A Method of Improving the Theoretical Basis of kp Determination from PLP‐SEC Measurements

Making use of hitherto ignored features (such as the peak width) contained in the chain‐length distributions of polymers prepared by pulsed‐laser polymerization (PLP), corrections are calculated from simulated chain‐length distributions for improving the accuracy of the “characteristic chain length” L₀ data on which the evaluation of the propagation rate constant k_p is based. These corrections refer to a wide range of chain lengths and primary radical production, slightly chain‐length‐dependent termination by disproportionation or combination, and a reasonable extent of axial dispersion introduced by the chromatographic device used in the evaluation of the chain‐length distribution. They can be applied to the point of inflection on the low‐molecular‐weight side of the extra peaks as well as to the peak maximum. The remaining mean error which, of course, concerns the evaluation of L₀ only, is shown to be of the order of 1.0–1.5%, if the mode of termination is unknown, and comes down to about half that value if information on the mode of termination is available. Although all the other errors inherent in the size exclusion chromatography (SEC) method are still present, this method constitutes substantial progress with respect to the accuracy of determining k_p data from PLP experiments followed by chromatographic analysis.

Hyper mass distributions calculated for L₀ = 200, C = 5 and b = 0.16 for termination by disproportionation considering Poissonian and Gaussian broadening.     

Improving kp Data Originating From Higher Order PLP‐SEC Peaks

Summary: Based on certain features, especially the width of the so‐called extra peaks in the simulated chain length distribution (CLD) of polymers prepared by pulsed laser polymerization (PLP), it is calculated by which factor the positions of the higher order points of inflections and maxima deviate from the theoretical L₀ data that are to be used for the evaluation of k_p. These corrections, which can be put into the form of master equations, are for slightly chain length dependent termination by disproportionation or combination and cover a wide range of chain lengths and primary radical production and a reasonable range of axial dispersion σ_ad,k, caused by the chromatographic device used in the evaluation of the chain length distribution. They can be applied either to the point of inflection on the low molecular weight side of the extra peaks as well as to the peak maximum. For usual extents of column broadening (σ_ad,k ≈ 0.05) the mean error that is about 7% for uncorrected data from second order points of inflection is reduced to the order of 1.5% even if no assumption concerning the mode of termination is made. The situation is a little less satisfactory for the correction of the positions of the second order peak maxima. Third order peak data are a priori less falsified and yield still better results after correction. Thus the proper treatment of higher order peaks helps to extend the range of chain lengths for which highly reliable k_p data can be gained from PLP experiments followed by chromatographic analysis.

Plots of lequation/tex2gif-stack-1.gif/(nL₀) versus lg(L₀) obtained from first order (circles), second order (triangles) and third order (squares) peaks showing uncorrected values in the left diagram and corrected values using correction functions X in the right one, both calculated for σ_ad,k = 0.05. (+) and (×) represent ill‐defined peaks.     

The Influence of Chain Length‐Dependent Propagation on the Evaluation of the Chain Length Dependence of the Rate Coefficient of Bimolecular Termination, 1. PLP–SEC Methods

Chain length distributions have been calculated for polymers prepared by pulsed laser polymerization (PLP) under the condition that not only chain termination but also chain propagation is subject to chain length dependence. The interplay between these two features is analyzed with the chain length dependence of the rate coefficient of termination k_t introduced in the form of a power law and that of propagation k_p modeled by a Langmuir‐type decrease from an initial value for zero chain length to a constant value for infinite chain lengths. The rather complex situation is governed by two important factors: the first is the extent of the decay of radical concentration [R] during one period under pseudostationary conditions, while the second is that termination events are governed by [R]² while the propagation goes directly with [R]. As a consequence there is no general recommendation possible as to which experimental value of k_p is best taken as a substitute for the correct average of k_p characterizing a specific experiment. The second point, however, is apparently responsible for the pleasant effect that the methods used so far for the determination of k_t and its chain length dependence (i.e., plotting some average of k_t versus the mean chain‐length of terminating radicals on a double‐logarithmic scale) are only subtly wrong with regard to a realistic chain length dependence. This is especially so for the quantity k_t* (the average rate coefficient of termination derived from the rate of polymerization in a PLP system) and its chain length dependence.     

Improving kp Data Originating from PLP Number Distributions

Summary: Based on certain features, especially the width of the so‐called extra peaks in the simulated number‐chain‐length distribution (CLD) of polymers prepared by pulsed laser polymerization (PLP), it is calculated by which factor the positions of the points of inflections and maxima deviate from the theoretical L₀ data that are to be used for the evaluation of k_p. These corrections are for slightly chain‐length‐dependent termination by disproportionation or combination and cover a wide range of chain‐lengths and primary radical production. They can be applied either to the point of inflection on the low‐molecular weight side of the extra peaks or to the peak maximum. On average, the mean error that is about −2.5% for uncorrected data from first‐order points of inflection is reduced to the order of less than 1% even if no assumption concerning the mode of termination is made. The situation is similar for the positions of the first‐order peak maxima where the mean error of about +7% likewise can be cut down to less than 1% if the proper correction function is chosen. Second‐ and third‐order peaks data, which are a priori less falsified, yield still better results after correction. Mass sensitivity of the detection process has comparatively little effect: it is only for unrealistically high extents of chain‐length dependence in detection that considerable falsifications are to be expected. As an additional result it turned out that correction functions obtained for number distributions are also applicable to mass spectrometry raw distributions and even for mass distributions x(l) · l provided Poissonian broadening is the only broadening process.

Number distribution x_C(l) calculated for termination by combination times attenuation function F₁(l).     

Monte Carlo simulation of the chain length distribution in pulsed-laser polymerization experiments in microemulsion

The Monte Carlo method has been used for numerically simulating pulsed-laser polymerization (PLP) in microemulsion, in order to establish if a shift from inflection point to peak maximum as the best measure of the propagation rate constant, k_p, will occur theoretically. Termination is assumed to be instantaneous in the simulations as droplet sizes can be very small in microemulsions. From the results of the simulations it is found that instantaneous termination indeed causes the peak maximum to become the best measure of k_p. From these results it can be deduced that in bulk it is not simply the Poisson-broadening that causes the peak maximum to yield an overestimation of k_p. This overestimation is rather caused by the fact that the termination rate is finite leading to an asymmetrical peak in the molecular weight distribution. In combination with broadening this yields the inflection point to be the best measure of k_p in the bulk.     

Termination rate coefficients derived from PLP-SEC data

A novel procedure is outlined by which the termination rate coefficient, k_t, may be deduced from molecular weight and monomer conversion data of pulsed laser polymerization (PLP) – size exclusion chromatography (SEC) experiments. For this k_t analysis only the central part of the molecular weight distribution (MWD) between the first point of inflection (POI), that is also used for k_p analysis, and the third such POI is taken into account. Within this region a characteristic ratio of areas under the MWD is fitted either by using PREDICI or by applying a lumping scheme method. The success of the lumping scheme procedure is demonstrated for the bulk polymerization of butyl methacrylate. The k_t values derived by this method refer to small initial degrees of monomer conversion as are typical of PLP-SEC investigations. The relatively fast and efficient lumping scheme technique is restricted to situations where k_t may be considered independent of chain length and where chain transfer processes are not important.     

Laser Single Pulse Initiated RAFT Polymerization for Assessing Chain‐Length Dependent Radical Termination Kinetics

Summary: A novel method combining RAFT polymerization with pulsed‐laser initiation for determining chain‐length dependent termination rate coefficients, k_t, is presented. Degenerative chain‐transfer in RAFT enables single‐pulse pulsed‐laser polymerization (SP‐PLP) traces to be measured on systems with a narrow radical distribution that remains essentially unchanged during the experiment. SP‐PLP‐RAFT experiments at different polymerization times allow for determining k_t as a function of chain length via classical kinetics assuming chain‐length independent k_t.

Single‐pulse pulsed‐laser polymerization trace for BMPT‐mediated RAFT polymerization of butyl acrylate.     

Pulsed Laser Polymerization at Low Conversions: Broadening and Chain Transfer Effects

Pulsed laser polymerization (PLP) is widely employed to measure propagation rate coefficients k_p in free radical polymerization. Various properties of PLP have been established in previous works, mainly using numerical methods. The objective of this paper is to obtain analytical results. We obtain the most general analytical solution for the dead chain molecular weight distribution (MWD) under low conversion conditions which has been hitherto obtained. Simultaneous disproportionation and combination termination processes are treated. The hallmarks of PLP are the dead MWD discontinuities located at integer multiples of n₀ = k_pt₀C_M, where t₀ is the laser period and C_M is the monomer concentration. We show that chain transfer reduces their amplitude by factors , consistent with numerical results obtained by other workers. Here c_tr is the chain transfer coefficient and Ln₀ (L = integer) are the discontinuity locations. Additionally, transfer generates a small amplitude continuous contribution to the MWD. These results generalize earlier analytical results which were obtained for the case of disproportionation only. We also considered two classes of broadening: (i) Poisson broadening of growing living chains and (ii) intrinsic broadening by the MWD measuring equipment (typically gel permeation chromatography, GPC). Broadening smoothes the MWD discontinuities. Under typical PLP experimental conditions, the associated inflection points are very close to the discontinuities of the unbroadened MWD. Previous numerical works have indicated that the optimal procedure is to use the inflection point to infer k_p. We prove that this is a correct procedure provided the GPC resolution σ is better than nequation/tex2gif-stack-1.gif. Otherwise this underestimates Ln₀ by an amount of order σ²/n₀.

Schematic of a chain transfer reaction with monomer as the transfer agent.     

Pulsed‐laser polymerization‐gel permeation chromatographic determination of the propagation‐rate coefficient for the methyl acrylate dimer: A sterically hindered monomer

The methyl acrylate dimer (MAD) is a sterically hindered macromonomer, and the propagating radical can fragment to an unsaturated end group. The propagation‐rate coefficient (k_p) for MAD was obtained by pulsed‐laser polymerization (PLP). The Mark–Houwink–Sakaruda parameters required for the analysis of the molecular weight distributions (MWDs) were obtained by multiple‐detector gel permeation chromatography (GPC) with on‐line viscometry. The small radical created by the fragmentation results in a short‐chain polymer that means the MWD may no longer be given by that expected for “ideal” PLP conditions; simulations suggest that the degree of polymerization required for “ideal” PLP conditions can be obtained from the primary point of inflection provided the GPC traces also show a clear secondary inflection point (radicals terminated by the second, rather than the first, pulse subsequent to initiation). Over the temperature range of 40–75 °C, the data can be best fitted by k_p/dm³ mol^?1 s^?1 = 10^6.1 exp(?29.5 kJ mol^?1), with a moderately large joint confidence interval for the Arrhenius parameters. The data are consistent with an increased activation energy and reduced frequency factor as compared with acrylate or methacrylate; both of these changes can be ascribed to hindrance. © 2001 John Wiley & Sons, Inc. J Polym Sci Part A: Polym Chem 39: 3902–3915, 2001     

Pseudostationary Polymerization: Improving the Determination of the Rate Constant of Propagation

Pulsed-laser initiated polymerization (PLP) leads to chain length distributions with characteristic extrapeaks. The low molecular weight side points of inflection L_LPI are located near to (multiples) of a specific chain length L₀ which is equal to the product of pulse separation t₀ and propagation frequency k_p[M], i.e. rate constant of propagation k_p times monomer concentration [M], allowing a direct determination of k_p. However, Poissonian broadening inherent in the polymerization process as well as Gaussian broadening due to axial dispersion caused by the size exclusion chromatographic (SEC) device leads to a shift of L_LPI as compared to L₀ – its extent depending on the experimental parameters chosen – which in turn causes an error up to 10–20% in the rate constants evaluated. Fortunately, comparison of the experimental peak width with some sort of theoretical peak width yields several types of correction factors and furthermore master-correction functions which are able to reduce the remaining error on average by at least a factor of 10.     

Absolute propagation rate coefficients in radical polymerization from gel permeation chromatography of polymers produced by intermittent initiation

The pulsed laser polymerization technique is now a well accepted method to determine propagation rate coefficients for radical polymerization from molar mass distributions resulting from intermittent initiation. A simplified apparatus for the periodic photoinitiation is used which is much less expensive than the laser equipment. The usefulness of the simplified equipment was proved by the determination of k_p for styrene at technically relevant temperatures up to 130°C for the first time. Furthermore, careful inspection of the molar mass distribution (mmd) reveals that depending on the reaction conditions, inflection points (L_i) can not only be found at integer multiples of k_p • t_o • [M] but also at 0.5ⁱ • k_p • t_o • [M], i = 1, 2, 3, … . A rule to find the inflection points leading to correct values for k_p is proposed. It is shown that the shape of the mmd inter alia depends on the amount of primary radical termination compared to the termination reaction between growing chains. With dominant primary termination, the maxima of the distribution will give the correct k_p, whereas in the absence of primary termination the inflection points should be used. Experimental conditions like initiator concentration, light intensity etc. may influence the position of the L_i at least to some extent, and so may give a small but principal error or uncertainty in k_p. A new mathematical method for the time-dependent simulation of the resulting mmd is presented which allows the calculations being performed on a PC within an acceptable time.     

Free‐Radical Propagation Rate Coefficients via n‐Pulse Periodic Polymerization

Aspects of applying n‐pulse periodic initiation in pulsed laser polymerization/size‐exclusion chromatography (PLP/SEC) experiments are studied via simulation of molecular weight distributions (MWDs). In n‐pulse periodic PLP/SEC, sequences of n laser pulses at successive time intervals Δt₁ up to Δt_n are periodically applied. With the dark time intervals being suitably chosen, n‐modal MWDs with n well separated peaks occur. The n‐pulse periodic PLP/SEC method has the potential for providing accurate propagation rate coefficients, k_p. Among several measures for k_p, the differences in molecular weights at the MWD peak positions yield the best estimate of k_p under conditions of medium and high pulse laser‐induced free‐radical concentration. Deducing k_p from n dark time intervals (corresponding to n regions of free‐radical chain length) within one experiment at otherwise identical PLP/SEC conditions allows addressing in more detail a potential chain‐length dependence of k_p. Simulations are compared with experimental data for 2‐pulse periodic polymerization of methyl methacrylate.

Measured MWD (solid line) and associated first derivative curve (dotted line) for a 2‐pulse periodic bulk polymerization experiment of MMA at 20 °C.     

On the (im)possibility of evaluating correct individual rate constants of chain propagation kp and chain termination kt by combining kp2/kt and kp/kt data for chain‐length dependent termination

On the basis of simulated data two ways of evaluating individual rate constants by combining k_p²/k_t and k_p /k_t (k_p , k_t = rate constants of chain propagation and termination, respectively) were checked considering the chain‐length dependence of k_t. The first way tried to make use of the fact that pseudostationary polymerization yields data for k_p²/k_t as well as for k_p /k_t referring to the very same experiment, in the second way k_p²/k_t (from steady state experiments) and k_p/k_t data referring to the same mean length of the terminating radical chains were compared. In the first case no meaningful data at all could be obtained because different averages of k_t are operative in the expressions for k_p /k_t and k_p²/k_t. In spite of the comparatively small difference between these two averages (≈15% only) this makes the method collapse. The second way, which can be regarded as an intelligent modification of the “classical” method of determining individual rate constants, at least succeeded in reproducing the correct order of magnitude of the individual rate constants. However, although stationary and pseudostationary experiments independently could be shown to return the same k_t for the same average chain‐length of terminating radicals within extremely narrow limits no reasonable chain‐length dependence of k_t could be derived in this way. The reason is an extreme sensitivity of the pair of equations for k_p/k_t and k_p²/k_t towards small errors and inconsistencies which renders the method unsuccessful even for the high quality simulation data and most probably makes it even collapse for real data. This casts a characteristic light on the unsatisfactory situation with respect to individual rate constants determined in the classical way, regardless of a chain‐length dependence of termination. As a consequence, all efforts of establishing the chain‐length dependence of k_t are recommended to avoid this way and should rather resort to methods based on inserting a directly determined k_p into the equations characteristic of k_p²/k_t or k_p/k_t, properly considering the chain‐length dependent character of k_t.     

Modeling Termination Kinetics of Non‐Stationary Free‐Radical Polymerizations

Pulsed‐laser induced polymerization is modeled via an approach presented in a previous paper.^[1] An equation for the time dependence of free‐radical concentration is derived. It is shown that the termination rate coefficient may vary significantly as a function of time after applying the laser pulse despite of the fact that the change in monomer concentration during one experiment is negligible. For the limiting case of t ≫ c^–1 (k_pM)^–1, where c is a dimensionless chain‐transfer constant, k_p the propagation rate coefficient and M the monomer concentration, an analytical expression for k_t is derived. It is also shown that time‐resolved single pulse‐laser polymerization (SP–PLP) experiments can yield the parameters that allow the modeling of k_t in quasi‐stationary polymerization. The influence of inhibitors is also considered. The conditions are analyzed under which M (t) curves recorded at different extents of laser‐induced photo‐initiator decomposition intersect. It is shown that such type of behavior is associated with a chain‐length dependence of k_t.     

New aspects of pseudostationary polymerization and their application

Contrary to the stationary state little thought has been given so far to the general principles of the pseudostationary state. In this discourse an attempt is made to demonstrate that — within wide limits — arbitrary initiation profiles may be used to determine k_p/k_t (k_p = rate constant of chain propagation, k_t = rate constant of chain termination) from the frequency dependence of rate of polymerization (in analogy to the rotating-sector technique) as well as to evaluate k_p from the chain-length distribution (CLD) of samples prepared under pseudostationary conditions. Adverse factors like nonspontaneous transformation of absorbed photons into primary radicals do not invalidate this result. The existence of a universal relationship (independent of the initiation profile) is proved to exist for the second moment of the CLD of samples prepared under pseudostationary initiation conditions for constant (chain-length independent) k_t. Pseudostationarity, however, might be also achieved if not the initiation but the termination is periodically varied. In this case the CLD has a completely different shape but allows determination of k_p likewise. Finally, the case of chain-length dependent k_t is shortly discussed in connection with pulsed-laser initiation. Although the general equation for the second moment of the CLD does not apply any longer for this case some generality appears to exist under these conditions, too.     

Modeling termination kinetics in free‐radical polymerization using a reduced number of parameters

An approach for modeling chain‐length dependent termination rate coefficients is presented. The method is based on the assumption that free‐radical chain length may be considered as a continuous variable. As compared to discrete numerical methods, in continuous modeling the number of independent dimensionless parameters can be significantly reduced. As a consequence, for a wide variety of monomers the conversion dependence of k_t can be predicted without extensive numerical calculations. The method may also be used to determine polymerization conditions under which simpler models of k_t (which neglect effects arising from the dependence of k_t on chain length) may be applied. Calculations for methyl methacrylate, styrene, and butyl acrylate bulk polymerizations up to high degrees of monomer conversion show that the impact of chain length on termination varies with conversion and strongly depends on the type of monomer.     

Simulation of pseudostationary radical polymerization, 1. Chain-length distributions for arbitrary initiation profiles

A procedure is developed which allows to treat arbitrary periodic initiation profiles (asymmetric and symmetric triangle profiles, sinusoidal profiles, Gaussian profiles etc.) in pseudostationary radical polymerization. Using an iterative method these profiles are transformed into the (likewise periodic) radical profiles and into the chain-length distributions of the resulting polymer in case of termination by disproportionation. These distributions are analysed for the position of their inflection points which may be used for experimental determination of the elementary rate constant of chain propagation k_p. It turned out that for all profiles that have at least one discontinuity (e.g. asymmetric triangle profiles) the position of the point of inflection is a correct measure of k_p for a conveniently wide range of experimental parameters. In case of profiles without discontinuity (symmetric triangle profiles, sinusoidal and Gaussian profiles) the position of the inflection point is shifted to lower values which means that the k_p values determined on this basis will be a little too small. In most cases, however, the error introduced by this fact will not exceed the overall error of the experiment so that in practice the method of determining k_p in pseudostationary polymerization is not restricted to those profiles which exhibit discontinuities.     

Modeling the Effects of Chain Length on the Termination Kinetics in Multivinyl Photopolymerizations

A model is presented that predicts photopolymerization kinetics over several orders of magnitude change in initiation rate. The model incorporates polymerization features that have long been assumed negligible when examining multivinyl photopolymerizations. The assumption that radical termination is chain‐length‐independent is relaxed by incorporating a chain‐length‐dependent termination (CLDT) parameter based on Random‐walk theory into the kinetic model. Experiments and modeling of multivinyl free‐radical photopolymerizations clearly demonstrate that CLDT is important at low conversions, where a deviation from the classical square‐root relationship between polymerization rate (R_p) and initiation rate (R_i) is observed (R_p ∝ R _i^α, α = ¹/₂, classically). At moderate conversions, when reaction diffusion dominates termination, a transition region is observed from a chain‐length‐dependent to a chain‐length‐independent region. During this transition, long chain – long chain termination is reaction diffusion controlled while the short chain – short chain termination event remains translational and segmental diffusion controlled. The scaling exponent, α, gradually increases throughout this region until achieving the classical value, where once attained, a plateau is observed. Chain‐length effects were also examined by including chain‐transfer (CT) reactions into the kinetic expressions. Upon CT agent addition, a transition region is still observed; however, at low conversion, α adheres more closely to the classical predictions. Most importantly, the model clearly demonstrates a transition from a CLDT region at low conversion to reaction diffusion controlled termination region at high conversion, where chain length is unimportant.     

Investigation of the Chain Length Dependence of kp: New Results Obtained with Homogeneous and Heterogeneous Polymerization

New experimental results were collected for the free radical polymerization of styrene by pulsed laser polymerization in solution or in microemulsion. The location of the point of inflection (on the low molecular weight side) and the maximum of the first peak in the chromatograms (measured by size-exclusion chromatography) was used to extract k_p data. The extent of band broadening was determined with narrow polystyrene standards with an assumed Poisson chain length distribution. For a given experiment both k_p values (obtained via the point of inflection and the maximum) were corrected and thus became identical in most cases. Even after the correction, the effect of chain length dependence persists to a higher chain length.