ARTICLE IN PRESS
Journal of Crystal Growth 289 (2006) 278–285 www.elsevier.com/locate/jcrysgro
A thermodynamic model for the precipitation of nanostructured copper oxalates Lucica Cristina Soare, Jacques Lemaıˆ tre, Paul Bowen, Heinrich Hofmann Powder Technology Laboratory, Institute of Materials, Swiss Federal Institute of Technology, EPFL, MX-D 336, Station 12, CH-1015 Lausanne, Switzerland Received 18 April 2005; received in revised form 10 November 2005; accepted 14 November 2005 Available online 30 January 2006 Communicated by M. Uwaha
Abstract A thermodynamic model is developed for the precipitation of copper oxalate from aqueous solutions of copper nitrate and sodium oxalate in a small well-mixed batch reactor. This model is then used to explain the discrepancy between experimental pH evolution and the predicted evolution for the precipitation of copper oxalate as the only solid phase. The experimental behaviour is simulated by taking into account the possible precipitation of small quantities of copper hydroxyl carbonates from adventitious CO2. Experiments with controlled CO2 partial pressures illustrate the power of these types of simulations in helping elucidate precipitation reactions. The thermodynamic model can help simulate the kinetics of precipitation by linking precipitation yield with real time pH measurements. r 2005 Elsevier B.V. All rights reserved. Keywords: A1. Kinetics; A1. Phase diagrams; A1. Solubility; A1. Thermodynamic modelling
1. Introduction The synthesis of inorganic materials of high quality, specific size and morphology is a key aspect in the development of new materials in fields such as catalysis, medicine, electronics, ceramics, pigments and cosmetics [1–4]. The synthesis step is decisive for particle properties such as the primary particle size, morphology, crystallinity and purity. These in turn determine the product properties and the quality of the end product [5]. Nanostructured materials in the range from 1 to 100 nm can be formed by molecules, clusters or crystallites. Nanoparticles are often obtained by precipitation at high levels of supersaturation, which is sensitive to temperature, reactant concentration and mixing conditions. The formation of the particles and the growth processes in precipitation directly depend on the supersaturation, and varies both with the feed concentration and flow rate. A modification of the particle morphology and the variation of particle size can result Corresponding author. Tel.: +0041 21 693 4907; fax: +0041 21 693 3089. E-mail address: paul.bowen@epfl.ch (P. Bowen).
0022-0248/$ - see front matter r 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.jcrysgro.2005.11.085
from the excess of a particular species in solution, which could be caused by non-stoichiometric reaction conditions [6]. The use of thermodynamic solubility calculations to identify the various species in solution during the precipitation process is one way to try and understand such processes, along with the correct choice of a thermodynamic model [7]. A high supersaturation produces small primary particles as it promotes high nucleation rates. Supersaturation can also play a role in determining the rate of aggregation, which is itself the result of the balance between the formation of aggregates and their almost immediate disruption by fluid shear. It is often suggested [8] that one role of supersaturation is to provide new material to bond or ‘‘cement’’ two colliding crystals together. Also, interactions between molecules and ions in solution and species that represent the termination of the bulk crystal structure are critical as they determine reactivity and free energy at the solid–liquid interface. Two important considerations related to processes occurring at surfaces are the mechanism and the rates of reactions. The mechanism is the series of steps involved in the process of reaction. One of these steps will be slower than another,
ARTICLE IN PRESS L.C. Soare et al. / Journal of Crystal Growth 289 (2006) 278–285
and will thus determine the rate-limiting step. Although a variety of distinct surface sites may contribute to the reaction, the overall rate is dominated by the most abundant of the most reactive sites. Rates for specific processes can be suppressed by diffusion of reactants and products to and from surfaces, and are affected by the characteristics of the solid, composition of the solution (pH, saturation, and ionic strength) and by temperature. With respect to the composition of the solution and its influence on precipitation, Rubattel et al. [9] calculated solubility isotherms for the system (Y(OH)3–Ba(OH)2– Cu(OH)2–H2C2O4–(HNO3/NaOH)–H2O). Their aim was to produce a powder with a well-defined stoichiometric cationic ratio Y1Ba2Cu3. Furthermore, a detailed evolution of the different ionic species in solution was considered, taking into account several equilibria and solid compounds that could precipitate from this complex chemical system. These authors [9] highlight the importance of complex formation in the studied system, their sensitivity to pH and the simultaneous presence of several species, all of which influence the final precipitate properties. Recently, Pujol et al. [10] carried out a detailed study of nanostructured cobalt oxalate precipitation using cryogenic electron microscopy techniques to characterise intermediate precipitation products. They showed a strong correlation between the particle evolution and the solution chemistry. The particle evolution was linked to the ionic concentration in solution, and consequently with pH. A possible mechanism was finally proposed, showing a disordered core and an ordered nanostructured shell directly related to the changing solution conditions during the precipitation. The aim of our work is to improve our capacity to understand quantitatively the effect of the solution evolution during the precipitation of copper oxalate by simulating the precipitation kinetics using a detailed description of the properties of the CuC2O4 electrolyte solution. These include the effect of various ions present in the solution, ion pairs and various complexes determined using an accurate and appropriate thermodynamic model for the chemical system under study. The correlation of precipitation kinetics is made with experiments carried out in a well-mixed mini-batch reactor of 20 mL where CuC2O4 precipitation occurs upon mixing Cu(NO3)2 and Na2C2O4. CuðNO3 Þ2ðaqÞ þ Na2 C2 O4ðaqÞ Ð CuC2 O4ðsÞ # þ2NaNO3ðaqÞ . The thermodynamic solubility data used for the precipitated powder in water include the effects of complexation (hydroxo and oxalato complexes), ionic strength and the simultaneous presence of several solids [11], which shows an important influence on the kinetic behaviour. The selected computing technique allows easy extension to other precipitates in the general system, e.g. hydroxy carbonates of copper.
279
2. Thermodynamic model for the aqueous system Cu2+/ + 2 NO2 3 /Na /C2O4 2.1. Solubility of copper oxalate in aqueous media The knowledge of the solubility equilibria in a large variety of experimental conditions provides a relatively efficient and simple method of following a precipitation reaction that cannot easily be made experimentally. These types of calculations take into account the effect of the presence of many species as described in a previous study [9]. The numerical calculations consider many different equilibria, including dissolution, protonation and complex formation that must satisfy electrical neutrality. For the reaction of ions in solution, the rate of reaction depends not only on the ion concentrations but moreover on the activity coefficient g. In dilute solutions, the activity coefficient is close to unity. The relationship between the concentration and the ion activity is given by ðM i Þ ¼ ½M i g,
(1)
where (Mi) is the ion activity and [Mi] the ion concentration. The activity coefficient g depends on the ionic charge, the ionic strength of the solution and the radius of the ionic species, and can be calculated using the Debye–Hu¨ckel approximation: pffiffiffi I 2 pffiffiffi , log gi ¼ AZ i (2) 1 þ Br I where A ¼ 1:8246 106 (eT)3/2 and B ¼ 502:9 (eT)3/2 (nm1 mol1/2 L1/2 K1/2), e is the solvent dielectric constant (for water e ¼ 251:629 0:803 T þ 0:000744 T 2 ), T the temperature in Kelvin, r the recovering radius of the ionic species in nm and I the ionic strength. Eq. (2) gives a good estimation up to ionic strengths around 0.01 M. The ionic diameters of ions considered for the calculation of the activity coefficients for the precipitation system Cu(OH)2–H2C2O4–(CO2)–(HNO3/NaOH)–H2O are given in Table 1. The appropriate model choice depends mainly on the ionic strength I of the solution under investigation, which is defined by X I ¼ 0:5 Z 2i M i , (3) i
where Zi and Mi are the valence of ion i and the molal concentration, respectively. Dissolution–precipitation of copper oxalate takes place following the reaction 2 ðCuC2 O4ðsÞ 2Cu2þ ðaqÞ þ C2 OðaqÞ Þ.
(4)
The solubility of a salt such as copper oxalate can be calculated by enforcing the equilibrium condition through the relationship, giving the ionic product Ps, which for nonideal electrolyte solutions is expressed in terms of the ion
ARTICLE IN PRESS L.C. Soare et al. / Journal of Crystal Growth 289 (2006) 278–285
280
activities as follows: Ps ¼ ðCu2þ ÞðC2 O2 4 Þ,
(5)
where Ps is the ionic product. The corresponding equilibrium is described by the following relationship: K s ¼ ðCu2þ Þeq ðC2 O2 4 Þeq .
(6)
The equilibrium constant Ks at 25 1C is log (Ks) ¼ 9.665 [12]. The supersaturation S is calculated from the activities of copper and oxalate ions and the equilibrium constant Ks. S ¼ ðPs =K s Þ1=x ,
(7)
where x is the number of ionic species in solution; in the present case of copper oxalate, x ¼ 2. Throughout this article, supersaturation is described by a reduced supersaturation, Sr, given by S r ¼ log ðPs =K s Þ1=x .
(8)
In order to calculate the activities of free copper and oxalate ions, associations between ionic species have to be taken into account, and the existence of several species in solution is postulated (Tables 2 and 3) [13]. Furthermore, these data permit us to estimate another important parameter of precipitation, the reaction yield of the precipitation x, which is defined as x¼
fCug0 fCug , fCug0 fCugeq
(9)
where {Cu}0 is the analytical concentration of copper in solution. Table 1 Diameters of hydrated ions at 298 K Ionic species
Diameter (nm) [13]
Cu2+ HO H+ C2O2 4 HC2O 4 Cu(OH)+ Cu(OH) 3 Cu(OH2 4 ) Cu2(OH2+ 2 ) Cu(C2O4)2 2 Cu(HC2O4)+
0.60 0.35 0.90 0.45 0.40 0.30 0.30 0.45 0.50 0.50 0.50
For example, in the present case, we have a 2:2 electrolyte, which is the case for the copper oxalate; a copper ion needs an oxalate ion for the precipitation of a mole of the oxalate. The reaction yield varies from 0 in the initial stage up to 1 where the reaction equilibrium is achieved. 2.2. Example of the thermodynamic model applied to the copper oxalate system Based on the above considerations, the equilibrium composition of CuC2O4 solutions can be calculated. The calculation was made for the system Cu(OH)2–H2C2O4– (CO2)–(HNO3/NaOH)–H2O at 25 1C, for an ion concentration of Cu ¼ C2O4 ¼ 0.005 M. In the first step, the pH value of the solution was calculated to be around 5.34 under a partial pressure of carbonic gas of pCO2 ¼ 3:3 104 atm (see experimental section for choice of pCO2 ). Then, the pH values were modified by adding HNO3 to the solution for an acidic pH or NaOH for a basic pH. The simulation results of the copper species distribution in solution as a function of pH are shown in Fig. 1. In a pH range from 2 to 7, the dominant species containing copper is the neutral oxalato, CuC2 O04 (440%). The two other significant species, both at around 20%, are the complex CuðC2 O4 Þð2Þ , or the free ion Cu2+. The other species are 2 present in minor amounts. From the solution chemistry and the final equilibrium state, the reaction yield was calculated for two distinct Table 3 Complex formation in solution for the system: Cu(OH)2–H2C2O4–(CO2)–[HNO3–NaOH]–H2O Log K298 [13]
Number
Reaction
1 2 3 4 5 6 7 8 9 10 11 12 13
HO +H -H2O + C2O2 4 +H -HC2O4 + HC2O +H -H C 4 2 2O4 Cu2++HO-Cu(OH)+ Cu(OH)++HO-Cu(OH)2 Cu(OH)2+HO-(Cu(OH)3) (Cu(OH)3)+HO-(Cu(OH)4)2 2Cu2++Cu(OH)2-(Cu2(OH)2)2+ 0 Cu2++C2O2 4 -CuC2O4 2 CuC2O4+C2O4 -(Cu(C2O4)2)2 + Cu2++HC2O 4 -Cu(HC2O4) 2 2 CuCO3+CO3 -(Cu(CO3)2) Cu2++CO2 3 -CuCO3
+
13.99 1.25 4.26 6.00 4.70 3.50 2.20 6.93 5.71 4.62 2.90 3.17 6.75
Table 2 Dissolution reactions taken into account for the copper oxalate system at 25 1C Chemical name Copper oxalate (moolite) Copper carbonate Azurite Malachite
Chemical formula
Reaction
CuC2O4 0.5 H2O CuCO3 Cu(OH)2 2CuCO3 Cu(OH)2 CuCO3
2+
Cu +C2O2 4 +H2O Cu2++CO2 3 3Cu2++2CO2 3 +2OH 2+ 2 2Cu +CO3 +2OH
Ks 9.65 9.63 33.78 45.95
ARTICLE IN PRESS L.C. Soare et al. / Journal of Crystal Growth 289 (2006) 278–285
Cu(2+) CuOx (0) Cu(HOx)(+) Cu(Ox)2(2-) Cu(NO3)(+) Cu(CO3)2(2-) CuCO3 (0)
Ions distribution in solution
100% 80% 60% 40%
1.00 CuC2O4 Cu(OH)2· Cu(CO3)
0.50
Cu(OH)2· 2 Cu(CO3)
0.00 -0.50 -1.00 5.40
5.60
5.80
6.00
6.20
6.40
pH
20% 0% 1.00
Fig. 4 shows the supersaturation evolution of these three solids as a function of pH when the only solid phase to precipitate is copper oxalate in the presence of carbonic gas. At the beginning of the precipitation, the system is supersaturated with respect to oxalate and malachite (S r 40). Then, as the copper oxalate precipitates, its Sr decreases and the supersaturation with respect to the hydroxy carbonates of copper increases (malachite and azurite). It is important to note that at a low enough partial pressure of carbonate of 6.2 106 atm, the supersaturation of the oxalate is the same, but for the two hydroxy carbonates it becomes 0.00 and 0.35 for malachite and azurite, respectively. Furthermore, at this low partial pressure of carbonate, the precipitation system is supersaturated with respect to another phase, copper hydroxide Cu(OH)2, which is not so at a pCO2 higher than 2.66 104 atm. Experimentally, it is possible to follow the pH behaviour as a function of time and compare it with the theoretical predictions. Using this thermodynamic model, it is possible to follow the kinetic pathway of copper oxalate precipitation as a function of pH, in
Supersaturation
situations—without or with carbonic gas at a partial pressure of 3.3 104 atm as illustrated in Fig. 2. These simulations were based on the hypothesis that the precursor solutions are stoichiometric fCug=fH2 C2 O4 g ! 1 at 25 1C. The initial pH was computed by considering all the possible compounds that may precipitate given by the ionic species present in the solution (Tables 1 and 2), then changing the supersaturation values as the copper oxalate is precipitated, in this case from an initial value of 0.70 to equilibrium at 0.00 which in turn modifies the pH. Fig. 2 shows the pH evolution as a function of the reaction yield at a concentration of 0.005 M after mixing the precursors. The pH increases from 5.35 to 6.20 or 5.65 without or with carbonic gas, respectively. Using this simulation approach in the present system at 25 1C, it is possible to plot the pH evolution as a function of supersaturation, S, in relation to the different compounds that may form as a function of the chemical composition of the solution as precipitation takes place (Fig. 3). With very low carbonic gas content (108 atm), the system is only supersaturated with respect to copper oxalate and is undersaturated with respect to the hydroxy carbonates of copper.
281
2.00
3.00
4.00
5.00
6.00
7.00
8.00
Fig. 3. Supersaturation (Sr) evolution of copper oxalate and the two hydroxy carbonates (malachite Cu(OH)2 CuCO3 and azurite Cu(OH)2 2CuCO3) during the copper oxalate precipitation as a function of pH made in the presence of carbonic gas at a pressure of 108 atm.
pH
Fig. 1. Copper species distribution in solution as a function of pH and their complex formation for the system Cu(OH)2–H2C2O4–(CO2)–HNO3–NaOH–H2O (where C2O4 ¼ Cu).
1.00
6.40
Supersaturation
0.50
Without CO2 With CO2, pCO2 = 3.3·10-4 atm
pH
6.00
0.00 -0.50 CuC2O4 CuCO3*Cu(OH)2 2CuCO3*Cu(OH)2
-1.00
5.60 -1.50 5.20
5.20 0.00
5.70
6.20
6.70
7.20
pH
0.25
0.50
0.75
1.00
Reaction Yield
Fig. 2. pH evolution as a function of reaction yield with and without the influence of the partial pressure of carbonic gas at an atmospheric pressure of 3.3 104 atm.
Fig. 4. Supersaturation (Sr) evolution of copper oxalate and the two hydroxy carbonates (malachite Cu(OH)2 CuCO3 and azurite Cu(OH)2 2CuCO3) during the copper oxalate precipitation as a function of pH made in the presence of a carbonate concentration of 105 M ðpCO2 ¼ 2:66 104 atmÞ.
ARTICLE IN PRESS 282
L.C. Soare et al. / Journal of Crystal Growth 289 (2006) 278–285
well-defined conditions of temperature and partial pressure of carbonic gas. From this initial snapshot of the copper oxalate system, we see that there is a complex behaviour with the formation of various complexes during the precipitation reaction, and even the possible presence of several solid compounds such as the basic carbonates. The chemistry of the solution and the supersaturation with respect to different solids will play an important role in determining the real route of precipitation, and to correlate this type of information with the various stages of nucleation, growth and aggregation should help towards a better understanding of precipitation reactions. 3. Experimental section
Table 4 pH values of standard and precursor solutions before and after equilibrium with the carbonic gas in the chamber Solution
Initial pH
Final pH
Na2C2O4 Cu(NO3)2 NaNO3–NaHCO3
7.65 4.64 8.19
7.14 4.69 8.88
Table 5 Carbonic gas dissolved in both precursor solutions and their stoichiometry Solution
Carbonic gas (atm)
Stoichiometry
2Na+/C2O2 4 Cu2+/2NO 3
1.48 104 1.83 105
1.01 1.00
3.1. Solution analysis 6.00
5.60 pH
CuC2O4 powders were synthesised in an aqueous solution by the reaction of copper nitrate (Merck 1.02753 p.a.) and sodium oxalate (Merck 1.06557 p.a.). The solutions were prepared by diluting the necessary amount of powder in ultrapure water. The ultrapure water was boiled and filtered through 200 nm membranes. Both stock solutions were then filtered using a ceramic membrane of 20 nm (Whatman). The exact concentration of the stock solution of copper nitrate was measured by inductively coupled plasma atomic emission spectroscopy (ICP-AES, Perkin-Elmer software Plasma 2000). The precipitation experiments were carried out by mixing 10 mL of each reactant of copper nitrate and sodium oxalate at a concentration of 0.010 M. The precipitation was accomplished at 25 1C, in a small volume of 20 mL, by rapid injection during 6 s, allowing us to assume that the precursors were well mixed. The mixing time was estimated to be at around 0.6 s [14]. For these conditions, the pH was measured every 10 s for times up to 90 min using a Labview software.
5.20
4.80 0.0
5.0
10.0
15.0
20.0
25.0
30.0
Time (min) Fig. 5. pH evolution as a function of time for several replicate precipitation experiments with a concentration of 0.005 M of copper nitrate and sodium oxalate.
Using the thermodynamic model, it was then possible to calculate the carbonate dissolved in each solution and their stoichiometry, knowing the partial pressure of carbonate as summarised in Table 5.
3.2. Solution control 3.3. pH evolution Both precursor solutions Cu(NO3)2 and Na2C2O4 were stored in a chamber where it was possible to control the partial pressure of carbonate using a carbonate/bicarbonate standard solution of 0.10 M NaNO3 and 0.01 M NaHCO3. The carbonate/bicarbonate solution was prepared under the same conditions as the precursors (with boiled water and filtered through 200 nm membranes). This solution at room temperature should have a CO2 partial pressure of 3.3 104 atm and a pH of 8.11. After reaching the equilibrium with CO2 of atmospheric pressure in the chamber (of known volume), the pH of the solution should increase to a value of 9.04. Under these conditions, it is possible to estimate the partial pressure of CO2 in the chamber. The pH values of the three solutions were measured just before and after an ageing time of 3 days in the chamber, as indicated in Table 4.
The pH was measured as a function of time for several precipitation experiments in order to estimate the reproducibility of the evolution of the experimental values (Fig. 5). The experimental pH shows an increase for a short period of around 2 min, then slowly decreases for the remainder of the precipitation experiment. The general behaviour of each experiment is similar with variation of 70.2 pH units in the 5–30 min region. 4. Experiment vs. theory The theoretical behaviour of pH for our standard conditions of precipitation has been discussed in Section 2 (0.01 M copper nitrate and 0.01 M sodium oxalate). The thermodynamic solubility data showed an increase in pH
ARTICLE IN PRESS L.C. Soare et al. / Journal of Crystal Growth 289 (2006) 278–285
7.50 7.00 experimental theoretical
pH
6.50 6.00 5.50 5.00 0
5
10
15 Time (min)
20
25
30
Fig. 6. Experimental and theoretical pH evolution as a function of time for the copper oxalate precipitation.
0.80
Supersaturation
0.60
CuC2O4 Cu(OH)2· CuCO3 Cu(OH)2· 2CuCO3
0.40 0.20 0.00 -0.20 5.50
5.70
5.90
6.10
5.90
6.10
pH
(a) 0.80
CuC2O4 Cu(OH)2· CuCO3
0.60 Supersaturation
for both the cases without or with carbonic gas at pCO2 ¼ 3:3 104 atm. The experimental and simulated curves plotted as a function of time show a significant discrepancy (Fig. 6). At the beginning, both curves show the same behaviour, and the pH increases. Then the experimental curve decreases instead of the continuous increase seen for the theoretical curve. The link between the theoretical curve of the pH and time has been made by measuring (ICP-ASE) the analytical concentration of copper in the supernatant for each experimental point marked on the curve (Fig. 6); then, by equating yield with time, a time axis for the simulated precipitation reaction could be computed. The above discrepancy between the simulated and experimental pH evolution was for a thermodynamic model where only a single precipitated phase was considered, i.e. copper oxalate. X-ray powder diffraction carried out on the precipitated powder indeed only showed reflections for copper oxalate (JCPDS no. 210297). The effect of dissolved carbonate in the system was also simulated (Fig. 2), but the presence of dissolved carbonate, although reducing the magnitude of the pH increase, is still well above the experimental result. This led us to hypothesise on the possible precipitation of solid phases other than copper oxalate, namely the hydroxy carbonates malachite (Cu(OH)2 CuCO3) and azurite (Cu(OH)2 2CuCO3). In Section 2 (Fig. 4), the possibility of these phases becoming supersaturated was demonstrated. Other sources for the discrepancy such as slight deviations from stoichiometry (1–2%) could not simulate the experimental results. The following hypothesis was made to help explain the pH evolution by simulating the precipitation reaction using the thermodynamic model set-up in Section 2, and the pCO2 is that calculated from the controlled chamber simulation. Initially, the high supersaturation of copper oxalate is relieved by precipitating only copper oxalate up to the maximum of the experimental pH evolution, equal to 6.17. A simulation of this copper oxalate precipitation reduces the supersaturation value of this phase from 0.69 to 0.42
283
Cu(OH)2· 2CuCO3
0.40 0.20 0.00 -0.20 5.50
5.70
(b)
pH
Fig. 7. Supersaturation (Sr) behaviour of copper oxalate, malachite and azurite as a function of pH: (a) period of pH increase 0–2 min, (b) period of pH decrease 2–30 min.
Table 6 Supersaturation value (Sr) of the three compounds that might be considered at specific values of experimental pH pH
Copper oxalate CuC2O4
Malachite Cu(OH)2 CuCO3
Azurite Cu(OH)2 2CuCO3
6.17 5.60
0.42 0.42
0.43 0.00
0.28 0.18
(Fig. 7a). During the precipitation of copper oxalate, the system starts to supersaturate with respect to the hydroxy carbonate phases of copper as summarised in Table 6, and illustrated in Fig. 7. Our hypothesis then considers that only malachite precipitates from this point onwards, which simulates very well the decreasing pH region (Fig. 8). At the beginning of this second stage, the malachite reaches a supersaturation of about S r ¼ 0:43 above that of the copper oxalate (S r ¼ 0:42). The azurite phase also saturates and shows a supersaturation but at much lower values, and becomes undersaturated as the malachite precipitates. It was therefore assumed that azurite does not precipitate under these simulation conditions. The evolution of supersaturation as a function of pH is given in Fig. 7b for this decreasing pH region. The precipitation of malachite shows the pH decreasing from 6.17 to 5.60. The supersaturation values at these pHs are given in Table 6.
ARTICLE IN PRESS L.C. Soare et al. / Journal of Crystal Growth 289 (2006) 278–285
284
5.00 6.20
Cu (mmol/L)
4.00
pH
6.00 Copper Oxalate precipitation 5.80
3.00 2.00 1.00
5.60
0.00 0
5.40 0
0.5
(a)
1
1.5
Time (min)
pH
Malachite Precipitation
5.80
5.60
5.40 2.00
(b)
7.00
12.00
17.00
50
75
100
Fig. 9. Copper concentration in the supernatant as a function of time; results obtained from ICP analysis.
6.20
6.00
25
Time (min)
2
22.00
27.00
Time (min)
Fig. 8. pH behaviour for the precipitation reaction with a pCO2 of 3.3 104 bar—comparison between theory (lines) and experiment (symbols); (a) copper oxalate precipitation, (b) malachite precipitation.
Thus, Fig. 8 shows two domains of pH evolution; the first can be explained by an initial precipitation of copper oxalate for the first 2 min or so, where 73% of the maximum precipitation yield is predicted with respect to the equilibrium solubility. Then, the second domain where the pH decreases slowly over a long period of time is explained by the precipitation of malachite only. The mass balance from the thermodynamic model indicates a low amount of malachite precipitation of about 0.40 wt% of the total precipitate. This very small amount has a very significant influence on the pH behaviour but such a small quantity of malachite could not be detected using XRD, TEM diffraction, thermogravimetric analysis or Fourier transform infrared spectroscopy (FTIR). Wet chemical methods were thought to be unsuitable as copper oxalate is sparingly soluble in acids, and with basic media, it is difficult to keep carbonate contamination below the 0.1% of carbonate predicted to be present in the copper oxalate. Fig. 9 shows the ICP analysis of the copper in the supernatant. At the maximum value of the pH, around 73 wt% of the total copper oxalate has already precipitated (from pH measurement and thermodynamic model), before the proposed malachite precipitation starts. Experimentally, after 30 min, ICP shows that around 23% of copper oxalate remains in solution compared to the expected equilibrium value. This is in close agreement with the 27% predicted by the thermodynamic model from the pH
measurement. If the copper oxalate continued to precipitate, then the pH should increase, which is not observed experimentally. Measurements of pH and copper concentration remaining in solution from 30 to 90 min show only very small changes (Fig. 9). The experimental values of copper in solution and the theoretical values show some discrepancies possibly because of the presence of nanoparticles that remain in the supernatant, and thus the experimental ICP results are higher than those predicted by the thermodynamic model, especially at early precipitation times. The decrease in copper in solution between 30 and 60 min seen in Fig. 9 is therefore attributed to nanosized particles (observed during in situ particle size measurement [12]) that continue to attach themselves to the micron-sized copper oxalates. Copper oxalate is a nanostructured material with a proposed precipitation mechanism of an ordered agglomeration of 70 nm primary particle units after an initial burst of nucleation and growth. Here, we have seen that, using thermodynamic modelling, even small quantities of adventitious carbonate may have a strong influence on solution properties, the resulting pH and hence the particle formation. The predicted malachite formation may influence the surface properties of the copper oxalate primary particles and thus contribute to the self-ordering growth mechanism previously proposed [15]. Further work on the kinetics and mechanism are underway, aiming to link the above information from the thermodynamic model to the physical evolution and substructure of copper oxalate precipitates. 5. Conclusions A thermodynamic model for the precipitation of copper oxalate has been developed, which can take into account the effect of different carbon dioxide partial pressures ðpCO2 Þ. With the experimental set-up presented, a small batch reactor, the precipitation kinetics of copper oxalate under well-defined mixing conditions and controlled pCO2 were followed by monitoring the pH evolution during the precipitation reaction. It was only possible to simulate the experimental pH behaviour by taking into account the
ARTICLE IN PRESS L.C. Soare et al. / Journal of Crystal Growth 289 (2006) 278–285
possible precipitation of secondary carbonate phases such as malachite and azurite. The thermodynamic model predicts that a small quantity (0.4 wt%) of malachite is sufficient to significantly alter the pH evolution during the precipitation process. This small amount of a secondary phase could not be detected experimentally, and thus the model gives a probable explanation of the solution behaviour but further work is needed to confirm the exact nature of the precipitated carbonate phase. However, using the thermodynamic equilibria of several possible complexation and dissolution reactions with several solid phases, it was possible to link experimental and simulated pH behaviour. This was done using a hypothesis of two regimes of precipitation; first, only the precipitation of copper oxalate, followed by the formation of a second compound, the hydroxy carbonate malachite. It was also possible to estimate the amount of copper oxalate and malachite precipitated from the thermodynamic model, and such approaches should prove a useful tool in helping elucidate further details of the precipitation mechanism in future work. Acknowledgements The authors would like to thank Dr. S. Rousseau for the ICP measurements, Dr. N. Jongen for fruitful discussions
285
and the Swiss Federal Office for Education and Science, Project COST 523 No. 90–0020 for funding.
References [1] S. Mann, G.A. Ozin, Nature 382 (1996) 313. [2] H. Yang, N. Coombs, G.A. Ozin, Nature 386 (1997) 692. [3] T. Ahmadi, Z.L. Wang, T.C. Green, et al., Science 272 (1996) 1924. [4] E. Matijevic, Curr. Opin. Colloid Interface Sci. 1 (1996) 176. [5] Y. Axmann, Thesis EPF Lausanne, no. 3029, 2004. [6] A.M. Donia, D. Dollimore, Thermochim. Acta 290 (1) (1997) 139. [7] L. Vicum, M. Mazzotti, J. Baldyga, Chem. Eng. Technol. 26 (3) (2003) 325. [8] M.J. Hounslow, A.S. Bramly, W.R. Paterson, J. Colloid Interface Sci. 203 (1998) 383. [9] S. Rubattel, J. Lemaitre, P. Bowen, T.A. Ring, J. Crystal Growth 135 (1994) 135. [10] O. Pujol, P. Bowen, P.A. Stadelmann, H. Hofmann, J. Phys. Chem. B 108 (2004) 13128. [11] M. Donnet, Thesis no. 2623, EPF Lausanne, Switzerland, 2002. [12] L.C. Soare, Thesis no. 3083, EPF Lausanne, 2004 (pdf available from http://library.epfl.ch/theses/). [13] S. Kotrly, L. Sucha, Handbook of Chemical Euqilibria in Analytical Chemistry, Ellis Horwood, England, 1985. [14] M. Donnet, P. Bowen, N. Jongen, J. Lemaıˆ tre, H. Hofmann, Langmuir 21 (2005) 100. [15] N. Jongen, P. Bowen, J. Lemaitre, J.C. Valmalette, H. Hofmann, J. Colloid Interface Sci. 226 (2002) 189.
