Online Monitoring of Styrene Polymerization in Miniemulsion byHyperpolarized 129Xenon NMR SpectroscopyMathis Duewel, Nicolas Vogel, Clemens K. Weiss, Katharina Landfester, Hans-Wolfgang Spiess,and Kerstin Munnemann*

Max Planck Institute for Polymer Research, Ackermannweg 10, 55128 Mainz, Germany

*S Supporting Information

ABSTRACT: Online monitoring of a miniemulsion polymerization of styrene by hyperpolarized 129Xe NMR spectroscopy ispresented. The chemical shift of 129Xe directly reports on the monomer/polymer ratio in the reaction mixture and therefore onthe conversion of the reaction. The method allows for monitoring the progress of a polymerization with high time resolution andwithout sample extraction. The results obtained by 129Xe NMR spectroscopy were successfully validated by comparison withresults of calorimetry. Further characterization of the polymer dispersions with regard to the solids content and the particlediameters proved the comparability of both methods. Small deviations in the kinetic data and the properties of the polymersuspension from both methods are explained by the loss of monomer during the 129Xe NMR experiments due to the used setup.In the future, 129Xe NMR might also be applied for online monitoring of other chemicals reactions, e.g., allowing for determiningthe kinetics of thermoneutral reactions.

■ INTRODUCTIONMany chemical and physical properties of polymers depend onthe numerous parameters of the polymerization process.Adjusting the polymer properties can thus be achieved bytailoring the polymerization process.1 Online monitoringmethods offer an efficient control of the reaction, which leadsto products with constant quality at minimal costs.2 The mostvaluable ones work without the need of sample extraction.Common techniques to monitor industrial polymerizationreactions are reaction calorimetry, optical spectroscopymethods like IR or Raman spectroscopy, and process gaschromatography.3,4 The widely used calorimetric measurementsare suitable for immediate reaction control, but only for exo- orendothermic reactions.NMR spectroscopy is another useful tool for online

monitoring of polymerization reactions, providing valuableinformation about the composition of the reaction mixture.5−7

It has been shown that online NMR spectroscopy formonitoring polymerizations can provide an acceptable timeresolution using 1H NMR and 13C NMR.2,8 However, theintrinsic low sensitivity of NMR measurements and the need oftime-consuming signal averaging often lead to problems in

achieving this, which is especially true for 13C at naturalabundance limiting the achievable time resolution to a fewminutes. Isotopic enrichment of 13C would help to overcomethis problem but is much too costly. Moreover, 1H NMRmostly requires the use of expensive deuterated solventshampering its application for industrial processes.Several so-called hyperpolarization methods can overcome

the lack of sensitivity of NMR spectroscopy, allowing for anexcellent time resolution in dynamic measurements. Theycreate a large nonequilibrium population of the Zeeman energylevels of the used nucleus thus achieving signal enhancementsof several orders of magnitude compared to the thermalcase.9−11 Different hyperpolarization methods like dynamicnuclear polarization (DNP),12 para-hydrogen induced polar-ization (PHIP),13,14 or the hyperpolarization of noble gases byspin-exchange optical pumping (SEOP)15,16 have beendeveloped until now. In this work, hyperpolarization of thenoble gas 129Xe via spin-exchange optical pumping was applied

Received: November 30, 2011Revised: January 22, 2012Published: February 8, 2012

Article

pubs.acs.org/Macromolecules

© 2012 American Chemical Society 1839 dx.doi.org/10.1021/ma202605n | Macromolecules 2012, 45, 1839−1846

using an apparatus described elsewhere.17 During the process ofSEOP, angular momentum is transferred from the photons ofcircular polarized laser light to the electron spins of valenceelectrons of rubidium and subsequently to the nuclear spins ofxenon gas.15,16 The hyperpolarization process takes placeoutside the NMR spectrometer, before the gas is broughtinto contact with the sample.The signal enhancement in NMR spectroscopy due to the

hyperpolarization of 129Xe offers the possibility of fastmeasurements and the online monitoring of dynamicprocesses.18−20 Furthermore, one can take advantage of twoother properties of the 129Xe nucleus: First, the largepolarizability of the xenon electron cloud by its physical andchemical environment leads to a large chemical shift range of129Xe NMR. This sensitivity allows for reporting of sampleproperties by the chemical shift of the Xe atom without theneed of covalent bonding between the Xe atom and the samplemolecules. Many examples of gaseous 129Xe NMR exploitingthis effect can be found in the material science of porousmedia.21−23 Another useful property of Xe is its solubility inliquid phases, both organic and aqueous, which allows for theuse of 129Xe in liquid reaction mixtures.The polymerization techniques of emulsion and miniemul-

sion polymerization reactions are widely used in industry andscience because of their good reaction heat dissipation, theconstant, low viscosity of the reaction mixture, and highachievable conversion rates. The droplets of a miniemulsionserve as a kind of nanoreactor for the reaction which equals to abulk polymerization inside the droplets. In addition to thepreparation of “simple” polymeric nanoparticles by radicalpolymerization, miniemulsion polymerization allows using abroad range of monomers, conducting polyreactions as e.g.polycondensation, or polyaddition reactions,24−26 which are notaccessible with conventional emulsion polymerization. Itfurther offers the possibility to encapsulate a variety ofmaterials to generate functional nanoparticles.27,28 Thepresence of the aqueous phase in emulsion and miniemulsionpolymerization reactions can pose severe difficulties for 1HNMR measurements due to the very prominent water signalwhich can superimpose other signals stemming from thereaction educts or products. By using 129Xe NMR, the presenceof a large proton background signal does not interfere with themeasurements.In this contribution, the real-time monitoring of a free radical

polymerization of styrene in a miniemulsion is demonstrated,using a continuous flow of hyperpolarized 129Xe through thereaction mixture. A short description of the monitoring of abatch polymerization of methyl methacrylate at 35 G was givenby Gloeggler et al.29 However, to our knowledge, the currentpaper gives the first full length report demonstrating thathyperpolarized 129Xe can be used for online monitoring of apolymerization reaction. By combining all properties ofhyperpolarized 129Xe as described above, it is used here as anNMR probe in a polymerization reaction of an industriallyimportant monomer in the complex environment of aminiemulsion. The absence of xenon from common chemicalsubstances leads to the absence of any undesired backgroundsignal rendering the interpretation of the sparse Xe spectra veryconvenient. In this work, only two distinct peaks of xenoninside the reaction mixture are recorded. The resulting simpleNMR spectra can be analyzed in a very straightforward mannerby the determination of the chemical shift of the 129Xe peaksgiving access to the conversion of the polymerization.

Therefore, the kinetics of two polymerization reactions wererecorded by hyperpolarized 129Xe NMR spectroscopy using twooil-soluble azo-initiators with different decomposition temper-atures. In order to prove the validity of the method, theprogress of the polymerization reaction observed by 129XeNMR was compared to results obtained by calorimetry.Calorimetric measurements on similar emulsion and mini-emulsion systems can also be found in the literature.30−33 Toallow for better comparison and to exclude any time-dependentchanges of the miniemulsion, small samples of the same batchof miniemulsion were measured simultaneously by 129Xe NMRspectroscopy and calorimetry. Thereafter, the conversion of thedifferent samples and reactions were determined by measuringthe solids content. To check for differences between thepolymer colloids obtained from the individual reactionsmonitored by NMR and calorimetry, the mean particlediameters were measured by photon cross-correlation spec-troscopy (PCCS) and scanning electron micrographs.

■ MATERIALS AND METHODSFor the miniemulsions, styrene, sodium dodecyl sulfate (SDS),hexadecane (all purchased from Sigma-Aldrich, Germany), theinitiators 2,2′-azobis(2-methylbutyronitrile) (V59, 10 h half-lifedecomposition temperature 340 K), and 2,2′-azobis(4-methoxy-2,4-dimethylvaleronitrile) (V70, 10 h half-life decomposition temperature303 K, both initiators purchased from Wako, Japan) were used. Theminiemulsions were prepared by mixing 24 g of ultrapure H2O (Milli-Q-grade), 6 g of distilled styrene (57.6 mmol), 60 mg of SDS assurfactant (0.21 mmol), 250 mg of hexadecane as hydrophobic agent(1.1 mmol), and either 100 mg of V59 (0.52 mmol) or 100 mg of V70(0.32 mmol) as initiator and the treatment with ultrasound (1/2 in.tip, 90% amplitude, Branson Sonifier 450D) for 2 min, applied in 10 spulses with 10 s break. During and after the emulsification, theminiemulsions were ice-cooled to reduce any possible polymerizationprior to the measurements.

For the hyperpolarization of 129Xe, xenon (purity 4.7, naturalabundance, Westfalen AG, Germany), nitrogen (purity 5.0, WestfalenAG, Germany), helium (purity 4.6, Westfalen AG, Germany), andrubidium (purity 99.8%, Alfa Aesar GmbH & Co KG) were used. TheXe was polarized in a polarizer built by the Research Center Julich in agas mixture consisting of 1% Xe, 8% N2, and 91% He. The polarizerwas used in a continuous mode during the whole reaction time with agas flow of around 70 mL/min at a pressure of 7 bar. The gas mixturecontaining the polarized 129Xe was directly used for the NMRmeasurements without pressure reduction or replenishment of thediluted Xe gas.

For dissolving the hyperpolarized 129Xe in the reaction mixture,hollow-fiber membranes (hydrophobic surface, made from poly-propylene, pore sizes ∼30 nm, Membrana GmbH, Germany) wereused.34−36 The experimental setup for the 129Xe NMR measurementsis sketched in Figure 1. The membranes allow for the moleculardissolution of Xe gas into the bulk of the sample without the loss ofhyperpolarization. The use of the membrane avoids the formation ofgas bubbles or foam and thus minimizes susceptibility artifacts. Afteran equilibration concentration of Xe in the sample is reached, Xe gasalso diffuses back through the membrane, thus providing a continuousexchange of hyperpolarized and due to NMR pulses and T1 relaxationdepolarized 129Xe atoms in the sample. The membranes were gluedinto homemade tube caps made from poly(ether ether ketone)(PEEK) providing in- and outlets for the gas mixture as well as a tightseal applicable for pressures of up to 8 bar and temperatures of up to atleast 350 K. In order to minimize any diffusional losses of organicmonomer across the hydrophobic membrane into the gas flow,remotely controlled nonmagnetic pneumatic valves were used near thesample tube inside the NMR magnet to create a bypass around thesample tube in between subsequent measurements.

Macromolecules Article

dx.doi.org/10.1021/ma202605n | Macromolecules 2012, 45, 1839−18461840

All NMR experiments were carried out in a 7 T NMR magnetequipped with a Tecmag console. Pressure-resistant NMR tubes (o.d.10 mm, i.d. 5 mm) were used for all experiments.For the time-resolved polymerization monitoring, 1.5 mL of the

miniemulsion was put into the sample tube together with themembranes (see Figure 1). After an initial pressurizing step, the gasflow containing hyperpolarized 129Xe was fed through the membranesinto the still cool sample tube for 2 min, allowing for the initialdissolution of Xe gas into the sample. Subsequently, the sample tubewas inserted into the NMR magnet heated to 343 K. The insertion ofthe sample tube was considered as the time t = 0 for thepolymerization reaction. Starting after 1 min of reaction, 129Xe NMRspectra were recorded. For each spectrum of the time-resolvedmeasurements, four scans with a repetition delay of 10 s were recordedto allow for a sufficient signal-to-noise ratio. For the first 10 min ofreaction time, spectra were recorded every minute, for the next 30 minevery 3 min, then until the first hour of reaction time every 5 min, andfinally for the second hour of reaction time every 10 min.All 129Xe NMR spectra were analyzed by using Matlab for fitting the

129Xe NMR spectra and determining the chemical shifts as the centerof gravity of the adsorbed or dissolved Xe peaks. The peak of the freeXe gas was used as a reference and set to 0 ppm. In the case of twodistinct peaks for the dissolved 129Xe (beginning of the reaction), bothpeaks have been fitted and the mean center of gravity of both peakswas calculated and used for the further evaluation.For the mole-fraction-dependent experiments, polystyrene (mole

weight 100 000 g mol−1, polymer standard, PSS, Germany) andstyrene (Sigma-Aldrich, Germany) were used. Nine mixtures ofstyrene and polystyrene were made with styrene mole fractionsranging from 0.95 to 0.3. Thermal 129Xe NMR spectra of pure styrene,pure polystyrene, and the nine different monomer−polymer mixtureswere recorded by averaging the thermal Xe NMR signal for severalhundred scans after evacuating the sample tube and pressurizing itwith up to 3 bar of Xe gas.The calorimetric measurements were carried out on a μRC-micro

reaction heat calorimeter (Thermal Hazard Technology, UK) at 343K. The sample volume was 1 mL. The solids content was determinedby evaporation of the liquid phase from the dispersion at 323 K undera pressure of 50 mbar and weighing the solid residue. Particle sizeswere measured by photon cross-correlation spectroscopy (PCCS)using a Nanophox PCCS (Sympatec GmbH, Germany). For the

measurement, 35 μL of the dispersion was diluted with 1.5 mL ofMilli-Q water. Scanning electron micrographs were recorded on aGemini 1530 microscope (Carl Zeiss AG, Germany). From theelectron micrographs, mean particle diameters were determined from150 polymer particles for each polymerization. Size exclusionchromatography (SEC) was performed at 303 K with a WatersAlliance 2000 autosampler and a Waters 510 HPLC pump using PSS-SDV columns with pore sizes of 500, 1 × 104, and 1 × 105 Å,respectively. The signals were detected on an Erma UV detector (λ =254 nm) and a SOMA RI detector. The eluent and the solvent for thesamples was THF, the elugrams were calibrated against PS standards.

■ RESULTS AND DISCUSSIONFigure 2 shows representative NMR spectra of 129Xe atomsdissolved in the miniemulsion after the cold sample was

introduced in the spectrometer and heated to the temperatureof the polymerization. Three distinct peaks can be observedcorresponding to Xe in different physical and chemicalenvironments. The peak of free Xe gas inside the hollow-fibermembranes has the lowest chemical shift and can be used as aninternal reference (0 ppm). The Xe atoms dissolved in thedifferent phases (aqueous and organic) of the miniemulsiongive rise to the peaks at 186 and 192 ppm, respectively.In the case of a relatively slow exchange of the xenon atoms

between the continuous aqueous phase and the discontinuousorganic phase, as it is the case for room temperature and below,two distinct peaks can be observed in the NMR spectra (lowerspectra in Figure 2). Heating the sample to 343 K (temperatureof the polymerization) results in higher exchange rates and thetwo distinct peaks start to merge until only one exchange peak

Figure 1. Sketch of the NMR sample tube containing theminiemulsion and the hollow fiber membranes inside the NMR coil.The hydrophobic membrane allows for the passage of hyperpolarized129Xe in and out the reaction mixture. The inlet and outlet of themembrane are connected to the gas flow of the polarizer.

Figure 2. NMR spectra of hyperpolarized 129Xe dissolved in theminiemulsion. According to the number of phases, up to three Xepeaks are observable. The right peak in the spectra corresponds to thefree Xe gas located in the hollow fiber membranes, whereas the middleand the left peak correspond to Xe dissolved in the aqueous (bluebackground) and organic (yellow) environment, respectively. The coldsample (278 K) was heated during the first minutes in the NMRspectrometer to 343 K giving rise to a faster exchange of the Xe atomsbetween the organic and aqueous phase and the appearance of anexchange peak (green).

Macromolecules Article

dx.doi.org/10.1021/ma202605n | Macromolecules 2012, 45, 1839−18461841

is visible (upper spectra in Figure 2; this process is also visiblein the spectra of the first 15 min in Figure 4, see below).In order to easily monitor polymerization reactions by 129Xe

NMR spectroscopy, a (preferably linear) dependence of thechemical shift of the dissolved 129Xe on the conversion of thepolymerization should exist. In miniemulsion polymerizations,however, four contributions to the chemical shift of dissolvedXe must be considered:First, there are the interactions between 129Xe atoms and

H2O molecules of the continuous phase whose contribution tothe total chemical shift can be considered constant during thereaction because the amount of water in the system does notchange substantially. Second, interactions exist between thedissolved 129Xe atoms themselves, but the contribution is verysmall due to the very low Xe density in these experiments (only1% of Xe in the gas mixture) and can be neglected.Furthermore, there are two chemical shift contributions inthe organic phase of the miniemulsion stemming from theinteractions between the 129Xe atoms and the monomer andpolymer molecules. The first contribution is decreasing duringthe reaction as the amount of the monomer molecules isdepleting; the latter contribution is growing during the reactionas the polymerization degree increases.By taking into account all the described contributions, the

behavior of the chemical shift can be calculated by theequation37−39

δ = −σ

− σ

− σ

− σ

−

−

−

−

T T

T T

T T

T T

( )[mono]( )

( )[poly]( )

( )[H O]( )

( )[Xe]( )

O

dissolved Xe 0Xe mono

0Xe poly

0Xe H

2

1Xe Xe

2

(1)

δdissolved Xe is the resulting chemical shift of 129Xe dissolved in theminiemulsion. The constants σ0

Xe‑mono, σ0Xe‑poly, σ0

Xe‑H2O, andσ1Xe‑Xe are the shielding contributions due to interactions of Xe

with the monomer, the polymer, the water of the aqueousphase, and itself, respectively. The terms [mono], [poly],[H2O], and [Xe] depict the amount of monomer, polymer,water, and xenon in the miniemulsion, and T is thetemperature.By omitting the Xe−Xe interactions as described above and

assuming the Xe−H2O interactions and the temperatureconstant, the equation simplifies to

δ = −σ − σ

−

− −[mono] [poly]

const

dissolved Xe 0Xe mono

0Xe poly

(2)

With [poly] = 1 − [mono], the equation can be furthersimplified, giving the following linear dependence on the molefraction of the monomer [mono]:

δ = σ − σ

− σ −

− −

−( )[mono]

const

dissolved Xe 0Xe poly

0Xe mono

0Xe poly

(3)

In order to check for the linear dependence of the chemicalshift on the monomer concentration, the chemical shifts of129Xe in pure styrene and pure polystyrene (Mw = 100 000 gmol−1) as well as in nine mixtures of styrene and polystyrenewith increasing monomer to polymer ratios were determined.In Figure 3, the chemical shift of dissolved 129Xe is plottedversus the mole fraction of the polymer and the mole fraction

of the monomer in the mixture. A linear dependence of thechemical shift on the sample composition is clearly observed.The values were fit linearly, yielding the following relationshipbetween chemical shift and the mole fraction of the monomer[mono]:

δ = − · +27.6 ppm [mono] 224.8 ppmXe,mix (4)

In this contribution, the studied polymerization reactionstook place in a miniemulsion, a heterophase system, which isnot only consisting of the mixture of styrene and polystyrenebut also of water. Thus, the fit values based on eq 4 cannot beused to directly determine the conversion of the reactionmixture in the following dynamic experiments because of thecontribution of the aqueous phase to the chemicals shift inminiemulsions. However, it will be demonstrated in thefollowing that the underlying dependence of the chemicalshift shown above is just as well applicable to the monitoring ofcomplex systems like miniemulsion polymerization reactions.A series of 129Xe NMR spectra recorded during the

miniemulsion polymerization of styrene are presented in Figure4. The plot shows a zoomed region containing the peaks of thehyperpolarized 129Xe dissolved in the reaction mixture in therange of 186−205 ppm. For the sake of visibility, not allrecorded spectra from the start to the end of the reaction aredepicted. During the course of the polymerization, a clear shiftof the 129Xe peaks to higher chemical shift values is observed.Thus, the progress of the reaction is directly observable fromthe evolution of the 129Xe NMR signal. In the beginning, atemperature effect is visible as the xenon peaks move to smallerchemical shift values (compare first and second spectra inFigure 4). This shift is attributed to the heating of the samplefrom 278 to 343 K. In the first spectra the heating effectcounters and partially compensates the effect of the startingpolymerization reaction, which simultaneously causes anincrease in chemical shift. The resulting “equilibration time”due to heating is in fact comparable with the time which isneeded for the heating of the sample in the calorimeter used inthese studies (see below).Figure 4 shows also the already mentioned merging of the

two 129Xe peaks stemming from the aqueous and the organicphase of the miniemulsion to only one exchange peak due tochanges in the exchange rate in the course of the experiment

Figure 3. Plot of the chemical shift of 129Xe versus the mole fraction ofa monomer/polymer mixture. The points are the chemical shift valuesof pure polystyrene (Mw = 100 000 g mol−1), nine mixtures of styreneand polystyrene with decreasing polymer ratio, and pure styrene. Alinear fit of the data points is shown as a solid line.

Macromolecules Article

dx.doi.org/10.1021/ma202605n | Macromolecules 2012, 45, 1839−18461842

(compare spectra before and after 10 min in Figure 2). Theexchange rate increases as the temperature inside the sampleincreases. However, substantial broadening and distortion ofthe dissolved 129Xe peak occurs between 30 and 60 minreaction time. The line broadening can be attributed to anincrease of viscosity inside the monomer/polymer dropletswhich leads to a decreased mobility of the Xe atoms and thepolymer chains. The decreased mobility cause a rise of dipolarinteractions between Xe atoms and protons resulting in a netdipolar broadening of the 129Xe NMR lines.40,41

In this study, time-resolved spectra for two differentpolymerization reactions for a reaction time of 120 min wereobtained by using two different azo-initiators, namely V59 andV70 (full names and 10 h half-life decomposition temperaturesare denoted in the Materials and Methods section). At thereaction temperature of 343 K, polymerization reactions startedby the two initiators follow the expected miniemulsionmechanism described by Bechthold et al.30 (see SupportingInformation). However, the two initiators used here decomposeat very different rates at 343 K. The very fast decomposition ofV70 compared to V59 leads to a shorter nucleation phase and afaster on/off mechanism for the polymerization reactioninitiated by V70, resulting in a higher reaction rate at thebeginning of the polymerization. The faster on/off mechanismleads to shorter polymer chains compared to the reactioninitiated by V59 (for SEC results, see Supporting Information).The different molecular weights of the polymer suspensions

allow us to the check for a dependence of the chemical shift of129Xe on the molecular weight of the polymers. A more detailedmechanistic insight into the two reactions from calorimetricdata and the molecular weight distributions can be found in theSupporting Information.For both polymerization reactions, the chemical shift of the

dissolved 129Xe was determined for each spectrum of the timeseries and plotted in Figure 5. The open and filled circles in

Figure 5 show the development of the chemical shift of thedissolved 129Xe for the reaction initiated by V70 and V59,respectively. As expected due to the shorter nucleation phase,the chemical shift of the dissolved Xe in the reaction with theinitiator V70 shows a faster increase than the one obtained inthe reaction employing the initiator V59. For longer reactiontimes, the chemicals shift values of both reactions become thesame due to the complete conversion of both polymerizations.Accordingly, the difference in the molecular weights of theproducts of the two polymerizations does not influence thechemical shift of 129Xe in our experiments.For comparison, Figure 5 also shows the two extremes for

the chemical shift of the dissolved 129Xe in theses polymer-izations (triangles). The lower limit at 185.5 ppm represents129Xe dissolved in a miniemulsion at a temperature of 343 Kcontaining only monomer, no polymer. It has been determinedusing a miniemulsion consisting of all reaction componentsexcept the initiator. Because of the lack of the initiator, noreaction could take place in the sample. The curve of thechemical shift of the reaction mixture nearly reaches the lowerboundary (counteraction between thermal effect and shift dueto the ongoing reaction).The upper limit at 207.0 ppm is given by the chemical shift

of dissolved Xe in a completely polymerized miniemulsion at atemperature of 343 K. It was measured by using a colloidalpolymer suspension of the same particle size as in theminiemulsions. During the monitored polymerization reactions,the chemical shift values do not reach this boundary but stay∼3 ppm lower. This chemical shift difference can be attributedto the manufacturing of the colloid particles. The particles havebeen subjected to a dialysis process which removes any residualmonomer, surfactant and charged, soluble oligomers from the

Figure 4. Time series of 129Xe NMR spectra recorded during aminiemulsion polymerization reaction at 343 K (initiated with V70).The plot shows the chemical shift range of the dissolved Xe. Theinsertion of the sample tube into the heated magnet was used as thestarting time t = 0 min. The spectra depict a strong dependence of theXe chemical shift on the progress of the polymerization.

Figure 5. Plot of the chemical shift of 129Xe versus reaction time withV70 as initiator (open circles), resulting in a faster increase in thechemical shift, and V59 as initiator (filled circles), resulting in a slowerincrease. The triangles correspond to the Xe chemical shift in a fullypolymerized colloid of the same particle size (207.0 ppm) and aminiemulsion containing only monomer droplets (185.5 ppm) at 343K. The lines are for guiding the eyes.

Macromolecules Article

dx.doi.org/10.1021/ma202605n | Macromolecules 2012, 45, 1839−18461843

system. Thus, the colloid particles obtained in this mannerdiffer from the in situ particles in the reaction mixture, givingrise to the observed difference in chemical shift.To check the validity of the 129Xe NMR monitoring method

presented here, calorimetric measurements of the polymer-ization reactions were also carried out for polymerizationreactions with the two initiators. In order to allow for a goodcomparison between the results obtained from calorimetry and129Xe NMR spectroscopy, two samples from the sameminiemulsion batch were simultaneously measured by thetwo methods for each polymerization reaction. Thus, anychange of the miniemulsion or any polymerization reactionprior to the start of the experiment can be ruled out for theinterpretation of the differences between 129Xe NMR andcalorimetry.The effect measured by a calorimeter is the heat flow in

dependence of time which directly correlates with the reactionkinetics of the observed reaction. Thus, reaction heatcalorimetry is a differential method. The chemical shift datameasured by the NMR experiments depend on the polymer-ization degree or reaction conversion. Therefore, NMR reactionmonitoring is an integral method. To compare the data fromthe two methods, the calorimetric data were numericallyintegrated.Figure 6 shows the integrated calorimetric data (red line)

and the corresponding 129Xe NMR data (black circles) for both

polymerization reactions. For the calorimetric data, the point ofzero conversion was chosen as the time from which the reactionheat becomes exothermic (the first point of the positive slope,at t = 500 s). Comparably, the lowest chemical shift valuesderived from the 129Xe NMR data for each of the twoexperimental series are assumed to be the points of zeroconversion (at t = 420 s for both 129Xe NMR experiments).After a reaction time of 7200 s, full conversion can be expectedfor both miniemulsion polymerization reactions (see calori-metric data in Supporting Information). Therefore, thecalorimetric data have been scaled to conversion = 100%after 7200 s. Accordingly, the highest chemical shift valuesderived from the 129Xe NMR data have been scaled to 100%conversion in order to reflect the anticipated full conversion.Both methods show a good qualitative agreement, demonstrat-ing the validity of our approach. For the polymerizationreaction initiated by V70 we even observe an excellentagreement for conversion rates up to 65%. However, smalldifferences occur for high conversion rates for both polymer-izations and the reactions seem to be slightly faster whenobserved by 129Xe NMR. In order to reveal the differences ofboth observation methods, a thorough characterization of thepolymerization products was performed.To this end, the solids content of the four miniemulsions as

well as the mean diameter of the particles were determined (seeTable 1). For both 129Xe NMR experiments, the solids contentis lower and the particle diameters are smaller than for thecorresponding calorimeter experiments. As there was no visiblecoagulum of the miniemulsion particles, this would suggest alower conversion in the polymerizations measured by 129XeNMR, thereby contradicting the previously mentionedobservation of a faster reaction observed by 129Xe NMR. Inorder to reconcile the observed differences, we consider a massloss of the monomer occurring during the 129Xe NMRexperiments. This mass loss is attributed to a diffusion of thehydrophobic monomer across the hydrophobic membrane intothe dry gas flow and was tried to be minimized by a bypass ofthe gas flow between subsequent measurements. However, theloss of monomer could not be totally avoided due to thenecessary Xe supply during the measurements. Such a loss ofmonomer during the polymerization reactions leads to both ahigher polymer−monomer ratio after shorter reaction times(higher chemical shift in 129Xe NMR) as well to smaller particlediameters.Despite the mass loss, the accordance between the

polymerizations observed by calorimetry and 129Xe NMR isquite good. Because of the large concentration of monomer inthe droplets at the start of the polymerization, the significanceof the monomer loss is small for short reaction times. Forlonger reactions times, the significance of monomer lossincreases which can lead to the larger difference in the observedconversions measured by calorimetry and 129Xe NMR (seereaction initiated by V59 in Figure 6a). However, the mass lossof the hydrophobic monomer can be suppressed in futureexperiments by optimization of the experimental setup, forexample, by the use of another type of hollow fiber membrane.Besides the described experimental imperfections in this work,the accuracy of the 129Xe NMR method is limited only by thesignal-to-noise ratio (SNR) which influences the accuracy ofthe determination of the chemical shift of 129Xe and theachievable time resolution. A higher SNRand thus a betteraccuracy of the presented methodcould be achieved by using

Figure 6. Comparison between the (integrated) calorimetry data (reddotted line) and the 129Xe NMR data (solid line with filled circlesshowing the measured chemical shift) for the polymerization initiatedby V59 (a) and V70 (b). The left y-axis shows the conversion, and theright y-axis shows the corresponding 129Xe chemical shift values.

Macromolecules Article

dx.doi.org/10.1021/ma202605n | Macromolecules 2012, 45, 1839−18461844

a polarizer setup giving a higher 129Xe polarization rate10 or byusing isotopically enriched Xe gas.The solids content in the calorimeter experiments is only

0.4% and 1.2% lower than the theoretical value for a completereaction which highlights the very high conversion achievableby miniemulsion polymerizations.24,31 According to thecalorimetric data (see Supporting Information), the conversionreached 100% for the calorimeter experiments after a reactiontime of 7200 s, which corresponds well with the very smalldeviation from the theoretical solids content.To obtain an insight into the particle size distributions for

the two methods, SEM images of the particles from eachpolymerization reaction were acquired. Normally, the dropletsize and the size distribution in a miniemulsion polymerizationof a given amount of monomer are determined by the amountof surfactant in the reaction mixture and fixed in theemulsification step. Because of the hydrophobic nature of themembrane surfaces present during the 129Xe NMR experiments,possible changes in particle size distribution were considereddue adsorption and desorption processes of the droplets to andfrom the membrane surface. Figure 7 shows two SEM images

for the fast polymerization from the reaction inside the NMRspectrometer and in the calorimeter. No major differences inthe particle size distribution can be found for all experiments.Besides the mentioned loss of monomer in the 129Xe NMRexperiments, the products and the course of the polymerizationreactions in both methods seem to be very well comparable.

■ CONCLUSION

In this work, the chemical shift dependence of 129Xe on theconversion of polymerization reactions has been demonstrated.Using this dependence and the large NMR signal ofhyperpolarized 129Xe, the online monitoring of a chemicalreaction by 129Xe NMR was successfully accomplished. Kineticdata for free-radical polymerization reactions of an industriallyimportant monomer with two different reaction rates wereobtained with good time resolution. In order to approve thevalidity of our method, the results of the 129Xe NMRmeasurements were compared to calorimetric measurements.Both methods showed a good qualitative agreement except forminor differences at higher conversion rates that could be

explained by a monomer loss during the 129Xe NMRexperiments which can most likely be circumvented byoptimization of the experimental setup. The comparability ofboth methods was proven by a thorough characterization of thepolymer products. Hyperpolarized 129Xe NMR can thusprovide an excellent method to investigate polymerizationsexhibiting very complicated 1H and 13C NMR spectra due tothe simplicity of the 129Xe NMR spectra and the simplerelationship of the 129Xe chemical shift on the composition ofthe reaction mixture. Furthermore, the application of hyper-polarized 129Xe NMR spectroscopy for the online monitoringof copolymerization reactions could allow to follow theconversion of each monomer independently which cannot bedone by calorimetry. However, this would be only possible infavorable cases where the two monomers exhibit very differentchemical shift values of the dissolved 129Xe and extensivecalibration measurements for the chemical shift of 129Xedissolved in mixtures of the used comonomers would benecessary. 129Xe NMR might also be well applicable for theonline monitoring of other chemical reactions apart frompolymerizations. It might, for example, allow for time-resolvedmeasurements of thermoneutral reactions which cannot beassessed by calorimetry. The presented method would veryprobably be applicable to larger reactors by implementing apump-around loop which could bring a sample volume inside a(high-field) NMR magnet. In this case already existing “flow-through membrane modules” could be used for dissolution ofthe hyperpolarized 129Xe prior to the NMR measurement.36

Given the indepence of the signal amplitude of hyperpolarizedsamples from the magnetic field strength, the use of a low-fieldNMR spectrometer for routine reactor monitoring seemsfeasible under the condition of sufficiently large chemical shiftdispersion.

■ ASSOCIATED CONTENT

*S Supporting InformationData from calorimetry, the calculated average number ofradicals, and the molecular weight distributions of the polymersuspensions obtained in the calorimeter experiments. Thismaterial is available free of charge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATION

Corresponding Author*E-mail: muenne@mpip-mainz.mpg.de.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

We thank S. Appelt and W. Hasing (FZ Julich) for thedevelopment of the xenon polarizer. N. Vogel acknowledgesfunding from the Materials Science in Mainz (MAINZ)graduate school.

Table 1. Characterization of the Polymer Dispersions after Polymerization

particle diameter [nm]

solids content [wt %] by PCCS from micrographs

calorimetry 129Xe NMR theory calorimetry 129Xe NMR calorimetry 129Xe NMR

V70 20.4 16.3 20.8 94.4 ± 12.0 87.7 ± 12.0 71.6 ± 13.4 66.6 ± 14.6V59 19.6 16.5 20.8 105.6 ± 12.0 93.9 ± 12.0 88.8 ± 11.0 76.9 ± 11.0

Figure 7. Scanning electron micrographs of the polymer suspensionsafter the polymerization. Left side: V70 129Xe NMR; right side: V70calorimetry.

Macromolecules Article

dx.doi.org/10.1021/ma202605n | Macromolecules 2012, 45, 1839−18461845

■ REFERENCES(1) Matyjaszewski, K.; Gnanou, Y.; Leibler, L. MacromolecularEngineering: Precise Synthesis, Materials Properties, Applications; Wiley-VCH Verlag GmbH & Co. KGaA: New York, 2007.(2) Guthausen, G.; Vargas, M. A.; Cudaj, M.; Hailu, K.;Sachsenheimer, K. Macromolecules 2010, 43 (13), 5561−5568.(3) Fonseca, G. E.; Dube, M. A.; Penlidis, A. Macromol. React. Eng.2009, 3 (7), 327−373.(4) Frauendorfer, E.; Wolf, A.; Hergeth, W. D. Chem. Eng. Technol.2010, 33 (11), 1767−1778.(5) Haddleton, D. M.; Perrier, S.; Bon, S. A. F. Macromolecules 2000,33 (22), 8246−8251.(6) Gramm, S.; Komber, H.; Schmaljohann, D. J. Polym. Sci., Part A:Polym. Chem. 2005, 43 (1), 142−148.(7) Abdollahi, M.; Sharifpour, M. Polymer 2007, 48 (1), 25−30.(8) Landfester, K.; Spiegel, S.; Born, R.; Spiess, H. W. Colloid Polym.Sci. 1998, 276 (4), 356−361.(9) Ruth, U.; Hof, T.; Schmidt, J.; Fick, D.; Jansch, H. J. Appl. Phys. B1999, 68 (1), 93−97.(10) Ruset, I. C.; Ketel, S.; Hersman, F. W. Phys. Rev. Lett. 2006, 96(5), xxxx.(11) Ardenkjaer-Larsen, J. H.; Fridlund, B.; Gram, A.; Hansson, G.;Hansson, L.; Lerche, M. H.; Servin, R.; Thaning, M.; Golman, K. Proc.Natl. Acad. Sci. U. S. A. 2003, 100 (18), 10158−10163.(12) Carver, T. R.; Slichter, C. P. Phys. Rev. 1956, 102 (4), 975−980.(13) Bowers, C. R.; Weitekamp, D. P. J. Am. Chem. Soc. 1987, 109(18), 5541−5542.(14) Munnemann, K.; Spiess, H. W. Nature Phys. 2011, 7 (7), 522−523.(15) Walker, T. G.; Happer, W. Rev. Mod. Phys. 1997, 69 (2), 629−642.(16) Appelt, S.; Baranga, A. B.; Erickson, C. J.; Romalis, M. V.;Young, A. R.; Happer, W. Phys. Rev. A 1998, 58 (2), 1412−1439.(17) Shah, N. J.; Unlu, T.; Wegener, H. P.; Halling, H.; Zilles, K.;Appelt, S. NMR Biomed. 2000, 13 (4), 214−219.(18) Kilian, W.; Seifert, F.; Rinneberg, H.Magn. Reson. Med. 2004, 51(4), 843−847.(19) Han, S.; Kuhn, H.; Hasing, F. W.; Munnemann, K.; Blumich, B.;Appelt, S. J. Magn. Reson. 2004, 167 (2), 298−305.(20) Schroder, L.; Meldrum, T.; Smith, M.; Schilling, F.; Denger, P.;Zapf, S.; Wemmer, D.; Pines, A. World Congr. Med. Phys. Biomed. Eng.2009, 25 (13), 176−179.(21) Sozzani, P.; Comotti, A.; Simonutti, R.; Meersmann, T.; Logan,J. W.; Pines, A. Angew. Chem., Int. Ed. 2000, 39 (15), 2695−2698.(22) Moudrakovski, I.; Soldatov, D. V.; Ripmeester, J. A.; Sears, D.N.; Jameson, C. J. Proc. Natl. Acad. Sci. U. S. A. 2004, 101 (52),17924−17929.(23) Emmler, T.; Heinrich, K.; Fritsch, D.; Budd, P. M.; Chaukura,N.; Ehlers, D.; Ratzke, K.; Faupel, F. Macromolecules 2010, 43 (14),6075−6084.(24) Schork, F. J.; Luo, Y. W.; Smulders, W.; Russum, J. P.; Butte, A.;Fontenot, K. Polym. Part. 2005, 175, 129−255.(25) Landfester, K. Annu. Rev. Mater. Res. 2006, 36, 231−279.(26) Landfester, K.; Crespy, D. Beilstein J. Org. Chem. 2010, 6, 1132−1148.(27) Landfester, K.; Weiss, C. In Modern Techniques for Nano- andMicroreactors/-reactions; Caruso, F., Ed.; Springer: Berlin, 2010; Vol.229, pp 1−49.(28) Weiss, C.; Landfester, K. In Hybrid Latex Particles; van Herk, A.M., Landfester, K., Eds.; Springer: Berlin, 2011; Vol. 233, pp 185−236.(29) Gloggler, S.; Blumich, B.; Appelt, S. Magn. Reson. Porous Media2011, 1330, 101−104.(30) Bechthold, N.; Landfester, K. Macromolecules 2000, 33 (13),4682−4689.(31) Li, D. H.; Sudol, E. D.; El-Aasser, M. S. J. Appl. Polym. Sci. 2006,101 (4), 2304−2312.(32) Blythe, P. J.; Klein, A.; Phillips, J. A.; Sudol, E. D.; El-Aasser, M.S. J. Polym. Sci., Part A: Polym. Chem. 1999, 37 (23), 4449−4457.

(33) Taratula, O.; Dmochowski, I. J. Curr. Opin. Chem. Biol. 2010, 14(1), 97−104.(34) Baumer, D.; Brunner, E.; Blumler, P.; Zanker, P. P.; Spiess, H.W. Angew. Chem., Int. Ed. 2006, 45 (43), 7282−7284.(35) Cleveland, Z. I.; Moller, H. E.; Hedlund, L. W.; Driehuys, B. J.Phys. Chem. B 2009, 113 (37), 12489−12499.(36) Amor, N.; Zanker, P. P.; Blumler, P.; Meise, F. M.; Schreiber, L.M.; Scholz, A.; Schmiedeskamp, J.; Spiess, H. W.; Munnemann, K. J.Magn. Reson. 2009, 201 (1), 93−99.(37) Buckingham, A. D.; Pople, J. A. Discuss. Faraday Soc. 1956, 22,17−21.(38) Jameson, C. J.; Jameson, A. K.; Cohen, S. M. J. Chem. Phys.1973, 59 (8), 4540−4546.(39) Jameson, C. J.; Dedios, A. C. J. Chem. Phys. 1992, 97 (1), 417−434.(40) Kentgens, A. P. M.; Vanboxtel, H. A.; Verweel, R. J.; Veeman,W. S. Macromolecules 1991, 24 (12), 3712−3714.(41) Saunders, M.; JimenezVazquez, H. A.; Khong, A. J. Phys. Chem.1996, 100 (39), 15968−15971.

Macromolecules Article

dx.doi.org/10.1021/ma202605n | Macromolecules 2012, 45, 1839−18461846