Origin of precious opal revisited: Possible quick formation of precious opal


 It has been well known that a sedimentary precious opal is composed of closely packed uni-size silica spheres with voids filled by an air or water, and that an interference among reflections from the boundaries of those spheres and filler yields play-of-colour (iridescence). So, occurrence of a play-of-colour means occurrence of natural selection in size of spheres, or suppression of further nucleation after initial outburst of spheres, during its formation process. We had been exploring the possibility if we can regard a Stöber process as an analogue of the formation process of precious opal. The key is the reason why variation in size is rarely found on both precious opal and Stöber colloid. To give a clue, we examined the internal structure of Stöber particles and how those particles were formed at very initial stage of the process. The answers for evenness in shape and size are a quick supersaturation of reactive silica species, consecutive formation of large and loose polymers by fast dehydration, and their quick aggregation as the initial burst of silica spheres in highly diffusive medium. These can be achieved in nature by quick but continuous decrease in temperature on “basic” (high pH) geothermal hot water moving upward through cracks in rocks. Sedimentary precious opal can thus be formed when such naturally occurring colloid is filtered by a permeable bed.


Sedimentary precious opal and a Stöber colloid
Amorphous silica spheres of desired size can easily be prepared, and play-of-colour can also be reproduced (Fig. 1a) in labs with a Stöber process, which utilizes hydrolysis of Sialkoxide and condensation of thus yielded silicic acid by dehydration in one reaction medium. This process has clear similarities with that of precious opal; both are segregation of solid from a liquid, both utilize coagulation and thus dehydration of silicic acid, and final products are monodispersed amorphous silica spheres. Their products also have similarities.
Particulate structure for each sphere has been reported on sedimentary precious opals by several authors [5][6][7] . Silica spheres produced with the Stöber process were porous 8 , and aggregation of solid primary particles (aggregation-growth model) has been proposed as their formation process 9 .

Structure of Stöber particle
Monodisperse silica spheres of virtually the same size were prepared with the method reported previously 10 . In the present experiments we had set concentration of Si-alkoxide (tetraethyl orthosilicate: TEOS) at 0.22 M to obtain larger spheres in shorter time.
Mixed chemicals had been gently stirred over sampling period (t < 960 min). However, continuous stirring was ample to attain mono-disperse colloid. We could attain such colloid even with 0.5 min of initial stirring to accomplish homogeneity of the reaction medium (Fig.   1a). The reaction medium started to be translucent blue at 3 min and became turbid at 8 min after mixing of chemicals. By settling down the medium for some weeks, or utilizing centrifugal separation, we could get a precipitate of silica spheres. Wet precipitates showed typically green and orange play-of-colour, and the colour rarely retained when dried. Dried white pellets of precipitates (inset of Fig. 1a) were easily broken by hand, indicating that the pellets were composed of loosely bound silica spheres stacked in close-packing manner with no solid filler. This was later confirmed under field emission-type scanning electron microscope (FE-SEM) (Fig. 1a). Sizes and morphologies of these spherical particles, some Surfaces of spheres sampled after completing their growths (t ≈ 120 min: Fig. 3) were smooth even under high resolution, while they showed sandy texture under higher acceleration voltage (e.g., t = 240 min, Fig. 2d). This reflected remnant of inhomogeneity inside the spherical particles. As one of the authors (H.O.) had reported previously on simultaneously formed thin film, this accretion was well dehydrated and thus hard even before the drying process 10 . Nonetheless, as-prepared particles sampled at early stage (t ≤ 5 min) had kept faint translucency for electron beam (Fig. 2c) and those after their growth (t = 480 min) had not been hard, inner part in particular, enough to be fractured (Fig. 2e). We could find a fracture surface through the centre of the sphere on a well-dried (kept for four months in desiccator) one (Fig. 2f). Interesting feature was found on this fracture surface; central part of the spherical particle was an agglomeration of finer particles, approximately 20 nm in diameter, instead of a uniform one expected for monomer-addition growth 11 based on a classic "nucleation-and-growth" model 12 . This "core" part was around 200 nm in diameter, while boundary of the core and mantle was ambiguous. Indeed, the texture looked as if the size of accretion got finer with reaction time after aggregation of "primary particles" which made the core part. Similar but multi-step aggregations had been proposed for a Stöber colloid 13 .
However, we could not find any sign of multi-step aggregation and our silica spheres were simple "secondary particles" composed of primary ones. This growth process, namely growth by accumulation of primary particles which reduce their sizes with time, had already been proposed for the growth of the silica thin film 10 . Present observation disclosed at the first time that silica sphere in the Stöber process was initially formed as an aggregation of unisized primary particles and grown by accumulation of later-formed, and smaller, primary particles.

Growth kinetics of silica spheres in Stöber process
Prior to an exploration of the identity of the "primary particles", we explore the growth process of these spheres with a little kinetics. Change in size of silica spheres in our Stöber colloid was linear in logarithmic time until 20 min, then gradually reduces its rate, and finally stopped at no later than 120 min (Fig. 3). Steady growth of the spheres at the early stage, from 2 to 20 min, can be explained by a simple diffusion-limited model 10,14 no matter what reactive silica species, monomers, polymers or nanosized primary particles, is accumulating on the surface. When a growth is diffusion-limited, in other words limited by a collision frequency between a reactive silica species and a particle, its growth rate depends on an amount of reactive silica species in the medium. In the present case hydrolysis of TEOS itself is a slow-rate step 15 and the rate of dehydration of reactive species is faster, too fast to allow the reactive species keep dissolving in the medium, than the rate of hydrolysis of TEOS 16 .
Linear fashion in particle growth therefore means continuous supply of reactive silica species by hydrolysis of TEOS with a fixed rate (this case amount of unhydrolyzed TEOS is inverse proportional to a duration time). The cause which slowed the growth down over 20 ~ 120 min is unclear but back reaction on hydrolysis is likely. Depression of TEOS in the medium finally ended the growth at no later than 120 min.

Identity of primary particles, and formation and growth of secondary particles
Particles sampled at 1 and 2 mins showed notable difference in their morphologies. In general, appearance of bright fringe on an outline of objects under SEM means presence of an edge or vertical face (the so-called edge effect). From this viewpoint, the particle sampled at 2 min (Fig. 2b) had certain vertical dimension, while those at 1 min (Fig. 2a) did not. The particles sampled at 1 min looked as nearly translucent white spots on the substrate, like a dried jellyfish. This shape indicated that density of the particles sampled at 1 min was quite low as a blob, namely loosely bound homogeneous spherical large polymers, and they shrunk anisotropically when dried.
The horizontal sizes of these dried polymers (≈ 35 nm) were comparable to the diameter of the particles sampled at 2 min (d ≈ 40 nm), and both of these were far larger than the "primary particles" (d ≈ 20 nm) found at the core part of the later-sampled one (Fig. 2f).
Their sizes indicated that each of the particles sampled at 2 min was not a further grown and dehydrated form of each large polymer sampled at 1 min but an aggregation of those, that is to say "secondary particles". This means that these silica spheres were initially formed as an assemblage of blobs which correspond to primary particles referred to above, and dehydrated simultaneously with their growth. This explains the reason why the core part was softer, namely less dehydrated, than the mantle on the as-prepared spherical particle (Fig. 2e).
Furthermore, this can explain also the reason why particle sizes had not been determined with a dynamic light scattering during the induction period (3 ~ 8 min in our experiments) and the smallest sizes detected by the light scattering had been already within 50-70% of their final values 15 . They had simply not been condensed enough to be detected by a light scattering in the reaction medium. We have no way to estimate actual size and number of the large polymers when they started to aggregate. However, their population should be very high otherwise no aggregation had occurred and they had grown individually.
Since the size of the core part (d ≈ 200 nm) is larger than the size of particles sampled at 5 min, these spherical particles were grown basically by adhesion of large polymers for some more minutes. However, homogeneity in size of the secondary particles was already attained at 5 min: Averaged particle diameter was 168 (7)  For high diffusivity of the reaction medium, later-formed polymers only collided onto secondary particles nearby and they seldom formed a new aggregation by themselves. While later-formed polymers could grow in size comparable to the initially formed ones for some more minutes, their sizes became smaller with time with reducing production of reactive silica species. Lack of distinct boundary between core and mantle, and improvement on evenness of their outer shapes too, are due to this gradual change in size of polymers.

Keys
Particulate structure of our Stöber particles suggests that the formation process of our artificial precious opal is virtually identical with, or at least similar to, that of a natural Population of later-formed polymers did not achieve high enough to form secondary particles by themselves and they accumulated onto already formed secondary particles. From above, the reaction system should obey requirements as follows. Firstly, fast dehydration of reactive species in a basic medium is essential for formation of large polymers (primary particles).
Population density of these large polymers should be raised quickly to make yielded polymers unavoidable to collide to each other and form an aggregation (secondary particles).
Their agglomeration, and accumulation of later formed ones, also should occur quickly prior to their solidification. To attain these, large amount of reactive species should be supplied in the system in a short period and the medium should have low viscosity to attain high diffusivity. Entire process should be finished in rather short period prior to precipitation of thus formed secondary particles in low viscosity medium. A reaction field should have certain dimensions for fractionation to accomplish uniformity in sizes of secondary particles.
Balances among speeds of dehydration (pH), aggregation (viscosity) and supersaturation (here this corresponds to a rate of hydrolysis of TEOS but in general this corresponds to change in solubility of silica species in a medium) determines whether precipitate is a particles of a few tens of nanometers, spherical particles with a particulate structure, or a bulk body.

Structure of sedimentary precious opal
Natural precious opal (Fig. 1b,c) gave us some insights on its formation process. While a particulate structure of each sphere was not visible in our observation since no chemical etching was applied, packing of spheres and inter-particle voids were apparent under FE-SEM (Fig. 1b). The bulk body was fractured along a two-dimensional array of six-coordinate voids (centre of regular octahedron with spheres on its apices). Observed aspect ratio (≑ 1) of the square array of spheres indicated cubic closest packing of those as had been analyzed in ref. 18 in detail. This also guarantees that the centres of these spheres were located virtually on the fracture surface. Yet another kind of voids, smaller four-coordinated ones, had not be seen under FE-SEM. Indeed, an amount of overlaps among connected spheres, namely difference among distances from the centre of a sphere to its surface and a half of that between centres of connected spheres was enough to fill tetrahedral voids around the spheres. Large curvature radius of the surface of each particle indicated occurrence of post-stack growth rather than shrinkage of void by compression.
Untextured slabs (veins) were found as insertions in the stack. The walls of such cracks were composed of non-spherical (deformed) particles with no sequence from the body of the stack (Fig. 1c). Presence of these slabs indicates occurrence of post-stack shrinkage of spherical particles and presence of continuous "flow" of a water which was supersaturated a little with a reactive silica species. This flow caused a post-stack growth of spheres until all tetrahedral voids were filled and the body lost its permeability. Once this was accomplished each of octahedral voids was isolated from the outside of the stack and left unfilled. This flow carried also soft spherical particles from the upstream side, which were adhered onto the wall of cracks. Here we avoid to get deeply involved into forms of dissolved silica species and their chemistry at actual geological sites, instead we give a simple and qualitative model which can yield a sedimentary precious opal.
It is needless to say that a solubility of silica in water is fairly low at ambient temperature, but the solubility increases with increasing temperature particularly under basic condition 19 . It is interesting to note that this enhancement in solubility is maximized at 225°C; at this temperature the solubility at pH = 10.2 is ten times greater than that at pH = 8.0 (ref. 19 ). This means that a steep increase in an amount of reactive silica species and their fast polymerization can happen in basic geothermal hot water moving in a low temperature environment. When hot water in aquifer is covered by an impermeable strata, and once the strata is fractured by a diastropic event, this hot water will be released for high pressure and moves upward through the cracks having certain dimensions. This motion will associate decrease in temperature of the flowing liquid and induces supersaturation of silicic acid.
When change in temperature is steep enough, a lot of large polymers will quickly be generated and spherical secondary particles grow in the flow. This naturally occurring colloid will be filtered by a permeable host rock and will fill the very end of the path by a stack of particles. Flowing water soak through the stack and the stack grows on the upstream side.
Simultaneously, remnant reactive silica species in the soak water fill tetrahedral voids in the stack and make the body hard. Filtered water soak into the permeable host rock too and may make the host rock harder.
The remaining issue is a possibility where nature attain an appropriate balances among speeds of dehydration, aggregation and supersaturation in subterranean water. It is intuitively understood that this seldom occurs in nature, and this makes precious opal precious.  The suspensions sampled at early stage (t ≤ 5 min) contain unignorable amount of reactive silica species and unhydrolized TEOS. These remnants covered the dried surface of the samples like a glue with objects beneath. To deal with this, these samples were rinsed immediately after spreading by dropping dehydrated ethanol on a tilted surface. Surface of the Si-substrate was roughen with a polishing paper (4000 grid; 3M) and then diamond paste (1 µmφ) for these samples so that uneven surface hold soft and tiny particles even on rinsing process. These substrates were used on experiments #11 and #12. In these experiments, the reaction vessel was placed in an incubator but its top was left open for sampling purpose and stirred by a magnetic stirrer over sampling period (3 and 5 min on experiment #11, 1, 2, 3 and 5 min on experiment #12). Particle diameters were measured on particles with no contact to its neighbours (experiment #11: n = 3 and 26 for 3 and 5 min, experiment #12: n = 6, 4, 4, 4 for 1, 2, 3 and 5 min).
In experiment #14 120 ml of the mixture of chemicals was stirred for 0.5 min and poured into a cylinder of 35 mmφ and 200 mm in hight. Top of the cylinder was covered with a polyethylene film and the cylinder was kept 1 month at 20°C to precipitate the particles. The precipitate was dedicated to examine variations in sizes of the spheres from the bottom to the top.
Mechanical grinding was applied also on well-dried silica spheres from preliminary experiment. An aliquot of suspension was taken after 240 min stirring and its precipitate was dried as-prepared for 4 months in desiccator. This precipitate was manually ground in an alumina mortar with dehydrated ethanol as a dispersion medium. metal plating was applied except the stack of silica spheres on which serious charge-up was unavoidable without plating (Fig. 1a). Au sputtering was applied on the stack with current of 50 mA for 1min. Thickness of Au plating was determined as 15 nm based on a difference in averaged diameters of silica spheres from same colloid with and without this plating.
Specimen of natural precious opal was taken from a semi-antique accessory of circa 1970's. Detail of locality is unknown. Fragments from this specimen were used for observations without any treatment.