A number of bioresorbable and bioactive materials with the required biocompatibility and sufficient mechanical properties have been the subject of much interest in recent studies [1] [2] [3] [4] [5] . The first bioactive glasses (Bioglass 45S5) developed by Hench et al. [6] [7] in 1970s, were presenting a specific biological response resulting in the formation of a bond between the bioactive glass surface and bone tissue, thus becoming the benchmark for novel silicate based bioactive glasses with application in dentistry and orthopaedics [8] [9] . However, partial degradation behaviour of Bioglass limits its application and has stimulated research for new bioactive and bioresorbable glasses as potential alternative [10] .

In the last two decades, phosphate based glasses have been considered as potential materials for the repair and reconstruction of bone [11] [12] [13] . As their hydrolytic degradation rate can be varied from hours to months by changing the glass composition which can closely match that of the inorganic phase of bone [14] [15] [16] , many phosphate glass formulations synthesised are biocompatible and subtle alterations to their composition permit a wide variation of mechanical and thermal properties [17] [18] [19] . Ahmed et al. [20] investigated phosphate based glass in the system 40P₂O₅-25CaO-20MgO-15Na₂O which showed good cellular response and Hasan et al. [21] studied the system 45P₂O₅-16CaO-24MgO-11Na₂O-4Fe₂O₃, which also showed good cytocompatibility and proved successful for fibre drawing.

In current publication, it was reported that addition of B₂O₃ into phosphate based glass could improve chemical durability and mechanical stability of phosphate based glass [22] [23] [24] [25] . Sharmin et al. [26] developed several borophosphate based glass compositions and reported that the glass transition temperature was increased whilst the degradation rate was reduced significantly with addition of B₂O₃ into the glass network. It was confirmed, and also reported by Lim et al., [27] that the formation of covalent B-O-P bonding produced a stronger, better chemically resistant glass network with higher packing density and lower crystallisation tendency.

Additionally, preliminary studies also show that the addition of MgO into phosphate based glass could connect chain structure of phosphate based glass and transfer open metaphosphate glass structure to more compact structure, thus improving the glass structure stability [20] . Moreover, Fe₂O₃ dropped into the glass network could show a much more durable degradation behaviour and favourable cytocompatibility [28] . However, till to date, the effect on thermal properties, degradation behaviour and glass structure of phosphate based glass by substitution of MgO with B₂O₃ and replacing MgO with Fe₂O₃ has not been investigated significantly.

Based on the established phosphate glass system P₂O₅-B₂O₃-CaO-MgO-Na₂O-Fe₂O₃ [21] [26] [29] , a range of novel complex PBGs in the system of 48P₂O₅-12B₂O₃-(25-X)MgO-14CaO-1Na₂O-(X)Fe₂O₃ and 45P₂O₅-(Y)B₂O₃-(32-Y)MgO-14CaO-1Na₂O-8Fe₂O₃ was considered in this study, and an investigation into their structural, thermal and degradation properties has been conducted. The effects on the glass density were also evaluated. Degradation studies of the glasses were conducted in phosphate buffer saline (PBS) solution.

The glasses in this study were manufacture in Sinoma Co., Ltd. (China) using phosphorous pentoxide (P₂O₅), boric acid (H₃BO₃), calcium hydrogen phosphate dehydrate (CaHPO₄∙2H₂O), magnesium hydrogen phosphate trihydrate (MgHPO₄∙3H₂O), sodium dihydrogen phosphate dehydrate (NaH₂PO₄∙2H₂O), potassium dihydrogen phosphate (KH₂PO₄), and iron (III)-phosphate Tetrahydrate (FePO₄∙4H₂O) (Sinopharm Group, China). The precursors were weighted out and placed into a large Pt crucible which was built in the customised furnace composed of Al₂O₃. When all the mixed precursors had been added, they were heated using a silicon carbide rod by direct resistance heating. After 24 hours heating at 1200˚C, the melted glass was poured into a stainless steel bucket water for cooling and collecting. The cooled glasses were then removed from the bucket and dried in oven for 12 hours. 5 kg glass batches were manufactured for each composition.

The bulk glasses were then re-melted and cast as 9 mm diameter rods by pouring into a graphite mould at 450˚C. After 2 hours at this temperature, the glass rods in the mould were cooled to room temperature at a cooling rate of 0.3˚C/min. Then, the rods were re-placed into the furnace for annealing process via heating at the temperature of T_a (T_g + 10˚C) for 90 minutes isothermally. After that, the rods were allowed to cool to room temperature at 0.3˚C/min cooling rate.

In order to confirm the amorphous state of all the glass compositions, powder X-ray diffraction spectra were obtained using a Bruker D8 advance X-ray diffractometer. The measurements were taken at room temperature and ambient atmosphere with Ni-filtered CuK α radiation (λ = 0.15418 nm), operated at 40 kV and 35 mA. The scans were performed with an angular range 2θ from 10˚ to 100˚, a step size of 0.01˚ and a step time of 0.1 s.

Fourier transform infrared spectroscopy was performed on around 1 - 2 mg samples of glass powder, using a Bruker Vertex 70 spectrometer (Bruker Optics, Germany). Spectra were recorded in the region of 500 to 4000 wavenumbers using a standard MKII Golden GateTM Single Reflection Attenuated Total Reflectance (ATR) system (Specac Ltd.) and analysed using OPUS software version 5.5.

Raman spectra were recorded at room temperature using inVia-Reflex Raman spectrometer (Renishaw, 633 nm laser). Bulk glass with good surface quality was placed on the sample stage and observed via objective turret in order to focus laser on glass surface. D0.3 filter was used to control transparency (50%) and Raman spectra from 400 cm⁻¹ to 2000 cm⁻¹ were taken into account in this study.

Bulk glasses of the different compositions were ground to fine powder using a pestle and mortar. The glass transition temperature T_g, crystallisation temperature T_c, melting point T_m and liquidus temperature T_l of the glasses was determined using a differential scanning calorimetry (DSC, TA Instruments SDT Q600, UK). Samples of approximately 30 mg of the glass powders were heated from room temperature to 1200˚C at 10˚C∙min⁻¹ in flowing nitrogen gas. A blank run was carried out to determine the baseline which was then subtracted from the traces obtained. The T_g was extrapolated from the midpoint in the endothermic reaction of the heat flow. The first deviation of the DSC curve from the base line above T_g and before the crystallisation peak was taken as the onset of crystallisation temperature T_onc. The thermal stability of the glass was indicated in terms of processing window T_pw by measuring the interval between T_g and T_onc [30] [31] :

T p w = T o n c − T g (1)

The density of the glasses was determined using Ultrapyc 1200e (Quantachrome, USA). The equipment provides an accurate measurement of volume and is calibrated by using a standard steel calibration ball (2.18551 cm³), with errors of ±0.05%. Triplicate, bubble-free bulk glass samples with an average weight of approximately 7 g were used for the volume measurements.

Three PBG glass rods of each composition, 9 mm diameter and 10 mm length, were put into 30 ml glass vials, immersed in phosphate buffer saline (PBS) and then placed into an incubator at 37˚C. The surface area of each disc was calculated from dimensional measurements taken using a micrometer (Mitutoyo, Japan) and their mass was measured to 4 decimal places using a digital balance (Mettler Toledo, USA). The period of degradation analysis was 60 days and the time points were days 1, 2, 3, 4, 7, 9, 12, 16, 20, 23, 26, 30, 33, 37, 40, 44, 47, 50, 54, 58 and 60. At each time point, the glass rods were taken out of the vials, excess moisture was removed by blotting the samples dry with tissue and they were then dried in an oven at 50˚C for 60 minutes. Dimension and mass measurements were taken and the pH of the solution was measured by using a microprocess pH meter (Mettler Toledo, USA) before returning the samples to the vials with fresh PBS solution. The rate of mass loss was calculated using the following equation:

Massloss ( % ) = M 0 − M t M 0 (2)

where M_o is the initial mass (g) and M_t is the mass at time t. The values of mass loss per unit area, determined via Equation (4), were plotted against time. The slope of the graph was determined by fitting a straight line through the data and passing through the origin point gave the dissolution rate in terms of kg∙m⁻²∙s⁻¹

Masslossperunitarea ( kg ⋅ m − 2 ⋅ s − 1 ) = M 0 − M t A o (3)

where A_o is surface area of glass disc at time t.

The average values and standard deviation of all data involved in this study were calculated and analysed using the Prism software (version 6.0, GraphpPad Software, San Diego, CA, USA). A one-way analysis of variance (ANOVA) was calculated with the Tukey multiple post-test to compare the significance of change in one factor with time. The error bars on all the data represent standard error of mean.

In this study, six glass formulations of phosphate based glasses were considered as two groups, MgP-Fe(X) for glass system 48P₂O₅-12B₂O₃-(25-X)MgO-14CaO-1Na₂O-(X)Fe₂O₃ and MgP-B(Y) for glass system 45P₂O₅-(Y)B₂O₃-(24-Y)MgO-14CaO-1Na₂O-8Fe₂O₃. The density analysis of bulk glass presented the value of density increased from 2.86 to 3.01 × 10³ kg∙m⁻³ with increase of Fe₂O₃ content for MgP-Fe glass system, whilst the decrease in the density was observed for MgP-B glass system when B₂O₃ content increased from 15 mol% to 20 mol%.

The thermal analysis of phosphate based glasses were summarised in Table 1,