spatially encoded diffusion-ordered nmr spectroscopy of

HAL Id: hal-02392883https://hal.archives-ouvertes.fr/hal-02392883

Submitted on 19 Jan 2021

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Spatially encoded diffusion-ordered NMR spectroscopyof reaction mixtures in organic solvents

Ghanem Hamdoun, Ludmilla Guduff, Carine van Heijenoort, ChristopheBour, Vincent Gandon, Jean-Nicolas Dumez

To cite this version:Ghanem Hamdoun, Ludmilla Guduff, Carine van Heijenoort, Christophe Bour, Vincent Gandon, etal.. Spatially encoded diffusion-ordered NMR spectroscopy of reaction mixtures in organic solvents.Analyst, Royal Society of Chemistry, 2018, 143 (14), pp.3458-3464. �10.1039/C8AN00434J�. �hal-02392883�

https://hal.archives-ouvertes.fr/hal-02392883

https://hal.archives-ouvertes.fr

JournalName

ARTICLE

Thisjournalis©TheRoyalSocietyofChemistry20xx J.Name.,2013,00,1-3|1

Pleasedonotadjustmargins


Received00thJanuary20xx,Accepted00thJanuary20xx

DOI:10.1039/x0xx00000x

www.rsc.org/

Spatiallyencodeddiffusion-orderedNMRspectroscopyofreactionmixturesinorganicsolventsGhanemHamdoun,aLudmillaGuduff,aCarinevanHeijenoort,aChristopheBour,bVincentGandon,bJean-NicolasDumeza

Diffusion-ordered NMR spectroscopy (DOSY) is a powerfulmethod for the analysis ofmixtures. Classic DOSYmethodsrequireseveralminutesofacquisition,andweshowherethatDOSYexperimentscanberecordedinlessthanonesecondforthechallengingcaseofsolutionmixturesinlow-viscositysolvents.Theproposedmethodreliesonaspatialencodingofthediffusiondimension,forwhichconvection-compensationandspectral-selectionstrategiesareintroduced.Themethodis illustratedwith theanalysisofa reactionmixture, forwhichmoreaccurateestimatesof thediffusioncoefficentsareobtained.

IntroductionThe analysis of chemical mixtures is a challenging task, forwhich nuclear magnetic resonance spectroscopy has manyadvantages. Diffusion-orderedNMR spectroscopy (DOSY) is apowerfulanalyticalmethodthatmakesitpossibletoseparatethe spectra of compounds from amixture according to theirdiffusion coefficient.1-3 DOSY NMR has been widely used fortheanalysisofcomplexmixturesandprovidesinformationonmolecular sizes4-7 and shapes.4 It can also be used toinvestigatemolecular interactions,8 aswell as to characterizereactiveintermediates.9-12

In DOSY, the diffusion information is obtained with apseudo-2D experiments in which field-gradient pulses ofincreasing area (PFG)13 are used to encode the effect of therandom Brownianmotion of particles as a signal decay. Themost common approach to analyseDOSYdata consists of anexponentialfitofpeakintensitiesorintegralsIasafunctionofthe diffusion-encoding gradient area (gδ), according to theStejskal-Tannerequation,14

𝐼 = 𝐼!𝑒!!!!!!!!Δ!

(1)

where𝐼! is thereference intensity in theabsenceofdiffusionweighting and 𝐼 , in the presence of a gradient of strength ganddurationδ.Disthediffusioncoefficient,Δ‘istheeffectivediffusiondelay,andγisthegyromagneticratioofthespin.This

analysis is only strictly valid if the concentrations of thecomponents of the mixture remain constant during theexperiment. For a mixture with a time-dependentcomposition, changes in concentrations during the DOSYexperimentswillresultinover-(orunder-)estimateddiffusioncoefficients for compounds that are consumed (or produced)over time. As a result, the standard experiment is notapplicable as such to reactionmixtures, as it yields apparentdiffusioncoefficientsthatdependontherateofthechemicaltransformation. Several approaches have been introduced to account forchanges in concentration in theprocessingofDOSYdata. Forexample, Nilsson et al15 proposed the use of a multivariateanalysis,wheretimeisusedasathirddimensioninamultiwaydecomposition, to simultaneously characterise both theconcentrationevolutionandthediffusioncoefficientduringachemical reaction. In the permutated-DOSY approach, arandompermutationof the gradient intensitiesmitigates theinfluenceofconcentrationchangesontheestimateddiffusioncoefficients.16 Alternatively,severalstrategiesareavailabletoaccelerateDOSYexperiments,which canhelp reactionmonitoring if theduration of the experiment can be made shorter than thecharacteristic time of the reaction. In the Oneshot pulsesequence,17 coherence-selection and diffusion-encodinggradientsaremergedtoreducethenumberofrequiredscansto one per increment. Schemes are also available for a jointreduced sampling of the indirect time and gradientdimensions.18, 19 A significant acceleration of DOSYexperimentsisobtainedwithaspatialparallelisation(SPEN)ofthediffusiondimension,introducedbyKeelerandco-workers,whichreducesthedurationoftheexperimenttolessthanonesecond.20-22 As shown by Frydman and co-workers, a single-scan 2D acquisition that separates the spectral and spatialdimension then helps to analysemore complex samples.We

ARTICLE JournalName

2 |J.Name.,2012,00,1-3 Thisjournalis©TheRoyalSocietyofChemistry20xx



recently showed how the SPEN DOSY concept can also begeneralisedtohigher-dimensionalexperiments.23

TheSPENDOSYexperimentisinprincipleagoodcandidateto analyse mixtures with time-dependent compositions. Theexperiment in itscurrent form,however,suffers fromvarioussources of errors that limit its applicability. Specifically, itssensitivity to sample convection makes it unsuitable for theanalysis of complex mixtures in important organic solventssuch as dichloromethane or acetonitrile, and signal overlapbecauseofspectralfoldingisalsoalimitation. Inthisarticle,weshowthatspatiallyencodedDOSYcanbeusedforthefastanalysisofmixturesinlow-viscositysolvents,using convection-compensation and spectral-selectionschemes. The resulting experiment is used to monitor anorganic reactionmixture,andSPENDOSY is found toprovideestimatesofthediffusioncoefficientsthatarelessaffectedbychangesofconcentrationsovertime.Inparticular,anaccuratediffusion coefficient is obtained for a transient intermediate.The reduced error in the diffusionmeasurement is found totranslate into more accurate estimates of molecular weightsforthecomponentsofthereactingmixture.

Methods

Samplepreparation

Two samples were prepared for DOSY experiments on non-reactingmixtures.Forthecomparisonofstimulated-echo(STE)SPEN DOSY and double-stimulated-echo (DSTE) SPEN DOSY,the mixture (M1) was composed of cyclododecane (1 mg),squalene(15µL)andadamantane(2mg)solubilizedin600µLof CD2Cl2.For the comparison of DSTE SPEN DOSY with andwithoutspectralselection,themixture(M2)wascomposedofcyclododecane (1mg), squalene (15µL), adamantane (2mg)and 1,3-butadiene (5µL) solubilized in 600µL of CD2Cl2. Foreachsampleastandard5mmtubewasused.

Reactionmonitoring

The reaction analysed here is a di-amination reaction (M3).Twodifferentinitialstoichiometrieswereconsidered:i/9.0010-4molofisobutyraldehyde(81µL)and9.2610-5molofp-phenylenediamine(10mg);ii/1.8510-4molofisobutyraldehyde(16µL)and9.2610-5molofp-phenylenediamine(10mg).Inbothcases,thereactionwascarriedoutdirectly ina5mmNMR tube. A solution of isobutyraldehyde in 600 µL ofacetonitrilewasprepared,andthesamplewasusedtosetthelockandshims.p-Phenylenediaminewasaddedwithasyringe,theNMR tubewas inverted several times, then inserted intotheNMRmagnet. Themeasurementswere started just afterinsertionofthesample.

NMRspectroscopy

AlltheexperimentswerecarriedoutonaBrukerspectrometeroperatingata1HLarmorfrequencyof600.13MHz,equippedwith a 5 mm triple-axis gradient probe. For all the

experiments, the temperaturewas set at a nominal value of283K.

ConventionalDOSY

Conventional DOSY data were recorded using a doublestimulated echo sequence with bipolar gradient pulses(dstebpgp3sBrukersequencewithadditionallockstabilisationgradients and spatial selective filter). The diffusion-encodinggradientvaluesconsistedofalinearramprangingfrom0.0065T/mto0.455T/m.Thedurationofthegradientpulseswas800µs.Thediffusiondelaywas120msfortheexperimentsonthereactingmixture(M3),and140msfortheexperimentsontheunreactingmixtures(M1,M2).Each2Ddatasetwasacquiredwith16384pointsinthedirectdimension,aspectralwidthof7000Hz,anacquisitiontimeof1.4s,arelaxationdelayof8sand8gradientstepswitha16-stepphasecycle,resultinginatotalexperimentdurationof22min.

SPENDOSYexperiment

FortheSPENexperiments, thediffusion-encodingparameterswerechosensothatthechirppulsewouldsweeparegionof10 mm. For the unreacting mixtures (M1, M2), the chirpbandwidthwas74000Hz, theencodinggradientwas0.1696T/m, the chirp duration was 1.5 ms, and the post-chirpgradient durationwas 1.6ms; the diffusion delaywas 50msfor SPEN STE and 84.5 ms for the DSTE experiment. Theacquisitionparameterswerea0.2881T/macquisitiongradientstrengthwith128gradientsloopsof256pointseachanda1.1µsdwelltime.Thisresultsinaspectralwidthof1800Hz.Forthereactingmixture(M3),thechirpbandwidthwas56000Hz,theencodinggradientwas0.1278T/m,thechirpdurationwas1.5ms,andthepost-chirpgradientdurationwas1.6ms;the diffusion delay was 150 ms. The acquisition parameterswere a 0.1921 T/m acquisition gradient strength with 128gradientsloopsof256pointseachanda1µsdwelltime.Thisresultsinaspectralwidthof2000Hz.Thegradientstrengthforcoherenceselectionaroundthefirstchirp pulse, during the first stimulated echo,were a = 0.064T/m,a+c=0.147T/mwithadurationof800µs.Thegradientstrength for coherence selection around the second chirppulsewereb=0.128T/m,b+c=0.211T/mwithadurationof800 µs. The gradient strength for the spoiler duringlongitudinalstoragewasf=0.3648T/mwithadurationof800µs.During the second stimulated echo, gradient strength forcoherenceselectionaroundthefirstchirppulsewerea=0.089T/m,a+c=0.160T/mwithadurationof800µs.Thegradientstrength for coherence selection around the second chirppulsewereb=0.153T/m,b+c=0.224T/mwithadurationof800 µs. The gradient strength for the spectral selectionscheme was 0.0585 T/m with a duration of 800 µs. Thegradient strength for the spoiler during longitudinal storageweref=0.2752T/mandg=0.3136T/mwithadurationof800µs.

Dataprocessing

JournalName ARTICLE




Data from conventional DOSY experiments were processedwith theDynamics Center software (Bruker, Billerica, U.S.A.).For all SPEN DOSY experiments, the data were processed in

MATLAB (TheMathworks, Natick, U.S.A.) using home-writtenroutines,adaptedinpartfromtheDOSYToolbox24.

Figure1a)SPENSTEDOSYb)SPENDSTEDOSYpulsesequences.Δisthetimethatelapsesbetweenthecentresofthechirppulses.TheSPENDSTEDOSYpulsesequencecontainsanoptionalspectralselectionblock,showninagreyrectangle.Theselectedcoherencetransferpathwayisshowninred/blueforthesequencewithout/withspectralselection.Gradientsaandbarecrusherssurroundingtherefocusingchirppulses;gradientcselectstheanti-echopathwayforthestimulatedecho;gradientsgandfareaspoilerduringlongitudinalstorage.(Gctp:gradientforcoherence-transfer-pathwayselectionalongxandyaxis)

Inshort,the2DSPENdatawasreadfromBrukerfiles,thensortedandreshaped inordertogeta2Dmatrix.Thedatawasmultipliedbyasinefunctionandzero-filledinthetimedimension. Only the odd echoes were retained, and 2DFourier transformation was performed. The spectra wereprocessed in magnitude mode. For SPEN 2D DOSY, peakmaximawereselectedmanuallyand thespatialprofile foreachpeakwasobtainedasarowofthe(z;ω)dataset.Thespatialprofileswere fittedwithaStejskal-Tannerequationfor all resonances. TheDOSY representationwasobtainedGaussian line shapes with maxima corresponding to thecalculatedDvalueandlinewidthssetbytheerrorofthefit.

Estimationofmolecularweights

ThediffusioncoefficientsDxmeasuredforsampleM3wereconvertedintoestimatesofmolecularweightsMWest,usingthe empirical relationships derived by Stalke6 and co-workers:

log 𝐷!,!"#$ = log 𝐷! − log 𝐷!"# + log (𝐷!"#,!"#)(2)and

𝑀𝑊!"# = 10!"# !!,!"#$ !!"# (!)

! (3)where Dref is the diffusion coefficient of an internalreference, chosen here to be the solvent (acetonitrile)25,log(Dref,fix)isatabulatedvalueofthediffusioncoefficientofthe reference, and log(k) andα are empirical parametersdeterminedbyStalkeandco-workersby fittinga large setof diffusion data to Eq. 3. Here the parameters for “DSE”(dissipatedspheresandellipsoid)shapeswereused.

ResultsanddiscussionSpatially-encodedDOSYinlow-viscositysolvents

InspatiallyencodedDOSY,thesequentialacquisitionofdiffusion-attenuated NMR spectra for varying gradientareasisreplacedbyaparallelacquisition,inwhichdifferentvirtual slices of the sample tube experience differentgradient areas. Following the work of Keeler and co-workers, as well as Frydman and co-workers,21, 22 werecently reported a stimulated-echo-based (STE) SPEN

DOSY experiment for the sub-second acquisition of DOSYdata,showninFig.1a.23

WhileSTESPENDOSYisappropriateforexperimentsinD2O, a severe limitation becomes readily apparent formeasurements carriedout in low-viscosity solvents. Figure2ashowstheSPENDOSYspectrumobtainedfromamixtureof three molecules that are frequently used as internalreferencesinanalysesofmolecularweightswithDOSY(M1:cyclododecane, squalene and adamantane) in deuterateddichloromethane CD2Cl2 at 283 K. The large values of theestimated diffusion coefficients (Table 1), as well as thequalitatively incorrect shape of the diffusion decay curves(Figure S1) reveals that the measurement is affected bysampleconvection,whichcanbesignificantina5mmNMRtubeforalow-viscositysolventsuchasdichloromethane.26InconventionalDOSYexperiments,theeffectofconvectionmay be mitigated by the use of two diffusion-encodingstepsratherthanone,withoppositecoherencepathways.27Forstimulatedechoes,thisreducesthesignalintensitybyafactor of 2 but providesmore robustmeasurements. Thisdouble-diffusion-encodingapproachwas recently found tobealsovalidforspin-echobasedspatiallyencodedDOSY28.Hereweshowthatconvection-compensationalsoworksforstimulated-echo-based sequences (shown in Fig. 1b).Indeedthedoublestimulated-echo(DSTE)SPENDOSYdata(Fig. 2b) yields diffusion coefficients that are in very goodagreementwiththoseobtainedwiththeconventionalDSTEDOSYexperiment(Table1).

Note that all the results for conventional DOSYexperiments are obtained here with a restricted samplelength of 10 mm to limit systematic errors that originatefrom the non-linearity of the zmagnetic-field gradient onourtriple-axisgradientprobe

A SPEN DOSY spectrum obtained with the DSTEapproach is shown in Figure2c, for a secondmixture thatadditionally include 1,3-butadiene. The corresponding 1D1HspectrumisshowninFigure3.Inthiscase,thereducedspectralwidth that canbeaccessedwith theMRI-inspiredacquisitionemployedinSPENDOSYresultsinsignalfoldingand overlap. As a result, inaccurate diffusion coefficientestimates are obtained from the single-exponential fit of

ARTICLE JournalName




overlapped peaks. This is illustrated with the signal forcyclododecaneat1.38ppm,whichyieldsanoverestimateddiffusioncoefficientbecauseoftheoverlapwiththefoldedsignal of 1,3-butadiene. Using a spectral-selection schemebased on a band-selective refocusing pulse flanked bycrusher gradients to suppress signals outside of theobserved spectral region (figure 1b and 2d), significantlybetterresultsareobtained inthediffusiondimension.Thecyclododecanesignalat1.38ppmnowyieldsanestimateddiffusion coefficient (Table 1) that is in good agreementwith the reference conventional DSTE DOSY on the same

sample.Forthismixture,agoodresolutioninthediffusiondomain is obtained, with the signals of the fourcomponentsclearlyseparatedaccordingtotheirmolecularweight distribution: Dsqualene (410 g.mol-1) > Dcyclododecane

(162 g.mol-1) > Dadamantane (138 g.mol-1) > D 1-3 butadiene (54g.mol-1). Note that the spectral-selection scheme onlyaddresses the cases of overlap due to spectral folding;peaks that already overlap in the 1H 1D spectra may beaddressed with a multi-exponential fit if sufficientsensitivityisavailable.

Figure2DOSYspectraofamixtureofcyclododecane,adamantaneandsqualene(a-b)andamixtureofcyclododecane,adamantane,squaleneand1,3-butadiene(c-d)inCD2Cl2.ThespectrawererecordedwithSPENSTE(a)SPENDSTE(b-c)andSPENDSTEwithspectrallyselectivefilter(d)DOSYpulsesequences.Thespectraldimensionofthe2DdataisshownaboveeachDOSYdisplay.OnlyresolvedpeakswereincludedintheDOSYprocessing.Alltheexperimentswereperformedona600MHzBrukerAvancespectrometerusinga5mmtriple-axisgradientprobe

Table 1 Diffusion coefficients obtained from a mixture of cyclododecane,adamantane and squalene (M1) and a mixture of cyclododecane,adamantane, squalene and 1,3-butadiene (M2) using several DOSY pulsesequences.

Diffusioncoefficient.1010m2s-1

M1 M2

compoundsδ

(ppm)SPENSTE

SPENDSTE

ConvDSTE

SPENDSTE

ss-SPENDSTE

cyclododecane 1.38 49.2 18.1 17.3 20.8 17.7squalene 1.64 41.2 10.1 9.8 9.5 9.4squalene 1.71 41.5 10.6 9.1 10.0 10.1

adamantane 1.81 51.8 20.5 19.5 19.9 19.6

Overall, the diffusion coefficients obtained with theSPEN DOSY display a systematic difference, compared tothe ones obtained with conventional DOSY pulsesequences,oflessthan5%here.Thismaybeaccountedfor

byresidualeffectsofsampleconvectionandgradientnon-linearity, aswell as imperfect coherence-transfer pathway(CTP) selection. Gradient-based CTP selection was usedhere todecrease thenumberof scans toone. In contrast,existing conventional convection-compensated pulsesequences rely on phase cycling for CTP selection, whichrequires several scans per gradient increment. The SPENDSTE pulse sequence overall provides a significantaccelerationofDOSYexperimentsinorganicmixtures,withcomplete diffusion information obtained in less than 1 s.These results open up prospects for the monitoring oforganic reactions and may help to characterizeintermediate species as well as measure their diffusioncoefficients.

JournalName ARTICLE




Figure 3 1D 1H NMR spectrum of a mixture of cyclododecane, adamantane.Squaleneand1,3-butadieneinCD2Cl2.

Reactionmonitoring

As an illustration of the possibility to analyse reactingmixtures, the reaction of isobutyraldehyde with 1,4-diaminobenzene29 (Figure 4) in acetonitrile-d3 wasmonitored by SPEN DSTE DOSY NMR experiment. Thereactionyieldsabis-imineandgoesthroughamono-imineintermediate. Depending on the initial stoichiometry, themono-imine compound is either i) completely consumed(belowtheNMRdetection limit)or ii) stilldetectable.Thismodelreactionprovidesthepossibilitytoboth ii)teststheaccuracy of a measurement on a transient species and i)recordreferencevaluesofthediffusioncoefficients(showninFigureS3).

Figure4NMRmonitoringof thereactionof isobutyraldehydewith1,4-diaminobenzene inacetonitrile-d3. (a)Reactionscheme. (b)Timeseriesof1D1HNMRspectrarecordedduringthereaction;thefirstspectrumwasrecoded2minafterthestartofthereaction,thenaspectrumwasrecordedevery1.5min;anexpansionofthearomaticregion isshown. (c) Time evolution of the peak areas generated using the 1HNMRdata. The sampling scheme for theDOSY experiments is shownqualitatively. Dashed vertical linesindicatesthetimesforwhichestimateddiffusioncoefficientsarereportedinTable2.

The reaction was conducted in an NMR tube, inacetonitrileat10°C.1D1HNMRspectrawererecordedatregular intervalsover the reaction time, and characteristicresonances for the intermediate (mono-imine) and bis-imineproductwereidentified(Figure4b).Thesignalat6.62ppm was used to follow the mono-imine intermediate.Similarly, the bis-imine product could be followed byintegration of its characteristic resonance at 7.01 ppm. Inthereactionprofileofthistransformation,showninFigure4c,thetransientformationofthemono-imineintermediateis apparent. The mono-imine then further reacts with asecondaldehydemoleculetoproducethebis-imine. The total duration of a conventional DSTE DOSYexperiment is of about 22 min in this context. This isbecause of the incremented gradient dimension togetherwith a 16-step phase cycle. This duration is of the sameorder or magnitude as the reaction time for thetransformationunder study.As a result, theacquisitionofdiffusion information during the course of the reaction isimpaired, as illustrated in Figure 5a with a conventionalDSTEDOSYspectrum(theacquisitionofwhichwasstarted1minafterthebeginningofthereaction).Althoughagood

separation of the components is observed, erroneousvaluesof thediffusioncoefficientsareobtained,as shownin Table 2. Specifically, the diffusion coefficient for themono-imine is largely overestimated: the value obtainedduring the reaction is 50 % larger the one obtained atequilibrium(20.4vs13x10-10m2.s-1). Theobservedsystematicerrormayalsobeappreciatedfrom thediffusiondecay curves (FigureS2),whicharenotsuitablydescribedby theStejskal-Tannerequation.Ontheother hand, the product that evolves in the oppositedirection displays an apparent diffusion coefficient slightlylowerthanexpected(Table2).

ARTICLE JournalName




Table 2 Diffusion coefficients measured by conventional and SPEN DOSY on areactingmixtureinacetonitrile-d3.FortheSPENexperiments,four-timepointsaregiven(correspondingtothedashedlineinFigure3c.

Diffusioncoefficient.1010m2.s-1

SPENDSTEDOSYat(min) ConvDSTEstarted

δppm

compound 3.5 5 8 11t=2min

eq.a

7.01 bis-imine 11.4 12.7 12.6 12.3 10.1 11.3

6.62 mono-imine 13.4 13.7 13.3 13.1 20.4 13.0

a:chemicalequilibrium

In a separate run, a time series of DOSY spectra wasrecorded using the SPEN DSTE experiment (Figure 4c). ArepresentativespectrumisshowninFigure5b.Inthiscase,the inter-scan delay is set to 15 s and the complete

diffusioninformationforeachspectralregionisobtainedinlessthan200ms,sothattheestimateddiffusioncoefficientismuchlessaffectedbythevariationofconcentration.ThisisapparentinTable2,withagoodagreementbetweenthediffusioncoefficientsobtainedwithSPENDOSY“onthefly”and those obtained with the conventional experiment onanequilibratedsample.

DiffusioncoefficientsmeasuredbyDOSYarefrequentlyused to estimate thehydrodynamic radii or themolecularweightsof compounds in solution.3-6 The consequencesoferroneousmeasurementsofdiffusioncoefficientsfortime-dependent concentrations are well illustrated here by anattempt to estimate molecular weights for the reactingmixture,usingconventionalDOSYandSPENDOSYdata.Theexternal-calibration-curve (ECC) method of Stalke and co-workerswasused,asitrequiresasingleinternalreferencethatcanbethesolvent.6,25

Figure5DOSYspectraofareactionmixtureofisobutyraldehydewith1,4-diaminobenzeneinacetonitrile-d3.(a)ConventionalDSTEDOSYspectrumrecordedover22minfromthebeginningofthereaction.(b)Concatenationof3spectrallyselectiveSPENDSTEDOSYspectrarecordedatt=3.5min.Aspectrumwasobtainedevery15s,andeachspectrumwasrecordedinlessthan200ms.A1H1DspectrumofthemixtureisshownabovetheDOSYdisplays.OnlyresolvedpeakswereincludedintheDOSYprocessing.

Table 3 Estimatedmolecularweights obtained on the reactionmixtureM3withexternal-calibrationcurves,usingacetonitrileasaninternalreference.

SPENDSTEDOSY ConvDSTEstartedat

t(3.5min) t(2min) eq.a

δ

ppmcompound

D

.1010

MW

g.mol-1

Δ

%

D

.1010

MW

g.mol-1%

D

.1010

MW

g.mol-1

Δ

%

7.0 bis-imine 11.4 244 11 10.1 314 31 11.3 243 11

6.6 mono-imine 13.4 185 12 20.4 96 69 13.0 192 16

1.9 acetonitrile 25.1 - - 25.8 - - 24.8 - -

Δ:differenceinMW

Table 3 shows estimated molecular weights obtainedfrom several data sets, for the mono-imine (MW = 162g.mol-1 ) and bis-imine (MW = 216 g.mol-1) compounds.WhiletheconventionalDOSYdatayieldsestimatesthatareerroneous by 50% or more, with SPEN DOSY data theestimatedmolecularweightsarewithin20%of theknownvalues.Theseresultsareduetothefactthattheuseofaninternal reference makes it possible to compensate forsome sources of error (such as gradient mis-calibration),butcannotcompensateforthechange inconcentrationofthe analysed compounds during the reaction. While themodel reaction analysed here serves as a validation, thepossibility to obtain a sensible estimate of the molecular

JournalName ARTICLE




weightof an intermediatewouldbeparticularlyuseful forthemechanisticstudyofanovelreaction.

ConclusionsIn summary, we have shown that the acquisition ofdiffusion-ordered NMR spectra can be acceleratedsignificantly for experiments carried out in low-viscositysolvents,which include importantorganicsolventssuchaschloroform, dichloromethane or acetonitrile. This studysheds light on the great potential of SPENDOSY to followchemical transformations and to obtain diffusioninformationfromreactionintermediates.

Conflictsofinterest“Therearenoconflictstodeclare”.

AcknowledgementsThisresearchwassupportedbytheLabExCharmmmat,theRégionIle-de-France,theAgenceNationaledelaRecherche(ANR-16-CE29-0012).

Notesandreferences1. C. S. Johnson, Prog. Nucl. Magn. Reson. Spectrosc.,

1999,34,203-256.2. B.Antalek,ConceptsMagn.Reson.,2002,14,225-258.3. G.Pages,V.Gilard,R.MartinoandM.Malet-Martino,

Analyst,2017,142,3771-3796.4. A. Macchioni, G. Ciancaleoni, C. Zuccaccia and D.

Zuccaccia,Chem.Soc.Rev.,2008,37,479-489.5. D. Li, G. Kagan, R. Hopson and P. G.Williard, J. Am.

Chem.Soc.,2009,131,5627-5634.6. R. Neufeld and D. Stalke, Chem. Sci., 2015, 6, 3354-

3364.7. R.Evans,Z.Deng,A.K.Rogerson,A.S.McLachlan,J.J.

Richards,M.Nilsson andG.A.Morris,Angew. Chem.Int.Ed.,2013,52,3199-3202.

8. S. Denis-Quanquin, F. Riobé, M.-A. Delsuc, O. MauryandN.Giraud,Chem.Eur.J.,2016,22,18123-18131.

9. N. E. Schlörer, E. J. Cabrita and S. Berger, Angew.Chem.Int.Ed.,2002,41,107-109.

10. D. Li, I. Keresztes, R. Hopson and P. G.Williard,Acc.Chem.Res.,2009,42,270-280.

11. P. García-Álvarez, R. E. Mulvey and J. A. Parkinson,Angew.Chem.Int.Ed.,2011,50,9668-9671.

12. R.Neufeld,T.L.Teuteberg,R.Herbst-Irmer,R.A.Mataand D. Stalke, J. Am. Chem. Soc., 2016, 138, 4796-4806.

13. K.F.MorrisandC.S.Johnson,J.Am.Chem.Soc.,1992,114,3139-3141.

14. E.O.StejskalandJ.E.Tanner,J.Chem.Phys.,1965,42,288-292.

15. M.Nilsson,M.Khajeh,A.Botana,M.A.BernsteinandG.A.Morris,Chem.Commun.,2009,0,1252-1254.

16. M. Oikonomou, J. Asencio-Hernández, A. H. VeldersandM.-A.Delsuc,J.Magn.Reson.,2015,258,12-16.

17. M. D. Pelta, G. A. Morris, M. J. Stchedroff and S. J.Hammond,Magn.Reson.Chem.,2002,40,S147-S152.

18. M. Urbańczyk, W. Koźmiński and K. Kazimierczuk,Angew.Chem. Int.Ed.(international),2014,53,6464-6467.

19. S.M.Pudakalakatti,K.Chandra,R.ThirupathiandH.S.Atreya,Chem.Eur.J.,2014,20,15719-15722.

20. N. M. Loening, J. Keeler and G. A. Morris, Magn.Reson.Chem.,2001,153,103-112.

21. M. J.Thrippleton,N.M.LoeningandJ.Keeler,Magn.Reson.Chem.,2003,41,441-447.

22. Y. Shrot and L. Frydman, J.Magn. Reson., 2008,195,226-231.

23. L. Guduff, I. Kuprov, C. van Heijenoort and J.-N.Dumez,Chem.Commun.,2017,53,701-704.

24. M.Nilsson,J.Magn.Reson.,2009,200,296-302.25. S. Bachmann, B. Gernert and D. Stalke, Chem.

Commun.,2016,52,12861-12864.26. E. Martínez-Viviente and P. S. Pregosin, Helv. Chim.

Acta,2003,86,2364-2378.27. A.JerschowandN.Müller,J.Magn.Reson.,1997,125,

372-375.28. L. Guduff, D. Kurzbach, C. vanHeijenoort, D. Abergel

andJ.-N.Dumez,Chem.Eur.J.,2017,23,16722-16727.29. D.A.Foley,E.Bez,A.Codina,K. L.Colson,M.Fey,R.

Krull, D. Piroli, M. T. Zell and B. L. Marquez, Anal.Chem.,2014,86,12008-12013.

spatially encoded diffusion-ordered nmr spectroscopy of

Documents