This paper has been submitted at Nov 4th, 1996, for publication in a
scientific journal and is now subject to the reviewing process.
You may ask for a hardcopy with the complete manuscript, which contains all figures and all
greek symbols.
Everybody
is invited to comment to this paper. I will consider every argument and
comment back.
This is the 3rd revised version.
Dr. Bernhard Wessling
Zipperling Kessler / Ormecon Chemie
D-22949 Ammersbek
5. 2. Other experimental results
There are even more hints to be found in literature, even though the authors might not come to the same conclusions as I do.
5. 2. 1. ICP morphology / particle size / coherence length
The debate we are making is not limited to the topic of "solution or dispersion". It starts with the question: "Are ICPs of a fibrillar[39] or of a globular morphology[40]?" It is still not yet accepted that even the very highly conducting polyacetylene, called "N-PAc" consists of very fine globular - although in this case elliptical - primary particles, as Theophilou found out later by STM[41]. They were around 20 nm in length in stretching direction.
Even more convincing: the coherence length generally accepted in the field ranges from 5 - 10 nm[42]. All determinations of the coherence length or the "metallic islands" in conductive polymers not depending from its chemical nature are leading to values between 3 and 7.5 nm, not far from the value determined by us (8 nm plus 1.6 nm amorphous shell, cf. 5.1.2.).
5. 2. 2. Concave viscosity curves
K. Levon et al. reported about a viscosity study of PAni (EB) in NMP and NMP with LiCl[43]. They found concave reduced viscosity curves, fig 11. In dispersions, viscosity curves of all shapes including concave forms can be found. The form is only dependent on the microscopic structure of the dispersion, i. e. if there are super-structures being formed, if the particles are flocculating or de-flocculating, or if phase transitions (from, e.g. bi-continous to normal 2-phase structures with 1 continuous phase) are happening at various concentrations.
However, Levon et al. interpreted their findings without going further into details by saying "aggregation of polyaniline chains in dilute solution was observed". Was it really what they observed? Wouldn't it have been more correct to state, that a concave reduced viscosity curve was observed, which can be interpreted by assuming something?
At least, the authors admit: "... aggregates still exist in solution as shown by light scattering. It may be that aggregates larger than dipolymers were not dissociated completely into the isolated chains but into small aggregates." That is what colloidal scientist are calling "dispersions".
5. 2. 3. "Solubilization" of PAni in water by interpolymer complexation
Y. Liao and K. Levon[44] claim to have made a dilute solution of PAni (EB) with LiCl in either poly(4-styrenesulfonic acid) or its monomer in NMP, which could be transferred into water without precipitation. However, the reduced viscosity curves (fig 12) are showing almost no dependence from concentration. This was attributed by the authors to a "dissolution by complexation" (of the PAni chains by the PSSA or SSA), without further explanation.
However, if that were the case, the viscosity must increase with concentration of dissolved PAni chains (PAni/PSSA complex), which is not the case. I would interprete the reported results with a rather poor and irreproducible dispersion of PAni in NMP, with PSSA at the interface, which is forming a very complex (and probably also dramatically changing) micelle structure in water. Such systems would be an exciting object to be studied in colloidal science, another type of microemulsion: a sol (PAni/PSSA in NMP) forming an emulsion in water (probably with PSSA at the NMP/water interface).
5. 2. 4. Wormlike micelles
L. Dai and J. White reported about "aggregation in conducting polymer solutions"[45]. The object of the study were polyacetylene-polyisoprene block copolymers and mixtures of these having different PAc-PI-block content in toluene. In viscosity measurements they observed non-ideal behaviour and assigned it (correctly) to intermolecular interactions. The temperature dependence of viscosity was also studied and it was shown, that - in contrast to different mixtures of comparable pure PI-polymer mixtures, the deviations came from interactions between the PAc-blocks.
From viscosity measurements with polymers containing different amount of PAc-blocks, they derived the Huggin's constant k', a measure for the solvent strength (degree of solution) in polymer solutions. The constant found was much higher (up to 1.1 with higher PAc content) than normal values for true solutions (0.3 - 0.4), which is a sign for intermolecular interactions and association/aggregation.
Their structure hypothesis of the polymer/solvent system is a "wormlike micelle" form.
By measuring the surface tension of their systems, they determined the critical micelle concentration (i.e. the concentration above which the surface tension does not decrease any more, or: below which the surface tension is increasing).
From SAXS[46] measurements they were deriving the statement, that the systems either contain a mixture of particles with different scattering length densities or multicomponent particles. They derived an aggregates size for undoped PAc blocks of 12 to 14 nm, and for the J2-doped lower values close to 4 ... 5 nm. They concluded that the undoped block copolymer forms worm-like micelles, the doped one a lamellar structure in the solvent systems due to aggregation of the PAc blocks.
5. 2. 5. Soap-like structures
P. Garrin, J. Aimé et al.[47] reported about another PAc block copolymer, the one made with polystyrene. Their study was focused on SANS[48] experiments in THF "solution" from where they derived clear evidence for aggregation (micelle formation), and - to their surprise - a demixing upon heating, where the copolymer formed a film at the solvent/gas interface. A layer structure was earlier found by Aldissi et al[49] for a similar Pac-block copolymer.
The authors complained that it were difficult to explain why the copolymer formed a film at the solvent/gas interface, "even at a qualitative level". They did not consider that the copolymer was not truly soluble (but was dispersed like a soap) and therefor has to form either micelles, when dispersed throughout the solvent, or will form a separate phase at the solvent surface, like soap.
5. 2. 6. Aggregation of polyoctyl-thiophene
Heffner and Pearson studied "solution processing of a doped conducting polymer"[50] (poly-3-octyl-thiophene, P3OT, FeCl3 or nitrosylhexafluorophosphate doped, in toluene). Beware: the dopants were not soluble in toluene, so that they were introduced to the polymer as an acetonitrile solution, which is miscible with toluene - but in the resulting system, it will play a complex role probably at the interfaces!
Upon introduction of the doping solution to the (undoped) polymer solution, some "aggregation" (as they called it) of the resulting doped "invariably occurred"[51]. They continue: "Although the spectral change which occurs in the sol phase upon doping indicates that some doped polymer must remain dissolved" - from where did they conclude this? - aggregates were still clearly visible in doped solutions with polymer concentrations as low as 0.01c* (c*: critical micelle concentration).[52] ... The degree of aggregation ranged from tiny clusters suspended in clear solution to the entire sample forming a cohesive gel[53]. ... We have found the gelation to be a reversible process" (cf. 5. 1. 3.).
The authors are still convinced of dealing with solutions (which only by mistake show some aggregates) and so were trying to improve the solution (prevent aggregation) by adding some salts, as they expected, that "the addition of an organic salt[54] to screen the interaction and reduce the degree of aggregation, as is observed in the 'salting in' of aqueous protein solutions. ... However, rather than creating a deaggregated doped solution as we hoped, interactions between the charged ions caused a complete dedoping of the solution."
As an interpretation of this phenomenon I would propose to consider that the original system is again a complex microemulsion consisting of doped phases, surrounded by the octyl-side-chains arranged as a shell (micelles), in toluene with some complex phase structure mediated by acetonitrile. The addition of the salt might have changed the phase structure in a way, that acetonitrile cannot any more facilitate this microemulsion and the micelle formation, but a phase separation between toluene/P3OT (undoped) and acetonitrile/doping/salt, leading to a microemulsion of this composition.
5. 2. 7. Viscoelasticity
R. Gregory presented[55] fascinating results from their viscosity studies of PAni (EB) in NMP "solutions". He showed convincing evidence that these solvent systems were viscoelastic, a property directly linked with their colloidal character. Most colloidal systems do not show Newtonian behaviour of the viscosity upon shear stress, and there are often time-dependent phenomena (relaxation) being observed.
These are due to dynamic structure formation and restructuring processes based on reversible long-range interaction between the particles.
5. 2. 8. Gel permeation chromatography: changes upon salt or co-solvent addition
M. Angelopoulos reported in the same conference some "magic" change of, as she interpreted, "the molecular weight of PAni (EB)". When comparing the retention time in GPC using NMP or NMP/5% LiCl, the "high molecular weight fraction" (at retention time 25 minutes) vanished, and the peak at 30 minutes retention time ("low molecular weight") increased.
When adding 10% m-cresol to a EB "solution" in DMPU[56], again the peak of the "high molecular weight fraction" vanished whereby the "low molecular weight" peak increased.
Her interpretation was, that the salt and the co-solvent, resp., were "breaking up the high Mw fraction". I would like to ask, how LiCl or m-cresol could be able to cut the chemical bonds in the PAni (EB) chains, which are supposed to occur fully solvated in the NMP. This is virtually impossible.
An alternative interpretation would be, that GPC of NMP-PAni (EB or ES) systems just tell us about the retention time (adsorption-desorption process) of NMP-dispersed PAni particles in the GPC column, whereby the "high molecular weight fraction" would be the bigger particles, here: secondary particles, i.e. aggregates of primary particles, and the "low molecular weight fraction" peak (higher retention time) represents the fraction of the smaller (primary?) particles.
Upon adding LiCl or m-cresol, we happen to desaggregate (= disperse) the bigger particles, the equivalent peak is disappearing, and the peak representing the smaller particle fraction becomes the only and increased peak.
5. 2. 9. Magnetic susceptibility of polyaniline in "solution"
Cao and Heeger [60] reported about electron spin resonance measurements of the protonated form of polyaniline (protonated with campher sulfonic acid and dodecyl sulfonic acid, resp.). First, we note that the room temperature magnetic susceptibility is more or less equal for PAni-CSA in m-cresol or -DBSA in xylene, in cast films or in polyblends with PMMA (around 3 - 4 10-5 e.m.u./mol). In free standing films, there is a small temperature dependence, in the solvent systems, there is none. The peak-to-peak linewidths of the ESR signals somewhat smaller for the PAni-CSA compound compared with the -DBSA, both are showing no temperature dependence. In the solvents (m-cresol, xylene) the respective compounds show the same magnetic susceptibility as in the free standing film, but now no temperature dependence at all. The ESR signal linewidths sharpened some-what, but not significantly.
The magnetic susceptibility of PAni-CSA in m-cresol is independent from the PAni concentration (concentration values between about 10-4 and 10-1 weight-%). The linewidth of the ESR signal sharpens with increasing concentration.
Also in polyblends with PMMA, the susceptibility is independent from PAni concentration, even though a conductivity breakthrough is observed at a critical concentration of about 1% due to dissipative structure formation (cf. [1b], [13], [14], [17]).
The results are virtually independent from the solvents used. Even so much different solvents like m-cresol, formic acid and xylene do not produce significantly different values.
Due to the stability of the metallic state in the solvents, the authors themselves are suspecting "that the 'solution' (quotation marks by Cao/Heeger) might actually consist of micro-aggregates ..." However, based on the observation, that the magnetic susceptibility in "solution" is constant over a wide concentration range they concluded that "micelle formation" seemed unreasonable, but without giving further reasoning.
The first important conclusion to be drawn from these results is: as there is no difference at all in the magnetic susceptibility between the PAni in solvents, as 100% pure film, or in a polyblend, there is no difference in the structure or size of the metallic unit, irrespective whether we call it "metallic island", or "metallic core", hence no change in the intra-chain electronic interactions when transferring PAni from the solid state into a solvent. This means, the solvent does not interact with PAni between the chains, where they form a metal. The coherence is not changed. The solvent does not change the chain state from being incorporated in a metallic primary particle to a truly dissolved, solvated chain (e.g. becoming a random coil). Here, we note an important contradiction between this clear conclusion and the conclusions made by the same group on the basis of light scattering (cf. par. 4.1.1 and ref. [29]).
In the light of the "dispersion hypothesis", these observations are easy to understand: PAni consists of primary particles of close to 10 nm in diameter containing a metallic core of about 8 - 9 nm size. These particles are unchanged as such, they are only separated from each other more or less completely by solvents or PMMA matrix, when transferred into other media than air. Solvents are acting principally in the same way as the polyblend matrix PMMA: PAni is dispersed in the medium and its surface is wetted by the solvent (or PMMA).
These conclusions are further supported by the fact, that so much different solvents as m-cresol or decalin do not change the magnetic susceptibility.
The only change one can observe is some sharpening of the linewidths, which one would have to assign to different interaction possibilities between the particles: if they cannot interact at all (too big of an interparticle distance at low concentrations), the linewidth is broader; if there are interparticle contacts (in the pure film, at higher concentrations in blends and the same in the dispersing solvents, then the linewidth gets somewhat sharper. The particle size itself does not change, only the flocculation degree. Note: also in solvents, dispersed particles are producing complex structures (flocculates), which may be elongated chains or complex 3-dimensional structures, cf [1b].
Also the presence of temperature dependence in the pure film in contrast to no T dependence in solvents can be explained by the same fact: in the pure film, one is measuring the temperature activated tunneling, which is absent when no tunneling can occur.
The work by Cao and Heeger is an important contribution to the understanding of polyaniline in solvents. It is a further clear hint to accepting such systems as being colloidal in nature, where the metallic properties do not change between the solid state and the dispersed state in solvents or blends (cf also [7]).
