- Email: [email protected]
Synthesis and characterization of different chain length sodium polyphosphates
Journal of Non-Crystalline Solids 382 (2013) 11–17
Contents lists available at ScienceDirect
Journal of Non-Crystalline Solids journal homepage: www.elsevier.com/ locate/ jnoncrysol
Synthesis and characterization of different chain length sodium polyphosphates Arash Momeni a, Mark Joseph Filiaggi a,b,⁎ a b
School of Biomedical Engineering, Dalhousie University, Halifax, Nova Scotia B3H 3J5, Canada Faculty of Dentistry, Department of Applied Oral Sciences, Dalhousie University, 5981 University Avenue, Halifax, Nova Scotia B3H 3J5, Canada
a r t i c l e
i n f o
Article history: Received 28 June 2013 Received in revised form 2 October 2013 Available online 23 October 2013 Keywords: Sodium polyphosphate; Kurrol salt; Graham salt; Chain length
a b s t r a c t The simple chemistry of sodium polyphosphate (NaPP) makes it a very safe inorganic polymer for bioapplications. Recently we have focused on an in-situ forming system comprised of NaPP and calcium chloride solutions which, on contact, form a gel-like material. This system could be used in a variety of bio-applications. The first step of our studies has been synthesis and characterization of NaPP with the average degree of polymeriza tion Dp from 50 up to 25,000, that is summarized in this manuscript. NaPP with Dp lower than 500 was obtained from fractionation of a NaPP glass. NaPP with Dp higher than 500 was obtained from an ion-exchange process on a potassium polyphosphate crystalline phase. The ion-exchange process was optimized maximizing the potassium exchange with sodium while minimizing resin waste. The Dp of prepared NaPPs was determined by titration, viscosity and 31P-liquid nuclear magnetic resonance measurements. The accuracy of these methods in determining the chain length is discussed. The effect of furnace time and temperature on Dp is also discussed. In addition, since both fractionation and ion-exchange methods introduce water, the resulting NaPPs were evaluated for residual water by differential scanning calorimetry and elemental analysis methods. © 2013 Elsevier B.V. All rights reserved.
1. Introduction In 1816 Berzelius wrote about amorphous phosphoric acids characteristically ranging from viscous liquids, to materials that would be described as either glasses or unusually rigid liquids at room temperature [1]. Seventeen years later, Thomas Graham prepared vitreous sodium phosphate exhibiting the metaphosphate composition (Na2O/P2O5 = 1) by heating sodium dihydrogen phosphate to high temperature, followed by chilling the melt suddenly [2]. Referred to as Graham salt, this compound was erroneously labeled a sodium hexametaphosphate (a ringed structure) for over 100 years until the work by Van Wazer [3] concluded that Graham salt solutions consisted, instead, of a mixture of linear polymers (i.e. polyphosphates) with the following general chemical formula:
where X+ is a monovalent cation such as H+, Na+ or K+ and n is the degree of polymerization (Dp). ⁎ Corresponding author at: School of Biomedical Engineering, Dalhousie University, Halifax, Nova Scotia B3H 3J5, Canada. Tel.: +1 902 494 7102; fax: +1 902 494 6621. E-mail address: fi[email protected] (M.J. Filiaggi). 0022-3093/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jnoncrysol.2013.10.003
Shortly after the discovery of Graham salt in 1833 [2], it was found that its solutions have the ability to prevent precipitation or redissolve precipitates of the alkaline earth metals. This complexing ability was an obscure piece of information of little interest to either science or applied technology until 1933, when Hall applied this effect to the softening of water [4]. Since then, polyphosphates have widely been used as water softening agent in industry and municipal water supplies. They are also widely used to aid food processing or to improve the eating quality of many foods, particularly meat and fish products. Interestingly, polyphosphates are also found in every cell in nature. Their functions include acting as an energy supply and ATP substitute, an osmotic inert reservoir for phosphate, a chelator of metals, a buffer against alkali, a channel for DNA entry, a cell capsule and a regulator of responses to stresses [5,6]. In 2004, Ruiz et al. showed that human platelets are also rich in polyphosphate, which is secreted upon thrombin stimulation [7]. Later studies showed that polyphosphate is a very strong pro-coagulant interconnecting intrinsic and extrinsic pathways of the blood clotting cascade [8]. Our group has recently focused on a novel injectable in-situ forming system, based on polyphosphates, with the potential to deliver and release drugs locally in procedures such as trans-arterial chemoembolization (TACE). In TACE, a biomaterial is administered directly into blood vessels feeding a tumor thereby blocking its nutrient supply and also releasing chemotherapeutics locally [9]. Our in-situ forming system is comprised of sodium polyphosphate (NaPP) and calcium chloride solutions which, on contact, form a coacervate or a gel-like material that is immiscible with water. NaPP chain length is an
12
A. Momeni, M.J. Filiaggi / Journal of Non-Crystalline Solids 382 (2013) 11–17
important factor determining the physical and biological properties of this in-situ forming system; therefore, preparing different chain length polyphosphates has formed a major part of our research. NaPP glass is a distribution of polyphosphate chains and by its fractionation different short chain length polyphosphates could be obtained. Synthesis of long chain polyphosphate, however, is more complicated as it requires dissolution of a crystalline polyphosphate phase. In this paper we comprehensively describe synthesis and of NaPP characterization with the average degree of polymerization Dp from 50 up to 25,000. 2. Material and method 2.1. Preparation of NaPP 2.1.1. Preparation of NaPP with Dp b500 Sodium phosphate monobasic monohydrate (NaH2PO4·H2O; SigmaAldrich) was used to prepare NaPP with Dp lower than ≈500. NaH2PO4· H2O was heated in a Thermolyne Type 46200 High Temperature Furnace in a platinum–5% gold crucible from 25 °C to 700 °C over a 90 min period, keeping the melt at 700 °C for 1, 3 or 9 h followed by quenching on a copper plate. The oxide and hydroxide layers on the copper plate were removed prior to quenching using acetic acid and sodium chloride solution. In order to fraction the NaPP glass, the frit was grounded and dissolved in deionized water to give a 10% (w/v) NaPP solution. The solution was stirred using a magnetic stirrer for 12 h and then fractioned by serial addition of acetone. After each acetone addition, the solution was mixed for 15 min and then centrifuged at 4200 rpm for 5 min to collect the precipitate, which was subsequently frozen in the freezer at −20 °C. The frozen precipitate was freeze-dried using FreeZone® 1 Liter Benchtop Freeze Dry System (Labconco Corp., USA) for 24 h at −43 °C in pressures lower than 133 × 10−3 mbar resulting in a lyophilized product that was kept in a desiccator until use. 2.1.2. Preparation of NaPP with Dp N500 To prepare NaPP with Dp higher than ≈500, potassium phosphate monobasic (KH2PO4; Sigma-Aldrich) was heated in the furnace in a platinum–5% gold crucible from 25 °C to 775 °C over a 90 min period, then held at 775 °C for 30 min, 2 or 50 h prior to cooling the crucible bottom in water. The resulting product is a water insoluble crystalline phase of KPO3 known as the potassium Kurrol salt. The Kurrol salt was washed by preparing a 20% (w/v) aqueous suspension of its powder which was magnetically stirred at 4 °C for 20 min. The Kurrol salt was then paper-filtered, washed with ethanol and dried in a desiccator for 24 h. Water soluble NaPP was obtained from Kurrol salt by an ion exchange process in which potassium cations were replaced by sodium cations. A slurry of Kurrol salt was prepared in previously boiled deionized water followed by dry Amberlite™ IRP69 sodium resin addition and rigorous mixing using a mechanical stirrer (2000 rpm) at 30 °C in pH range of 8–8.5, continually adjusted by 0.1 M tetramethylammonium hydroxide solution. After mixing, the solution was paper-filtered to remove the resin, frozen and then freeze-dried for 72 h at −43 °C in pressures lower than 133 × 10−3 mbar to obtain NaPP with Dp higher than ≈500. The Kurrol salt/resin weight ratio, mixing time and concentration of the Kurrol salt in the slurry were varied in order to determine the optimum conditions under which a maximum amount of potassium could be replaced by sodium while minimizing resin waste. 2.2. Characterization of NaPP 2.2.1. Compositional analysis The K/P mole ratio of our stock KH2PO4 ðK=PÞKH2 PO4 was determined by Energy-Dispersive X-ray Spectroscopy (EDS) (Hitachi S-4700 FEG Scanning Electron Microscope equipped with Inca EDX system, Japan), Inductively Coupled Plasma Optical Emission Spectroscopy (ICP) (Optima™ 7300 V ICP-OES, PerkinElmer Instruments, USA) and a pH measurement method [10]. For the latter method, the pH of 0.1 M KH2PO4 in CO2-free
water under nitrogen gas at 25 °C was measured and its variation from the theoretical pH of KH2PO4 under similar conditions was used to determine the ðK=PÞKH2 PO4 . The amount of phosphate in all NaPP samples was assessed by dissolving 50 mg of each sample in 10 ml 6 N HCl, maintaining it in boiling water for 4 h, followed by 20 h cool down period and then titration by 1 M NaOH. The moles of 1 M NaOH required between two inflection points, pH 3.5 and 8, are equal to the moles of phosphate in 50 mg of a sample. The efficiency of the ion exchange process was also evaluated by measuring the remaining potassium to phosphorus mole ratio in the NaPP ((K/P)IE). Here, the amounts of potassium and phosphorus in ion-exchanged samples were measured by PerkinElmer 2380 Atomic Absorption Spectroscopy (AAS) and titration, respectively. The (K/P)IE was also verified by ICP only in some of the samples. 2.2.2. Thermal analysis Thermal behavior and amount of residual water in all NaPP samples after lyophilization was determined by Differential Scanning Calorimetry and Thermal Gravity Analysis (DSC/TGA) using a STA 409 PC LUXX® thermal analyzer (Netzsch, Germany). The baseline was obtained using an empty platinum crucible and then the instrument was calibrated. Approximately 20 mg of each sample was added to the platinum crucible and the DSC/TGA data was collected from 25 °C to 1000°C at the heating rate of 10°C/min. DSC/TGA data of NaH2PO4·H2O, KH2PO4 and Kurrol salt were also obtained. All DSC/TGA data were analyzed using Proteus® Software. 2.2.3. Dp analysis Liquid 31P nuclear magnetic resonance (NMR), titration and viscosity studies were carried out to determine the chain length of NaPP. 100 mg of NaPP was dissolved in 0.75 ml D2O to be analyzed by a Bruker AV 300 MHz NMR spectrometer at 101.26 MHz, using a 151 pulse, 65,536 (64 k) data points giving a 0.8 s acquisition time, a 7.0 s repetition rate and greater than 100 scans. Spectra were reported using the d scale, with positive values downfield, and were referenced to an 85% solution of H3PO4 in H2O. The peaks around 0, −9, −21, −23 and −35ppm represent the Q0 (orthophosphates), Q1 (end of chain phosphorus atoms), Q2 (middle of chain phosphorus atoms in polyphosphates), Meta-Q2 (middle of chain phosphorus atoms in rings) and Q3 (branched phosphorus atoms), respectively [11]. The Dp of NaPP was determined by the following equation based on the area under the peaks:
Dp ¼
2 Q1 þ Q2 þ Q3 Q1
:
ð1Þ
To determine the chain length by titration, 100 mg of NaPP was dissolved in 10 ml deionized water and titrated by 1 M and 0.1 M NaOH solutions. The moles of NaOH that were used to titrate the NaPP solution between two inflection points, pH 4.5 and 9, were used to determine the Dp based on the following equation [12]: Dp ¼
20 weight ðgÞ of NaPP dissolved in solution : ml of 1 N NaOH required between two inflection points
ð2Þ
Both NMR and titration give number average Dp and therefore, the number average molecular weight Mn is equal to 102ñDp , where 102 is the molecular weight of NaPO3 monomer. In contrast, viscosity measurements give the weight average molecular weight Mw . NaPP with Dp lower than 500 and those with chain lengths higher than 500 were dissolved in 0.035 N and 0.35 N NaBr solutions, respectively, at concentrations lower than 0.015 g/ml. The viscosity of these solutions was measured using DV3™ cone–plate rheometer (Brookfield, USA) at 25°C and the linear graph of (specific viscosity/concentration) is plotted against concentration. The extrapolation of this graph to zero
A. Momeni, M.J. Filiaggi / Journal of Non-Crystalline Solids 382 (2013) 11–17
concentration yields intrinsic viscosity ([η]) that is used to calculate the Mw based on the following equations [13,14]: −5 ½ηðin 0:035 N NaBrÞ ¼ 10 Mw þ 0:025;
for NaPP with Dp b 500 at 25 C ð3Þ
−4
½ηðin 0:35 N NaBrÞ ¼ 0:65 10 Mw
0:69
;
for NaPP with Dp N 500 at 25 C:ð4Þ
3. Results 3.1. NaPP with Dp b500 Fig. 1 shows the DSC/TGA graph of NaH2PO4·H2O. From ~70 °C to 150 °C, the compound loses 12.76% of its weight implying the loss of one molecule of water per phosphate unit and formation of NaH2PO4. Around 220 °C, there is a further 6.52% mass loss suggesting the release of one water molecule per two phosphate units and formation of Na2H2P2O7, with a subsequent 6.47% mass loss again at approximately 350 °C, implying the loss of another molecule of water per two phosphate units and formation of trimetaphosphate rings (i.e. Na3P3O9). Finally, the compound melts around 640 °C and NaPP glass is obtained by quenching this melt. Quenching on a stainless steel oxide plate formed a black layer on the surface of the glass that was in contact with the plate, suggesting a reaction between the melt and the stainless steel oxide layer. In contrast, glasses quenched on the copper plate were visually clear, with EDS analysis showing no sign of copper contamination in the glass. Fig. 2(a) shows the liquid 31P NMR spectra of NaPP glasses kept at 700 °C for 1, 3 or 9 h, corresponding to a Dp of 186 ± 17, 131 ± 12 or 174 ± 15 (n = 3), respectively. Based on a two-tailed Student's t-test, there is no significant difference between the Dp of 1 and 9 h samples, but both groups are significantly (p b 0.05) longer than the 3 h samples. A NaPP glass was also prepared by keeping the melt at the higher temperature of 970°C for 1h. This glass had a Dp of 183±10 (n=3), which is not significantly different from the glass that was prepared by keeping the melt at 700 °C for 1 h. Based on the NMR results, the fraction of ring metaphosphates (Meta-Q2) was consistently 4.5–5% of the total phosphorus atoms in all of the NaPP glasses. A very small peak representing Q3 sites was also detected in dissolved NaPP glasses, with
13
its area implying less than one phosphate branching point for every 1000 phosphate units. The NaPP glasses that were kept at 700°C for 1 or 3h were fractioned by serial addition of acetone to their solutions. As shown in Fig. 2(b), long chains precipitated early with a small amount of acetone, but precipitation of shorter chains required much more acetone. Fig. 2(b) also shows the Dp of each of the fractions. The original NaPP glass that was fractioned had a Dp of 197, but based on its fractions, it constitutes a wide distribution of different polyphosphate chains with Dp starting from 56 up to 398. The fraction with Dp of 270 contributes almost 30% to the original NaPP glass mass. Fig. 2(c) shows the corresponding liquid 31 P NMR spectra for these eight fractions. As the chains get longer the area under the Q1 peak becomes smaller while the area under the Q2 peak becomes larger. Fig. 2(d) shows the titration of the same fractioned NaPPs; the required volume of NaOH increases as the chains become shorter. The Dp determined by NMR and titration correlates as shown in Fig. 3(a). For short chains, NMR and titration results are the same as shown by the solid line with the slope of 1.04. However, as the chains get longer the accuracy and precision of titration method decreases and these two methods do not give the same results. Also, as the chains get longer more NMR scans are required to achieve an accurate result. Theoretically, the distribution of chains in a NaPP glass with an original Dp of n can be determined with the random distribution equation [15,16]:
fx ¼
x n−2 x−1 ; nðn−2Þ n−1
for x N 2; n N 2
ð5Þ
where fx is the fraction of total number of phosphorus atoms having the Dp of x. Using this equation, a comparison between theoretical and experimental distributions of our polyphosphates was carried out in two NaPP glasses prepared by keeping the melt at 700 °C for 1 and 3 h. In Fig. 3(b), the separation between stepwise curves (experiment) and solid line curves (theory) becomes more prominent as degree of polymerization of fractions (y-axis) increases. For instance, in the polyphosphate with the Dp of 197, 36% of phosphates are theoretically in chains with Dp larger than 400, but experimentally this value is only 23%. This indicates that the experimental distribution fits the theoretical distribution perfectly at low Dp, but as chains get longer the
Fig. 1. DSC (solid lines) and TGA (dashed lines) graphs. With the exception of NaH2PO4·H2O, DSC/TGA graphs of all other compounds are shifted down for clarity.
14
A. Momeni, M.J. Filiaggi / Journal of Non-Crystalline Solids 382 (2013) 11–17
b)
a)
d)
c) Fig. 2. (a) Liquid 31P NMR spectra of NaPP glasses prepared by keeping the melt at 700 °C for 1, 3 and 9 h; (b) fractionation result of a NaPP glass with Dp of 197: the graph shows the amount of acetone required for precipitation of each fraction and columns show the percentage weight of each fraction; (c) liquid 31P NMR spectra and (d) titration graphs of NaPP fractions obtained from a NaPP glass with Dp of 197: numbers show the measured Dp based on the method. With the exception of lowest spectrum, all other NMR spectra are shifted downfield for clarity.
theoretical curve overestimates the percentage of total number of phosphorus atoms in polyphosphates with large Dp. Table 1 summarizes the fractionation results of two NaPP glasses. The M n value measured by the NMR for the original glasses correlates with the Mn value calculated using the following equation based on the fractionation results. X
Ni Mi Mn ¼ X Ni
ð6Þ
The M w of NaPPs as determined using viscosity measurements is also displayed in Table 1 along with Mw and poly-dispersity values calculated using the following equations based on the fractionation results: X 2 Ni Mi Mw ¼ X N i Mi Poly‐dispersity ¼
ð7Þ Mw Mn
:
ð8Þ
A. Momeni, M.J. Filiaggi / Journal of Non-Crystalline Solids 382 (2013) 11–17
15
due to the residual water. There is a 2.66% weight loss between 80 and 135 °C, followed by a slow weight loss of 1.19% until 350 °C, where a sharp weight loss of 6.87% occurs. There is a reaction happening between residual water and NaPP glass between 80 and 135 °C because the weight loss is accompanied by an exothermic rather than an endothermic peak. Presumably polyphosphate chains are being degraded into sodium pyrophosphate, which then transforms into sodium trimetaphosphate ring at 350 °C. The percentage of weight loss upon heating is reported in Table 1 for all fractioned NaPP samples. Since water is reacting with the NaPP glass, TGA is not an accurate method for determining the percentage of residual water in the fractioned samples. Instead, the amount of sodium phosphate in each sample was determined by titration and subtracted from the original weight of the sample to determine the weight percentage of residual water. As reported in Table 1, fractioned NaPP glasses contain between 3 and 11% residual water by weight. 3.2. NaPP with Dp N 500
Fig. 3. (a) Comparing the chain length measurements by titration and liquid 31P NMR (n = 3 and error bars show STD). The solid line is a trendline passing the first five points while the dashed line passes all of the seven points; (b) comparing the theoretical and experimental distributions of polyphosphate in NaPP glasses prepared by keeping the melt at 700 °C for 1 and 3 h with original Dp of 197 and 135, respectively. Stepwise curves show experimental Dp of fractions plotted against cumulative percentage of phosphorus and solid line curves show theoretical plots of Dp against cumulative percentage of phosphorus calculated from random distribution equation.
Here, Mw measured by these two methods did not correlate. One should note that fractioned NaPPs contain residual water after freeze-drying. Fig. 1 compares the DSC/TGA graph of un-fractioned NaPP glass with that of fractioned NaPP. Since the un-fractioned NaPP has no residual water, there is no weight loss in the TGA graph. There is, however, an exothermic peak around 350 °C in the DSC graph representing transformation into sodium trimetaphosphate ring or long chain NaPO3-II [17]. This is followed by melting around 630 °C. In contrast, the DSC/TGA graph of fractioned NaPP glass is more complex
Small variations of the ðK=PÞKH2 PO4 from unity is reported to significantly affect the chain length of NaPP that is obtained through potassium Kurrol salt [10]. The ðK=PÞKH2 PO4 of our stock material was 0.92 ± 0.02 based on EDS, while ICP gave a value of 1.05 ± 0.01. However, neither of these methods is sensitive enough to confirm the exact value of ðK=PÞKH2 PO4 . Instead, the pH of 0.1M KH2PO4 in CO2-free water under nitrogen gas at 25 °C was measured, and its variation from the theoretical pH of KH2PO4 under similar conditions was used to determine the ðK=PÞKH2 PO4 . Based on this method the ðK=PÞKH2 PO4 was 0.9996 ± 0.00106, essentially equal to one, and therefore no ðK=PÞKH2 PO4 adjustment of the stock material was performed. Fig. 1 shows the DSC/TGA graph of KH2PO4 where a sudden weight loss of 13.13% occurs between 182 and 358 °C, implying the loss of one water molecule per phosphate unit and formation of long chain KPO3, potassium Kurrol salt. For this process, an ion exchange resin was used to dissolve the Kurrol salt, release the polyphosphate chains and replace the potassium with sodium. The challenge was to replace the maximum amount of potassium with sodium in long chain polyphosphate solutions. Fig. 4(a) shows the dependence of (K/P)IE to the amount of resin that was used during the ion exchange process. The higher the amount of resin, the more potassium is replaced by sodium. However, (K/P)IE values lower than 10% could only be achieved with a Kurrol salt to resin weight ratio of 1 to 10. Diluting the solution or increasing the mixing time did not decrease the (K/P)IE. In contrast, using the same amount of resin (1 to 6 weight ratio) but adding it in 3 steps (1 to 3:1.5:1.5) instead of one step (1:6) decreased the (K/P)IE to
Table 1 Summary of fractionation results for two NaPP glasses prepared by keeping the melt at 700 °C for either 1 or 3 h. Fractions of 700 °C for 1 h
Mn a (g/mol)
Moles in 52 g glass
Mw b (g/mol)
Residual waterc (wt.%)
Fraction1 46,002 1.5 70,520 3.5 Fraction2 37,536 3.4 49,960 4.1 Fraction3 28,458 2.7 26,270 7.4 Fraction4 19,278 5.5 21,980 7.0 Fraction5 11,934 6.8 13,410 5.3 Fraction6 6018 6.5 6990 3.7 Fraction7 N/A N/A N/A N/A M n of original NaPP glass based on the fractionation result (g/mol) M n of original NaPP glass based on its NMR measurement (g/mol) M w of original NaPP glass based on the fractionation result (g/mol) M w of original NaPP glass based on its viscosity measurement (g/mol) Poly-dispersity of original NaPP glass based on the fractionation result a b c d
Mn is determined based on liquid 31P NMR. M w is determined based on viscosity measurements. Residual water is determined based on phosphate concentration analysis. Weight loss is determined based on TGA analysis between 25 and 500 °C.
Mn a (g/mol)
Moles in 52 g glass
Mw b (g/mol)
Residual waterc (wt.%)
Weight lossd (wt.%)
Fractions of 700 °C for 3 h
11.09 10.27 10.69 10.08 9.6 13.51 N/A 19,019 20,072 26,908 33,550 1.41
Fraction1 27,744 5.2 38,770 10.6 Fraction2 19,584 4.6 24,580 9 Fraction3 16,830 3.8 18,320 8.6 Fraction4 12,648 7.1 12,620 5.8 Fraction5 8160 8.6 10,130 10.6 Fraction6 4794 7.3 5960 5.8 Fraction7 4182 4.5 3550 8.6 M n of original NaPP glass based on the fractionation result (g/mol) M n of original NaPP glass based on its NMR measurement (g/mol) M w of original NaPP glass based on the fractionation result (g/mol) M w of original NaPP glass based on its viscosity measurement (g/mol) Poly-dispersity of original NaPP glass based on the fractionation result
Weight lossd (wt.%) 10.18 10.92 10.4 14.11 10.22 11.96 8.98 12,469 13,770 17,186 24,630 1.38
16
A. Momeni, M.J. Filiaggi / Journal of Non-Crystalline Solids 382 (2013) 11–17
Fig. 1 shows the DSC/TGA graph of ion-exchanged NaPP and compares it to its precursor, Kurrol salt. Kurrol salt displayed three very small endothermic peaks at 247, 446 and 639 °C implying three phase transformations of Kurrol salt [18,19], followed by melting around 810 °C. In contrast, ion-exchanged NaPP melts at around 600 °C similar to shorter chain NaPPs. In the DSC graph of the ion-exchanged NaPP, a peak around 150 °C resembles a similar peak detected for the fractioned NaPP, while the peak at 330 °C resembles the peak that was observed in the NaPP glass. The TGA graph also revealed a 6–9% weight loss in ion-exchanged NaPPs after heating to 400 °C. Presumably, as the ionexchanged NaPP is heated, some of its residual water is evaporated, with the remainder reacting with the chains transforming them to shorter chains. The accurate amount of residual water in long chain ion-exchanged NaPPs was determined to be 8–12% based on phosphate concentration analysis. 4. Discussion
Fig. 4. (a) Dependence of the K/P ratio of ion-exchanged NaPPs on the experimental conditions; (b) viscosity results of three different ion-exchanged NaPPs in 0.35 N NaBr solution at 25 °C.
values lower than 5%. Based on these observations, the final procedure for obtaining NaPP with Dp higher than ≈500 and (K/P)IE lower than 5% was as follows: first, 4.5 g of dried Kurrol salt was added to 13.5 g of dry Amberlite™ IRP69 sodium resin in 100 ml water. The solution was rigorously mixed for 90 min at 2000 rpm at 30 °C in pH range of 8–8.5, continually adjusted by 0.1 M tetramethylammonium hydroxide solution. After mixing, the solution was paper-filtered to remove the resin and then in the second step, 6.75 g of fresh resin was added to the filtered solution and mixed under similar conditions for 30 min. This step was repeated, and the final solution paper-filtered and freezedried. Using this method the weight of NaPP obtained after freeze drying was always higher than 3.3 g, or a yield of higher than 80%. Even dilute solutions of these polyphosphates are viscous, a characteristic feature of long chain polymers. Both liquid P NMR and titration failed to determine the chain length of these long chain polyphosphates because both methods rely on their ability to detect the number of end group phosphates (Q1). Even after 2300 scans, no Q1 peak was detected (Fig. 2), suggesting Dp is at least higher than 3000 (M n ~306,000 g/mol) based on the signal to noise ratio. The only feasible way to determine the chain length was found to be viscosity measurements. Fig. 4(b) shows the viscosity results of these long chain polyphosphate. The intercept of the trendlines with the specific viscosity/concentration axis gives intrinsic viscosity that is related to M w based on Eq. (4). According to the viscosity measurements, NaPPs prepared by the ion exchange process are very long, many exceeding M w of 1 million g/mol. Furnace time indeed affected their chain lengths, with KH2PO4 maintained for 30 min, 2 h and 50 h at 775 °C yielding NaPPs with Dp of 8836 ± 333 (n = 3), 7017 ± 1495 (n = 5) and 19,454 ± 2234 (n = 3), respectively. Based on a twotailed Student's t-test, there was no significant difference between the Dp of 30 min and 2 h samples; however, both groups were significantly (p b 0.05) shorter than the 50 h sample group.
The most widely used method to obtain NaPPs with different Dp is the solubility fractionation with acetone described here which was first reported by Van Wazer [16]. The other method is chromatographic fractionation which results in more narrowly size distributed polyphosphates [20–23]. Nevertheless, chromatographic methods lose their ability to separate polyphosphates as the chains get longer. For instance, high performance liquid chromatography using an anion exchange column is only capable of separating chains up to 30 phosphates in length [20,21]. We found that the random distribution equation overestimates the percentage of total number of phosphorus atoms that are in polyphosphates with large Dp. The random distribution equation represents the distributions corresponding to a completely random process of chain making and scission, ignoring the increasing negative charge that is being built up as the phosphate chains grow in the melt. This growing negative charge could become an increasing repulsive force for the next phosphate anions that are seeking to bind with the chain and, as a result, a higher percentage of polyphosphate anions remain shorter than what is expected from theoretical estimation. In NaPPs produced through melting NaH2PO4·H2O the average chain length depends on the furnace time and temperature. Long chain phosphates are already formed when the melt is kept at 700 °C for 1 h, but when the furnace time is increased to 3 h their average chain length drops. When the melt is kept for the longer 9 h period or at the higher temperature of 970 °C, the average chain length grows again but never exceeds the average chain length which was already obtained at the lower temperature and the shorter furnace time. In theory the average chain length of a NaPP glass with the Na/P mole ratio of one should grow to infinity [17,19], but in practice several factors prevent chain growth; first, the Na/P mole ratio of the precursor material could deviate from the stoichiometric mole ratio of one due to impurities; second, as the chains grow the increasing negative charge on the chains prevent further growth; and third, the evaporation of the last traces of water from the melt becomes extremely difficult and these water molecules could act as chain terminators. For instance, if one water molecule remains for every 500-phosphate molecule, this will restrict growth to no higher than 500 phosphates per chain. There are two sources for water in the melt: water that is the byproduct of the condensation reaction, and water that is in the atmospheric air. In fact, Thilo [24] has shown the inverse relation between sodium polyphosphate chain length and water vapor pressure inside of the furnace. The Mw as measured by viscosity appears relatively inaccurate for the shorter chain NaPP fractions. This inaccuracy could be caused by significant errors that exist in determining the accurate intrinsic viscosity value. These errors may be the result of the low sensitivity of viscometers in measuring low viscosity solutions, or it could simply be a result of inaccuracy of constants which were determined by Strauss [13,14] for Eq. (3) based on his light scattering results.
A. Momeni, M.J. Filiaggi / Journal of Non-Crystalline Solids 382 (2013) 11–17
Even after fractionation the longest fractioned NaPP obtained by melting sodium phosphate monobasic had a Dp of ≈400. In contrast, when potassium phosphate monobasic is heated it crystallizes into potassium Kurrol salt, which is known to be constituted of polyphosphates with Dp much longer than 500 phosphorus per chain. There are many other crystalline compounds possessing a O/P ratio of 3 that share these characteristically long polyphosphate chains [18]. Examples include Maddrell salt, calcium polyphosphate crystalline phases, Zn(PO3)2, and Ba2K(PO3)5; however, none are water soluble. Potassium Kurrol salt is also not water soluble but, interestingly in the presence of small monovalent cations such as H+, Na+ or Li+, it becomes water soluble because these cations replace some of the potassium ions, distorting the crystalline structure and releasing the chains into the solution. Based on this phenomenon very long chain NaPPs were obtained from potassium Kurrol salt. The ion exchange process was optimized to efficiently remove the maximum amount of potassium from the solutions without wasting too much resin. The resulting NaPPs were very long, as evidenced from liquid 31P NMR and viscosity measurements. Due to the high viscosity of ion-exchanged NaPP solutions, the Mw determined by viscosity measurements is reliable in contrast to short chain NaPPs. TheM w of the ion-exchanged NaPP was directly correlated to the furnace time of KH2PO4. Although this dependence was previously suggested by Malmgren [25,26], his work did not include replicates to check reproducibility, lacked statistical analysis and overall was not convincing. Achieving ion-exchanged NaPP includes multiple steps and we took a great care to ensure every step of the process was identical between replicates. In fact, our preliminary studies when such care was not considered led to non-reproducible chain lengths. Polyphosphates grow into extremely long chains in crystalline phases because they are not affected by the factors that prevent the chain growth in melts; in the crystalline state the charge of the chains remain neutral and also any impurity such as water or excess cation is driven to the surface of crystal as they do not fit the crystal template [27]. For instance, in potassium Kurrol salt it is known that if a slight excess of K2O is present in the system, it won't act as the chain terminator and instead forms K5P3O10, crystallizing separately from Kurrol salt [10,17,27]. In such a system it is expected that longer heat treatments result in longer chains since the only factor limiting the chain growth is the kinetics of the system, that is, the ability of monomers to migrate to the surface of the crystals. Since in both fractionation and ion-exchange methods polyphosphates are introduced to water, the resulting NaPPs will contain residual water. This residual water could only be removed by heating, but as shown in the DSC results, the heating process will also degrade the chains. If this water is not removed, its exact amount should be determined for each NaPP sample and taken into account in experiments. Also, we have observed that over time this residual water degrades the polyphosphate chains that are kept under vacuum inside a desiccator
17
at room temperature; it is highly recommended to preserve NaPPs inside a freezer to slow down the degradation rate. 5. Conclusion NaPPs with chain lengths shorter than 500 phosphates per chain were produced successfully and reproducibly by fractionation of NaPP glass that was prepared from NaH2PO4·H2O. NaPPs with chain lengths longer than 500 phosphates per chain were produced successfully by ion exchange of Kurrol salt that was prepared from KH2PO4. The amount of remaining potassium in ion-exchanged NaPPs was reduced to K/P mole ratios lower than 5%. NaPP with its ability to gel in the presence of multivalent cations (e.g. calcium) is a promising inorganic polymer for bio-applications, and manipulating its chain length has been the key to optimizing its use. Acknowledgments The authors gratefully acknowledge the funding from the Natural Sciences and Engineering Research Council of Canada (NSERC) and NSERC CREATE BioMedic program. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27]
J. Berzelius, Ann. Phys. 53 (1816) 393–446. T. Graham, Philos. Trans. R. Soc. Lond. 123 (1833) 253–284. C.F. Callis, J.R. Van Wazer, P.G. Arvan, Chem. Rev. 54 (1954) 777–796. R.E. Hall, Water Softening and Washing, Hall Laboratories, Inc., USA, 1934. A. Kornberg, J. Bacteriol. 177 (1995) 491–496. I. Kulaev, T. Kulakovskaya, Annu. Rev. Microbiol. 54 (2000) 709–734. F.A. Ruiz, C.R. Lea, E. Oldfield, R. Docampo, J. Biol. Chem. 279 (2004) 44250–44257. J.H. Morrissey, S.H. Choi, S.A. Smith, Blood 119 (2012) 5972–5979. K.Y. Tam, K.C.-F. Leung, Y.-X.J. Wang, Eur. J. Pharm. Sci. 44 (2011) 1–10. R. Pfanstiel, R.K. Iler, J. Am. Chem. Soc. 74 (1952) 6059–6064. J.R. Van Wazer, C.F. Callis, J.N. Shoolery, R.C. Jones, J. Am. Chem. Soc. 78 (1956) 5715–5726. R.C. Mehrotra, Pure Appl. Chem. 44 (1975) 201–220. U.P. Strauss, P.L. Wineman, J. Am. Chem. Soc. 80 (1958) 2366–2371. U.P. Strauss, E.H. Smith, P.L. Wineman, J. Am. Chem. Soc. 75 (1953) 3935–3940. J.R. Van Wazer, J. Am. Chem. Soc. 72 (1950) 644–647. J.R. Van Wazer, J. Am. Chem. Soc. 72 (1950) 647–655. J.R. Van Wazer, Phosphorus and Its Compounds, Interscience Publishers, Inc., 1958 A. Durif, Crystal Chemistry of Condensed Phosphates, Plenum Press, 1995. D.E.C. Corbridge, The Structural Chemistry of Phosphorus, Elsevier Scientific Pub. Co., 1974 B.C. Sales, L.A. Boatner, J.O. Ramey, J. Non-Cryst. Solids 232–234 (1998) 107–112. B. Tischendorf, J.U. Otaigbe, J.W. Wiench, M. Pruski, B.C. Sales, J. Non-Cryst. Solids 282 (2001) 147–158. T. Miyajima, K. Yamauchi, S. Ohashi, J. Liq. Chromatogr. 4 (1981) 1891–1901. J. Liu, K. Shirahama, T. Miyajima, J.C.T. Kwak, Colloid Polym. Sci. 276 (1998) 40–45. V.E. Thilo, Poly-phosphate und -arsenate, International Symposium on Inorganic Polymers, Chemical Society, University of Nottingham, Great Britain, 1961. H. Malmgren, Acta Chem. Scand. 2 (1948) 147–165. H. Malmgren, Acta Chem. Scand. 6 (1952) 1–15. Edward J. Griffith, Factors influencing the chain lengths of inorganic polyphosphates, Phosphorus Chemistry, American Chemical Society, 1992, pp. 86–101.
Contents lists available at ScienceDirect
Journal of Non-Crystalline Solids journal homepage: www.elsevier.com/ locate/ jnoncrysol
Synthesis and characterization of different chain length sodium polyphosphates Arash Momeni a, Mark Joseph Filiaggi a,b,⁎ a b
School of Biomedical Engineering, Dalhousie University, Halifax, Nova Scotia B3H 3J5, Canada Faculty of Dentistry, Department of Applied Oral Sciences, Dalhousie University, 5981 University Avenue, Halifax, Nova Scotia B3H 3J5, Canada
a r t i c l e
i n f o
Article history: Received 28 June 2013 Received in revised form 2 October 2013 Available online 23 October 2013 Keywords: Sodium polyphosphate; Kurrol salt; Graham salt; Chain length
a b s t r a c t The simple chemistry of sodium polyphosphate (NaPP) makes it a very safe inorganic polymer for bioapplications. Recently we have focused on an in-situ forming system comprised of NaPP and calcium chloride solutions which, on contact, form a gel-like material. This system could be used in a variety of bio-applications. The first step of our studies has been synthesis and characterization of NaPP with the average degree of polymeriza tion Dp from 50 up to 25,000, that is summarized in this manuscript. NaPP with Dp lower than 500 was obtained from fractionation of a NaPP glass. NaPP with Dp higher than 500 was obtained from an ion-exchange process on a potassium polyphosphate crystalline phase. The ion-exchange process was optimized maximizing the potassium exchange with sodium while minimizing resin waste. The Dp of prepared NaPPs was determined by titration, viscosity and 31P-liquid nuclear magnetic resonance measurements. The accuracy of these methods in determining the chain length is discussed. The effect of furnace time and temperature on Dp is also discussed. In addition, since both fractionation and ion-exchange methods introduce water, the resulting NaPPs were evaluated for residual water by differential scanning calorimetry and elemental analysis methods. © 2013 Elsevier B.V. All rights reserved.
1. Introduction In 1816 Berzelius wrote about amorphous phosphoric acids characteristically ranging from viscous liquids, to materials that would be described as either glasses or unusually rigid liquids at room temperature [1]. Seventeen years later, Thomas Graham prepared vitreous sodium phosphate exhibiting the metaphosphate composition (Na2O/P2O5 = 1) by heating sodium dihydrogen phosphate to high temperature, followed by chilling the melt suddenly [2]. Referred to as Graham salt, this compound was erroneously labeled a sodium hexametaphosphate (a ringed structure) for over 100 years until the work by Van Wazer [3] concluded that Graham salt solutions consisted, instead, of a mixture of linear polymers (i.e. polyphosphates) with the following general chemical formula:
where X+ is a monovalent cation such as H+, Na+ or K+ and n is the degree of polymerization (Dp). ⁎ Corresponding author at: School of Biomedical Engineering, Dalhousie University, Halifax, Nova Scotia B3H 3J5, Canada. Tel.: +1 902 494 7102; fax: +1 902 494 6621. E-mail address: fi[email protected] (M.J. Filiaggi). 0022-3093/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jnoncrysol.2013.10.003
Shortly after the discovery of Graham salt in 1833 [2], it was found that its solutions have the ability to prevent precipitation or redissolve precipitates of the alkaline earth metals. This complexing ability was an obscure piece of information of little interest to either science or applied technology until 1933, when Hall applied this effect to the softening of water [4]. Since then, polyphosphates have widely been used as water softening agent in industry and municipal water supplies. They are also widely used to aid food processing or to improve the eating quality of many foods, particularly meat and fish products. Interestingly, polyphosphates are also found in every cell in nature. Their functions include acting as an energy supply and ATP substitute, an osmotic inert reservoir for phosphate, a chelator of metals, a buffer against alkali, a channel for DNA entry, a cell capsule and a regulator of responses to stresses [5,6]. In 2004, Ruiz et al. showed that human platelets are also rich in polyphosphate, which is secreted upon thrombin stimulation [7]. Later studies showed that polyphosphate is a very strong pro-coagulant interconnecting intrinsic and extrinsic pathways of the blood clotting cascade [8]. Our group has recently focused on a novel injectable in-situ forming system, based on polyphosphates, with the potential to deliver and release drugs locally in procedures such as trans-arterial chemoembolization (TACE). In TACE, a biomaterial is administered directly into blood vessels feeding a tumor thereby blocking its nutrient supply and also releasing chemotherapeutics locally [9]. Our in-situ forming system is comprised of sodium polyphosphate (NaPP) and calcium chloride solutions which, on contact, form a coacervate or a gel-like material that is immiscible with water. NaPP chain length is an
12
A. Momeni, M.J. Filiaggi / Journal of Non-Crystalline Solids 382 (2013) 11–17
important factor determining the physical and biological properties of this in-situ forming system; therefore, preparing different chain length polyphosphates has formed a major part of our research. NaPP glass is a distribution of polyphosphate chains and by its fractionation different short chain length polyphosphates could be obtained. Synthesis of long chain polyphosphate, however, is more complicated as it requires dissolution of a crystalline polyphosphate phase. In this paper we comprehensively describe synthesis and of NaPP characterization with the average degree of polymerization Dp from 50 up to 25,000. 2. Material and method 2.1. Preparation of NaPP 2.1.1. Preparation of NaPP with Dp b500 Sodium phosphate monobasic monohydrate (NaH2PO4·H2O; SigmaAldrich) was used to prepare NaPP with Dp lower than ≈500. NaH2PO4· H2O was heated in a Thermolyne Type 46200 High Temperature Furnace in a platinum–5% gold crucible from 25 °C to 700 °C over a 90 min period, keeping the melt at 700 °C for 1, 3 or 9 h followed by quenching on a copper plate. The oxide and hydroxide layers on the copper plate were removed prior to quenching using acetic acid and sodium chloride solution. In order to fraction the NaPP glass, the frit was grounded and dissolved in deionized water to give a 10% (w/v) NaPP solution. The solution was stirred using a magnetic stirrer for 12 h and then fractioned by serial addition of acetone. After each acetone addition, the solution was mixed for 15 min and then centrifuged at 4200 rpm for 5 min to collect the precipitate, which was subsequently frozen in the freezer at −20 °C. The frozen precipitate was freeze-dried using FreeZone® 1 Liter Benchtop Freeze Dry System (Labconco Corp., USA) for 24 h at −43 °C in pressures lower than 133 × 10−3 mbar resulting in a lyophilized product that was kept in a desiccator until use. 2.1.2. Preparation of NaPP with Dp N500 To prepare NaPP with Dp higher than ≈500, potassium phosphate monobasic (KH2PO4; Sigma-Aldrich) was heated in the furnace in a platinum–5% gold crucible from 25 °C to 775 °C over a 90 min period, then held at 775 °C for 30 min, 2 or 50 h prior to cooling the crucible bottom in water. The resulting product is a water insoluble crystalline phase of KPO3 known as the potassium Kurrol salt. The Kurrol salt was washed by preparing a 20% (w/v) aqueous suspension of its powder which was magnetically stirred at 4 °C for 20 min. The Kurrol salt was then paper-filtered, washed with ethanol and dried in a desiccator for 24 h. Water soluble NaPP was obtained from Kurrol salt by an ion exchange process in which potassium cations were replaced by sodium cations. A slurry of Kurrol salt was prepared in previously boiled deionized water followed by dry Amberlite™ IRP69 sodium resin addition and rigorous mixing using a mechanical stirrer (2000 rpm) at 30 °C in pH range of 8–8.5, continually adjusted by 0.1 M tetramethylammonium hydroxide solution. After mixing, the solution was paper-filtered to remove the resin, frozen and then freeze-dried for 72 h at −43 °C in pressures lower than 133 × 10−3 mbar to obtain NaPP with Dp higher than ≈500. The Kurrol salt/resin weight ratio, mixing time and concentration of the Kurrol salt in the slurry were varied in order to determine the optimum conditions under which a maximum amount of potassium could be replaced by sodium while minimizing resin waste. 2.2. Characterization of NaPP 2.2.1. Compositional analysis The K/P mole ratio of our stock KH2PO4 ðK=PÞKH2 PO4 was determined by Energy-Dispersive X-ray Spectroscopy (EDS) (Hitachi S-4700 FEG Scanning Electron Microscope equipped with Inca EDX system, Japan), Inductively Coupled Plasma Optical Emission Spectroscopy (ICP) (Optima™ 7300 V ICP-OES, PerkinElmer Instruments, USA) and a pH measurement method [10]. For the latter method, the pH of 0.1 M KH2PO4 in CO2-free
water under nitrogen gas at 25 °C was measured and its variation from the theoretical pH of KH2PO4 under similar conditions was used to determine the ðK=PÞKH2 PO4 . The amount of phosphate in all NaPP samples was assessed by dissolving 50 mg of each sample in 10 ml 6 N HCl, maintaining it in boiling water for 4 h, followed by 20 h cool down period and then titration by 1 M NaOH. The moles of 1 M NaOH required between two inflection points, pH 3.5 and 8, are equal to the moles of phosphate in 50 mg of a sample. The efficiency of the ion exchange process was also evaluated by measuring the remaining potassium to phosphorus mole ratio in the NaPP ((K/P)IE). Here, the amounts of potassium and phosphorus in ion-exchanged samples were measured by PerkinElmer 2380 Atomic Absorption Spectroscopy (AAS) and titration, respectively. The (K/P)IE was also verified by ICP only in some of the samples. 2.2.2. Thermal analysis Thermal behavior and amount of residual water in all NaPP samples after lyophilization was determined by Differential Scanning Calorimetry and Thermal Gravity Analysis (DSC/TGA) using a STA 409 PC LUXX® thermal analyzer (Netzsch, Germany). The baseline was obtained using an empty platinum crucible and then the instrument was calibrated. Approximately 20 mg of each sample was added to the platinum crucible and the DSC/TGA data was collected from 25 °C to 1000°C at the heating rate of 10°C/min. DSC/TGA data of NaH2PO4·H2O, KH2PO4 and Kurrol salt were also obtained. All DSC/TGA data were analyzed using Proteus® Software. 2.2.3. Dp analysis Liquid 31P nuclear magnetic resonance (NMR), titration and viscosity studies were carried out to determine the chain length of NaPP. 100 mg of NaPP was dissolved in 0.75 ml D2O to be analyzed by a Bruker AV 300 MHz NMR spectrometer at 101.26 MHz, using a 151 pulse, 65,536 (64 k) data points giving a 0.8 s acquisition time, a 7.0 s repetition rate and greater than 100 scans. Spectra were reported using the d scale, with positive values downfield, and were referenced to an 85% solution of H3PO4 in H2O. The peaks around 0, −9, −21, −23 and −35ppm represent the Q0 (orthophosphates), Q1 (end of chain phosphorus atoms), Q2 (middle of chain phosphorus atoms in polyphosphates), Meta-Q2 (middle of chain phosphorus atoms in rings) and Q3 (branched phosphorus atoms), respectively [11]. The Dp of NaPP was determined by the following equation based on the area under the peaks:
Dp ¼
2 Q1 þ Q2 þ Q3 Q1
:
ð1Þ
To determine the chain length by titration, 100 mg of NaPP was dissolved in 10 ml deionized water and titrated by 1 M and 0.1 M NaOH solutions. The moles of NaOH that were used to titrate the NaPP solution between two inflection points, pH 4.5 and 9, were used to determine the Dp based on the following equation [12]: Dp ¼
20 weight ðgÞ of NaPP dissolved in solution : ml of 1 N NaOH required between two inflection points
ð2Þ
Both NMR and titration give number average Dp and therefore, the number average molecular weight Mn is equal to 102ñDp , where 102 is the molecular weight of NaPO3 monomer. In contrast, viscosity measurements give the weight average molecular weight Mw . NaPP with Dp lower than 500 and those with chain lengths higher than 500 were dissolved in 0.035 N and 0.35 N NaBr solutions, respectively, at concentrations lower than 0.015 g/ml. The viscosity of these solutions was measured using DV3™ cone–plate rheometer (Brookfield, USA) at 25°C and the linear graph of (specific viscosity/concentration) is plotted against concentration. The extrapolation of this graph to zero
A. Momeni, M.J. Filiaggi / Journal of Non-Crystalline Solids 382 (2013) 11–17
concentration yields intrinsic viscosity ([η]) that is used to calculate the Mw based on the following equations [13,14]: −5 ½ηðin 0:035 N NaBrÞ ¼ 10 Mw þ 0:025;
for NaPP with Dp b 500 at 25 C ð3Þ
−4
½ηðin 0:35 N NaBrÞ ¼ 0:65 10 Mw
0:69
;
for NaPP with Dp N 500 at 25 C:ð4Þ
3. Results 3.1. NaPP with Dp b500 Fig. 1 shows the DSC/TGA graph of NaH2PO4·H2O. From ~70 °C to 150 °C, the compound loses 12.76% of its weight implying the loss of one molecule of water per phosphate unit and formation of NaH2PO4. Around 220 °C, there is a further 6.52% mass loss suggesting the release of one water molecule per two phosphate units and formation of Na2H2P2O7, with a subsequent 6.47% mass loss again at approximately 350 °C, implying the loss of another molecule of water per two phosphate units and formation of trimetaphosphate rings (i.e. Na3P3O9). Finally, the compound melts around 640 °C and NaPP glass is obtained by quenching this melt. Quenching on a stainless steel oxide plate formed a black layer on the surface of the glass that was in contact with the plate, suggesting a reaction between the melt and the stainless steel oxide layer. In contrast, glasses quenched on the copper plate were visually clear, with EDS analysis showing no sign of copper contamination in the glass. Fig. 2(a) shows the liquid 31P NMR spectra of NaPP glasses kept at 700 °C for 1, 3 or 9 h, corresponding to a Dp of 186 ± 17, 131 ± 12 or 174 ± 15 (n = 3), respectively. Based on a two-tailed Student's t-test, there is no significant difference between the Dp of 1 and 9 h samples, but both groups are significantly (p b 0.05) longer than the 3 h samples. A NaPP glass was also prepared by keeping the melt at the higher temperature of 970°C for 1h. This glass had a Dp of 183±10 (n=3), which is not significantly different from the glass that was prepared by keeping the melt at 700 °C for 1 h. Based on the NMR results, the fraction of ring metaphosphates (Meta-Q2) was consistently 4.5–5% of the total phosphorus atoms in all of the NaPP glasses. A very small peak representing Q3 sites was also detected in dissolved NaPP glasses, with
13
its area implying less than one phosphate branching point for every 1000 phosphate units. The NaPP glasses that were kept at 700°C for 1 or 3h were fractioned by serial addition of acetone to their solutions. As shown in Fig. 2(b), long chains precipitated early with a small amount of acetone, but precipitation of shorter chains required much more acetone. Fig. 2(b) also shows the Dp of each of the fractions. The original NaPP glass that was fractioned had a Dp of 197, but based on its fractions, it constitutes a wide distribution of different polyphosphate chains with Dp starting from 56 up to 398. The fraction with Dp of 270 contributes almost 30% to the original NaPP glass mass. Fig. 2(c) shows the corresponding liquid 31 P NMR spectra for these eight fractions. As the chains get longer the area under the Q1 peak becomes smaller while the area under the Q2 peak becomes larger. Fig. 2(d) shows the titration of the same fractioned NaPPs; the required volume of NaOH increases as the chains become shorter. The Dp determined by NMR and titration correlates as shown in Fig. 3(a). For short chains, NMR and titration results are the same as shown by the solid line with the slope of 1.04. However, as the chains get longer the accuracy and precision of titration method decreases and these two methods do not give the same results. Also, as the chains get longer more NMR scans are required to achieve an accurate result. Theoretically, the distribution of chains in a NaPP glass with an original Dp of n can be determined with the random distribution equation [15,16]:
fx ¼
x n−2 x−1 ; nðn−2Þ n−1
for x N 2; n N 2
ð5Þ
where fx is the fraction of total number of phosphorus atoms having the Dp of x. Using this equation, a comparison between theoretical and experimental distributions of our polyphosphates was carried out in two NaPP glasses prepared by keeping the melt at 700 °C for 1 and 3 h. In Fig. 3(b), the separation between stepwise curves (experiment) and solid line curves (theory) becomes more prominent as degree of polymerization of fractions (y-axis) increases. For instance, in the polyphosphate with the Dp of 197, 36% of phosphates are theoretically in chains with Dp larger than 400, but experimentally this value is only 23%. This indicates that the experimental distribution fits the theoretical distribution perfectly at low Dp, but as chains get longer the
Fig. 1. DSC (solid lines) and TGA (dashed lines) graphs. With the exception of NaH2PO4·H2O, DSC/TGA graphs of all other compounds are shifted down for clarity.
14
A. Momeni, M.J. Filiaggi / Journal of Non-Crystalline Solids 382 (2013) 11–17
b)
a)
d)
c) Fig. 2. (a) Liquid 31P NMR spectra of NaPP glasses prepared by keeping the melt at 700 °C for 1, 3 and 9 h; (b) fractionation result of a NaPP glass with Dp of 197: the graph shows the amount of acetone required for precipitation of each fraction and columns show the percentage weight of each fraction; (c) liquid 31P NMR spectra and (d) titration graphs of NaPP fractions obtained from a NaPP glass with Dp of 197: numbers show the measured Dp based on the method. With the exception of lowest spectrum, all other NMR spectra are shifted downfield for clarity.
theoretical curve overestimates the percentage of total number of phosphorus atoms in polyphosphates with large Dp. Table 1 summarizes the fractionation results of two NaPP glasses. The M n value measured by the NMR for the original glasses correlates with the Mn value calculated using the following equation based on the fractionation results. X
Ni Mi Mn ¼ X Ni
ð6Þ
The M w of NaPPs as determined using viscosity measurements is also displayed in Table 1 along with Mw and poly-dispersity values calculated using the following equations based on the fractionation results: X 2 Ni Mi Mw ¼ X N i Mi Poly‐dispersity ¼
ð7Þ Mw Mn
:
ð8Þ
A. Momeni, M.J. Filiaggi / Journal of Non-Crystalline Solids 382 (2013) 11–17
15
due to the residual water. There is a 2.66% weight loss between 80 and 135 °C, followed by a slow weight loss of 1.19% until 350 °C, where a sharp weight loss of 6.87% occurs. There is a reaction happening between residual water and NaPP glass between 80 and 135 °C because the weight loss is accompanied by an exothermic rather than an endothermic peak. Presumably polyphosphate chains are being degraded into sodium pyrophosphate, which then transforms into sodium trimetaphosphate ring at 350 °C. The percentage of weight loss upon heating is reported in Table 1 for all fractioned NaPP samples. Since water is reacting with the NaPP glass, TGA is not an accurate method for determining the percentage of residual water in the fractioned samples. Instead, the amount of sodium phosphate in each sample was determined by titration and subtracted from the original weight of the sample to determine the weight percentage of residual water. As reported in Table 1, fractioned NaPP glasses contain between 3 and 11% residual water by weight. 3.2. NaPP with Dp N 500
Fig. 3. (a) Comparing the chain length measurements by titration and liquid 31P NMR (n = 3 and error bars show STD). The solid line is a trendline passing the first five points while the dashed line passes all of the seven points; (b) comparing the theoretical and experimental distributions of polyphosphate in NaPP glasses prepared by keeping the melt at 700 °C for 1 and 3 h with original Dp of 197 and 135, respectively. Stepwise curves show experimental Dp of fractions plotted against cumulative percentage of phosphorus and solid line curves show theoretical plots of Dp against cumulative percentage of phosphorus calculated from random distribution equation.
Here, Mw measured by these two methods did not correlate. One should note that fractioned NaPPs contain residual water after freeze-drying. Fig. 1 compares the DSC/TGA graph of un-fractioned NaPP glass with that of fractioned NaPP. Since the un-fractioned NaPP has no residual water, there is no weight loss in the TGA graph. There is, however, an exothermic peak around 350 °C in the DSC graph representing transformation into sodium trimetaphosphate ring or long chain NaPO3-II [17]. This is followed by melting around 630 °C. In contrast, the DSC/TGA graph of fractioned NaPP glass is more complex
Small variations of the ðK=PÞKH2 PO4 from unity is reported to significantly affect the chain length of NaPP that is obtained through potassium Kurrol salt [10]. The ðK=PÞKH2 PO4 of our stock material was 0.92 ± 0.02 based on EDS, while ICP gave a value of 1.05 ± 0.01. However, neither of these methods is sensitive enough to confirm the exact value of ðK=PÞKH2 PO4 . Instead, the pH of 0.1M KH2PO4 in CO2-free water under nitrogen gas at 25 °C was measured, and its variation from the theoretical pH of KH2PO4 under similar conditions was used to determine the ðK=PÞKH2 PO4 . Based on this method the ðK=PÞKH2 PO4 was 0.9996 ± 0.00106, essentially equal to one, and therefore no ðK=PÞKH2 PO4 adjustment of the stock material was performed. Fig. 1 shows the DSC/TGA graph of KH2PO4 where a sudden weight loss of 13.13% occurs between 182 and 358 °C, implying the loss of one water molecule per phosphate unit and formation of long chain KPO3, potassium Kurrol salt. For this process, an ion exchange resin was used to dissolve the Kurrol salt, release the polyphosphate chains and replace the potassium with sodium. The challenge was to replace the maximum amount of potassium with sodium in long chain polyphosphate solutions. Fig. 4(a) shows the dependence of (K/P)IE to the amount of resin that was used during the ion exchange process. The higher the amount of resin, the more potassium is replaced by sodium. However, (K/P)IE values lower than 10% could only be achieved with a Kurrol salt to resin weight ratio of 1 to 10. Diluting the solution or increasing the mixing time did not decrease the (K/P)IE. In contrast, using the same amount of resin (1 to 6 weight ratio) but adding it in 3 steps (1 to 3:1.5:1.5) instead of one step (1:6) decreased the (K/P)IE to
Table 1 Summary of fractionation results for two NaPP glasses prepared by keeping the melt at 700 °C for either 1 or 3 h. Fractions of 700 °C for 1 h
Mn a (g/mol)
Moles in 52 g glass
Mw b (g/mol)
Residual waterc (wt.%)
Fraction1 46,002 1.5 70,520 3.5 Fraction2 37,536 3.4 49,960 4.1 Fraction3 28,458 2.7 26,270 7.4 Fraction4 19,278 5.5 21,980 7.0 Fraction5 11,934 6.8 13,410 5.3 Fraction6 6018 6.5 6990 3.7 Fraction7 N/A N/A N/A N/A M n of original NaPP glass based on the fractionation result (g/mol) M n of original NaPP glass based on its NMR measurement (g/mol) M w of original NaPP glass based on the fractionation result (g/mol) M w of original NaPP glass based on its viscosity measurement (g/mol) Poly-dispersity of original NaPP glass based on the fractionation result a b c d
Mn is determined based on liquid 31P NMR. M w is determined based on viscosity measurements. Residual water is determined based on phosphate concentration analysis. Weight loss is determined based on TGA analysis between 25 and 500 °C.
Mn a (g/mol)
Moles in 52 g glass
Mw b (g/mol)
Residual waterc (wt.%)
Weight lossd (wt.%)
Fractions of 700 °C for 3 h
11.09 10.27 10.69 10.08 9.6 13.51 N/A 19,019 20,072 26,908 33,550 1.41
Fraction1 27,744 5.2 38,770 10.6 Fraction2 19,584 4.6 24,580 9 Fraction3 16,830 3.8 18,320 8.6 Fraction4 12,648 7.1 12,620 5.8 Fraction5 8160 8.6 10,130 10.6 Fraction6 4794 7.3 5960 5.8 Fraction7 4182 4.5 3550 8.6 M n of original NaPP glass based on the fractionation result (g/mol) M n of original NaPP glass based on its NMR measurement (g/mol) M w of original NaPP glass based on the fractionation result (g/mol) M w of original NaPP glass based on its viscosity measurement (g/mol) Poly-dispersity of original NaPP glass based on the fractionation result
Weight lossd (wt.%) 10.18 10.92 10.4 14.11 10.22 11.96 8.98 12,469 13,770 17,186 24,630 1.38
16
A. Momeni, M.J. Filiaggi / Journal of Non-Crystalline Solids 382 (2013) 11–17
Fig. 1 shows the DSC/TGA graph of ion-exchanged NaPP and compares it to its precursor, Kurrol salt. Kurrol salt displayed three very small endothermic peaks at 247, 446 and 639 °C implying three phase transformations of Kurrol salt [18,19], followed by melting around 810 °C. In contrast, ion-exchanged NaPP melts at around 600 °C similar to shorter chain NaPPs. In the DSC graph of the ion-exchanged NaPP, a peak around 150 °C resembles a similar peak detected for the fractioned NaPP, while the peak at 330 °C resembles the peak that was observed in the NaPP glass. The TGA graph also revealed a 6–9% weight loss in ion-exchanged NaPPs after heating to 400 °C. Presumably, as the ionexchanged NaPP is heated, some of its residual water is evaporated, with the remainder reacting with the chains transforming them to shorter chains. The accurate amount of residual water in long chain ion-exchanged NaPPs was determined to be 8–12% based on phosphate concentration analysis. 4. Discussion
Fig. 4. (a) Dependence of the K/P ratio of ion-exchanged NaPPs on the experimental conditions; (b) viscosity results of three different ion-exchanged NaPPs in 0.35 N NaBr solution at 25 °C.
values lower than 5%. Based on these observations, the final procedure for obtaining NaPP with Dp higher than ≈500 and (K/P)IE lower than 5% was as follows: first, 4.5 g of dried Kurrol salt was added to 13.5 g of dry Amberlite™ IRP69 sodium resin in 100 ml water. The solution was rigorously mixed for 90 min at 2000 rpm at 30 °C in pH range of 8–8.5, continually adjusted by 0.1 M tetramethylammonium hydroxide solution. After mixing, the solution was paper-filtered to remove the resin and then in the second step, 6.75 g of fresh resin was added to the filtered solution and mixed under similar conditions for 30 min. This step was repeated, and the final solution paper-filtered and freezedried. Using this method the weight of NaPP obtained after freeze drying was always higher than 3.3 g, or a yield of higher than 80%. Even dilute solutions of these polyphosphates are viscous, a characteristic feature of long chain polymers. Both liquid P NMR and titration failed to determine the chain length of these long chain polyphosphates because both methods rely on their ability to detect the number of end group phosphates (Q1). Even after 2300 scans, no Q1 peak was detected (Fig. 2), suggesting Dp is at least higher than 3000 (M n ~306,000 g/mol) based on the signal to noise ratio. The only feasible way to determine the chain length was found to be viscosity measurements. Fig. 4(b) shows the viscosity results of these long chain polyphosphate. The intercept of the trendlines with the specific viscosity/concentration axis gives intrinsic viscosity that is related to M w based on Eq. (4). According to the viscosity measurements, NaPPs prepared by the ion exchange process are very long, many exceeding M w of 1 million g/mol. Furnace time indeed affected their chain lengths, with KH2PO4 maintained for 30 min, 2 h and 50 h at 775 °C yielding NaPPs with Dp of 8836 ± 333 (n = 3), 7017 ± 1495 (n = 5) and 19,454 ± 2234 (n = 3), respectively. Based on a twotailed Student's t-test, there was no significant difference between the Dp of 30 min and 2 h samples; however, both groups were significantly (p b 0.05) shorter than the 50 h sample group.
The most widely used method to obtain NaPPs with different Dp is the solubility fractionation with acetone described here which was first reported by Van Wazer [16]. The other method is chromatographic fractionation which results in more narrowly size distributed polyphosphates [20–23]. Nevertheless, chromatographic methods lose their ability to separate polyphosphates as the chains get longer. For instance, high performance liquid chromatography using an anion exchange column is only capable of separating chains up to 30 phosphates in length [20,21]. We found that the random distribution equation overestimates the percentage of total number of phosphorus atoms that are in polyphosphates with large Dp. The random distribution equation represents the distributions corresponding to a completely random process of chain making and scission, ignoring the increasing negative charge that is being built up as the phosphate chains grow in the melt. This growing negative charge could become an increasing repulsive force for the next phosphate anions that are seeking to bind with the chain and, as a result, a higher percentage of polyphosphate anions remain shorter than what is expected from theoretical estimation. In NaPPs produced through melting NaH2PO4·H2O the average chain length depends on the furnace time and temperature. Long chain phosphates are already formed when the melt is kept at 700 °C for 1 h, but when the furnace time is increased to 3 h their average chain length drops. When the melt is kept for the longer 9 h period or at the higher temperature of 970 °C, the average chain length grows again but never exceeds the average chain length which was already obtained at the lower temperature and the shorter furnace time. In theory the average chain length of a NaPP glass with the Na/P mole ratio of one should grow to infinity [17,19], but in practice several factors prevent chain growth; first, the Na/P mole ratio of the precursor material could deviate from the stoichiometric mole ratio of one due to impurities; second, as the chains grow the increasing negative charge on the chains prevent further growth; and third, the evaporation of the last traces of water from the melt becomes extremely difficult and these water molecules could act as chain terminators. For instance, if one water molecule remains for every 500-phosphate molecule, this will restrict growth to no higher than 500 phosphates per chain. There are two sources for water in the melt: water that is the byproduct of the condensation reaction, and water that is in the atmospheric air. In fact, Thilo [24] has shown the inverse relation between sodium polyphosphate chain length and water vapor pressure inside of the furnace. The Mw as measured by viscosity appears relatively inaccurate for the shorter chain NaPP fractions. This inaccuracy could be caused by significant errors that exist in determining the accurate intrinsic viscosity value. These errors may be the result of the low sensitivity of viscometers in measuring low viscosity solutions, or it could simply be a result of inaccuracy of constants which were determined by Strauss [13,14] for Eq. (3) based on his light scattering results.
A. Momeni, M.J. Filiaggi / Journal of Non-Crystalline Solids 382 (2013) 11–17
Even after fractionation the longest fractioned NaPP obtained by melting sodium phosphate monobasic had a Dp of ≈400. In contrast, when potassium phosphate monobasic is heated it crystallizes into potassium Kurrol salt, which is known to be constituted of polyphosphates with Dp much longer than 500 phosphorus per chain. There are many other crystalline compounds possessing a O/P ratio of 3 that share these characteristically long polyphosphate chains [18]. Examples include Maddrell salt, calcium polyphosphate crystalline phases, Zn(PO3)2, and Ba2K(PO3)5; however, none are water soluble. Potassium Kurrol salt is also not water soluble but, interestingly in the presence of small monovalent cations such as H+, Na+ or Li+, it becomes water soluble because these cations replace some of the potassium ions, distorting the crystalline structure and releasing the chains into the solution. Based on this phenomenon very long chain NaPPs were obtained from potassium Kurrol salt. The ion exchange process was optimized to efficiently remove the maximum amount of potassium from the solutions without wasting too much resin. The resulting NaPPs were very long, as evidenced from liquid 31P NMR and viscosity measurements. Due to the high viscosity of ion-exchanged NaPP solutions, the Mw determined by viscosity measurements is reliable in contrast to short chain NaPPs. TheM w of the ion-exchanged NaPP was directly correlated to the furnace time of KH2PO4. Although this dependence was previously suggested by Malmgren [25,26], his work did not include replicates to check reproducibility, lacked statistical analysis and overall was not convincing. Achieving ion-exchanged NaPP includes multiple steps and we took a great care to ensure every step of the process was identical between replicates. In fact, our preliminary studies when such care was not considered led to non-reproducible chain lengths. Polyphosphates grow into extremely long chains in crystalline phases because they are not affected by the factors that prevent the chain growth in melts; in the crystalline state the charge of the chains remain neutral and also any impurity such as water or excess cation is driven to the surface of crystal as they do not fit the crystal template [27]. For instance, in potassium Kurrol salt it is known that if a slight excess of K2O is present in the system, it won't act as the chain terminator and instead forms K5P3O10, crystallizing separately from Kurrol salt [10,17,27]. In such a system it is expected that longer heat treatments result in longer chains since the only factor limiting the chain growth is the kinetics of the system, that is, the ability of monomers to migrate to the surface of the crystals. Since in both fractionation and ion-exchange methods polyphosphates are introduced to water, the resulting NaPPs will contain residual water. This residual water could only be removed by heating, but as shown in the DSC results, the heating process will also degrade the chains. If this water is not removed, its exact amount should be determined for each NaPP sample and taken into account in experiments. Also, we have observed that over time this residual water degrades the polyphosphate chains that are kept under vacuum inside a desiccator
17
at room temperature; it is highly recommended to preserve NaPPs inside a freezer to slow down the degradation rate. 5. Conclusion NaPPs with chain lengths shorter than 500 phosphates per chain were produced successfully and reproducibly by fractionation of NaPP glass that was prepared from NaH2PO4·H2O. NaPPs with chain lengths longer than 500 phosphates per chain were produced successfully by ion exchange of Kurrol salt that was prepared from KH2PO4. The amount of remaining potassium in ion-exchanged NaPPs was reduced to K/P mole ratios lower than 5%. NaPP with its ability to gel in the presence of multivalent cations (e.g. calcium) is a promising inorganic polymer for bio-applications, and manipulating its chain length has been the key to optimizing its use. Acknowledgments The authors gratefully acknowledge the funding from the Natural Sciences and Engineering Research Council of Canada (NSERC) and NSERC CREATE BioMedic program. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27]
J. Berzelius, Ann. Phys. 53 (1816) 393–446. T. Graham, Philos. Trans. R. Soc. Lond. 123 (1833) 253–284. C.F. Callis, J.R. Van Wazer, P.G. Arvan, Chem. Rev. 54 (1954) 777–796. R.E. Hall, Water Softening and Washing, Hall Laboratories, Inc., USA, 1934. A. Kornberg, J. Bacteriol. 177 (1995) 491–496. I. Kulaev, T. Kulakovskaya, Annu. Rev. Microbiol. 54 (2000) 709–734. F.A. Ruiz, C.R. Lea, E. Oldfield, R. Docampo, J. Biol. Chem. 279 (2004) 44250–44257. J.H. Morrissey, S.H. Choi, S.A. Smith, Blood 119 (2012) 5972–5979. K.Y. Tam, K.C.-F. Leung, Y.-X.J. Wang, Eur. J. Pharm. Sci. 44 (2011) 1–10. R. Pfanstiel, R.K. Iler, J. Am. Chem. Soc. 74 (1952) 6059–6064. J.R. Van Wazer, C.F. Callis, J.N. Shoolery, R.C. Jones, J. Am. Chem. Soc. 78 (1956) 5715–5726. R.C. Mehrotra, Pure Appl. Chem. 44 (1975) 201–220. U.P. Strauss, P.L. Wineman, J. Am. Chem. Soc. 80 (1958) 2366–2371. U.P. Strauss, E.H. Smith, P.L. Wineman, J. Am. Chem. Soc. 75 (1953) 3935–3940. J.R. Van Wazer, J. Am. Chem. Soc. 72 (1950) 644–647. J.R. Van Wazer, J. Am. Chem. Soc. 72 (1950) 647–655. J.R. Van Wazer, Phosphorus and Its Compounds, Interscience Publishers, Inc., 1958 A. Durif, Crystal Chemistry of Condensed Phosphates, Plenum Press, 1995. D.E.C. Corbridge, The Structural Chemistry of Phosphorus, Elsevier Scientific Pub. Co., 1974 B.C. Sales, L.A. Boatner, J.O. Ramey, J. Non-Cryst. Solids 232–234 (1998) 107–112. B. Tischendorf, J.U. Otaigbe, J.W. Wiench, M. Pruski, B.C. Sales, J. Non-Cryst. Solids 282 (2001) 147–158. T. Miyajima, K. Yamauchi, S. Ohashi, J. Liq. Chromatogr. 4 (1981) 1891–1901. J. Liu, K. Shirahama, T. Miyajima, J.C.T. Kwak, Colloid Polym. Sci. 276 (1998) 40–45. V.E. Thilo, Poly-phosphate und -arsenate, International Symposium on Inorganic Polymers, Chemical Society, University of Nottingham, Great Britain, 1961. H. Malmgren, Acta Chem. Scand. 2 (1948) 147–165. H. Malmgren, Acta Chem. Scand. 6 (1952) 1–15. Edward J. Griffith, Factors influencing the chain lengths of inorganic polyphosphates, Phosphorus Chemistry, American Chemical Society, 1992, pp. 86–101.