Controlled Perturbation of the Thermodynamic Equilibrium by Microfluidic Separation of Porphyrin-based Aggregates in Multi-component Self-Assembling System

ISSN 1359-7345

Chemical Communications

www.rsc.org/chemcomm Volume 49 | Number 18 | 4 March 2013 | Pages 1773–1872

1359-7345(2013)49:18;1-5

COMMUNICATIONFloris Helmich and E. W. MeijerControlled perturbation of the thermodynamic equilibrium by microfl uidic separation of porphyrin-based aggregates in a multi-component self-assembling system

1796 Chem. Commun., 2013, 49, 1796--1798 This journal is c The Royal Society of Chemistry 2013

Cite this: Chem. Commun.,2013,49, 1796

Controlled perturbation of the thermodynamicequilibrium by microfluidic separation ofporphyrin-based aggregates in a multi-componentself-assembling system†

Floris Helmich and E. W. Meijer*

In a microfluidic H-cell, a multi-component self-assembled system is

brought out-of-equilibrium by changing the bimodal composition of

porphyrin stacks and pyridine-capped dimers. Driven by their different

diffusivities, diffusion-controlled separation in methylcyclohexane

reveals different compositions when detected in-line and off-line,

which demonstrates the kinetic behaviour of this metastable system.

The microfluidic technique also proves to be highly equipped to

determine diffusion constants of the different assemblies.

Supramolecular polymers comprised of p-conjugated chromo-phores form an interesting bottom-up approach for the devel-opment of functional, nanometer-sized materials.1 Theirmacroscopic properties are highly tuneable, reversible anddynamic due to the wide versatility of self-assembled architec-tures. Structural control herein strongly depends on the mono-mer design and the thermodynamic state imposed by externalstimuli such as temperature and concentration.2 Furthermore,the introduction of additional supramolecular interactionscauses a shift of equilibria between non-covalently linkedbuilding blocks.3 Their highly dynamic nature makes controlover self-assembly via non-thermodynamic pathways very chal-lenging. Reduced supramolecular dynamics already affordedmemory effects,4 interconversions of different aggregate types5

and size exclusion chromatography of discrete assemblies.6 Theuse of multiple building blocks is an important aspect in orderto create functional materials, while such assemblies becomeinherently more complex. Therefore, multi-component self-assembly requires in-depth analyses, especially when theyoperate away from thermodynamic equilibrium.7

In earlier reports, microfluidic devices have been tailored toinduce stable gradients that stimulate self-assembly processes;for instance by diffusive mixing of solvents8 and pH.9 Besidesthermodynamic parameters such as temperature and monomer

design,10 the formation of liposomes with controlled size hasalso been achieved by the appropriate hydrodynamic condi-tions.11 In this communication, we introduce microfluidictechniques as an approach to control the distribution ofkinetically metastable porphyrin aggregates.

Recently, we reported on the cooperative self-assembly of (S)-Zn-porphyrin S-Zn in the presence of pyridine (Fig. 1).12 Thedilution-induced self-assembly of this system was fully describedin a thermodynamic model. In methylcyclohexane (MCH), theporphyrins form helical, cofacial stacks upon the formation ofstrong 4-fold intermolecular hydrogen bonds. Axial coordinationof pyridine sterically blocks one porphyrin face for hydrogenbonding, which results in the formation of hydrogen bonded,pyridine-capped dimers.12 The pyridine titration shows a sharptransition from stacks to dimers and at high porphyrin concentra-tions a pseudo-bimodal distribution between the two is observed asshown by an apparent isosbestic point in their UV-vis spectra.13

AFM-micrographs of the elongated stacks clearly reveal thepresence of curled fibrils, which fully disappear after the depoly-merization process (Fig. 1). High association constants weredetermined for both these different hydrogen-bonded aggregates;hence, we conceived slow aggregate interconversion upon chan-ging an external stimulus. And indeed upon the addition of MCH,

Fig. 1 Thermodynamic model for the S-Zn–pyridine system with thecorresponding UV-vis titration experiment at 2.0 � 10�5 M in MCH and AFM-micrographs for the stacks (left) and dimers (right).

Institute for Complex Molecular System, Laboratory of Macromolecular and Organic

Chemistry, Eindhoven University of Technology, P.O. Box 513, 5600 MB, Eindhoven,

The Netherlands. E-mail: [email protected]

† Electronic supplementary information (ESI) available: Microfluidic procedures,DOSY and DLS measurements. See DOI: 10.1039/c2cc36887k

Received 22nd September 2012,Accepted 8th November 2012

DOI: 10.1039/c2cc36887k

www.rsc.org/chemcomm

ChemComm

COMMUNICATION

Publ

ishe

d on

08

Nov

embe

r 20

12. D

ownl

oade

d by

Uni

vers

ity o

f G

roni

ngen

on

25/0

8/20

15 1

3:15

:19.

View Article OnlineView Journal | View Issue

This journal is c The Royal Society of Chemistry 2013 Chem. Commun., 2013, 49, 1796--1798 1797

dilution-induced self-assembly was observed with slow kinetics(Bhours). This kinetic property as well as the considerable differ-ence in size between stacks and dimers encouraged us to investi-gate the possibility to control their distribution ratio by diffusivemass transport in a microfluidic H-cell,14 thereby temporarilycreating an out-of-equilibrium situation of kinetically metastablestructures. By detection at different timescales we probe their slowredistribution to the thermodynamically stable state.

In our experimental setup,15 an equivolumetric flow rate ismaintained over both the inlets and outlets, by using well-calibratedsyringe pumps and flow regulators, respectively (Fig. 2A). As a result,the contact time of the laminar interface between both flows equalsthe diffusion time (tdiffusion), which is determined by the appliedflow rate. Additionally, in-line UV-vis detection provides quantitativeand qualitative analysis of the diffused species. By combining thesefeatures, we first use the H-cell as a tool to estimate diffusioncoefficients of the pure aggregates and then we continue to createout-of-equilibrium situations by the separation of a mixture of stacksand dimers.

In order to validate the experimental setup, we performed adiffusion experiment on the porphyrin monomer model com-pound S-Zn–Me.16 At a concentration of 2.0 � 10�5 M in MCH,S-Zn–Me was eluted against pure MCH at flow rates between 0.50and 0.05 ml s�1 (0.35 o tdiffusion o 3.52 s, respectively) and thesteady-state UV-vis spectra were recorded in the extraction stream(Fig. 2B). The absorbance at lmax = 420 nm was normalized on thespectrum of S-Zn–Me at 1.0 � 10�5 M and subsequently plottedagainst the diffusion time (Fig. 2C). The obtained diffusionprofile was fitted to a 1-dimension-unsteady-state diffusionmodel,15 from which we estimated a diffusion coefficient of3.2 � 10�10 m2 s�1. The latter closely resembles the valueobtained in a diffusion ordered 1H-NMR (DOSY) experiment ofS-Zn–Me in D14-MCH (2.2 � 10�10 m2 s�1).15,17

Next, we performed identical diffusion studies on purestacks and pure dimers of S-Zn at 2.0 � 10�5 M, in which thelatter solution has a pyridine excess of 250. The normalized

diffusion profiles for the stacks and dimers (constructedfrom the normalized absorbance at lmax = 390 and 427 nm,respectively)15 are highly distinctive; hardly any mass transfer isobserved for stacks, whereas a considerable amount of dimershave diffused over (Fig. 2C). Additionally, the latter diffusionprofile approaches the curve of S-Zn–Me; hence we estimate adiffusion coefficient of 2.7 � 10�10 m2 s�1 for the dimers and7.8 � 10�12 m2 s�1 for the stacks, both of which correspond totheir diffusivities observed in dynamic light scattering (DLS)measurements.15 Despite the fact that the porphyrins are solelyheld together by supramolecular interactions, the differentdiffusion behaviour of stacks and dimers is obviously relatedto their aggregate size.14,18

As a next step herein, we exploit the microfluidic setup toseparate a mixture of stacks and dimers. At a pyridine excess of45, the molar distribution of S-Zn monomers over stacks anddimers (stack/dimer-ratio) at thermodynamic equilibrium is31/69, as determined from the Soret band intensity ratio of0.54 at 390 and 427 nm, respectively (Fig. 1).13 This initialmixture is eluted against MCH at a flow rate of 0.10 ml s�1

(tdiffusion = 1.76 s) and the in-line UV-vis spectra are acquired forboth the extraction and residual stream (Fig. 3A). Compared tothe initial state, the latter spectrum shows a drop in dimerabsorbance, while the stack absorbance remains unchanged. Asa result, the stack/dimer-ratio increases to 36/64, whereas theextraction stream reveals that predominantly dimers havediffused over; a stack/dimer-ratio of 16/84 is estimated fromthe spectrum. Based on their different mass transport, in-lineUV-vis detection reveals that the bimodal distribution betweenstacks and dimers can be moderately controlled by this micro-fluidic technique.

Besides porphyrin stacks and dimers, free pyridine and(pyridine-complexed) monomers also diffuse over at relativelyhigh rates. In addition, the diffusion process in this particularconfiguration leads to inhomogeneity of a two-fold diluted

Fig. 2 (A) Schematic representation of the H-cell, in which two 110 mm channelswith a depth of 50 mm merge and split (FM = flow meter, FR = flow regulator,flow regulator is placed in the residual stream and in-line UV-vis measurementsare performed in the extraction stream). (B) In-line UV-vis spectra of S-Zn–Meacquired at different flow rates. (C) Diffusion profiles for S-Zn–Me, S-Zn stacksand S-Zn : pyridine (1 : 1) dimers at lmax = 420, 390 and 427 nm, respectively,with the corresponding fits and simulation at DDOSY = 2.2 � 10�10 m2 s�1

(normalized diffusion = Cdiff/1.0 � 10�5 M).

Fig. 3 In-line (A) and off-line (B) UV-vis spectra of S-Zn at 2.0� 10�5 M with 9.0�10�4 M pyridine of the initial solution, residual stream and extraction streamrecorded tdet B20 s and B3 h after separation, respectively. (C) Kinetic profile ofthe dilution-induced self-assembly upon 2-fold dilution of the initial solutionprobed by CD (black axis, left at 390 nm) and UV-vis (red axis, right at 390 and427 nm).

Communication ChemComm

Publ

ishe

d on

08

Nov

embe

r 20

12. D

ownl

oade

d by

Uni

vers

ity o

f G

roni

ngen

on

25/0

8/20

15 1

3:15

:19.

View Article Online

1798 Chem. Commun., 2013, 49, 1796--1798 This journal is c The Royal Society of Chemistry 2013

solution. With these features, the system is being pushed out-of-equilibrium, leading to a complex reaction–diffusion event,which is characterized by the supramolecular dynamics forregaining thermodynamic equilibrium. In order to obtain moreinsight into this process, we compare the in-line UV-vis spectrawith the spectra obtained after sample collection, corres-ponding to detection times of B15 s and B3 h after separation,respectively.15

Directly after separation, the in-line spectra seem to revealthe sole diffusion process of dimers to the other stream.However after sample collection, highly distinctive spectra areobserved that cannot be explained by mass transport only(Fig. 3B). Compared to the in-line spectra, both the residualand extraction streams show an increased stack/dimer-ratio of58/42 and 47/53, respectively. In the residual stream, thesignificant enhancement of stacks is mainly due to loss ofdimers and the diffusion of pyridine and pyridine-complexedmonomers that source dimerization. Despite the pyridineenrichment in the low-concentration extraction stream, themodest enhancement of stacks can be explained by thedilution-induced self-assembly effect, which operates stronglyat high dilutions.12

Both short and long timescale analyses can be rationalizedon the basis of mass transport and thermodynamic equilibria.In order to investigate how the supramolecular dynamics spanthese distinctive timescales, we perform a 2-fold dilutionexperiment on the initial mixture, which closely resemblesthe microfluidic experiment. In this kinetic measurement, thedilution-induced self-assembly leads to the formation of stacks,which can exclusively be probed by CD spectroscopy while thestack/dimer-ratio can be probed by UV-vis at 390 and 427 nm.Directly after injecting an equivolumetric amount of MCHto the initial mixture, the kinetic profiles clearly reveal theredistribution over stacks and dimers. Approximately 4 hoursafter dilution the new thermodynamically stable state isreached (Fig. 3C). Likely due to the relatively slow supra-molecular dynamics, thermodynamic equilibrium is achievedwhile collecting the extraction and residual streams.

With reference to detection, these kinetic aspects are impor-tant to be considered when changing the thermodynamic statee.g. by mixing-in multiple components or concentration- andtemperature jumps.19 Noteworthily, this feature has also beenrecognized in the 1 : 1 complexation of Zn–porphyrin andpyridine in a microfluidic mixer20 and probing fast self-assemblykinetics of peptide ligands and quantum dots.21 On the otherhand, besides its properties to estimate size and separatesupramolecular aggregates, this microfluidic technique affordsthe controlled preparation of non-thermodynamically stableintermediates.22 Herewith, we envision the creation of kinetictraps by ‘freezing-in’ the intermediates and related experimentsare underway.

This work was supported by the Council of ChemicalSciences of The Netherlands Organization for Scientific

Research (CW-NWO). We thank Dr Thomas Hermans, NielsBrankaert and Joost van Dongen for experimental contributionsto the project. Dr Takashi Hirose is acknowledged forcurve-fitting and Paul Schlebos for the DOSY experiment.

Notes and references1 (a) A. Schenning and E. W. Meijer, Chem. Commun., 2005,

3245–3258; (b) E. R. Kay, D. A. Leigh and F. Zerbetto, Angew. Chem.,Int. Ed., 2007, 46, 72–191.

2 T. F. A. De Greef, M. M. J. Smulders, M. Wolffs, A. Schenning,R. P. Sijbesma and E. W. Meijer, Chem. Rev., 2009, 109, 5687–5754.

3 (a) R. Hoogenboom, D. Fournier and U. S. Schubert, Chem. Commun.,2008, 155–162; (b) J. Hamacek, M. Borkovec and C. Piguet, Chem.–Eur.J., 2005, 11, 5217–5226.

4 (a) F. Helmich, C. C. Lee, A. P. H. J. Schenning and E. W. Meijer,J. Am. Chem. Soc., 2010, 132, 16753–16755; (b) L. Rosaria, A. D’Urso,A. Mammana and R. Purrello, Chirality, 2008, 20, 411–419;(c) L. J. Prins, F. De Jong, P. Timmerman and D. N. Reinhoudt,Nature, 2000, 408, 181–184.

5 R. F. Pasternack, C. Fleming, S. Herring, P. J. Collings, J. dePaula,G. DeCastro and E. J. Gibbs, Biophys. J., 2000, 79, 550–560.

6 (a) W. S. Li, K. S. Kim, D. L. Jiang, H. Tanaka, T. Kawai, J. H. Kwon,D. Kim and T. Aida, J. Am. Chem. Soc., 2006, 128, 10527–10532;(b) M. J. Ahrens, R. F. Kelley, Z. E. X. Dance and M. R. Wasielewski,Phys. Chem. Chem. Phys., 2007, 9, 1469–1478; (c) D. Furutsu,A. Satake and Y. Kobuke, Inorg. Chem., 2005, 44, 4460–4462.

7 B. A. Grzybowski and C. J. Campbell, Chem. Eng. Sci., 2004, 59,1667–1676.

8 A. Jahn, W. N. Vreeland, M. Gaitan and L. E. Locascio, J. Am. Chem.Soc., 2004, 126, 2674–2675.

9 L. Brown, S. L. McArthur, P. C. Wright, A. Lewis and G. Battaglia,Lab Chip, 2010, 10, 1922–1928.

10 J. M. Zook and W. N. Vreeland, Soft Matter, 2010, 6, 1352–1360.11 A. Jahn, W. N. Vreeland, D. L. DeVoe, L. E. Locascio and M. Gaitan,

Langmuir, 2007, 23, 6289–6293.12 F. Helmich, C. C. Lee, M. M. L. Nieuwenhuizen, J. C. Gielen,

P. C. M. Christianen, A. Larsen, G. Fytas, P. Leclere, A. Schenningand E. W. Meijer, Angew. Chem., Int. Ed., 2010, 49, 3939–3942.

13 Due to their high association constants and cooperativity, both thehydrogen-bonded states (stacks and pyridine-capped dimers) arethe most abundant species in solution. As a result, they barelycontribute to the UV-vis absorbance thus an isosbestic point isobserved. For the current studies, we do not consider the presenceof free porphyrin monomers and monomeric adducts.

14 J. P. Brody and P. Yager, Sens. Actuators, A, 1997, 58, 13–18.15 See ESI†.16 S-Zn–Me contains a tertiary amide and due to this methyl substitu-

tion on the nitrogen, it cannot donate in intermolecular hydrogenbonding. Therefore, this molecule does not aggregate in MCH andserves as a discrete reference compound.

17 The difference in diffusion coefficient of S-Zn–Me determined byDOSY and the microfluidic experiment is likely due to minorcontributions of convective mass transfer as a consequence ofpressure-driven flow in the latter, which therefore shows an over-prediction of the diffusion coefficient.

18 B. H. Weigl and P. Yager, Science, 1999, 283, 346–347.19 (a) E. T. Pashuck and S. I. Stupp, J. Am. Chem. Soc., 2010, 132,

8819–8821; (b) A. Lohr and F. Wuerthner, Chem. Commun., 2008,2227–2229; (c) F. Helmich, M. M. J. Smulders, C. C. Lee, A. P. H. J.Schenning and E. W. Meijer, J. Am. Chem. Soc., 2011, 133,12238–12246.

20 M. Brivio, R. E. Oosterbroek, W. Verboom, A. van den Berg andD. N. Reinhoudt, Lab Chip, 2005, 5, 1111–1122.

21 J. H. Wang, P. J. Jiang, Z. Y. Han, L. Qiu, C. L. Wang, B. Zheng andJ. Xia, Langmuir, 2012, 28, 7962–7966.

22 (a) C.-W. Wang, D. Sinton and M. G. Moffitt, J. Am. Chem. Soc., 2011,133, 18853–18864; (b) A. Jahn, S. M. Stavis, J. S. Hong, W. N. Vreeland,D. L. Devoe and M. Gaitan, ACS Nano, 2010, 4, 2077–2087.

ChemComm Communication

Publ

ishe

d on

08

Nov

embe

r 20

12. D

ownl

oade

d by

Uni

vers

ity o

f G

roni

ngen

on

25/0

8/20

15 1

3:15

:19.

View Article Online