Article pubs.acs.org/IECR
Origin of Impurities Formed in a Polyurethane Production Chain. Part 2: A Route to the Formation of Colored Impurities June Callison,† Franziska Betzler,† Kimberly de Cuba,‡ Willem van der Borden,‡ Klaas van der Velde,‡ Robert H. Carr,‡ Hans M. Senn,† Louis J. Farrugia,† John M. Winfield,† and David Lennon*,† †
School of Chemistry, Joseph Black Building, University of Glasgow, Glasgow G12 8QQ, Scotland, U.K. Huntsman (Europe) BVBA, Everslaan 45, 3078 Everberg, Belgium
‡
S Supporting Information *
ABSTRACT: The quality of methylene diphenyl diisocyanate (MDI) products, which are valuable feedstocks in the industrial manufacture of polyurethanes, can be compromised by the presence of color, presumed to arise from trace impurities. One undesired branch in the synthesis chain originates with phosgenation of diaryl ureas, formed from reactions between aryl isocyanates and polyamine precursors. Subsequent key steps include, (i) breakdown of the primary compounds, substituted chloroformamidine-N-carbonyl chlorides (CCC), to give aryl isocyanide dichlorides, ArNCCl2, (ii) an apparent equilibrium connecting CCC with aryl carbodiimides, and (iii) the thermolysis of ArNCCl2 in the presence of MDI. Color formation is associated directly with the last process; it involves several events, including HCl elimination from reaction of ArNCCl2 and MDI, formation of carbon-centered radicals, and a contribution from oxidation at the methylene bridge.
1. INTRODUCTION Polyurethanes are a diverse group of polymers which find use in a wide range of applications.1 They are formed by a reaction between polyfunctional isocyanates and either polyether or polyester polyols. Commercially produced isocyanates can be divided into the high volume aromatics and low volume aliphatics and, of these, by far the most important is methylene diphenyl diisocyanate (MDI) which accounts for more than 60% of all global isocyanate production. MDI is used to describe both diisocyanate products and those based on the homologous isomeric mixture of polyisocyanate oligomers formed from the corresponding polyamine mixture, itself produced by acid-catalyzed condensation of aniline and formaldehyde, Scheme 1. In particular, the main diisocyanate isomer can be isolated from the oligomeric mixture to give socalled “pure MDI”, typically 98−99% 4,4′-MDI, the various commercial mixtures of di-isocyanates and higher oligomers being termed “polymeric MDI”. For example, a typical polymeric MDI has a composition of 4,4′-MDI (50%), 2,4′MDI (5−10%), and higher species with n = 1, 2 (Scheme 1); the average functionality of the mixture is near to 2.8.2 Despite recent interest in finding alternative routes for the production of isocyanates and related compounds,3−5 the phosgenation of primary aromatic amines remains the most widely used industrially.6 This article therefore concentrates on aspects of the isocyanate process chemistry connected with phosgenation routes. Within the industrial process, side reactions occur, which involve aryl poly(urea) formation and its subsequent phosgenation. These have been postulated as the route to the unwanted contamination of the MDI by strongly colored impurities.2 Ureas are formed from reactions between product isocyanates and reagent amines. The primary product from phosgenation of a urea is a chloroformamidine-N-carbonyl chloride, which can break down to form an isocyanide © 2012 American Chemical Society
dichloride. The latter is potentially a source of chlorine radicals which in turn can attack the backbone of the MDI polymeric chain at the methylene bridge, abstracting hydrogen, and forming a highly conjugated carbon-centered radical, which is strongly absorbing in the visible region. This postulated pathway was described to an extent some years ago,2 but to date, definitive evidence for its operation has been lacking. Recent work in these laboratories has validated the concept that aryl isocyanide dichloride species (ArNCCl2) can act as a source of chlorine radicals.6 The present work extends those studies and examines how phosgene facilitates the formation of ArNCCl2 species and other byproducts. By focusing on key model intermediates in the sequence outlined above and by undertaking several related physicochemical and spectroscopic studies, we have been able to substantiate the color-forming hypothesis and, furthermore, identify the chromophores, chromophore-precursors, and auxochromes active in the chemical system under investigation.
2. EXPERIMENTAL SECTION 2.1. Syntheses Involving Phosgene. CAUTION! Phosgene and triphosgene are extremely toxic. All personnel must be adequately protected with proper clothing and monitoring devices. Experimental protocols should involve the smallest possible inventories of OCCl2 Syntheses requiring phosgene were carried out in a fournecked Pyrex, round bottomed flask reactor that could be modified to suit the specific reactions. The temperature of the solution was measured using a Type K thermocouple inserted Received: Revised: Accepted: Published: 11021
April 15, 2012 July 20, 2012 August 2, 2012 August 2, 2012 dx.doi.org/10.1021/ie300987v | Ind. Eng. Chem. Res. 2012, 51, 11021−11030
Industrial & Engineering Chemistry Research
Article
Scheme 1. Synthesis of Methylene Diisocyanate (MDI)/Poly(methylene Diisocyanate, Poly(MDI)
For the second method, cartridges for safe generation of phosgene (Sigma-Aldrich) designed by Eckert7 were used. They consisted of a capped plastic test tube containing triphosgene and a deactivated amine or imine-based catalyst, used to stimulate the decomposition of the triphosgene into three moles of phosgene. The process was activated at the temperature which the triphosgene began to melt (∼353 K) and proceeded cleanly up to 383 K.8 The rate of reaction increases with temperature, therefore the release of phosgene could be controlled using a relationship describing the rate of phosgene production with time at different temperatures, provided by the suppliers.9 Background information relating to phosgene handling is given in the Supporting Information (section S.1). 2.2. Characterization of Products and Components of Mixtures. Except where described below, instruments and analytical procedures followed those described in Part 1.6 Electronic spectra were recorded using a Perkin-Elmer Lambda 850, double beam, double monochromator spectrometer with a photomultiplier R6872 detector, over the range 175−900 nm with a resolution of ≤0.05 nm. A sample of the solution to be analyzed was added to a rectangular quartz cell (Apollo Scientific), path length 10 mm. A second cell containing the solvent was used as a reference. Gel permeation chromatography was carried out on a Hewlett-Packard HP1550 instrument with a Polymerlab PLgel column, 5 μm, 100 Å, 300 × 7.5 mm. The UV detector used was a Spectroflow 757. Dichloromethane was the eluant with a flow rate of 0.75 cm3 min−1 for 40 min. A product mixture (1 g) was made up to 10 g with CH2Cl2. 2.3. Synthesis and Structure of Bis(4-benzylphenyl)urea. A solution of 4-benzylaniline (0.55 g, 3.0 mmol, Alfa Aesar, 98%) in chlorobenzene (15 cm3) was added to the phosgenation reactor described above. A solution of 4benzylphenyl isocyanate (0.52 g, 2.5 mmol, Sigma-Aldrich, 97%) in chlorobenzene (15 cm3) was added dropwise to the solution using a syringe. The reaction was carried out under N2 with an ice bath to ensure all the material would react. On addition of the isocyanate, the solution turned from clear to cloudy. After 2 h the product was recovered by vacuum filtration to remove any trace of the starting material, leaving
into a glass sleeve in one of the necks, making direct contact with the solution. Dinitrogen was flowed into the reactor via a Pyrex manifold, which comprised a series of PTFE/Pyrex stopcocks (J. Young), a regulator, trap, oil bubbler, and glass sparge tube. A side arm to the manifold served to connect an Eckert cartridge (see below) where required. The third neck was closed with a SubaSeal rubber septum through which the reagents could be added by means of a syringe. A stirrer hot plate (IKA) with an oil bath was used to heat the flask, using a contact thermometer to control the temperature of the oil. Magnetic stirrer bars were placed in both the four-necked flask and the oil bath. The solution could also be cooled by replacing the oil bath with a cooling bath. Different mixtures of solvent, ice, or solid CO2 were used to produce the desired temperatures depending on the situation. A condenser was fitted to the top of the flask, via the fourth neck with the output connected first to an expansion vessel, to avoid unexpected pressure increase, then to a scrubber system. The latter’s function was to neutralize any excess OCCl2 or product HCl produced. A peristaltic pump was used to circulate a 10% w/v aqueous NaOH solution through a Pyrex column containing Raschig rings at the point where the off-gas from the reactor entered the scrubbing system. A second condenser was included in the scrubber in order to stop the solution overheating; a thermocouple measured any change in temperature of the NaOH solution contained in the reservoir. The complete apparatus was housed within a fume cupboard, which was continuously vented during use. Reactions involving triphosgene as one of the reactants were carried out by first dissolving the triphosgene (Fluka, ≥99%) in the solvent, usually chlorobenzene (Sigma-Aldrich, ≥99%) and then adding this directly to the reactor using a funnel. The solution was purged for 5 min with N2, and a cyclohexane/CO2 bath was used to keep the solution below 283 K, while the other starting material was added using the syringe. The reaction mixture was brought to the desired temperature and left for a specified time. On completion, the solution was transferred to a single-necked round-bottom flask and the solvent removed by rotary evaporation. This procedure also ensured that no OCCl2 was left in the product. 11022
dx.doi.org/10.1021/ie300987v | Ind. Eng. Chem. Res. 2012, 51, 11021−11030
Industrial & Engineering Chemistry Research
Article
Table 1. Analysis and Characterisation Data for Bis(4-benzylphenyl)urea method elemental analysis (%) melting point (K) MS, FAB+ (m/z) IR, ν̃/ cm−1 (intensity)a 1
H NMR (400 MHz, DMSOd6) δ/ ppmb 13 C{1H}-NMR (100.5 MHz, DMSO-d6) δ/ ppm a
data C27H24N2O: found (required) C, 82.04 (82.65); H, 6.12 (6.12); N, 7.09 (7.14) 521 184 (C6H5CH2C6H4NH+), 392 (M+), 393 (M+H+) 3301 (vs), 3023 (w), 1636 (m), 1588 (m), 1543 (m), 1511 (w), 1414 (w), 1303 (m), 1231 (m), 848 (w), 793 (m), 772 (m), 741 (m), 725 (m), 694 (m), 650 (m), 612 (m) 3.91 (2H, s, CH2), 7.17 (4H, d, aromat. CH), 7.27 (10H, m, aromat. CH), 7.39 (4H, d, aromat. CH), 8.60 (2H, s, NH). 40.47 (CH2), 118.35 (aromat. CH), 124.88 (aromat. CH), 128.41 (aromat. CH), 128.61 (aromat. CH), 129.01 (aromat CH), 134.63 (aromat. Cq), 137.71 (aromat. Cq), 141.67 (aromat. Cq), 152.56 (NHCqONH)
vs, very strong; w, weak; m, medium. bs, singlet; d, doublet; m, multiplet.
behind a colorless powder. On the basis of its analysis and spectra (Table 1) it was identified as bis(4-benzylphenyl)urea. A suitable crystal of bis(4-benzylphenyl)urea was cooled from ambient temperature to 100 K over a period of 1 h, using an Oxford Instruments cryostream. Data were collected on a Bruker Apex-II diffractometer. Data reduction, structure solution, and structure refinement were carried out using the WinGX package10 of crystallographic programs. Since the absolute structure could not be determined from the diffraction data, all Friedel pairs were merged using mmm symmetry. Refinement with the SHELXL97-211 program was undertaken using full-matrix least-squares on F2 and all the unique data. All non-H atoms were allowed anisotropic thermal motion. The aromatic hydrogen atoms were included at calculated positions, and their positional parameters were tied to the attached C atom, with isotropic displacement parameters 1.2 times that of the attached C atom. The position of the amine hydrogen atom H(1) was determined from a difference Fourier map and its positional and isotropic displacement parameters were freely refined. Further details of the data collection and refinement are given in Table 2. Thermal ellipsoid plots were obtained using the program ORTEP-3 for Windows.12 The structure has been deposited with the Cambridge Crystallographic Data Centre, with deposition code CCDC 865560. The interaction energy between 2 bis(4-benzylphenyl)urea molecules in vacuum, stacked along the crystallographic c axis as observed in the solid state, was calculated using densityfunctional theory. Coordinates were taken from the experimental structure, with X−H distances set to standard values. The dispersion-corrected B97-D exchange-correlation functional was used with the def2-TZVPP basis set. The calculations were performed with the TURBOMOLE program13 making use of the highly efficient multipole-accelerated RI-J method. 2.4. Phosgenation of Ureas. Reactions between phosgene and 1,3-diphenylurea (Aldrich, 98%) or bis(4-benzylphenyl)urea, prepared as described above, were carried out using triphosgene or OCCl2 liberated from an Eckert cartridge, using a variety of reaction stoichiometry and temperatures. Full details are given in the Supporting Information, section S.2. At the end of each reaction, the solvent, usually chlorobenzene, was removed by rotary evaporation and the products, normally a combination of several compounds, were subjected to an initial examination using FTIR spectroscopy. Characteristic IR bands used to identify the species encountered in the mixtures are given in Table S.2 of the Supporting Information. In the light of the results from the initial IR examination, other physical and spectroscopic examinations were made to supplement the IR data. 2.5. Preparation of 1,3-Bis(4-methylphenyl)chloroformamidine-N-carbonyl Chloride. This model
Table 2. Experimental Crystallographic and Refinement Data for Bis(4-benzylphenyl)ureaa compound formula Mr space group cryst syst a (Å) b (Å) c (Å) V (Å−3) Z Dcalcd (g cm−3) F(000) λ/Å μ(Mo Ka) (mm−1) cryst size (mm) transm coeff (range) θ range/deg no. of data used for merging no. of unique data hkl range Rint Rσ No. of data in refinement no. of refined parameters Final R [I > 2σ(I)] (all data) Rw2 [I > 2σ(I)] (all data) GOF S max features in electron density map (eÅ−3) max shift/esd in last cycle
C27H24N2O 392.48 Fdd2 orthorhombic 18.6632(12) 46.730(3) 4.6397(3) 4046.4(5) 8 1.289 1664 0.71073 0.079 0.40 × 0.10 × 0.05 0.897−0.991 1.74−27.01 23616 1229 −21 → 23; −58 → 56; −5 → 5 0.041 0.035 1229 142 0.0294 (0.0347) 0.0641 (0.0655) 1.143 0.175 (max) −0.178 (min)
