2726
Ind. Eng. Chem. Res. 2007, 46, 2726-2734
Catalyst Solubility and Experimental Determination of Equilibrium Constants for Heterogeneous Atom Transfer Radical Polymerization Santiago Faucher, Paul Okrutny, and Shiping Zhu* Department of Chemical Engineering, McMaster UniVersity, Hamilton, Ontario, Canada L8S 4L7
The solubilities of heterogeneous atom transfer radical polymerization (ATRP) catalysts are determined for the first time and used to calculate previously unquantified ATRP equilibrium constants (KATRP). These new data are essential for understanding, modeling, and designing ATRP processes with low catalyst concentrations. KATRP values for CuIBr/1,1,4,7,10,10-hexamethyltriethylenetetramine (CuIBr/HMTETA) and CuIBr/N,N,N′,N′,N′′pentamethyldiethylenetriamine (CuIBr/PMDETA) are 8.66 × 10-6 and 1.44 × 10-6, respectively. The limited solubility of the CuIBr/HMTETA catalyst explains why large reductions in metal salt concentration can be made without affecting polymerization rates. Catalyst solubility and polymerization rate increase with ligand concentration. In contrast, increasing the metal salt concentration in excess of the ligand’s causes a drop in catalyst solubility. This unexpected observation is attributed to the formation of insoluble catalyst networks (gels). The solubility data point to differences in the ionic character of the catalyst complexes formed. The following solubility trend is observed at ATRP conditions (toluene, 90 °C): CuIBr/PMDETA . CuIIBr2/ PMDETA > CuIIBr2/HMTETA > CuIBr/HMTETA. Introduction
Scheme 1. ATRP Mechanism
Free radical polymerization is widely used on account of its mild reaction conditions, wide range of polymerizable monomers, and tolerance to impurities. It however lacks the ability to precisely control polymer molecular weights, a trait desired for the design of macromolecules with specific architectures and functionalities. Living ionic polymerization allows for the production of controlled polymer architectures but requires a high degree of purity in its raw materials and stringent reaction conditions. A living free radical polymerization mechanism is therefore desired to simplify the synthesis of controlled polymers. Atom transfer radical polymerization (ATRP) is one of several recently discovered living free radical polymerization mechanisms that are filling this void.1,2 In ATRP, the polymeric chain growth is mediated by a catalyst. The mechanism is outlined in Scheme 1.1,2 A metal salt complex (Mt-X/L), the catalyst, abstracts a halide from the initiator, an alkyl halide (R-X), thereby activating the polymerization. In so doing the catalyst is oxidized (X-Mt+X/L) and a radical (R•) is formed from the initiator. The radical can propagate with monomer (M), terminate with other radicals/ impurities, or be deactivated by the oxidized catalyst. The deactivated polymer (R-X) can be reversibly activated and deactivated following the same cycle. Termination is minimized by an equilibrium that favors the dormant alkyl halide and therefore a low radical concentration. The minimization of termination results in a quasi-living polymerization where molecular weight is a linear function of conversion. The polymer produced is of controlled molecular weight and narrow molecular weight distribution.1,2 While ATRP allows the production of tailored polymers, it requires high catalyst concentrations and this remains the main challenge to its commercialization. The catalyst must be removed from the product by laborious and costly purification methods. Purification can be minimized by selecting the minimum amount of catalyst necessary to achieve the kinetic * To whom correspondence should be addressed. Tel.: (905) 5259140, ext 24962. Fax: (905) 521-1350. E-mail:
[email protected].
requirements of the process. This in turn requires knowledge of the catalyst solubility and the relationship between catalyst concentration and polymerization rate. In homogeneous catalyst systems this relationship is simple as all of the added catalyst is soluble and partakes in the reaction process. For this reason, much of the founding work in ATRP has been focused on investigating homogeneous catalysts. Some partially soluble ATRP catalysts, henceforth referred to as heterogeneous catalysts, are however known to offer faster kinetics than their fully soluble homogeneous catalyst counterparts.3 Unfortunately, their complexity has made them less attractive to study in the laboratory and few have been optimized to eliminate excess and product-degrading catalyst additions. For example, a recent optimization undertaken on the heterogeneous catalyst CuIBr/HMTETA showed that only the soluble (homogeneous) fraction of the catalyst complex participated in the polymerization.3 The insoluble fraction of the catalyst did not participate in the polymerization. This meant that a significant reduction in the metal salt addition (more than 90%) could be made without affecting polymerization rate.3 Catalyst solubility data would have permitted a more direct route to this conclusion. Knowledge of catalyst solubility also allows determination of ATRP equilibrium constants (KATRP ) ka/kd) for heterogeneous catalyst systems. These constants are fundamental to the selection and use of ATRP catalysts. For homogeneous catalysts, KATRP values have been calculated based on kinetic data for systems where the ATRP equilibrium is preestablished.4-6 The equilibrium in these systems is preestablished by adding deactivating catalyst (CuII/L) in addition to the activating catalyst (CuI/L). Under these conditions the catalyst (CuI/L and CuII/L) and dormant polymer (R-X) concentrations can be assumed
10.1021/ie061506e CCC: $37.00 © 2007 American Chemical Society Published on Web 03/29/2007
Ind. Eng. Chem. Res., Vol. 46, No. 9, 2007 2727 Scheme 2. Ligands Used
to equal those at the start of the polymerization ([CuI/L]t ) [CuI/ L]0, [CuII/L]t ) [CuII/L]0, [R-X]t ) [R-X]0). While this method allows the calculation of KATRP for homogeneous systems, it is not applicable to heterogeneous catalyst systems since not all of the catalyst added is soluble. For the calculation of the equilibrium constant in heterogeneous catalyst systems, solubility data is necessary in addition to kinetic data. Model studies measuring ka and kd, independently, provide a second route to KATRP estimates.4,5,7-16 However, the conditions and model compounds used for their determination generally differ from those of ATRP. Solubility data may also find use in explaining differences in ATRP kinetics with solvent type.17-21 Increased rates may be due to changes in catalyst solubility and/or catalyst structure. Solubility data may provide some insight into which is the primary factor. Despite the advantages gained from solubility data, no solubility data exist in the literature. Electron spin resonance (ESR) has been used to measure the concentration of deactivating catalyst in heterogeneous systems (CuII), but these experiments were not designed to test saturated concentrations.22-25 For the activating catalyst (CuI), no solubility data has been reported. Again, this lack of data results from the use of simpler homogeneous ATRP catalysts in academe. However, many heterogeneous ATRP catalyst are commercially interesting because of their high activities and low cost. Their efficient use, i.e., the use of the minimum concentration of catalyst required to saturate the reaction system, therefore requires us to determine their solubilities. Such optimized systems can facilitate or eliminate the need for catalyst removal in the ATRP product, the main challenge in ATRP. We report here the first solubility data for two widely used ATRP catalysts, CuIBr/N,N,N′,N′,N′′-pentamethyldiethylenetriamine (PMDETA (1)) and CuIBr/1,1,4,7,10,10-hexamethyltriethylenetetramine (HMTETA (2)), and their oxidized counterparts CuIIBr2/HMTETA and CuIIBr2/PMDETA (see Scheme 2) that are necessary for the design of ATRPs with low residual catalyst concentrations in polymer. Many variables influencing solubility are studied including the solvent type, metal salt concentration, and ligand concentration. From these data the ATRP equilibrium constants (KATRP) for these heterogeneous catalysts, another key parameter in modeling and designing industrial ATRP processes, are determined for the first time. Experimental Section Materials. Methyl methacrylate (MMA; Aldrich, 99.9%) is distilled from CaH2 under vacuum and stored at 4 °C before use. Methyl isobutyrate (MIB; Aldrich 99%) is distilled from MgSO4 under vacuum and stored at 4 °C prior to use. Toluene is distilled from CaH2. Tetramethylsilane (TMS; 99.9%+ NMR grade), 1,1,4,7,10,10-hexamethyltriethylenetetramine (HMTETA; 99%), N,N,N′,N′,N′′-pentamethyldiethylenetriamine (PMDETA; 99%), CuIBr (98%), CuIIBr2 (99%), dimethyl malonate
(99%), and methyl R-bromophenylacetate (MBP, 97%) are used as received from Aldrich. Preparation of CuIBr from CuIIBr2. For the solubility studies, a high-purity CuIBr is prepared via reduction of CuIIBr2 by dimethyl malonate as described elsewhere.26 The fine white CuIBr powder is stored under nitrogen atmosphere. Characterization. 1H NMR spectra are recorded on a Bruker ARX-200 spectrometer at 200 MHz. For the polymerizations, the chemical shifts in CDCl3 are reported downfield from 0.00 ppm using the residual CHCl3 signal at 7.26 ppm as an internal reference. Samples are diluted in CDCl3. Monomer conversion is calculated from the intensity ratio of OCH3 signals from the polymer (3.60 ppm) and monomer (3.75 ppm). The polymer number- and weight-average molecular weights (Mn and Mw, respectively) are determined by gel permeation chromatography relative to narrow polystyrene standards. A Waters 710 sample autoinjector, three linear columns in series (Waters Styragel HR 5E, 2× Shodex KF-804L), a Waters 600 pump system, and a 410 RI detector are used for the assays. The eluent, THF, is pumped through the system at a fixed flow rate of 1 mL/min. The columns and detector are heated to 30 and 35 °C, respectively. Data are recorded and manipulated using a Waters Millennium software package. Copper concentrations are measured using inductively coupled plasma atomic emission spectroscopy (ICP-AES) on a JarrellAsh ICAP 9000. Five analyses are performed for each sample and averaged. The measured intensities are converted to copper concentrations using a 12 point calibration curve ranging from 0 to 60 ppm. Solubility Experiments. Toluene and MIB are loaded into separate Schlenk flasks sealed with poly(tetrafluoroethylene) (PTFE) stopcocks equipped with latex septums. Each flask is submitted to five vacuum-nitrogen purge cycles. The flask contents are then bubbled with nitrogen for 1 h prior to use. Ligands are purged similarly and then bubbled with nitrogen for 15 min prior to use. The desired CuIBr or CuIIBr2 is loaded into a 5 mL Schlenk flask with a stir bar. The flask is sealed by a PTFE stopcock equipped with a latex septum. A retaining clip and Parafilm wrapping secures the stopcock-flask ground glass connection. The flask is submitted to five vacuum-nitrogen purge cycles. Toluene, MIB, and ligand (previously purged as described) are added to the flask via the septum using nitrogen-purged syringes. For CuIBr, the solution will remain colorless so long as air has been excluded from the flask. Air ingress will cause the solution to turn green as CuI is oxidized to CuII. The flask is placed in an oil bath preset at 90 °C with stirring for the desired period of time. The flask contents are then hot filtered through a 0.1 µm PTFE syringe filter. The filtrate is homogeneous until transferred to a 25 mL flask open to the atmosphere; there it becomes heterogeneous with the formation of a green precipitate. The contents of the flask are rotaryevaporated under vacuum, leaving only the precipitate. To this, 1 mL of concentrated HCl is added to liberate the precipitate from the flask walls. The sample is diluted with deionized water, and its total copper content is determined by ICP-AES analysis. The soluble copper concentration in the organic system is then calculated. Polymerization. A typical polymerization using CuIBr/ HMTETA is as follows: CuIBr (Aldrich 98% purity) (64.5 mg, 0.45 mmol), is added to a 25 mL Schlenk flask. The flask is sealed by a PTFE stopcock whose stem is fitted with a latex septum. The flask is submitted to five vacuum-nitrogen purge cycles. Previously degassed MMA (4.5 g, 45 mmol) and toluene
2728
Ind. Eng. Chem. Res., Vol. 46, No. 9, 2007 Table 1. Soluble Copper for PMDETA- and HMTETA-Based Catalysts in Toluene PMDETA solubilitya mM ppm %b
HMTETA
CuI
CuII
CuI
CuII
19.4 1460 66
5.2 390 18
1.9 140 7
4.3 310 15
a Conditions are toluene at 90 °C and metal salt:ligand molar ratio of 1:1. b Percentage of metal salt added as CuIBr or CuIIBr2.
Figure 1. Solubility of CuIBr/HMTETA (4) and of CuIIBr2/HMTETA (2) as a function of equilibration time. Error bars show 95% confidence intervals. Experimental conditions as indicated.
(8.86 g) are added to the flask via a nitrogen-purged syringe. Previously purged HMTETA (122.3 µL, 0.45 mmol) and degassed initiator (MBP, 70.8 µL, 0.45 mmol) are then added dropwise to the flask with stirring. The flask is placed in an oil bath at 90 °C with stirring. Kinetic samples, 0.3 mL, are withdrawn with a nitrogen-purged syringe. The samples are stored in hermetic vials and placed in a freezer for future assay. Results and Discussion
Figure 2. Solubility of CuIBr/HMTETA (4) and of CuIIBr2/HMTETA (2) as a function of solvent composition (methyl isobutyrate and toluene). Error bars show 95% confidence intervals. Experimental conditions as indicated.
Experimental Conditions and Observations. The experiments presented herein were designed to investigate the solubility of the copper catalyst under ATRP conditions. Thus metal salt and ligand concentrations, the solvent used (toluene), and the experimental temperature (90 °C), are as found in ATRP. In some experiments methyl methacrylate is used to observe its effect on the solubility of the catalyst complex. In other experiments methyl isobutyrate, the saturated equivalent of methyl methacrylate, is used to investigate the effect of solvent polarity on catalyst solubility. In all the experiments, catalyst solubility is determined by measuring the total copper concentration (the catalyst metal center) in solution by ICP-AES. The two ligands studied (PMDETA (1) and HMTETA (2)) vary only in the number of tertiary amine groups they contain, as shown in Scheme 2. Both ligands are widely used in ATRP to form catalyst complexes with CuIBr and CuIIBr2. HMTETA is of primary interest in our research and for this reason the main focus of this study.3,27-29 Also presented are the solubility data of PMDETA catalysts. The catalyst complexes dissolved/suspended in solvents have distinct appearances. CuIBr/HMTETA-containing solutions appear opaque and white due to catalyst precipitates. In contrast, CuIIBr2/HMTETA solutions are transparent with a slight yellowbrown color. Precipitates are also present but adhere to the walls of the flask as brown-green oily residues. PMDETA catalysts impart similar characteristics to their solutions. CuIIBr2/PMDETA colors the solution and forms oily precipitates as observed for CuIIBr2/HMTETA. CuIBr/PMDETA gives a colorless solution with suspended white precipitates. The quantity of precipitates formed with PMDETA is however much less than that with HMTETA. As a result, the CuIBr/PMDETA solution appears only slightly cloudy. In all cases, the precipitates are the insoluble fraction of the catalyst complexes. For all the catalysts, the hot-filtered solutions are clear and colorless. Once exposed to air and cooled, however, solutions turn green with green precipitates. Visual observations provide
some idea of the catalyst solubility prior to assay. The darkest and most opaque samples correspond to the highest copper concentration. The darkest to lightest samples are CuIBr/ PMDETA, CuIIBr2/PMDETA, CuIIBr2/HMTETA, and CuIBr/ HMTETA. As discussed below, the catalyst solubilities follow the same trend. Solubility Equilibrium. We investigated the effect of experimental time on catalyst solubility to determine the time required to reach solubility equilibrium. Experiments were run with time periods varying from 15 min to 6 h for both CuIBr/ HMTETA and CuIIBr2/HMTETA. The Schlenk flask, containing degassed ATRP ingredients, was placed in an oil bath preset at 90 °C and stirred for the set time. The solution in the flask was then hot filtered into another flask for total copper analysis by ICP-AES. Figure 1 shows CuIBr/HMTETA solubility and CuIIBr2/ HMTETA solubility as a function of the time spent in the oil bath. Solubilities do not vary with time, indicating that an equilibrium between dissolved and precipitated catalyst has been attained. A convenient experimental time of 1 h was therefore selected for the study. The reproducibilities of the 1 h experiments are good: the 95% confidence intervals are (10% and (12% of the average values for the CuI and CuII catalysts, respectively. For PMDETA, a 1 h solubility period is also sufficient to reach equilibrium. The solubility of CuIBr/ PMDETA is 19.4 and 22.8 mM following 1 and 6 h, respectively. For CuIIBr2/PMDETA it is 5.2 and 5.4 mM following 1 and 6 h, respectively. CuIBr/HMTETA and CuIIBr2/HMTETA Solubility in Toluene. HMTETA is a tetradentate ligand, as shown in Scheme 2, that forms partially soluble catalyst complexes with CuIBr and CuIIBr2 for ATRP. The soluble fractions of these catalysts are highly active for the polymerization of methyl methacrylate, and this allows large reductions in catalyst usage.3 ATRPs with catalyst concentrations as low as 0.3 mM (20 ppm) yield
Ind. Eng. Chem. Res., Vol. 46, No. 9, 2007 2729 Table 2. Soluble Copper for HMTETA-Based Catalysts in Various Solvents CuI
a
solventa
mM
toluene toluene + MIBc toluene + MMAc
1.9 2.6 3.9
CuII
ppm
%b
mM
ppm
%b
140 191 279
7 9 13
4.3 5.0 3.3
310 374 233
15 17 11
Conditions are 90 °C, HMTETA as ligand, and metal salt:ligand molar ratio of 1:1. b Percentage of metal added as CuIBr or CuIIBr2. c 66 wt % toluene.
controlled polymerizations.3 These catalysts are therefore of great industrial interest. Figure 1 presents the solubilities of CuIBr/HMTETA and CuIIBr2/HMTETA in toluene at 90 °C, which are 1.9 and 4.3 mM, respectively. These soluble copper concentrations are much lower than the total copper added in these experiments and in ATRP (29.4 mM). Only a fraction of the added copper is therefore soluble: 6.7% for CuIBr/HMTETA and 14.5% for CuIIBr2/HMTETA. Differences in catalyst solubility are explained by differences in the lyophilic nature of the complexes. A more aliphatic and less ionic complex is usually more soluble in an organic nonpolar medium such as toluene. Given that CuIBr and CuIIBr2 are expected to form complexes with HMTETA in a oneto-one molar ratio, large differences in aliphatic character between these complexes are not anticipated. It is more probable that the observed solubility differences result from differences in the ionic character of the complexes. The small solubility differences between CuIBr/HMTETA and CuIIBr2/HMTETA suggest that CuIBr/HMTETA is slightly more ionic in character than CuIIBr2/HMTETA. Solid state X-ray crystallography studies have shown both complexes to be ionic.30-32 Solution state studies have not however been undertaken, so solid state models only provide an indication of what structures are to be found in solution. In the solution state other neutral structures may exist in equilibrium with those found in the solid state. The low solubilities measured here however suggest that both structures retain their ionic character in solution. Extended X-ray absorption fine structure (EXAFS) spectroscopy studies are needed to characterize the solution state structures of these complexes. Knowledge of the solution structures may provide an explanation as to why CuIBr/HMTETA is slightly less soluble than CuIIBr2/HMTETA. Pintauer et al. studied the structures of CuIBr/PMDETA and CuIIBr2/PMDETA in toluene but were unable to characterize the latter due to its poor solubility in the solvent.33 Considering the similarly low solubilities of CuIIBr2/PMDETA, CuIBr/ HMTETA, and CuIIBr2/HMTETA, determining the HMTETA complex structures by EXAFS may also be problematic. Resolving the interesting question of why CuIBr/HMTETA is slightly less soluble than CuIIBr2/HMTETA lies outside our needs in engineering design and requires further effort from the research community. CuIBr/PMDETA and CuIIBr2/PMDETA Solubility in Toluene. PMDETA is a commercially available tridentate ligand, as shown in Scheme 2, that forms active catalyst complexes with CuIBr and CuIIBr2 for ATRP. It is widely used in the ATRP literature because of its high availability and low cost.34 Table 1 compares the solubilities of these complexes to the corresponding HMTETA complexes in toluene. CuIBr/PMDETA is more soluble than CuIBr/HMTETA. PMDETA solubilizes 19.4 mM or 66% of the CuI added compared to 1.9 mM (6.7%) with HMTETA. CuIIBr2/PMDETA is slightly more soluble than CuIIBr2/HMTETA, 5.2 vs 4.3 mM, respectively.
Figure 3. Solubility of CuIBr/HMTETA (4) and of CuIIBr2/HMTETA (2) as a function of ligand concentration for a fixed metal salt concentration. Error bars show 95% confidence intervals. Experimental conditions as indicated.
While CuIIBr2/PMDETA, CuIBr/HMTETA, and CuIIBr2/ HMTETA have similar solubilities (1.9-5.2 mM Cu), CuIBr/ PMDETA is much more soluble (19.4 mM). This suggests that CuIBr/PMDETA forms a more neutral complex than those formed by CuIIBr2/PMDETA, CuIBr/HMTETA, and CuIIBr2/ HMTETA. This is supported by the work of Pintauer et al., who studied the structures of CuIBr/PMDETA and CuIIBr2/ PMDETA at room temperature in toluene using EXAFS.33 Under these conditions CuIBr/PMDETA was best modeled as a neutral complex with the structure [CuI(PMDETA)Br].33 Our solubility results support such a neutral complex model. Pintauer et al. also attempted to study the solubility of CuIIBr2/PMDETA in toluene; however, its low solubility prevented them from determining the structure.33 In a more polar solvent (toluene/ methanol ) 4:1) EXAFS results did suggest the ionic structure [CuII(PMDETA)Br]+[Br]-.33 For the HMTETA complexes, solid state X-ray crystallography studies of copper halide/ HMTETA complexes have shown them to be ionic.30 Becker et al.31 isolated and characterized CuICl/HMTETA complexes, reporting the molecular structure [CuI(HMTETA)]+[CuICl2]-. An analogous complex is likely formed with CuIBr. CuIIBr2/ HMTETA complexes have the molecular formula [CuII(HMTETA)Br]+[Br]-.32 These X-ray crystallographic studies along with our solubility data suggest that CuIBr/PMDETA is more soluble than CuIIBr2/PMDETA, CuIBr/HMTETA, and CuIIBr2/HMTETA because of its more neutral structure. Effect of Solvent Type. Solvents, including monomers, can improve catalyst solubility by modifying the polarity of the medium and/or by coordinating with the catalyst complex itself. The solubilities discussed so far are in toluene. The presence of monomer, as in ATRP, may alter these results directly through coordination of the monomer with the catalyst metal center or indirectly through changes in solvent polarity. Coordination is known to occur between methyl methacrylate and the metal center of the multidentate amine complex, but to a very low level.35 This is because coordination to the bromide
2730
Ind. Eng. Chem. Res., Vol. 46, No. 9, 2007
Table 3. Soluble Copper and Polymerization Rates as a Function of HMTETA Concentration soluble CuI a [HMTETA] (mM) 29 147 294
Mt:La
(molar ratio) 1:1 1:5 1:10
mM
ppm
kappb (M-1 s-1)
4.3 13.2 19.6
310 960 1430
5.00 × 10-5 1.20 × 10-4 1.60 × 10-4
a Conditions are 90 °C in toluene, HMTETA as ligand, and metal salt:ligand (Mt:L) molar ratio as shown. b Conditions are 90 °C, toluene/MMA ) 2 (w/w), MBP as initiator, Mt:L as shown, and [MMA]:[MBP]:[CuIBr] ) 100:1:1; kapp data from ref 3.
anion is highly favored over coordination with the monomer.35 For example, under the base case ATRP conditions investigated in this study ([MMA] ) 3 M, [CuBr] ) [L] ) 0.03 M, 90 °C) and using the equilibrium constants for the anion and monomer coordination derived by Braunecker et al.,35 which are respectively KBr ) 3.8 × 106 M-1 and KM ) 0.97 M-1, we calculate that only 0.4% of the catalyst complex is coordinated with the monomer. That is, the majority of the cationic complex (99.6%) will be coordinated with the anion (Br-) rather than with the monomer. Monomer should therefore primarily affect catalyst solubility through polarity effects rather than through coordination. Methyl isobutyrate (MIB) provides a convenient model for methyl methacrylate (MMA). MIB is the saturated counterpart of MMA and is therefore not prone to polymerization under experimental conditions. Furthermore, MIB does not π-coordinate with the catalyst metal center. CuIBr/HMTETA and CuIIBr2/HMTETA solubility were measured at varying volume fractions of MIB in toluene (see Figure 2). The solubilities of both complexes increase with increasing MIB content. Methyl methacrylate was also used to measure catalyst solubility, but this caused the system to polymerize. Table 2 compares catalyst solubility in the presence of MMA, MIB, and toluene. CuIBr/HMTETA solubility is higher with MMA (3.9 mM) than with MIB (2.6 mM). This difference in solubility is not large, given experimental error ((10%) and the variance caused by unwanted polymerization. A large difference in solubility between MMA and MIB is not expected since bromide complexation is favored over π-coordination as discussed above. CuIIBr2/HMTETA solubility is lower with MMA (3.3 mM) than with MIB (5 mM). The former result may however be erroneous since polymerization takes place, resulting in the reduction of CuIIBr2/HMTETA to CuIBr/HMTETA (see reverse reaction in Scheme 1). The assay cannot distinguish between CuIBr/HMTETA and CuIIBr2/HMTETA and measures the total copper. A reduction of the catalyst would cause an apparent lowering of the catalyst solubility since CuIBr/HMTETA has a lower solubility than CuIIBr2/HMTETA. In 33% MIB and 33% MMA only 9 and 13% (2.6 and 3.9 mM) respectively of the CuI added is soluble. Our prior kinetic work suggests a solubility limit below 2.9 mM.3 The kinetic and solubility data are therefore consistent and together explain why a 10-fold reduction in the amount of CuIBr typically used in ATRP can be made without significantly affecting the polymerization rate. This is because of the limited catalyst solubility in the reaction medium. Such data are useful in the design of ATRP processes for reduction of catalyst use and minimization of purification. Metal Salt:Ligand < 1:1. The solubility of CuI and of CuII increases with increasing ligand concentration as shown in Figure 3. Polymerization rates are also found to increase with ligand concentration.3 This is distinct from the trend observed for homogeneous catalysts. In homogeneous catalysis, maximum polymerization rates are attained at metal salt:ligand molar ratios of 1:1 or 1:2.5,36-40 Further increasing the ligand concentration
Figure 4. Apparent rate constants (kapp) (1/s) vs soluble copper concentration (M) for ATRPs of MMA with varying HMTETA concentrations (29, 147, and 294 mM), toluene/MMA ) 2 (w/w), [MMA]/[initiator]/[CuIBr] ) 100:1:1 (molar), 90 °C, stirred. Soluble copper concentrations are from solubility experiments with varying concentrations of HMTETA (29, 147, and 294 mM) in toluene at 90 °C, stirred.
Figure 5. Solubility of CuIBr/HMTETA (4) and of CuIIBr2/HMTETA (2) as a function of added metal salt concentration at a fixed ligand concentration. Error bars show 95% confidence intervals. Experimental conditions as indicated.
does not result in an increased polymerization rate since all of the catalyst is already dissolved. For heterogeneous catalysts, increasing ligand concentration does augment polymerization rates as a result of an increased concentration of soluble catalyst sites. We measured CuIBr/HMTETA catalyst solubility and polymerization rates as a function of ligand concentration and report the data in Table 3. Polymerization rates correlate well (R2 ) 1.0000) with CuIBr/HMTETA solubility. Figure 4 plots the apparent rate constant, defined by eq 1, as a function of the soluble catalyst concentration. From the slope of this plot the
Ind. Eng. Chem. Res., Vol. 46, No. 9, 2007 2731 Table 4. Soluble Copper as a Function of PMDETA Concentration soluble CuI [PMDETA] (mM) 29 147 294
Mt:La
(molar ratio) 1:1 1:5 1:10
soluble CuII
mM
ppm
%b
19.4 27.5 26.5
1460 2070 1990
66 94 90
mM
ppm
%b
5.2 13.9 14.3
390 1050 1070
18 47 49
a Conditions are 90 °C in toluene, PMDETA as ligand, and metal salt:ligand (Mt:L) molar ratio as shown. b Percentage of metal added as CuIBr or CuIIBr2.
rate law shown as eq 2 is derived. The order (0.56) is intermediate to those theoretically predicted by Matyjaszewski (1.00) and Fisher (0.33).4,6,41,42
RP ) kP[R•][M] ) kapp[M]
(1)
RP ∝ [CuI]soluble0.56
(2)
For PMDETA-based catalysts, increasing the ligand concentration also causes an increase in catalyst solubility. The data for the PMDETA catalysts are summarized in Table 4. Metal Salt:Ligand > 1:1. At metal salt:ligand (Mt:L) molar ratios greater than 1:1, the complexes of CuIIBr2 with HMTETA or PMDETA have low solubility (see Figure 5). A sharp drop in solubility is observed when this ratio increases toward 4:1. The soluble copper concentrations are 4.28 and 0.06 mM (310 and 4 ppm) for Mt:L molar ratios of 1:1 and 4:1, respectively. A similar drop is observed for complexes of PMDETA with CuIIBr2; solubilities are 5.20 and 0.00 mM (390 and 0 ppm) at Mt:L molar ratios of 1:1 and 4:1, respectively. The drop in solubility with increasing copper content is remarkable. A 100-fold reduction in the soluble CuII concentration is attained with a 4-fold increase in the amount of CuIIBr2 added. The effect is easily observed. At the high Mt:L ratios the filtered samples do not turn green and do not form precipitate once exposed to the atmosphere; this is in contrast to the green samples collected at the lower ratios. With excess CuIIBr2, a black insoluble mass, having viscoelastic properties, forms on the flask bottom. When tapped quickly with a glass rod, the material behaves as a solid. However, the application of a constant pressure will cause it to flow. The same observations are made for PMDETA at high Mt:L molar ratios: black mass and filtrates free of copper precipitates upon air exposure. We conclude that adding CuIIBr2 in excess of the ligand concentration causes the formation of a polymeric network consisting of the ligand cross-linked by CuIIBr2 (see Scheme 3). This network is insoluble. This discovery has important implications for the recovery of copper catalyst from ATRP polymerizations as investigated and reported elsewhere.27 Based on the stoichiometry at which the drop in solubility is observed, the structure is of the form [CuII3L]n and/or [CuII4L]n. Similar structures have been proposed by Scalbert to explain the complexation and subsequent precipitation of CuII using plant polyphenols.43 Insoluble complexes between soluble CuII ions and polyphenols form when the metal concentration is in excess of that of the polyphenol’s chelating sites.43 Similar behavior is observed here (Figure 5). The formation of polymeric complex species from CuCl2 and tetrahydrothiophene has also been reported elsewhere with the molecular structure [CuI3CuIILCl5]n.44 The trend observed with excess CuIIBr2 is not observed with CuIBr. Increasing the concentration of CuIBr over that of the ligand does not cause a reduction in the soluble CuI concentration (Figure 5). At CuIBr:HMTETA molar ratios of 1:1 and
Figure 6. Conversion vs time (filled symbols) and first-order rate plots (open symbols) for the ATRP of MMA using catalysts CuIBr/PMDETA (b, O) and CuIBr/HMTETA (2, 4). Polymerization conditions: toluene/ MMA ) 2 (w/w), [MMA]/[initiator]/[CuIBr]/[ligand] ) 100:1:1:1 (molar), 90 °C, stirred.
Figure 7. Molecular weight (Mn) vs conversion (filled symbols) and molecular weight distribution (Mw/Mn) (open symbols) for the ATRP of MMA using catalysts CuIBr/PMDETA (b, O) and CuIBr/HMTETA (2, 4). Dashed line indicates theoretical Mn. Polymerization conditions as per Figure 6.
4:1, the soluble CuI concentrations are 1.9 and 1.8 mM, respectively. Determination of ATRP Equilibrium Constant. ATRPs of methyl methacrylate, using CuIBr/HMTETA or CuIBr/PMDETA as catalyst, were run to collect the necessary kinetic data for calculation of ATRP equilibrium constants (KATRP). Figure 6 shows the monomer conversion versus time and first-order rate plot for the polymerizations. The apparent rate constants, derived from the slope of the first-order rate plots, are 1.37 × 10-4 and
2732
Ind. Eng. Chem. Res., Vol. 46, No. 9, 2007
Scheme 3. Formation of Insoluble Metal/Ligand Complex Network27
1.13 × 10-4 s-1 for HMTETA and PMDETA, respectively. Figure 7 shows the molecular weights (Mn) and molecular weight distributions (Mw/Mn) of the poly(methyl methacrylate) produced. Initiator efficiencies, defined by eq 3, where the
Iteff ) [RX]t/[I]0 ) Mn,theoretical/Mn,SEC
(3)
subscripts “0” and “t” refer to the initial and final conditions, are 64% and 54% for HMTETA and PMDETA, respectively. They remain constant for the duration of the polymerization. The polymer produced has a low average polydispersity, Mw/ Mn ) 1.14 for HMTETA and 1.19 for PMDETA. The ATRP equilibrium constant can be calculated from the kinetic and solubility data based on the accepted ATRP mechanism shown in Scheme 1. Polymerization rates are described by eq 1, while the equilibrium is
KATRP )
[CuII]t[R•]t
(4)
[CuI]t[RX]t
Substituting for the radical concentration in eq 1 yields
RP )
[
]
kPKATRP[CuI]t[RX]t [CuII]t
[M] ) kapp[M]
(5)
which is rearranged to give KATRP:
KATRP )
kapp[CuII]t
(6)
kP[CuI]t[RX]t
Knowledge of the apparent rate constant (kapp), propagation rate constant (kP), and dormant polymer ([RX]t) and soluble catalyst ([CuI]t and [CuII]t) concentrations thus allow KATRP to be calculated. The slope of the first-order kinetic plot (ln[M]0/ [M]t vs t) provides the apparent rate constant (kapp) as outlined previously. The propagation rate constant is calculated from literature data to be 1619 L/(mol s) for methyl methacrylate at 90 °C.45 The dormant polymer concentration ([RX]t) is calculated from the initiator efficiency via eq 3. The initiator
efficiencies remain constant throughout the polymerization, as indicated by the linear relationship between polymer molecular weight (Mn) and monomer conversion. A constant dormant polymer concentration therefore exists. The catalyst concentrations remain the only unknowns to complete the calculation of KATRP via eq 6. With knowledge of the solubility limits for the CuI and CuII catalysts, the effective catalyst concentrations are now determinable. For example, the CuIBr/HMTETA and CuIIBr2/HMTETA solubility limits are 2.6 and 5.0 mM, respectively (Table 2). In contrast, the total concentrations of catalysts (soluble + insoluble) in the system are calculated by mass balance to be [CuIBr/L]t ) Ieff[CuIBr/L]0 ) 18.8 mM and [CuIIBr2/L]t ) [CuIBr/L]0 - [CuIBr/L]t ) 10.6 mM. Since the total catalyst concentrations exceed the solubility limits, the system is saturated in both catalyst types. The saturated (solubility) concentrations (2.6 and 5.0 mM) are therefore used to complete the calculation of KATRP. The dormant polymer concentration for the HMTETA system is 18.8 mM via eq 3. KATRP for CuIBr/HMTETA is calculated to be 8.66 × 10-6. The calculated KATRP lies within the expected range of equilibrium constants determined for other homogeneous catalysts (10-4-10-9).4-6,10,15 As expected from the fast polymerization rate obtained from this catalyst at low concentrations,3 the equilibrium constant is on the upper end of this range. The equilibrium constant of CuBr/Me6TREN, another highly active tetradentate catalyst, has been estimated to be on the order of 1.54 × 10-4.10,15 The KATRP we calculate here for CuBr/ HMTETA is close to this value. A similar calculation is undertaken for the KATRP of CuIBr/ PMDETA. In this case, however, while the system is saturated in the deactivating catalyst CuIIBr2/PMDETA, the activating catalyst concentration (CuIBr/PMDETA) is lower than the solubility limit. The solubility of CuIBr/PMDETA and that of CuIIBr2/PMDETA are 19.4 and 5.2 mM, respectively (Table 4), while the total catalyst concentrations, by mass balance, are [CuIBr/L]t ) 15.9 mM and [CuIIBr2/L]t ) 13.5 mM. The total CuIBr/PMDETA concentration (15.9 mM) and saturated CuIIBr2/PMDETA concentration (5.2 mM) are therefore used in the calculation of KATRP. The dormant polymer concentration is 15.9
Table 5. Calculated ATRP Equilibrium Constants (KATRP) for CuIBr/HMTETA and CuIBr/PMDETA Catalysts solubility limit (mM)
total catalysta (mM)
ligand
CuI
CuII
CuI
CuII
Ieff (%)
kapp × 104 (s-1)
KATRPb × 106
HMTETAc
2.6 19.4
5.0 5.2
18.8 15.9
10.6 13.5
64 54
1.37 1.13
8.66 1.44
PMDETAd
a Calculated by mass balance from polymerization data. b Calculated via eq 6 from solubility data and polymerization kinetics for the system [MMA]: [MBP]:[CuIBr]:[ligand] ) 100:1:1:1, [CuIBr] ) 29.4 mM, toluene/MMA ) 2 (w/w), 90 °C. c Conditions for solubility data are 90 °C in toluene/MIB ) 2 (w/w), [CuIBr]0/[HMTETA]0 ) 29.4 mM/29.4 mM. d Conditions for solubility data are 90 °C in toluene [CuIBr]0/[PMDETA]0 ) 29.4 mM/29.4 mM.
Ind. Eng. Chem. Res., Vol. 46, No. 9, 2007 2733
mM. The KATRP for CuIBr/PMDETA is calculated to be 1.44 × 10-6. This value is in the range of KATRP estimated by others for CuIBr/PMDETA through kinetic and model studies (10-510-8).10,12,15 Table 5 summarizes the solubility data, kinetics, and calculated equilibrium constants. Conclusions Solubilities of the ATRP catalysts CuIBr/HMTETA, CuIIBr2/ HMTETA, CuIBr/PMDETA, and CuIIBr2/PMDETA in toluene at 90 °C have been measured using ICP-AES. CuIBr/PMDETA has a higher solubility (19.4 mM soluble copper) than CuIIBr2/ PMDETA (5.2 mM), CuIIBr2/HMTETA (4.3 mM), and CuIBr/ HMTETA (1.9 mM). This difference in solubility suggests that CuIBr/PMDETA forms a more neutral and therefore more soluble complex than the other catalysts studied. The solubility of CuIBr/HMTETA in toluene/MIB ) 2/1 (w/w) at 90 °C is 2.6 mM. This solubility is comparable to that suggested by prior ATRP kinetic studies (