Biomimetic synthesis of silica nanospheres depends on the aggregation and phase separation of polyamines in aqueous solution

Long-chain polyamines extracted from the highly siliceous cell walls of diatoms are known to precipitate silica nanospheres from aqueous, silicic-acid containing solutions at near-neutral pH in vitro.

The same is true for synthetic polyamines such as polyallylamine.

In the present contribution we show that the microscopic phase separation of polyallylamine in aqueous solution is strictly correlated with the silica precipitation activity of polyallylamine/silicic acid solutions.

Multivalent anions such as phosphate or sulfate efficiently induce this microscopic phase separation.

At higher anion concentrations, macroscopic phase separation occurs.

In contrast to the multivalent phosphate and sulfate ions, the monovalent chloride ions are much less efficient in polyallylamine aggregate formation.


Diatom cell walls are outstanding examples for nanoscale structured materials in nature.1

They consist of an amorphous composite material containing silica as well as certain biomolecules.

The latter are assumed to play a crucial role in diatom cell wall formation.

Examples for such biomolecules are the so-called silaffins isolated from the diatom Cylindrotheca fusiformis2 as well as the long-chain polyamines extracted from the cell walls of species such as Nitzschia angularis, Stephanopyxis turris, and others.3

Both these classes of molecules were also shown to precipitate silica from aqueous solutions containing silicic acid in vitro.

The precipitates formed by the long-chain polyamines from S. turris consist of nanospheres exhibiting a surprisingly narrow size distribution.

The particle diameter was shown to be strictly controlled by the concentration of phosphate ions added to the solutions.4

Initial NMR and dynamic light scattering experiments carried out on long-chain polyamines isolated from S. turris have shown that these molecules self-assemble in aqueous solutions.4

Silaffins from C. fusiformis behave very similarly.5

It is tempting to speculate that the aggregates (droplets) formed by the biomolecules in solution act as templates during cell wall formation.

A model based on phase separation processes could be developed which explains the pattern formation in diatom cell walls.6

Inspired by these observations, a number of synthetic amino acids, peptides, and polyamines were shown to precipitate silica from aqueous solutions as well.7–11

New opportunities for the so-called biomimetic silica synthesis are likely to arise from these discoveries.

One particularly interesting example is polyallylamine (PAA), a commercially available compound.

It exhibits a long-chain structure similar to the polyamines found in diatoms.

Polyallylamine was also shown to catalyze the polycondensation of silicic acid8 and to precipitate silica.9

The topic of the present paper is the detailed physico-chemical characterisation of PAA in aqueous solution especially with respect to the question whether or not an aggregation of the PAA molecules is a prerequisite for their silica-precipitating function as it is the case for the above-mentioned biomolecules extracted from diatoms.



Polyallylamine hydrochloride (PAA) was purchased from Aldrich.

The PAA molecule exhibits a repeated unit (ru) of the following structure: [–CH2CH(CH2NH2˙HCl)–]n with n ∼ 160 corresponding to a molecular weight of ca. 15 kDa.

NMR and dynamic light scattering experiments

NMR spectra were acquired on a DMX-500 spectrometer (Bruker, Karlsruhe, Germany) operating at 500 MHz 1H resonance frequency.

The calibration of the 31P NMR spectra was performed using H3PO4 (85%) as an external reference.

Dynamic light scattering (DLS) experiments were carried out on a MALVERN HPPS5001 nanosizer.

For the NMR and dynamic light scattering analyses, the polyallylamine hydrochloride was dissolved in millipore water.

A concentration of 1 mM PAA was chosen.

The desired concentration of orthophosphate ions (Pi) was obtained by the addition of the corresponding amount of sodium dihydrogen phosphate to the samples.

The final pH value was adjusted to 5.8 by the addition of NaOH.

Sulfate containing samples were prepared by the addition of sodium sulfate and chloride containing samples by the addition of sodium chloride.

Before the measurements, all solutions were allowed to equilibrate for at least 2 days.

In vitro precipitation of silica and SEM analysis

For the silica precipitation experiments, a freshly prepared solution of 1 M tetramethoxysilane in 1 mM HCl was incubated at 20 °C for 15 min and immediately used as a source of monosilicic/disilicic acid.12

A typical precipitation assay contained in 50 µl: 0.2 mM polyallylamine and phosphate (5 to 25 mM).

Silica formation was initiated by the addition of 2 µl silicic acid, prepared as described above.

After 12 min at 20 °C, precipitated silica was collected by centrifugation (4 min, 12.000 rpm).

The precipitate was washed twice with water, then suspended in water, applied to an aluminium sample holder, and air dried.

Silica precipitates were analyzed without sputter-coating with a LEO1530 field-emission scanning electron microscope equipped with energy dispersive X-ray analysis (EDXA, Oxford Instruments).

Colorimetric determination of silica precipitates was performed by the molybdenum blue method12 after dissolving precipitated silica in 20 µl 2 M NaOH (85 °C, 5 min).

Results and discussion

The influence of phosphate ions

Fig. 1 summarizes the results of silica precipitation experiments carried out for samples containing various amounts of phosphate.

The amount of silica precipitated from the PAA/silicic acid solution is plotted as a function of the phosphate concentration.

Silica precipitation takes place only above a threshold value of ca. 0.3 [Pi]/[ru].

Below, no silica precipitates at all indicating a crucial role of the phosphate ions.

It is interesting to note in this context that the PAA-induced silica precipitation experiments described by Patwardhan et al9. were carried out in a phosphate-containing buffer solution.

Furthermore, the particle diameter of the silica nanospheres precipitated by PAA is correlated with the phosphate concentration in complete analogy to the observations made for the long-chain polyamines extracted from S. turris.4

The maximum particle diameter of ca. 3 µm strongly exceeds the maximum value of about 1 µm obtained for the polyamines from S. turris.4

A possible explanation for this behaviour is the much smaller size of the polyamines from S. turris (1.55 kDa molecular weight).

On the other hand, the particles precipitated by the synthetic polyallylamine exhibit a much broader size distribution than the nanospheres obtained by using polyamines from S. turris.

In order to understand this behaviour, we have studied aqueous PAA solutions of various phosphate concentrations by dynamic light scattering (see Fig. 2).

In the absence of phosphate, aggregates of an average diameter of about 1 nm could be detected.

For a single PAA molecule, one would expect a diameter of the same order of magnitude.

This leads to the conclusion that the PAA molecules exist as single molecules (monomers) or small clusters in the absence of phosphate ions.

Addition of phosphate results in a continuous but slow increase of the aggregate diameter up to about 5 nm at 0.25 [Pi]/[ru].

Note, that this phosphate concentration is close to to the above-mentioned threshold value of ca. 0.3 [Pi]/[ru] necessary for silica precipitation.

A rapid increase of the aggregate diameter is observed beyond this threshold value.

At 0.31 [Pi]/[ru], aggregates of about 120 nm diameter could be detected.

Visual inspection shows that the samples become cloudy at this concentration which confirms the existence of larger aggregates (see Fig. 3).

If the phosphate concentration is further increased, the samples remain cloudy up to about 0.5 [Pi]/[ru].

DLS measurements show that the aggregate diameter increases up to almost 600 nm at 0.44 [Pi]/[ru] while a decreased value could be found at 0.5 [Pi]/[ru].

This decrease can be explained as follows: 1H NMR studies show that more than 2/3 of the polyamine has disappeared from the part of the sample inserted into the NMR tube.

Obviously, this amount of polyamine already sediments at the bottom of the sample without forming a visible separate phase.

The decreased concentration of polyamine in the top part of the sample must be responsible for the decreasing diameter of the aggregates.

At 0.63 [Pi]/[ru] and higher concentrations, the samples become clear after equilibration but exhibit a macroscopic phase separation: A separate phase containing the polyamine is formed at the bottom of the tube.

The aqueous phase on top does no longer contain a detectable amount of polyamine aggregates.

The observed aggregation and phase separation can be explained by the attractive electrostatic interactions between the positively charged polyamine molecules and the negatively charged phosphate ions13 and/or the formation of hydrogen bonds.

Comparison of Figs. 1 and 2 shows that the formation of the described large polyamine aggregates, i.e., microscopic phase separation is correlated with the capability of the system to precipitate silica from a PAA/silicic acid solution.

If, however, macroscopic phase separation takes place, i.e. for phosphate concentrations significantly exceeding 0.5 [Pi]/[ru], the solution looses its silica precipitation activity.

That means, microscopic phase separation is a necessary prerequisite for silica precipitation by the polyamine solution.

In order to characterize the physico-chemical properties of this microscopically phase separated solution in more detail, extended 31P and 1H NMR spectroscopic studies were carried out.

For all samples, only one single 31P signal is observed.

The chemical shift, δ, as well as the linewidth, Δν1/2, (full-width-at-half-maximum) of this signal, however, change.

The chemical shift is plotted as a function of phosphate concentration in Fig. 4 (top).

It remains almost constant up to about 0.3 [Pi]/[ru].

Beyond this phosphate concentration, δ continuously decreases.

At phosphate concentrations higher than 0.6 [Pi]/[ru], that means in the range of macroscopic phase separation, the chemical shift is again nearly constant and amounts to ca. 1.2 ppm.

As the solution in the NMR tube is almost completely free of polyamine at high phosphate concentrations, the latter chemical shift value represents free phosphate ions.

From the observation that only a single signal is observed for all concentrations it is concluded that phosphate ions interacting with the polyamine rapidly exchange with free phosphate ions (on the time scale of NMR spectroscopy).14,15

The increased and constant chemical shift at low phosphate concentrations indicates that phosphate is preferentially bound to the polyamine molecules up to about 0.3 [Pi]/[ru].

Above this concentration, an increasing amount of free phosphate exists which results in the described decrease of chemical shift.

The linewidth of the 31P NMR signals (Fig. 4, bottom) also exhibits an interesting behaviour.

From an initial value of about 4 Hz it increases up to about 12 Hz for 0.4 [Pi]/[ru].

At higher concentrations, Δν1/2 decreases until a constant value of 1.6 Hz is found for the aqueous part of the macroscopically phase-separated samples.

This behaviour can be explained as follows: Due to the rapid exchange between phosphate bound to the polyamine aggregates and free phosphate, an average linewidth is observed.14,15

It is known from the chemical shift analysis that the phosphate ions are preferentially bound to the polyamine aggregates at low phosphate concentrations.

The linewidth, therefore, corresponds to the linewidth in the polyamine-bound state for low phosphate concentrations.

Since an increasing aggregate diameter (see Fig. 2) results in a lower correlation time of rotational diffusion, an increasing linewidth is expected for increasing phosphate concentrations.

This is in agreement with the experimentally observed behaviour up to 0.4 [Pi]/[ru] (Fig. 4, bottom).

For concentrations above 0.4 [Pi]/[ru], the influence of the increasing amount of free phosphate is observed.

Since the linewidth of free phosphate is smaller than that of polyamine-bound phosphate, the measured average linewidth then decreases with increasing phosphate concentration.

After macroscopic phase separation took place, only free phosphate exists in the NMR samples giving rise to the small and constant linewidth of 1.6 Hz observed for high phosphate concentrations.

The influence of sulfate and chloride ions

The influence of sulfate and chloride ions upon aqueous solutions of PAA was also investigated.

The behaviour of sulfate, another multivalent anion, turned out be very similar compared to phosphate: The solutions become cloudy at about 0.3 [sulfate]/[ru] and exhibit macroscopic phase separation above 0.5 [sulfate]/[ru].

This behaviour shows that the ability to phase-separate aqueous PAA solutions must not be considered to be a specific property of phosphate ions.

It is striking that both, phosphate and sulfate are multivalent anions.

The question arises whether or not monovalent anions such as chloride can also be used to induce the described effects.

Fig. 5 shows the diameter of the polyamine aggregates determined by dynamic light scattering as a function of chloride ion concentration.

Obviously, aggregates of increasing diameter are observed at increasing chloride concentrations.

However, even for the highest chloride concentration (ca. 20 [Cl]/[ru]), the obtained aggregate diameter of ca. 9 nm is almost two orders of magnitude smaller than the maximum aggregate diameter of ca. 600 nm observed for a phosphate concentration of only 0.44 [Pi]/[ru].

It is furthermore important to note that the macroscopic phase separation found for phosphate and sulfate does not occur for chloride ions even at about 20 [Cl]/[ru].

That means, the monovalent chloride ions are much less efficient in attracting the PAA molecules compared to the multivalent phosphate and sulfate ions.

In agreement with these observations, no silica is precipitated at all from the chloride-containing PAA solutions even at chloride concentrations up to ca. 20 [Cl]/[ru] which is far beyond the physiologically relevant level.

In summary, it can be stated that microscopic phase separation is necessary for the PAA-induced silica precipitation from silicic-acid containing aqueous solutions.

This microscopic phase separation can be induced by the addition of multivalent anions such as phosphate or sulfate to the aqueous solutions.

Polysilicic acid molecules may be adsorbed on and/or dissolved in the polyamine microdroplets thereby forming a coacervate which hardens by silica formation.