were carried out in pure argon
flow utilizing alumina crucibles.
Glass rods were drawn into fibers’ form at fiber-­drawing
tower. Temperature, feeding speed, and pulling speed were
adjusted for each glass sample based on the data obtained
from DIL and DSC measurements. Using this method glass
fibers within 150−600 μm range could be obtained for each
glass composition.
One meter of each fiber type was cut into small pieces and
immersed in SBF solution (prepared according to ISO 23317
standard) up to 14  days in an incubator with shaking plat-
form at 37°C. Based on the diameter of each fiber, measured
using micrometer screw gauge, the amount of SBF was cal-
culated, considering the surface area to volume of SBF ratio
of 0.5  cm2/ml as suggested by Popa et al.34 Aged samples
were filtrated using 63-­micron polymer sieve, washed with
deionized water, and slowly dried on a Petri dish in a closed
desiccator filled with silica gel at the bottom. Afterward, the
intact fibers were analyzed by XRD. The surface morphology
of fibers before and after in vitro testing was observed using
SEM Hitachi SU-­70. Before observation, all samples were
covered with carbon using sputter coater.
The FTIR measurements were performed using Vertex
80v FTIR spectrometer (Bruker). Glass samples, as-­drawn
glass fibers, and fibers after immersion in SBF were ground
in order to obtain fine powders. The as obtained powders were
then mixed with KBr in 1:30 weight ratio. All FTIR measure-
ments were carried out at room temperature within the range
of 360–­4500 cm−1 with a resolution of 2 cm−1. Obtained data
were examined using OPUS software (Bruker).
The 29Si and 31P MAS-­NMR spectra were collected on a
Bruker Avance III HD 400 MHz (9.4 T) spectrometer, while
11
B MAS-­ NMR spectroscopy was performed on Bruker
Avance III HD 700  MHz (16.4  T) narrow-­bore spectrome-
ter. For 29Si MAS-­NMR spectra, the Larmor frequency of
79.5 MHz was used while samples were spun at 12 kHz in
7  mm ZrO2 rotors and the nuclei were excited with a 10°
pulse at relaxation delay times of 60 s. For 31P MAS-­NMR
spectra, the Larmor frequency of 161.9 MHz was used while
samples were spun at 5 kHz in 4 mm ZrO2 rotors and the nu-
clei were excited with a 90° pulse at relaxation delay times of
60 s. For 11B MAS-­NMR Hahn echo spectra, the Larmor fre-
quency of 224.6 MHz was used while samples were spun at
15 kHz in 4 mm ZrO2 rotors and the nuclei were excited with
a 90° soft pulse at relaxation delay times of 2 s. The samples,
Si(CH3)4 (0  ppm), H3PO3 (0  ppm) and H3BO3 (19.6  ppm),
were used as chemical shift references for 29Si, 31P and 11B,
respectively. The NMR spectra were decomposed using the
DMFit software.35
The presence of crystalline phases in bulk glasses was
evaluated by XRD. Measurements were performed using a
Rigaku SmartLab 3  kW X-­ray diffractometer with Cu Kα
(λ = 1.5418 Å) radiation and silicon semiconductor strip (D/
teX Ultra 250) detector. Diffraction patterns of glass samples
were recorded in the Bragg-­Brentano reflection geometry
(θ/2θ scan) and in continuous scanning mode. Qualitative
phase analysis of crystalline phases (if observed) was per-
formed using PDF4+ 2019 database and PDXL2 Software
supplied by Rigaku.

T A B L E 1   Compositions of the glasses employed in the current

study
3  |  RESULTS
mol.% B0.0 B0.5 B1.0 B2.0 B6.0 B10.0 The experimental conditions used in the preparation of
SiO2 38.5 38.0 37.5 36.5 32.5 28.5 glasses resulted in amorphous samples except for B10.0,
CaO 36.1 36.3 36.6 37.1 39.1 41.1 which showed a small amount of devitrification due to sur-
MgO 19.2 19.2 19.2 19.2 19.2 19.2
face crystallization (Figure 1). The crystalline phase was
identified as hydroxyapatite substituted by silicate units.36
B2O3 0.0 0.3 0.5 1.0 3.0 5.0
But the samples with less than 10.0 mol% B are thermally
P2O5 5.6 5.6 5.6 5.6 5.6 5.6
more stable. Notwithstanding, the B10.0 sample was still
CaF2 0.6 0.6 0.6 0.6 0.6 0.6
considered in the rest of the study even with the small
GADDAM et al.
amount of surface crystallization, which is only confined to
the surface of bulk cast glasses and could be mechanically
removed.
F I G U R E 1   X-­ray diffraction patterns of the glasses. Glass B10.0
shows small crystallization, which is identified as hydroxyapatite
substituted with silicate group (Ca10(PO4)4.92(SiO4)1.08(OH)O0.664,
ICDD ref. code 01-­080-­6260)
F I G U R E 2   Stacked plots of (A) DSC
and (B) DIL curves of the glasses. (ΔL is the
change in length from the original length,
Lo)
  
|
   4495
3.1  |  Thermal characterization
Figure 2A presents DSC curves of the glasses and it can
be noticed that all glasses show two glass transition tem-
peratures, viz., Tg1 and Tg2. The presence of the dual glass
transition temperatures is due to the occurrence of liquid-­
liquid phase separation (LLPS) in the current glasses. The
LLPS was confirmed by SEM observation of a polished
and acid etched (with 2 vol% HF solution) B0.0 glass sam-
ple (Figure S1). The DSC characteristic temperatures, Tg1,
Tg2, Tc (crystallization onset temperature), Tp (peak crys-
tallization temperature), and Tm (melting point) were ob-
tained according to the procedure presented in the Figure
S2a; which was repeated 10 times to obtain averages and
standard errors. From which, the parameters, glass stability
(ΔT1, ΔT2) and Hrubý (KH1, KH2)37,38 were calculated ac-
cording to the formulae,
ΔT1 = Tc − Tg1 ; ΔT2 = Tc − Tg2 (1)
( ) ( )
Tc − Tg1 Tc − Tg2
KH1 = ( ) ; KH2 = ( ). (2)
Tm − T c Tm − T c
All the thermal parameters obtained from DSC are pre-
sented in Table 2. The DIL curves of the glasses are pre-
sented in Figure 2B. As glasses are isotropic materials, the
linear (αL) and volumetric (αV) thermal expansion coeffi-
cients are related to each other by, αV = 3αL. The αV values,
dilatometric glass transition temperatures (Tg–­DL) and glass
softening points (Ts) were obtained by DIL. The αL values
were determined by a best-­ fit line between the tempera-
tures 100−650°C, and 500−650°C. The Tg–­DL values were
 |
4496       GADDAM et al.

T A B L E 2   Thermal and physical properties of glasses

B0.0 B0.5 B1.0 B2.0 B6.0 B10.0

Tg1 (°C) 716.8 ± 0.3 715.8 ± 0.3 711.6 ± 0.5 698.9 ± 0.2 680.0 ± 0.5 672.6 ± 0.2
Tg2 (°C) 821.8 ± 0.2 825.8 ± 0.2 819.0 ± 0.2 812.3 ± 0.3 794.0 ± 0.3 777.8 ± 0.2
Tc (°C) 900.8 ± 0.3 906.7 ± 0.2 900.2 ± 0.3 886.6 ± 0.3 862.4 ± 0.3 830.8 ± 0.1
Tp (°C) 927 932 928 916 887 859
Tm (°C) 1274.1 ± 0.2 1255.4 ± 1.5 1247.6 ± 0.1 1257.7 ± 0.3 1208.0 ± 0.6 1138.1 ± 0.6
ΔT1 (°C) 184.0 ± 0.6 190.9 ± 0.5 188.6 ± 0.8 187.7 ± 0.5 182.4 ± 0.8 158.2 ± 0.4
ΔT2 (°C) 79.0 ± 0.4 80.9 ± 0.4 81.2 ± 0.4 74.3 ± 0.6 68.4 ± 0.5 53.0 ± 0.3
KH1 0.493 ± 0.002 0.547 ± 0.004 0.543 ± 0.003 0.506 ± 0.002 0.528 ± 0.003 0.515 ± 0.003
KH2 0.212 ± 0.001 0.232 ± 0.002 0.234 ± 0.001 0.200 ± 0.002 0.198 ± 0.002 0.172 ± 0.002
–­1
ΔH (J g ) 214.2 251.1 239.8 255.3 279.6 227.6
–­1 −6 −6 −6 −6 −6
αV 100-­650 (K ) 30.5 × 10 29.6 × 10 31.3 × 10 29.1 × 10 30.5 × 10 34.2 × 10−6
–­1 −6 −6 −6 −6 −6
αV 500-­650 (K ) 33.3 × 10 32.7 × 10 32.8 × 10 31.9 × 10 33.7 × 10 37.8 × 10−6
Tg–­DL (°C) 728.0 ± 0.4 727.2 ± 0.3 721.8 ± 0.8 718.8 ± 0.4 697.6 ± 0.4 688.4 ± 0.9
Ts (°C) 766 765 755 750 730 719

The Ts were obtained taking the peak maximum of the curve.

The values of αV, Tg–­DL, and Ts are presented in Table 2.
Using the DSC curves, the configurational entropy
(ΔSconf.) of the samples as a function of temperature (T or T ′)
relative to Tg1, was calculated using the formula:

T liquid ( �) glass ( � )
cp T − cp T
∫
dT� (3)
( )
ΔSconf. T, Tg1 =
T�
Tg1

The parameters cp and cp are the specific heat capac-

liquid glass

ities for liquid and glass phases, respectively. At glass tran-

sition temperature, the sample changes from glass to liquid.
From DSC, the specific heat capacities for glass, cp , could
glass

be obtained at temperatures below Tg1. Extrapolating cp for

glass

temperatures above Tg1 and until Tc, the ΔSconf. can be calcu-
lated using Equation (3) in the range Tg1 to Tc. The calculated
values are presented in Figure 3. It can be seen that sam-
ples B0.0 and B1.0 exhibit larger values of ΔSconf. compared to
other samples, with sample B1.0 showing the largest values in
the entire range.

3.2  |  NMR and FTIR spectroscopy

F I G U R E 3   Configuration entropy of the samples calculated from
DSC: (A) normal scale, (B) reduced scale [Color figure can be viewed
at wileyonlinelibrary.com]
determined as intersection of two best fitting lines before
and after glass transition, an example of the procedure is pre-
sented in Figure S1b; this procedure was repeated 10 times
and the averages, and the standard errors were calculated.
Figure 4 presents NMR spectra of the glasses for the nuclei
29
Si, 31P, and 11B. The 29Si NMR spectra can give informa-
tion about the distribution of QnSi (mT). According to the
QnSi (mT) notation, which is a standard notation for silicate
units, QSi represents central Si tetrahedron, and n represents
the number of bridging oxygens (BOs) and m represents
the number of other tetrahedral or triangular units of type T
(P or B) attached to the central Si tetrahedron through the
GADDAM et al.   
|
   4497

T A B L E 3   Decomposition parameters of 31P NMR spectra and

their assignment

δiso width Amount

Type (ppm) (ppm) (%) Assignment
B0.5 Line 1 1.90 7.72 94.2 Q0P
Line 2 −5.87 7.00 5.8 Q1P
B1.0 Line 1 1.93 7.66 94.7 Q0P
Line 2 −6.00 7.00 5.3 Q1P
B2.0 Line 1 1.92 7.67 94.9 Q0P
Line 2 −5.83 7.00 5.1 Q1P
B6.0 Line 1 2.00 7.51 95.8 Q0P
Line 2 −5.31 7.00 4.2 Q1P

through charge dispersal effect40; or form BPO4 units similar

to AlPO4 units that mutually charge compensate.41 In both
scenarios, QnP (mB) type phosphate units are created. The 11B
nuclei are quadrupolar with spin 3/2 and thus experience sec-
ond order quadrupolar effects resulting in line broadening.
However, using the high magnetic field, such as in the cur-
rent study (16.4 T), the quadrupolar effects would be reduced
and useful structural information could be obtained. The 11B
MAS NMR could provide structural information about dif-
ferent three-­and four-­coordinated borate units given by the
notation, BIII and BIV, respectively.
In case of 29Si NMR spectra (Figure 4A), one can ob-
serve broad signals with maximum at c. 80 ppm, which cor-
responds to Q2Si (0T) units. The broad peak is skewed to the
right suggesting the presence of other QnSi (mT) units. With
increasing boron oxide and calcium oxide substitution, the
29
Si nuclei experience a deshielding effect causing an overall
shift of the spectrum towards left. This effect could be caused
due to the depolymerization of the silicate network or some
other complex phenomenon. For example, the formation of
every Si−O−BIV on a given Si tetrahedron, causes a desh-
ielding of( ~5 ppm of( 29Si )chemical(shift.)27,42 Therefore, the
units Q3Si 3BIV , Q2Si 2BIV and Q1Si 1BIV would experience
)

a deshielding of ~15, ~10, and ~5 ppm, respectively.

F I G U R E 4   NMR spectra of (A) 29Si, (B) 31P, and (C) 11B nuclei
[Color figure can be viewed at wileyonlinelibrary.com]
BOs. The 31P NMR spectra can give information about the
distribution of QnP (mT). The QnP (mT) notation for phosphate
units is slightly different from QnSi (mT) (for silicate units).39
Here, QnP (mT) represents a central phosphate tetrahedron that
is connected to n other phosphate units through BOs and m
represents the number of T (B or Si) tetrahedral or triangu-
lar units that are connected through the BOs. Moreover it is
known that phosphate units could charge compensate tet-
rahedral borate units (BIV) by forming P−O−BIV linkages
The results of 31P NMR (Figure 4B) show that in the in-
vestigated glass system, compositional changes have very
small influence on the phosphate structure as the spectra
look nearly identical with only small differences. All the
glasses show a slightly asymmetric spectra that are peaked
at ~1 ppm. This suggests that majority of the phosphate net-
work is composed of Q0P (0T) units. This is due to the high
affinity of phosphorus oxide (which is an acid according to
Lewis theory) towards alkaline earth oxides. Where, P2O5 re-
acts with CaO and/or MgO to form orthophosphates.43 The
slightly asymmetrical 31P NMR spectra was decomposed into
two gaussian lines (Lines 1 and 2) (Figure S3) and the results
of the parameters are presented in Table 3. It can be noticed
that with the increasing substitution of B units, the amount
 |
4498       GADDAM et al.

of Line 2 decreases with a corresponding increase of Line
1. The possible assignment of Lines 1 and 2 are presented in
Table 3 and is discussed in greater detail in the subsequent
sections.
Figure 4C presents results of 11B NMR, exhibiting two dis-
tinct broad peaks. These peaks at ~15 and ~2 ppm correspond
to BIII and BIV units, respectively. To quantify the amounts of
different units, the spectra were decomposed using minimum
number of lines that give the best possible fit. For BIII spe-
cies, two lines with second order quadrupolar effects were
used. For BIV units, one or two mixed Gaussian/Lorentzian
line shapes were used. The results of the decomposition: iso-
tropic chemical shift (δiso), Quadrupolar Coupling Constants
(CQ) and the chemical shift anisotropy (η) are presented in
Table 4 and Figure S2. From this, the total proportions of
BIII and BIV were calculated. Going from B0.5 to B6.0, the BIV
units experience a slight deshielding effect resulting in the
peak shifting to left. Consequently, the decomposition of 11B
NMR spectra show two peaks separated by ~0.5  ppm. The
variation in the 11B chemical shift in this region could be
caused by different BIV−O−P, BIV−O−Si, BIV−O−BIV, and
BIV−O−BIII bonds. The part of the 11B NMR spectrum cor-
responding to the BIII units was decomposed into two peaks
viz. BIII‒­a, and BIII‒­b. The BIII‒­a peak is the major peak with
δiso of ~19.3  ppm, according to the NMR parameters, this
peak could be assigned to T2 species, which are BIII units
with one NBO.44 The second peak for BIII units, BIII‒­b, has a
lower δiso value compared to BIII‒­a units and shifts to lower
values with increasing boron content. Moreover the chemi-
cal shift anisotropy of BIII‒­b peak decreases with increasing
boron content. This BIII‒­b peak could be assigned to T3 spe-
cies, which are BIII units with no NBO.44 However, the δiso
value of these units is much lower (~9  ppm) than what is
expected (~13 ppm). This could possibly be due to the forma-
tion of BIII−O−P bonds.45,46 The larger electron cloud on P
units would cause greater chemical shielding on these units.
Overall, with increasing boron content, the amount BIII units
increase relative to BIV units.
Figure 5A presents results of FTIR spectra of glasses.
All the glass compositions exhibit a broad band within wav-
enumber region from 850 to 1010  cm−1, due to the over-
lap of the vibrational bands of Q0Si (855  cm−1), Q1Si and Q2Si
(921 cm−1), Q2Si and Q3Si (1003 cm−1), and a band positioned
at ~1030–­1050  cm−1), attributed (Si–­O–­Si) in all QnSi sili-
cate species.34 Figure 5B shows spectra of boron containing
glasses with the spectra of B0.0 being subtracted from them.
It can be seen from Figure 5B that absorption bands grad-
ually increase with increasing boron oxide contents. The
absorption band with maximum at 1438 cm−1 corresponds
to B−O symmetric stretching vibrations of different borate
groups in BIII units. The intensity of the band at 1274 cm−1
increases, and this may be related to the asymmetric stretch-
ing vibrations of B−O bonds present in BIII units. Both above
mentioned bands suggest increasing contents of threefold
coordinated boron, which agrees with the 11B NMR spec-
tra. Furthermore, the absorption spectra exhibit a band with
maximum at ~724 cm−1, correlated with B−O−B bending
vibration. And the bands at 807 and 585 cm−1, correspond
to symmetrical vibrations in BIV units and bending vibration
O−B−O bridges, respectively. However, the last two bands
overlap with symmetrical SiO4 and O−Si−O bands respec-
tively, thus their intensity may be reduced.47 As a result of
decrease in silica molar fraction, one can also observe de-
creasing intensity of bands at 1172 and 454  cm−1, which
correspond to Si−O−Si stretching vibration and Si−O rock-
ing vibration, respectively.

T A B L E 4   Decomposition parameters of 11B NMR spectra

Amount
Sample Type δiso (ppm) CQ (GHz) η Amount Assignment Type (%)
B0.5 BIII‒­a 19.21 2.82 0.64 74 T2 BIII 84
III‒­b 3
B 10.14 2.69 0.10 10 T BIV 16
IV‒­a
B 5.49 —­ —­ 16
B1.0 BIII‒­a 19.1 2.65 0.71 85 T2 BIII 90
III‒­b 3 IV
B 10.0 2.86 0.05 5 T B 10
IV‒­a
B 3.7 —­ —­ 6
BIV‒­b 2.4 —­ —­ 4
III‒­a
B2.0 B 19.2 2.61 0.77 89 T2 BIII 92
III‒­b 3
B 8.9 2.52 0.09 2 T BIV 8
IV‒­a
B 2.8 —­ —­ 5
IV‒­b
B 2.4 —­ —­ 4
B6.0 BIII‒­a 19.8 2.75 0.65 92 T2 BIII 94
III‒­b 3 IV
B 8.7 2.23 0.05 1 T B 6
BIV‒­a 3.7 —­ —­ 6
GADDAM et al.
F I G U R E 5   FTIR spectra: absorption spectra of obtained glass
series (A), absorption spectra with B0.0 spectra subtracted from other
glasses spectra (B) [Color figure can be viewed at wileyonlinelibrary.
com]
3.3  |  Characterization of the fibers
Of all the samples drawn to fibers on the fiber drawing tower,
samples B0.0, B0.5 and B1.0 performed well. Other glass sam-
ples, undergone partial non-­uniform surface crystallization
during the fiber drawing hampering the drawing process
(Figure S4). Thus, only short time interval was available for
these glasses to be completely drawn. Nevertheless, sample
B1.0 performed the best with minimum crystallization until
the end of the process. Similarly, B0.0 sample performed well,
however, there were signs of partial crystallization on this
sample. Sample B0.5 showed small amount of crystallization
and its performance was in between B0.0 and B1.0. Therefore,
the performance of the samples during the fiber drawing pro-
cess showed a non-­monotonic trend.
XRD measurements were performed on powders received
by grinding glass fibers made by drawing process. XRD pat-
terns are presented in Figure 6. In case of samples B0.0, B0.5,
and B1.0, crystalline portion did not match any phase in data-
base. In case of B2.0 and B6.0, the amounts of crystalline phases
were slightly noticeable. However, further analysis resulted
in finding residual, insignificant amount of hydroxyapatite
substituted with silicate group (Ca10(PO4)4.92(SiO4)1.08(OH)
O0.664, ICDD ref. code 01-­080-­6260). Finally, B10.0 sample
  
|
   4499
F I G U R E 6   XRD patterns of as-­drawn fibers
exhibited further crystallization towards silicate substituted
apatite phase.
Figure 7 shows SEM micrographs of fibers before and
after immersion in SBF. Fibers immersed in SBF became
covered uniformly with a cracked layer resembling “tree
bark.” Such appearance is very common in bioactive glasses
after in vitro testing and results from the surface dissolution
of the glass substrate and the concomitant precipitation of
carbonated apatite. Under higher magnitude, the morphology
of obtained layers can be easily observed. It consists of small
(c. 150  nm) particles connected together, forming porous
structure.
The FTIR spectra of the SBF-­treated fibers are shown in
Figure 8A. As a result of the fibers ground to powders, the
surface composition became too diluted to be clearly detected
by FTIR. Therefore, the glass-­component of the spectrum
was removed by subtracting the baseline spectrum in each
case. The results are presented in Figure 8B. From Figure 8B,
the presence of the phosphate groups, namely, the asymmet-
ric ν4 bending mode of (PO4)3− groups (~561 and 602 cm−1),
symmetric ν1 stretching of (PO4)3− groups (~963 cm−1), and
the bands arising from ν3 modes of PO4 groups (~1090 and
1040  cm−1).48 Considering that fibers of similar diameters
(200−250 μm) were chosen for the in vitro tests, these results
may be representative of amount of HA formed on the sur-
face of the fibers.
 |
4500       GADDAM et al.
4  |   D IS C U SS ION
As mentioned in the introduction, in the current study, the
glasses were designed by replacing equimolar amounts SiO2
units with RBO2 units, where R ≡ Ca0.5 or Mg0.5. In this case
if each borate unit, that is, BIV or BIII, has one R associated
with it, either forming NBOs on BIII units or charge compen-
sating BIV units, the silicate and phosphate network connectiv-
ity should not be affected. Thus, preserving the chemistry and
bioactivity of B0.0, which are the much desirable properties of
B0.0 composition. The addition of RBO2 units would rise the
overall configurational entropy of the glass system, which, in
turn, lowers its viscosity and possibly increases its stability.
Nevertheless, according to the 11B NMR results (Table 4),
there is a formation of T3 species of BIII units without any R
associated, likely enhancing the depolymerization of the sili-
cate and phosphate networks. Accordingly, 29Si NMR spectra
show increasing deshielding effect with the addition of boron.
This deshielding effect is caused by the direct depolymeriza-
tion of the silicate network as mentioned above. However, it
is to be noted that part of this deshielding effect could also be
caused by the formation of Si‒­O‒­BIV bonds. The presence of
a single BIV unit in the next nearest neighborhood of Si causes
an increase of the 29Si-­δiso values by ~5 ppm.42 Moreover, part
F I G U R E 7   SEM micrographs of fibers
before and after immersion in SBF solution
for 14 days
of the BIV units that are present in the current glasses could
have been generated by the charge compensation by P tetrahe-
dra instead of R. This would further enhance the silicate and
phosphate network with increased modifiers. The quantifica-
tion of these bonds is difficult based on the current 1D NMR
techniques. However, the presence of two peaks for BIV units
in 11B NMR results suggest the formation of these units. It is
worth noting that sample B0.5 contains a greater amount of T3
units compared to any other sample. As a result, the 29Si NMR
spectra of B0.5 show a slightly greater deshielding effect that
B1.0 (Figure 4A).
The configurational entropy (ΔSconf.) values from the
DSC show a non-­monotonic trend with the increasing boron
oxide addition. The structural interpretation of this behavior
is essential, which would provide deeper insights into the
materials’ behavior. Therefore, in this paper, we propose a
method to extract configurational entropy from the NMR
spectra. An NMR spectrum essentially contains short-­range
structural information, that is, types of atoms and their po-
sitions around a given nucleus. For one atomic nucleus, this
information could be represented on multidimensional phase
space.49 For a large number of atoms, there will be a distri-
bution of information. However, in a 1D MAS-­NMR, this in-
formation is resolved onto a 1D line and is given by chemical
GADDAM et al.
F I G U R E 8   (A) FTIR spectra of the glasses immersed in SBF
for 14 days and (B) FTIR spectra with the subtracted baseline
corresponding to the base glass spectra shown in Figure 5A [Color
figure can be viewed at wileyonlinelibrary.com]
shift or frequency (ω). As a result, part of the information is
lost, nevertheless, NMR still provides sufficient useful struc-
tural information to probe the short-­range structure of the
glasses. From this information, we propose a method for the
calculation of the Shannon entropy, HNMR, corresponding to
the short-­range structure given by Equation (4)
∞
∫
∑ ( )
HNMR = − nT fT (𝜔)log2 fT (𝜔) d𝜔
T = Si,B,P
−∞
∞
∫
I(𝜔)
f(𝜔) = ; A= I(𝜔)d𝜔
A B0.5  has highest glass stability, it also has larger viscosity
−∞
The nT is the atomic fraction of network formers of type T.
The function fT (𝜔) is the distribution function that is obtained
from the NMR spectra by normalizing the NMR intensity
(I(𝜔)) as shown in Equation (5). Here, the entropy is calcu-
lated using frequency instead of chemical shift. However, as
the function fT (𝜔) is normalized, this would not affect the
final result. This procedure was applied to all NMR spectra
and the information entropy was calculated and is presented
in Figure 9. These results also show a non-­monotonic trend
with B1.0 having the largest amount of entropy. Interestingly,
  
|
   4501
F I G U R E 9   Shannon entropy calculated from NMR spectra
this is in accordance with the configuration entropy calcu-
lated from the DSC measurements (Figure 3B). The HNMR
only accounts for the short-­range structure. Thus, the agree-
ment between the trends in HNMR and ΔSconf. suggest that the
NMR short range structure should be reliable for correlating
the trends in the relaxation dynamics of the current glasses.
The relaxation dynamics in glasses depends on the
available thermal energy and the volume of the configura-
tional space available to the system to explore.50 Therefore,
the configurational entropy, which is the measure of the
size of the configuration space, obtained from the DSC and
NMR can provide insights into the dynamics/viscosity of
the glasses. That means, larger the entropy, the lower the
viscosity. This was also used in phenomenological mod-
elling of viscosity of the glasses.51 According to Figure
3A, the viscosity of the glasses should steadily decrease
with increasing boron content with exception of B1.0. This
would facilitate the fiber drawing by lowering the draw-
ing temperature. However, the fiber drawing also depends
on the glass stability. The glass stability values (ΔT and
KH) calculated from DSC measurements (Table 2; Figure
(4)
S5) show that for lower additions of boron, the stability
increases; and at higher additions, it starts to decrease. The
combined effects of glass stability and viscosity dictate the
performance of the fiber drawing. Even though the sample
(5)
(based on the entropy argument) at higher temperatures.
Thus, fibers for this sample can only be drawn at higher
temperatures. Therefore, this glass did not perform as good
as B0.0 or B1.0. The thermal stability increases slightly in
case of B0.5, B1.0, B2.0, a trend that is then reversed with
further substitution of silica by B2O3. The most plausible
explanation for this reversion is the enhanced diffusivity
of the species within the silicate network, as the silicate
network gets depolymerized. That results in continued
decreasing viscosity of the melts with further increasing
B2O3 content, facilitating atomic arrangements and crys-
tallization. Despite the variation of the variation in glass
 |
4502       GADDAM et al.
composition and structure, all the glasses performed excep-
tionally well in the SBF solution resulting in the formation
of hydroxy apatite layer (Figures 7 and 8). Deeper studies
on the performance of the fibers in SBF and cell cultures
will be reported in a future paper devoted to the biological
aspects of the performance of these fibers.
5  |   CO NC LUS ION S
The current study establishes the structure-­property rela-
tionships and their influence on fiber drawing ability of the
boron containing bioactive glasses. Small additions of boron
oxide improve the fiber drawing ability. Samples with lower
contents of boron (B0.5, B1.0, B2.0) exhibit improved ther-
mal stability in comparison with the original FastOs®BG.
However, with further increase of boron oxide content, the
lowering melt viscosity facilitates diffusion and atomic re-
arrangement, therefore decreasing the thermal stability of
the glasses. This explains the presence of hydroxyapatite
substituted with silicate group identified in the as-­cast B10.0
glass. Fiber drawing led to slight crystallization trends in
some glass samples. Nevertheless, B2.0 and B6.0 were drawn
with hardly detectable traces of crystallization as confirmed
by XRD. These results show great potential of this glass
series to be applied in production of bioactive glass fibers.
Further optimization of technological process may result in
avoiding crystallization during fibers fabrication process
entirely.
ACKNOWLEDGMENTS
This work was developed in the framework of Portugal-­
Poland (FCT-­ NAWA) Scientific and Technological
Cooperation, 2019/2020 –­Project no. 6818 -­311 and Polish
National Agency for Academic Exchange, grant number
PN/BIL/2018/1/00253. The work was developed within the
scope of the project CICECO-­Aveiro Institute of Materials,
UIDB/50011/2020 & UIDP/50011/2020, financed by na-
tional funds through the FCT/MEC and when appropri-
ate co-­financed by FEDER under the PT2020 Partnership
Agreement. The NMR spectrometers are part of the National
NMR Network (PTNMR) and are partially supported by
Infrastructure Project No 022161 (co-­financed by FEDER
through COMPETE 2020, POCI and PORL and FCT through
PIDDAC).
ORCID
Anuraag Gaddam  https://orcid.org/0000-0002-4266-6092
Przemysław Gołębiewski  https://orcid.
org/0000-0002-3892-2188
Hugo R. Fernandes  https://orcid.org/0000-0002-9689-3127
Ana S. Neto  https://orcid.org/0000-0002-8083-5758
Ryszard Diduszko  https://orcid.org/0000-0003-0590-6824
Agnieszka Malinowska  https://orcid.
org/0000-0003-4835-6180
Ryszard Stępień  https://orcid.org/0000-0003-2352-6382
Jarosław Cimek  https://orcid.org/0000-0002-1886-7986
Ryszard Buczyński  https://orcid.org/0000-0003-2863-725X
José M. F. Ferreira  https://orcid.org/0000-0002-7520-2809
R E F E R E NC E S
1. Hench LL. The story of Bioglass®. J Mater Sci Mater Med.
2006;17(11):967–­78.
2. Hench LL, Splinter RJ, Allen WC, Greenlee TK. Bonding mech-
anisms at the interface of ceramic prosthetic materials. J Biomed
Mater Res Symp. 1971;5(6):117–­41.
3. Vale AC, Carvalho AL, Barbosa AM, Torrado E, Mano JF,
Alves NM. Novel antibacterial and bioactive silicate glass
nanoparticles for biomedical applications. Adv Eng Mater.
2018;20(5):1700855.
4. Fagerlund S, Hupa L. Chapter 1: Melt-­derived bioactive silicate
glasses. In: Boccaccini AR, Brauer DS, Hupa L, editors. Bioactive
glasses: fundamentals, technology and applications. Smart mate-
rials series. Cambridge: Royal Society of Chemistry; 2016. p. 1–­
26. https://pubs.rsc.org/en/conte​nt/chapt​erhtm​l/2016/bk978​17826​
29764​-­00001​?isbn=978-­1-­78262​-­976-­4&serco​de=bk. Accessed
24 Apr 2021.
5. Cole KA, Funk GA, Rahaman MN, McIff TE. Characterization
of the conversion of bone cement and borate bioactive glass
composites. J Biomed Mater Res Part B Appl Biomater.
2020;108(4):1580–­91.
6. Rahimnejad Yazdi A, Torkan L, Stone W, Towler MR. The impact
of gallium content on degradation, bioactivity, and antibacterial
potency of zinc borate bioactive glass. J Biomed Mater Res Part B
Appl Biomater. 2018;106(1):367–­76.
7. Sanz CK, dos Santos AR, da Silva MHP, Marçal R, Tute EM,
Meza EL, et al. Niobo-­phosphate bioactive glass films produced
by pulsed laser deposition on titanium surfaces for improved cell
adhesion. Ceram Int. 2019;45(14):18052–­8.
8. Jones JR. Porous bioactive ceramic and glass scaffolds for bone re-
generation. In: Hench LL, editor. An introduction to bioceramics.
2nd ed. London: Imperial College Press; 2013. p. 463–­85. https://
www.world​scien​tific.com/doi/10.1142/97819​08977​168_0032.
Accessed 26 Apr 2021.
9. Leppäranta O, Vaahtio M, Peltola T, Zhang D, Hupa L, Hupa
M, et al. Antibacterial effect of bioactive glasses on clinically
important anaerobic bacteria in vitro. J Mater Sci Mater Med.
2008;19(2):547–­51.
10. Fagerlund S, Massera J, Moritz N, Hupa L, Hupa M. Phase compo-
sition and in vitro bioactivity of porous implants made of bioactive
glass S53P4. Acta Biomater. 2012;8(6):2331–­9.
11. Sepulveda P, Jones JR, Hench LL. In vitro dissolution of melt-­
derived 45S5 and sol-­gel derived 58S bioactive glasses. J Biomed
Mater Res. 2002;61(2):301–­11.
12. Cannillo V, Chiellini F, Fabbri P, Sola A. Production of Bioglass®
45S5 -­Polycaprolactone composite scaffolds via salt-­leaching.
Compos Struct. 2010;92(8):1823–­32.
13. Rottensteiner U, Sarker B, Heusinger D, Dafinova D, Rath S,
Beier J, et al. In vitro and in vivo biocompatibility of alginate
dialdehyde/gelatin hydrogels with and without nanoscaled bio-
active glass for bone tissue engineering applications. Materials.
2014;7(3):1957–­74.
GADDAM et al.
14. Groh D, Döhler F, Brauer DS. Bioactive glasses with improved
processing. Part 1. Thermal properties, ion release and apatite for-
mation. Acta Biomater. 2014;10(10):4465–­73.
15. Brauer DS. Bioactive glasses-­
structure and properties. Angew
Chemie Int Ed. 2015;54(14):4160–­81.
16. Lindfors NC, Koski I, Heikkilä JT, Mattila K, Aho AJ. A prospec-
tive randomized 14-­year follow-­up study of bioactive glass and
autogenous bone as bone graft substitutes in benign bone tumors. J
Biomed Mater Res -­Part B Appl Biomater. 2010;94(1):157–­64.
17. Goel A, Kapoor S, Rajagopal RR, Pascual MJ, Kim H-­W, Ferreira
JMF. Alkali-­free bioactive glasses for bone tissue engineering: a
preliminary investigation. Acta Biomater. 2012;8(1):361–­72.
18. Brito AF, Antunes B, dos Santos F, Fernandes HR, Ferreira JMF.
Osteogenic capacity of alkali-­free bioactive glasses. In vitro studies.
J Biomed Mater Res -­Part B Appl Biomater. 2017;105(8):2360–­5.
19. Cortez PP, Brito AF, Kapoor S, Correia AF, Atayde LM, Dias-­
Pereira P, et al. The in vivo performance of an alkali-­free bioac-
tive glass for bone grafting, FastOs®BG, assessed with an ovine. J
Biomed Mater Res Part B Appl Biomater. 2017;105(1):30–­8.
20. Shi Q, Wang J, Zhang J, Fan J, Stucky GD. Rapid-­setting, mes-
oporous, bioactive glass cements that induce accelerated in vitro
apatite formation. Adv Mater. 2006;18(8):1038–­42.
21. Poologasundarampillai G, Wang D, Li S, Nakamura J, Bradley R,
Lee PD, et al. Cotton-­wool-­like bioactive glasses for bone regener-
ation. Acta Biomater. 2014;10(8):3733–­46.
22. Zhang K, Wang Y, Hillmyer MA, Francis LF. Processing and
properties of porous poly(L-­lactide)/bioactive glass composites.
Biomaterials. 2004;25(13):2489–­500.
23. Kolan KCR, Semon JA, Bindbeutel AT, Day DE, Leu MC.
Bioprinting with bioactive glass loaded polylactic acid composite
and human adipose stem cells. Bioprinting. 2020;18:e00075.
24. Brink M, Turunen T, Happonen R-­P, Yli-­Urpo A. Compositional
dependence of bioactivity of glasses in the system Na2O-­K2O-­MgO-­
CaO-­B2O3-­P2O5-­SiO2. J Biomed Mater Res. 1997;37(1):114–­21.
25. Fu Q, Rahaman MN, Bal BS, Huang W, Day DE. Preparation and
bioactive characteristics of a porous 13–­93 glass, and fabrication
into the articulating surface of a proximal tibia. J Biomed Mater
Res Part A. 2007;82A(1):222–­9.
26. Fu Q, Rahaman MN, Bal BS, Bonewald LF, Kuroki K, Brown
RF. Silicate, borosilicate, and borate bioactive glass scaffolds with
controllable degradation rate for bone tissue engineering applica-
tions. II. In vitro and in vivo biological evaluation. J Biomed Mater
Res Part A. 2010;95A(1):172–­9.
27. Gaddam A, Fernandes HR, Ferreira JMF. Glass structure and crys-
tallization of Al and B containing glasses belonging to the Li2O–­
SiO2 system. RSC Adv. 2015;5(51):41066–­78.
28. Gaddam A, Fernandes HR, Pascual MJ, Ferreira JMF. Influence of
Al2O3 and B2O3 on sintering and crystallization of lithium silicate
glass system. J Am Ceram Soc. 2016;99(3):833–­40.
29. Huang W, Day DE, Kittiratanapiboon K, Rahaman MN. Kinetics
and mechanisms of the conversion of silicate (45S5), borate, and
borosilicate glasses to hydroxyapatite in dilute phosphate solu-
tions. J Mater Sci Mater Med. 2006;17(7):583–­96.
30. Yao A, Wang D, Huang W, Fu Q, Rahaman MN, Day DE. In vitro
bioactive characteristics of borate-­based glasses with controllable
degradation behavior. J Am Ceram Soc. 2007;90(1):303–­6.
31. Gorustovich AA, López JMP, Guglielmotti MB, Cabrini RL.
Biological performance of boron-­ modified bioactive glass
particles implanted in rat tibia bone marrow. Biomed Mater.
2006;1(3):100–­5.
  
|
   4503
32. Gorustovich AA, Steimetz T, Nielsen FH, Guglielmotti MB.
Histomorphometric study of alveolar bone healing in rats fed a
boron-­deficient diet. Anat Rec. 2008;291(4):441–­7.
33. Haro Durand LA, Góngora A, Porto López JM, Boccaccini
AR, Zago MP, Baldi A, et al. In vitro endothelial cell response
to ionic dissolution products from boron-­ doped bioactive
glass in the SiO2-­CaO-­P2O5-­Na2O system. J Mater Chem B.
2014;2(43):7620–­30.
34. Popa AC, Stan GE, Husanu MA, Mercioniu I, Santos LF,
Fernandes HR, et al. Bioglass implant-­coating interactions in syn-
thetic physiological fluids with varying degrees of biomimicry. Int
J Nanomedicine. 2017;12:683–­707.
35. Massiot D, Fayon F, Capron M, King I, Le Calvé S, Alonso B,
et al. Modelling one-­and two-­dimensional solid-­state NMR spec-
tra. Magn Reson Chem. 2002;40:70–­6.
36. Gomes S, Nedelec JM, Jallot E, Sheptyakov D, Renaudin G.
Silicon location in silicate-­ substituted calcium phosphate ce-
ramics determined by neutron diffraction. Cryst. Growth Des.
2011;11(9):4017–­26.
37. Hrubý A. Evaluation of glass-­forming tendency by means of DTA.
Czechoslov J Phys. 1972;22(11):1187–­93.
38. Cabral A Jr, Fredericci C, Zanotto E. A test of the Hrubÿ pa-
rameter to estimate glass-­ forming ability. J Non Cryst Solids.
1997;219:182–­6.
39. Lang DP, Alam TM, Bencoe DN. Solid-­
state 31P/27AI and
31 23
P/ Na TRAPDOR NMR investigations of the phosphorus en-
vironments in sodium aluminophosphate glasses. Chem Mater.
2001;13(2):420–­8.
40. Funke LM, Eckert H. Charge compensation in sodium boro-
phosphate glasses studied by 11B{23Na} and 31P{23Na} rota-
tional echo double resonance spectroscopy. J Phys Chem C.
2016;120(6):3196–­205.
41. Yu Y, Stevensson B, Edén M. Medium-­range structural organization
of phosphorus-­bearing borosilicate glasses revealed by advanced
solid-­state NMR experiments and MD simulations: consequences
of B/Si substitutions. J Phys Chem B. 2017;121(41):9737–­52.
42. Nanba T, Nishimura M, Miura Y. A theoretical interpretation of
the chemical shift of 29Si NMR peaks in alkali borosilicate glasses.
Geochim Cosmochim Acta. 2004;68(24):5103–­11.
43. Yu Y, Edén M. Structure-­composition relationships of bioactive
borophosphosilicate glasses probed by multinuclear 11B, 29Si, and
31
P solid state NMR. RSC Adv. 2016;6(103):101288–­303.
44. Kroeker S, Stebbins JF. Three-­
coordinated boron-­ 11 chemical
shifts in borates. Inorg Chem. 2001;40(24):6239–­46.
45. Tricot G, Raguenet B, Silly G, Ribes M, Pradel A, Eckert H. P-­
O-­B3 linkages in borophosphate glasses evidenced by high field
11 31
B/ P correlation NMR. Chem Commun. 2015;51(45):9284–­6.
46. Muñoz-­
Senovilla L, Tricot G, Muñoz F. Kinetic fragility and
structure of lithium borophosphate glasses analysed by 1D/2D
NMR. Phys Chem Chem Phys. 2017;19(34):22777–­84.
47. Lai Y, Zeng Y, Tang X, Zhang H, Han J, Su H. Structural investi-
gation of calcium borosilicate glasses with varying Si/Ca ratios by
infrared and Raman spectroscopy. RSC Adv. 2016;6(96):93722–­8.
48. Markovic M, Fowler BO, Tung MS. Preparation and comprehen-
sive characterization of a calcium hydroxyapatite reference mate-
rial. J Res Natl Inst Stand Technol. 2004;109 [6]:553.
49. Gaddam A, Montagne L, Ferreira JMF. Statistics of silicate units
in binary glasses. J Chem Phys. 2016;145(12):124505.
50. Mauro JC, Smedskjaer MM. Statistical mechanics of glass. J Non
Cryst Solids. 2014;396–­397:41–­53.
 |
4504       GADDAM et al.

51. Adam G, Gibbs JH. On the temperature dependence of coopera-

tive relaxation properties in glass-­forming liquids. J Chem Phys. How to cite this article: Gaddam A, Gołębiewski P,
1965;43(1):139–­46. Fernandes HR, et al. Development of microfibers for
bone regeneration based on alkali-­free bioactive glasses
doped with boron oxide. J Am Ceram Soc.
SUPPORTING INFORMATION
2021;104:4492–4504. https://doi.org/10.1111/
Additional supporting information may be found online in
jace.17853
the Supporting Information section.