Thermodynamic Measurement of Di-calcium Phosphate TAKASHI NAGAI, YUSUKE TANAKA, and MASAFUMI MAEDA The thermodynamic properties of the CaO-P2O5 system are important to develop an effective refining process in the iron and steel industry. In this study, the thermodynamic properties of (CaO)2P2O5 were investigated because the properties are necessary to develop a new dephosphorization process. The vapor of gaseous phosphorus and phosphorus oxide in equilibrium with a mixture of (CaO)2P2O5 and (CaO)3P2O5 at 1373 K to 1498 K (1100 °C to 1225 °C) were detected directly as an ion current by double Knudsen cell mass spectrometry. Comparing the ion currents with those from a mixture of Al2O3P2O5 and Al2O3, which is used as a reference mixture in this study, the Gibbs energy change of the following reaction was calculated: 5 2CaOðsÞ þ P2 ðgÞ þ O2 ðgÞ ¼ ðCaOÞ2 P2 O5 ðsÞ 2 DG ¼ 2180000 þ 590Tð28000ÞJ:

DOI: 10.1007/s11663-011-9509-2 Ó The Minerals, Metals & Materials Society and ASM International 2011

I.

INTRODUCTION

HOT metal pretreatment for dephosphorization is a common practice in the iron and steel industry, and it has been developed as an effective refining process. The thermodynamic properties of the CaO-P2O5 system are important because a flux based on this system is used in this process. Figure 1 shows a phase diagram of the CaO-P2O5 binary system with its many intermediate compounds between CaO and P2O5.[1] Many researchers have reported the thermodynamic data of calcium phosphates such as tetracalcium phosphate (CaO)4P2O5 and tri-calcium phosphate (CaO)3P2O5, which is determined by chemical equilibrium methods[2–6] or electromotive force (EMF),[7,8] because the compounds are products of a traditional dephosphorization reaction in steelmaking. Their thermodynamic properties are critical in understanding the behavior of phosphorus in steelmaking thermodynamically. The authors also reported the thermodynamic properties of these compounds determined by double Knudsen mass spectrometry.[9] Although CaF2 is added to the flux to decrease its melting point, the process should be improved because of regulations on the emission of florin. A new dephosphorization process has been tried, using a flux with higher TAKASHI NAGAI, Research Assistant, and MASAFUMI MAEDA, Professor, are with the Institute of Industrial Science, The University of Tokyo, Meguro-ku, Tokyo 153-8505, Japan. Contact e-mail: nagait@ iis.u-tokyo.ac.jp YUSUKE TANAKA, formerly Graduate Student with the Institute of Industrial Science, The University of Tokyo, is Lead Engineer with the JFE Steel Corporation, Chiyoda-ku, Tokyo 260-0835, Japan. Manuscript submitted September 8, 2010. Article published online April 22, 2011. METALLURGICAL AND MATERIALS TRANSACTIONS B

concentration of FetO, which has a melting point lower than that in the traditional process, or a multiphase flux, which involves two or more phases.[10–15] This makes it necessary to investigate the thermodynamic properties of the CaO-P2O5 system under conditions not investigated previously. The thermodynamic properties of di-calcium phosphate (CaO)2P2O5 were limited. Although three groups reported the properties estimated by chemical equilibrium methods[16,17] or EMF,[18,19] the only data available at the temperature of the process, which is higher than 1423 K, the melting point of carbon saturated iron, has been provided by Hoshino et al.[19] It is necessary to reexamine the properties by another method because thermodynamic properties of other compounds by EMF were different from those by the other methods.[9,20] In this study, thermodynamic properties of di-calcium phosphate, (CaO)2P2O5, were investigated by measuring gaseous species in equilibrium with a mixture of (CaO)3P2O5 and (CaO)2P2O5 using double Knudsen cell mass spectrometry. 5 2ðCaOÞ3 P2 O5 ðsÞ þ P2 ðgÞ þ O2 ¼ 3ðCaOÞ2 P2 O5 ðsÞ ½1 2

II.

EXPERIMENTAL PROCEDURE

A. Double Knudsen Cell Mass Spectrometry Knudsen cell mass spectrometry is developed to measure vapor pressures in equilibrium with condensed phases. In this method, vapor pressure is measured by a mass spectrometer as an ion current, which is proportional to the pressure. The pressure pi and ion current of VOLUME 42B, AUGUST 2011—685

in a nearby heating element and a PID-type thermoregulator. The temperatures of cells containing mixtures were measured by three thermocouples placed in holes drilled at the bottom of the cell holder. Knudsen cells were made of molybdenum. The dimensions of the outer cell were 10 mm outer diameter, 8 mm inner diameter, and 18 mm height. The lid of the outer cell had an orifice with a diameter of 0.4 mm. B. Preparation of Oxide Mixture

Fig. 1—Phase diagram of CaO-P2O5 binary system.

i species Ii measured by the mass spectrometer are related through the following fundamental equation[21]: pi ¼

T Ii Si

½2

where T is the absolute temperature of a specimen and Si is the device dependent constant that includes such factors as ionization cross-section and efficiency of the ion detector. The usual accuracy of the data determined by single Knudsen cell mass spectrometry is not always adequate, because the constancy of the device dependent constant Si is difficult to maintain over a long period, and the reproducibility of each experimental run might not be adequate. To overcome this problem, a technique using double Knudsen cells has been developed. The double Knudsen cells allow the measurement of ion currents of evaporated species from two specimens under identical conditions in a single experiment: One is an experimental specimen and the other taken as a reference. Details of the equipment for double Knudsen cell mass spectrometry are described elsewhere,[9,20,22–25] and therefore, they are explained only briefly here. A chamber was vacuumed by rotary pumps (RPs) and turbo molecular pumps (TMPs), and residual gas pressure in this chamber was kept below 1 9 10–4 Pa. A cell holder in the chamber held two Knudsen cells. An experimental substance and a reference substance were charged separately in the Knudsen cells and were installed in a high-vacuum chamber. Two Knudsen cells at a time were set and heated in this chamber; one of them was used for the measured substances and one was for reference. A quadrupole mass spectrometer (QMS; Leybold Inficon H200M, Inficon, East Syracuse, NY) installed on top of the chamber detected the atomic beam of evaporated species from the Knudsen cells. The cells were heated by a Ta electric resistance-heating element in the vacuum chamber. The temperature was controlled using a thermocouple placed 686—VOLUME 42B, AUGUST 2011

Aluminum phosphate (Al2O3ÆP2O5) was prepared by mixing alumina (Al2O3, 99.9 pct) and diammonium hydrogen phosphate ((NH4)2HPO4, 99 pct) to yield a mole ratio of Al2O3/P2O5 = 1/1. The mixture charged in a platinum crucible was dehydrated on a hot plate at approximately 473 K (200 °C), then heated slowly to 1473 K (1200 °C) and kept 8 hours at this temperature in an inert argon atmosphere by an electric resistance furnace. The X-ray diffraction (XRD) pattern of the product revealed that the product was a mixture of Al2O3ÆP2O5 and aAl2O3.[20] Tri-calcium phosphate ((CaO)3P2O5) and di-calcium phosphate ((CaO)2P2O5) were prepared by mixing calcium carbonate and phosphoric acid to yield mole ratios of Ca/P = 3/2 or 1/1. The desiccated mixture was charged in a Pt crucible and heated in an electric resistant furnace at 1473 K (1200 °C) for 8 hours. Calcium and phosphorus concentrations of the products were determined by inductively coupled plasma atomic spectroscopy (ICP-AES). The phase of the products was identified by XRD analysis and single-phase compounds were synthesized. C. Principle of Measurement Thermodynamic properties of (CaO)2P2O5 would be estimated by measuring pressures of one or more gases in equilibrium with the mixture of (CaO)2P2O5 and (CaO)3P2O5. According to Thermochemical Data of Pure Substance,[26] P, P2, P4, PO, PO2, P4O6, and P4O10 can be in equilibrium with the mixture at the experimental temperature of 1373 K to 1498 K (1100 °C to 1225 °C). It was reported that P2, PO, and PO2 in equilibrium with P2O5 in some oxide systems such as CaO-P2O5,[9] Al2O3-P2O5,[20] FetO-P2O5, and PbOP2O5[27] were measured as ion currents by a mass spectrometer. In exploratory experiments, the ion currents of P2, PO, and PO2 could be detected by the equipment used in this study. Therefore, the thermodynamic properties of (CaO)2P2O5 were estimated here by measuring the ion currents of these gaseous species. When (CaO)2P2O5 and (CaO)3P2O5 are in equilibrium, the chemical potential of both sides of the reaction should be equal. Therefore, a relationship exists between the chemical potentials of component i, lis . 2ðCaOÞ3 P2 O5 ðsÞ þ 3PO2 ðgÞ ¼ 3ðCaOÞ2 P2 O5 ðsÞ þ POðgÞ

½3

2lðCaOÞ3 P2 O5 þ 3lPO2 ¼ 3lðCaOÞ2 P2 O5 þ lPO

½4

METALLURGICAL AND MATERIALS TRANSACTIONS B

lis can be written by li s and activities of i, ai or vapor pressure of species i, pi as follows:   ½5 li ¼ li þ RT ln ai ¼ li þ RT lnðpi =p Þ where R is the universal gas constant. Because the Gibbs energy change of reaction [3], DG3 , can be expressed as Eq. [6], this is also expressed as Eq. [7] using ais, and pis. DG3 ¼ 3lðCaOÞ2 P2 O5 þ lPO  2lðCaOÞ3 P2 O5  3lPO2

½6

0

  1 a3ðCaOÞ P2 O5  pPO from specimen p 2 DG3 ¼ RTln@   3 A ½7 a2ðCaOÞ P2 O5  pPO2 fromspecimen p

DG14 ¼ DG3  DG11 0 !3 1   pPO from specimen p pPO2 from reference =p A  ¼ RT ln@  pPO from reference =p pPO2 from specimen p ½15 Because there is a relationship between the vapor pressure and the ion current as Eq. [2], the Gibbs energy change can be expressed by the ion currents of PO and PO2 from the specimen and reference mixtures.  ! IPO from specimen IPO2 from reference 3  DG14 ¼ RT ln  IPO from reference IPO2 from specimen ½16

3

Because (CaO)2P2O5 and (CaO)3P2O5 are stoichiometric compounds,[1] these activities in the mixtures can be set as unity. Therefore, DG3 can be written as Eq. [8] using pressures of PO and PO2. !  pPO from specimen p  DG3 ¼ RTln  ½8  3 pPO2 from specimen p To use double Knudsen mass spectrometry, a reference specimen is necessary. A mixture of Al2O3ÆP2O5 and Al2O3 was selected in this study because the activity of P2O5 in the mixture is close to that in the mixture of (CaO)2P2O5 and (CaO)3P2O5, and the Gibbs energy change of chemical reaction [9] was measured in previous study.[20] 5 Al2 O3 ðsÞ þ P2 ðgÞ þ O2 ðgÞ ¼ Al2 O3  P2 O5 ðsÞ 2 DG9 ¼ 1970000 þ 574Tð28000Þ

½9 ½10

Gibbs energy change of reaction (11), DG11 , can also be expressed as Eq. [12]. Al2 O3 ðsÞ þ 3PO2 ðgÞ ¼ Al2 O3  P2 O5 ðsÞ þ POðgÞ ½11

DG11

¼ RT ln

! aAl2 O3 P2 O5  ðpPO from reference =p Þ aAl2 O3  ðpPO2 from reference =p Þ3

Thus, DG14 can be estimated by measuring ion currents of PO and PO2 from the specimen and reference mixtures. Then, DG1 can be obtained using DG9 . DG1 ¼ DG9 þ DG14

III.

½17

RESULTS AND DISCUSSION

Approximately 0.4 g of the mixtures of (CaO)2P2O5 and (CaO)3P2O5, and of Al2O3ÆP2O5 and Al2O3 were placed in a molybdenum Knudsen cell, respectively. Ion currents from these mixtures were measured at 1373 K to 1498 K (1100 °C to 1225 °C) by double Knudsen cell mass spectrometry in this study. Specimen and reference mixtures after measurement were analyzed by XRD. The XRD pattern of an experimental mixture after measurements is showed in Figure 2. This indicates the mixture is composed of (CaO)2P2O5 and (CaO)3P2O5. It was confirmed that the same phases existed as those prior to the measurement because the mixture was prepared by mixed these single-phase compounds before measurement. The XRD pattern of the reference mixture was showed in a previous study.[20] The ion currents detected by the mass spectrometer during the measurement at 1498 K (1225 °C) are shown

½12

Because Al2O3ÆP2O5 and Al2O3 are stoichiometric compounds,[20] these activities in the mixtures can be set to unity. Then, the Gibbs energy change of reaction [11] can be written by pressures of PO and PO2. ! pPO from reference =p  DG11 ¼ RT ln ½13 ðpPO2 fromreference =p Þ3 Assuming a reaction [14], the Gibbs energy change of the reaction [14], DG14 , can be obtained easily by the difference between Eqs. [3] and [11]. 2 ðCaOÞ3 P2 O5 ðsÞ þ Al2 O3 ðsÞ  P2 O5 ¼3ðCaOÞ2 P2 O5 ðsÞ þ Al2 O3 ðsÞ ½14

METALLURGICAL AND MATERIALS TRANSACTIONS B

Fig. 2—XRD pattern of an experimental mixture after measurements.

VOLUME 42B, AUGUST 2011—687

uncertainty of Eq. [21] contains the errors from Eqs. [10] and [23], which were used to estimate DG20 . 5 2CaOðsÞ + P2 ðgÞþ O2 ðgÞ¼ðCaOÞ2 P2 O5 ðsÞ 2 DG20 ¼ 2180000 þ 590Tð28000Þ J 5 3CaOðsÞ + P2 ðgÞþ O2 ðgÞ¼ðCaOÞ3 P2 O5 ðsÞ 2 DGo22 ¼ 2204000 þ 526Tð16000Þ J

Fig. 3—Ion currents during measurement at 1473 K (1200 °C).

in Figure 3. These currents were taken by scanning a mass-to-charge ratio, m/z, from 1 to 100, and the scan was repeated approximately 50 times for each measurement of the specimen and the reference mixture; the 50 scans required approximately 5 minutes. Before measurement of the specimen and reference, the background of the ion currents was measured in the same way. The logarithmic value of ion currents of m/z = 47, 62, and 63 in each scan are plotted against scan times here. These ion currents were detected clearly from the two mixtures. The currents of m/z = 47, 62, and 63 would be caused by 31P16O+, 31P2+, and 31P16O2+, respectively. The value obtained by subtracting an average background current from an average current observed from a specimen in the 50 scans was used as the net current from the specimen for subsequent calculations. Although the ion currents from 101 to 300 in m/z were also measured in the same manner in some measurements, no ion currents were detected from these mixtures. The ion currents of PO, P2, and PO2 from the specimen and reference mixtures are summarized in Table I. Current intensities vary considerably in the measurements of the same mixture under the same experimental conditions; this is because the oxygen potential in a Knudsen cell is different in each measurement. A calculation was conducted for DG14 using ion currents of PO and PO2, based on Eq. [16]. DG14 is expressed as follows, and is shown in Figure 4:

½20 ½21 ½22 ½23

The Gibbs energy change in the literature is also shown in Figure 5. The value reported by Egan and Wakefield[16,17] or Hoshino et al.[19] is approximately 10 or 50 kJ lower than that in this study and is within the error range estimated. In measuring the ion current of gaseous molecular, the effects of fragment ion on the measurement should be considered. The fragment ion is generated by breaking the molecular at a constant rate when the molecular is ionized.[28] Therefore, the ion currents of PO, P2, and PO2 detected in this study should contain the fragment ions generated from other larger molecules, such as gaseous P4, P4O6, and P4O10 in equilibrium with the specimen and reference mixtures.[26] The effects of fragment ions from P4, P4O6, and P4O10 are insignificant in this study because the vapor pressures of these three in equilibrium with the mixtures are much lower than those of PO, P2, and PO2 thermodynamically.[26] However, the ion current of PO detected in this study contains the fragment ion from PO2. To confirm that the effects of this fragment ion on the thermodynamic measurement were limited here, the following verification was made. Assuming the following two reactions, the Gibbs energy can be estimated using ion currents of P2 and PO, or P2 and PO2 in the same manner. 3 2ðCaOÞ3 P2 O5 ðsÞ + 5POðgÞ¼ 3ðCaOÞ2 P2 O5 ðsÞþ P2 ðgÞ 2 ½24

DG14

 3   ! IP2 from specimen 2 IPO from reference 5 ¼ RT ln  IP2 from reference IPO from specimen

DG14 ¼ 162000 þ 143Tð26000Þ J ð1373