Nucleation and Particle Growth in Precipitates

A chemist adds barium chloride to a hot, stirred, dilute sulfate solution. A dense, sand-like precipitate settles within minutes. A student in the next hood adds the same reagent differently. They work quickly, cold, and unstirred, with a concentrated solution. The result is a gelatinous, colloidal mess that clogs the filter paper for an hour. Same chemical reaction, same stoichiometry, wildly different outcomes. The difference is not chemistry in the sense of bonds and charges. It is kinetics instead. Specifically, it is the competition between two rival processes. Both consume the same pool of dissolved ions: nucleation, the birth of new solid particles, and particle growth, the enlargement of particles that already exist. This competition sits at the heart of gravimetric analysis. Precipitation Gravimetry depends on collecting all of an analyte as a solid. That solid must be filterable, washable, dryable, and weighable without loss. A precipitate made of a few large, well-formed crystals is a gravimetric chemist's dream. It filters fast and has low surface area. This means it has low adsorbed-impurity load. It also approaches the ideal described in Properties of the Ideal Precipitate. A precipitate made of billions of sub-micron particles is a nightmare. It passes through filter pores. It drags down solvent and foreign ions by adsorption. It can also behave as a colloid that never settles at all. Both particle populations can form from identical moles of the same salt. What differs is how the supersaturated solution partitions its excess solute. It must divide that solute between forming new nuclei and feeding existing crystals. The big picture is a race with a threshold. A precipitating solution becomes supersaturated the instant reagent is added. This happens because the ionic product $Q$ momentarily exceeds the solubility product $K_{sp}$ . Supersaturation is an unstable, energy-rich state. The system relieves it by depositing solid. But depositing solid onto an existing crystal face is kinetically easy and thermodynamically cheap. Creating a brand-new nucleus is different. It requires paying an energetic penalty to build a surface out of nothing. This penalty imposes a barrier. The height of that barrier depends steeply on how supersaturated the solution is. At low supersaturation the barrier is high, and nucleation is rare. Dissolved ions overwhelmingly join the few nuclei that do form. This produces large crystals. At high supersaturation the barrier collapses, and nuclei appear everywhere at once. The ions get divided among an enormous number of tiny particles. This article works through the physical picture of that barrier. It derives the classical nucleation-rate expression and a growth-rate expression. It then shows how their relative steepness in supersaturation matters. This steepness explains why analytical chemists dilute, heat, stir slowly, add reagent dropwise, and digest precipitates before filtering. It sets up the quantitative relative-supersaturation criterion of Relative Supersaturation and von Weimarn, without yet deriving it in full. It also explains why fast-formed precipitates trap impurities. That trapping is the subject of Coprecipitation and Contamination.

Key takeaways

Precipitation always starts from a supersaturated solution, where the ionic product $Q$ exceeds $K_{sp}$ ; this excess free energy is the driving force for both nucleation and growth.
Nucleation must build new solid–liquid interface from nothing, so it faces a free-energy barrier $\Delta G^{\ddagger}$ that scales as $1/(\ln S)^2$ , where $S = Q/K_{sp}$ is the supersaturation ratio.
Growth onto an existing crystal face has no such barrier; it proceeds by diffusion of ions to the surface and orderly incorporation into the lattice, and its rate rises only linearly (or near-linearly) with supersaturation.
Because the nucleation rate depends on $S$ through a much steeper (exponential-of-a-power) function than the growth rate, small increases in $S$ tip the balance sharply toward more, smaller particles.
Practical levers — dilute reagents, slow addition, heat, digestion, controlled pH-driven precipitation — all work by keeping $S$ low long enough for growth to dominate over nucleation.
Particle size distribution set during precipitation cannot be fully repaired later; digestion narrows and coarsens crystals via Ostwald ripening but cannot undo massive over-nucleation.
Key idea: the ratio of nucleation rate to growth rate — both functions of the same supersaturation — determines whether a precipitate ends up as few large crystals or as a swarm of fine particles.

Where this lives today

Controlling the nucleation-versus-growth balance is not a museum piece confined to undergraduate gravimetric labs — it is a live engineering problem across chemistry and materials science. In pharmaceutical manufacturing, active ingredient crystallization is deliberately seeded and cooled at controlled rates specifically to suppress spontaneous nucleation and grow uniform crystals with reproducible dissolution behavior; polymorph control (which crystal form nucleates) is a major regulatory and patent battleground, and it hinges on exactly the $(\ln S)^{-2}$ barrier sensitivity derived above. In nanomaterials synthesis, the LaMer mechanism for monodisperse nanoparticle production deliberately engineers a short, sharp nucleation burst (a brief spike above the nucleation threshold) followed by a long, slow growth phase at sub-nucleation supersaturation, producing gold, silver, and quantum-dot nanoparticles with tight size distributions — a direct, intentional exploitation of the same rate competition, but run in reverse from gravimetric practice (chemists there want small particles, uniformly sized). In water treatment and scale prevention, engineers deliberately keep supersaturation of $\mathrm{CaCO_3}$ and $\mathrm{CaSO_4}$ low in pipes and boilers to prevent nucleation-driven scale formation, using antiscalants that raise the effective nucleation barrier. In biomineralization research, understanding how organisms nucleate calcium phosphate (bone) or calcium carbonate (shells) at controlled sites, avoiding runaway homogeneous nucleation elsewhere in the organism, draws on identical thermodynamic bookkeeping. And within analytical chemistry itself, the technique of homogeneous precipitation — generating the precipitating anion or cation slowly throughout the solution via a controlled chemical reaction (such as the hydrolysis of urea to slowly raise pH, precipitating hydroxides gradually) — is a direct engineering response to everything derived here: it keeps local $S$ low everywhere at once, sidestepping the burette-tip hot-spot problem and yielding the coarse, pure, easily filtered crystals gravimetric analysis depends on.

Ask the tutor: How does the von Weimarn ratio quantify "relative supersaturation" in a way that lets you predict particle size before you even start precipitating?

The physical picture — thorough prose

Consider dissolved $\mathrm{Ba^{2+}}$ and $\mathrm{SO_4^{2-}}$ ions in solution just after mixing, at a total ionic product $Q = [\mathrm{Ba^{2+}}][\mathrm{SO_4^{2-}}]$ that exceeds $K_{sp} = 1.1\times10^{-10}$ at 25 °C. The solution is thermodynamically unstable with respect to solid $\mathrm{BaSO_4}$ : precipitation is downhill in free energy. But "downhill" does not mean "instantaneous." Two competing routes exist for the system to relax toward equilibrium. The first route is to assemble a tiny cluster of ions — perhaps only 10–100 ion pairs — into a crystalline embryo suspended in solution. This costs energy, because every embryo has a surface, and creating solid–liquid interface has a positive surface free energy per unit area (an interfacial tension $\gamma$ , in J m⁻²). For very small embryos, the unfavorable surface term dominates over the favorable bulk term (crystallizing releases free energy per unit volume), so tiny embryos are less stable than the dissolved ions and tend to redissolve. Only once an embryo grows past a critical size $r^{*}$ does the bulk term win and further growth becomes spontaneous. Reaching $r^{*}$ requires climbing over a free-energy hill $\Delta G^{\ddagger}$ — this is homogeneous nucleation, and it is a genuine kinetic bottleneck, not merely a formality. The second route is for ions to attach to a crystal face that already exists — whether a stray dust particle, the vessel wall, or (once the first nuclei have formed) other crystals of the same precipitate. Here there is no new surface to create from scratch; an ion arriving at a kink site on an existing lattice is simply extending a structure that is already thermodynamically favorable. This route, crystal growth, has no comparable nucleation barrier — its rate is governed by how fast ions can diffuse to the surface and be incorporated, which is a much gentler function of supersaturation.

Figure

Free energy of a growing embryo versus its radius, showing the nucleation barrier ΔG‡ at critical radius r*

Once a handful of stable nuclei exist, every subsequently arriving ion faces the same choice again: join an existing crystal (growth), or gather with other ions to start a brand-new nucleus (further nucleation). The macroscopic outcome — few large crystals versus many small ones — is decided by which process wins this ongoing race as the precipitation proceeds and as $S$ falls back toward 1.

Figure

Two precipitation trajectories: low supersaturation (few, large crystals) vs high supersaturation (many, small crystals)

The core ideas — governing equations

The driving force behind both processes is expressed through the relative supersaturation ratio,

S = \frac{Q}{K_{sp}}

where:

$Q$ is the instantaneous ionic product of the precipitating ions raised to their stoichiometric powers (units matching $K_{sp}$ ),
$K_{sp}$ is the thermodynamic solubility product at the working temperature,
$S=1$ at saturation; $S>1$ signals supersaturation, the regime in which precipitation can occur.

Reading: $S$ is dimensionless and measures how far the system sits above equilibrium; it is the single control variable that both nucleation rate and growth rate respond to, though they respond with very different sensitivity. The classical nucleation rate (number of stable nuclei formed per unit volume per unit time) follows an Arrhenius-type law with the barrier height built in:

J = A \exp\!\left(-\frac{16\pi \gamma^{3} v^{2}}{3 k_B^{3} T^{3} (\ln S)^{2}}\right)

where:

$J$ is the nucleation rate (nuclei m⁻³ s⁻¹),
$A$ is a pre-exponential kinetic factor (collision/attachment frequency, roughly $10^{25}$ – $10^{35}$ m⁻³ s⁻¹ depending on the system),
$\gamma$ is the solid–liquid interfacial tension (J m⁻²),
$v$ is the volume occupied by one formula unit in the solid (m³),
$k_B$ is Boltzmann's constant, $T$ the absolute temperature,
$\ln S$ is the natural log of the supersaturation ratio (dimensionless), appearing squared in the denominator.

Reading: the exponent is negative and contains $(\ln S)^2$ in the denominator, so as $S \to 1^{+}$ ( $\ln S \to 0^{+}$ ), the exponent diverges to $-\infty$ and $J \to 0$ : at low supersaturation nucleation is essentially shut off. As $S$ increases, $\ln S$ grows, the denominator inflates, the exponent becomes less negative, and $J$ rises with extraordinary steepness — often over many orders of magnitude for a doubling of $S$ . The growth rate of an existing crystal face, in the diffusion-controlled or surface-reaction-controlled regime relevant to most precipitation gravimetry, is far gentler:

R_{\text{growth}} = k_g \,(S - 1)^{n}

where:

$R_{\text{growth}}$ is the linear growth rate of a crystal face (m s⁻¹) or equivalently a mass deposition rate,
$k_g$ is a growth rate constant folding in diffusion coefficient, surface area, and temperature,
$n$ is a small positive order, typically $1$ (diffusion-limited) or $2$ (surface-integration-limited) — never the extreme exponential sensitivity seen in $J$ .

Reading: growth is at most a low-order polynomial function of supersaturation; it never develops the catastrophic threshold behavior of nucleation. This asymmetry — one process exponentially explosive in $(\ln S)^{-2}$ , the other merely linear or quadratic in $(S-1)$ — is the entire reason supersaturation control governs particle size.

Derivation AThe nucleation free-energy barrier from surface and bulk terms

Goal. Derive the critical radius $r^{*}$ and barrier height $\Delta G^{\ddagger}$ for a spherical embryo forming from a supersaturated solution, and show how $\Delta G^{\ddagger}$ depends on $S$ .

Step 1 — write the total free energy of forming a spherical embryo of radius $r$ . Forming an embryo costs surface free energy (positive, opposing formation) and releases bulk free energy (negative, favoring formation, since the solid is more stable than the supersaturated solution):

\Delta G(r) = 4\pi r^{2}\gamma - \frac{4}{3}\pi r^{3}\,\Delta G_v

Here $\gamma>0$ is the interfacial tension and $\Delta G_v>0$ is the magnitude of the bulk free-energy change per unit volume of solid formed (a driving force, so it enters with a minus sign in the total).

Step 2 — express $\Delta G_v$ in terms of supersaturation. The molar free energy released on precipitation is $-RT\ln S$ per mole of formula units transferred from solution to solid (this follows from the chemical potential difference between the supersaturated solution and the saturated one, $\mu_{\text{soln}} - \mu_{\text{solid}} = k_BT\ln S$ per ion pair). Dividing by the molar volume $V_m = N_A v$ converts this to a volumetric driving force:

\Delta G_v = \frac{RT \ln S}{V_m} = \frac{k_B T \ln S}{v}

where $v = V_m/N_A$ is the volume per formula unit in the solid. Note $\Delta G_v > 0$ whenever $S>1$ , as required.

Step 3 — find the critical radius by maximizing $\Delta G(r)$ . Differentiate and set to zero:

\frac{d\Delta G}{dr} = 8\pi r\gamma - 4\pi r^{2}\Delta G_v = 0

Solving for the nonzero root:

r^{*} = \frac{2\gamma}{\Delta G_v} = \frac{2\gamma v}{k_B T \ln S}

This is the Gibbs–Thomson / Kelvin radius: embryos smaller than $r^{*}$ shrink back into solution (net decrease in $\Delta G$ on dissolving), embryos larger than $r^{*}$ grow spontaneously (net decrease in $\Delta G$ on growing). Note $r^{*}$ shrinks as $S$ rises — at high supersaturation even very small embryos are already past the critical size, so nucleation becomes easy.

Step 4 — substitute $r^{*}$ back into $\Delta G(r)$ to get the barrier height.

\Delta G^{\ddagger} = \Delta G(r^{*}) = 4\pi (r^{*})^{2}\gamma - \frac{4}{3}\pi (r^{*})^{3}\Delta G_v

Substituting $r^{*} = 2\gamma/\Delta G_v$ and simplifying (the two terms combine to one-third of the surface term evaluated at $r^{*}$ ):

\Delta G^{\ddagger} = \frac{16\pi \gamma^{3}}{3\,\Delta G_v^{2}} = \frac{16\pi\gamma^{3}v^{2}}{3\,k_B^{2}T^{2}(\ln S)^{2}}

This confirms the $(\ln S)^{-2}$ dependence quoted in the rate law: the barrier is inversely proportional to the square of $\ln S$ , so it collapses rapidly as supersaturation climbs, and diverges as $S\to1$ .

Step 5 — insert the Boltzmann factor to get the rate. Treating nucleation as an activated process with attempt frequency $A$ ,

J = A\exp\!\left(-\frac{\Delta G^{\ddagger}}{k_BT}\right) = A\exp\!\left(-\frac{16\pi\gamma^{3}v^{2}}{3k_B^{3}T^{3}(\ln S)^{2}}\right)

recovering the nucleation-rate law used above, with one extra factor of $k_BT$ absorbed from converting $\Delta G^{\ddagger}$ (in energy units) into the exponent (dimensionless).

Go deeper · Derivation B · 1 minWhy nucleation outpaces growth so much faster than S grows — a comparative sensitivity argument

Derivation BWhy nucleation outpaces growth so much faster than S grows — a comparative sensitivity argument

Goal. Show quantitatively that a modest increase in $S$ boosts $J$ by many orders of magnitude while boosting $R_{\text{growth}}$ only mildly, using the logarithmic sensitivity (elasticity) of each rate to $S$ .

Step 1 — define the logarithmic sensitivity. For any rate $R(S)$ , the fractional response to a fractional change in $S$ is captured by the elasticity

\varepsilon_R = \frac{d\ln R}{d\ln S} = \frac{S}{R}\frac{dR}{dS}

A large $\varepsilon_R$ means the rate is extremely sensitive to supersaturation; $\varepsilon_R \approx n$ (a small constant) means mild, near-linear sensitivity.

Step 2 — compute $\varepsilon_J$ for the nucleation rate. Let $B = \dfrac{16\pi\gamma^{3}v^{2}}{3k_B^{3}T^{3}}$ so that $J = A\exp(-B/(\ln S)^{2})$ . Then

\ln J = \ln A - \frac{B}{(\ln S)^{2}}

Differentiating with respect to $\ln S$ (treating $\ln S$ as the variable):

\frac{d\ln J}{d\ln S} = \frac{2B}{(\ln S)^{3}}

This is $\varepsilon_J$ . Because $B$ is typically enormous (see the worked example below, where $B \sim 10^{2}$ – $10^{3}$ for real $\gamma, v, T$ ) and $\ln S$ is often a modest number like $0.5$ – $2$ , $\varepsilon_J$ can reach values of hundreds to thousands: nucleation rate is ferociously sensitive to $S$ near the threshold.

Step 3 — compute $\varepsilon_{R_g}$ for the growth rate. From $R_{\text{growth}} = k_g(S-1)^n$ ,

\frac{dR_{\text{growth}}}{dS} = n\,k_g(S-1)^{n-1}

\varepsilon_{R_g} = \frac{S}{R_{\text{growth}}}\cdot n k_g (S-1)^{n-1} = \frac{nS}{S-1}

For $S$ well above 1 (say $S = 3$ ), $\varepsilon_{R_g} = n\cdot 3/2 = 1.5n$ — of order 1 to a few, utterly dwarfed by $\varepsilon_J$ .

Step 4 — compare.

\frac{\varepsilon_J}{\varepsilon_{R_g}} = \frac{2B/(\ln S)^{3}}{nS/(S-1)}

Even conservatively, with $B\sim 200$ , $\ln S \sim 1$ , $n=1$ , $S=3$ : $\varepsilon_J \approx 400$ while $\varepsilon_{R_g}\approx 1.5$ , a ratio of roughly 270-fold. This is the mathematical statement of the qualitative rule every gravimetric chemist learns: nucleation "turns on" far more abruptly than growth as supersaturation rises, so controlling $S$ is controlling the nucleation/growth split, and hence the final particle count and size.

Worked examples

Example 1 (foundational) — estimating the critical nucleus size and barrier for BaSO₄. Take barium sulfate precipitating at $T = 298\ \text{K}$ with an assumed interfacial tension $\gamma = 0.126\ \mathrm{J\,m^{-2}}$ (a literature-typical value for sparingly soluble sulfates) and molar volume $V_m = 52.0\ \mathrm{cm^3\,mol^{-1}} = 5.20\times10^{-5}\ \mathrm{m^3\,mol^{-1}}$ , so

v = \frac{V_m}{N_A} = \frac{5.20\times10^{-5}\ \mathrm{m^3\,mol^{-1}}}{6.022\times10^{23}\ \mathrm{mol^{-1}}} = 8.64\times10^{-29}\ \mathrm{m^3}

Suppose the working supersaturation, right after mixing dilute reagents, is $S = 12.5$ (an asymmetric, realistic value, not a round number like 10 or $e$ ). Then

\ln S = \ln(12.5) = 2.526

Compute $\Delta G_v$ :

\Delta G_v = \frac{k_BT\ln S}{v} = \frac{(1.381\times10^{-23}\ \mathrm{J\,K^{-1}})(298\ \mathrm{K})(2.526)}{8.64\times10^{-29}\ \mathrm{m^3}}

\Delta G_v = \frac{1.0397\times10^{-20}\ \mathrm{J}}{8.64\times10^{-29}\ \mathrm{m^3}} = 1.203\times10^{8}\ \mathrm{J\,m^{-3}}

Critical radius:

r^{*} = \frac{2\gamma}{\Delta G_v} = \frac{2(0.126\ \mathrm{J\,m^{-2}})}{1.203\times10^{8}\ \mathrm{J\,m^{-3}}} = 2.095\times10^{-9}\ \mathrm{m} = 2.10\ \mathrm{nm}

That is a nucleus only about 6–7 unit cells across — consistent with the "few tens of ion pairs" picture given in the physical description. Barrier height:

\Delta G^{\ddagger} = \frac{16\pi\gamma^{3}}{3\Delta G_v^{2}} = \frac{16\pi(0.126)^{3}}{3(1.203\times10^{8})^{2}}

Numerator: $16\pi = 50.27$ ; $(0.126)^3 = 2.000\times10^{-3}$ ; product $= 0.1005\ \mathrm{J^3\,m^{-6}}$ . Denominator: $3(1.203\times10^8)^2 = 3(1.447\times10^{16}) = 4.342\times10^{16}\ \mathrm{J^2\,m^{-6}}$ .

\Delta G^{\ddagger} = \frac{0.1005}{4.342\times10^{16}}\ \mathrm{J} = 2.315\times10^{-18}\ \mathrm{J}

Compare to thermal energy $k_BT = 4.116\times10^{-21}\ \mathrm{J}$ :

\frac{\Delta G^{\ddagger}}{k_BT} = \frac{2.315\times10^{-18}}{4.116\times10^{-21}} = 562

An exponent of $-562$ makes $J$ vanishingly small — physically, this says that at $S=12.5$ with this interfacial tension, homogeneous nucleation is still improbable in bulk solution over ordinary lab timescales, and observed precipitation is dominated by heterogeneous nucleation on dust, scratches, or added seed crystals, which have a much lower effective barrier. This is why real precipitations rarely wait for classical homogeneous nucleation — but the same $(\ln S)^{-2}$ scaling still governs the relative ease of nucleation as $S$ is pushed higher.

Limiting-case check: as $S \to 1^{+}$ , $\ln S \to 0^{+}$ , so $\Delta G_v \to 0$ , $r^{*}\to\infty$ , and $\Delta G^{\ddagger}\to\infty$ — nucleation becomes essentially impossible exactly at the solubility boundary, as physically required (no supersaturation, no driving force, no nucleation).

Example 2 (applied) — comparing two precipitation protocols for BaSO₄ gravimetry. An analyst must precipitate $\mathrm{BaSO_4}$ from a solution that will contain, at the moment of complete mixing, $[\mathrm{Ba^{2+}}] = 0.0150\ \mathrm{M}$ and $[\mathrm{SO_4^{2-}}] = 0.0180\ \mathrm{M}$ if reagent is added quickly and locally (protocol A: cold, fast, concentrated addition, a common beginner's mistake), versus effective local concentrations of $[\mathrm{Ba^{2+}}] = 8.0\times10^{-4}\ \mathrm{M}$ and $[\mathrm{SO_4^{2-}}] = 9.6\times10^{-4}\ \mathrm{M}$ if reagent is added dropwise with vigorous stirring into a large volume of hot, dilute, slightly acidified solution (protocol B, textbook-correct technique). $K_{sp}(\mathrm{BaSO_4}) = 1.1\times10^{-10}$ at 25 °C (protocol A) — assume for simplicity a similar $K_{sp}$ applies locally in protocol B despite the higher temperature, since we are comparing $S$ ratios, not doing rigorous thermodynamics at two different temperatures. Protocol A:

Q_A = (0.0150)(0.0180) = 2.70\times10^{-4}

S_A = \frac{2.70\times10^{-4}}{1.1\times10^{-10}} = 2.45\times10^{6}

\ln S_A = \ln(2.45\times10^6) = 14.71

Protocol B:

Q_B = (8.0\times10^{-4})(9.6\times10^{-4}) = 7.68\times10^{-7}

S_B = \frac{7.68\times10^{-7}}{1.1\times10^{-10}} = 6.98\times10^{3}

\ln S_B = \ln(6.98\times10^3) = 8.850

The nucleation rate exponent scales as $(\ln S)^{-2}$ , so the ratio of barriers is

\frac{\Delta G^{\ddagger}_B}{\Delta G^{\ddagger}_A} = \frac{(\ln S_A)^{2}}{(\ln S_B)^{2}} = \frac{(14.71)^{2}}{(8.850)^{2}} = \frac{216.4}{78.3} = 2.76

Protocol B's barrier is nearly three times higher than protocol A's — meaning nucleation in protocol B is dramatically suppressed relative to protocol A, even though $S_B$ is still enormous in absolute terms (locally, right at the point of reagent addition, both protocols are far above saturation; the difference protocol B buys is in how quickly the local excess is diluted and consumed by growth on existing particles before fresh nuclei can form). Using the barrier ratio inside the exponential (with the bulk numbers from Example 1's $B$ -type prefactor as an illustrative order of magnitude). If we take a representative $\Delta G^{\ddagger}_A/k_BT \approx 10$ near this much higher $S_A$ (barriers fall steeply as $S$ grows, unlike the small- $S$ case in Example 1) and scale by the ratio above, $\Delta G^{\ddagger}_B/k_BT \approx 27.6$ . The nucleation-rate ratio is then

\frac{J_A}{J_B} = \exp\!\left(\frac{\Delta G^{\ddagger}_B - \Delta G^{\ddagger}_A}{k_BT}\right) = \exp(27.6-10) = \exp(17.6) \approx 4.4\times10^{7}

Protocol A generates on the order of ten million times more nuclei per unit volume per unit time than protocol B under these assumptions. Every one of those extra nuclei divides up the same finite pool of $\mathrm{Ba^{2+}}$ and $\mathrm{SO_4^{2-}}$ , so protocol A's crystals end up correspondingly smaller — exactly the gelatinous, hard-to-filter precipitate described at the outset, while protocol B yields the coarse, filterable crystals gravimetric analysis requires.

Example 3 (exam/competition level) — designing a digestion strategy using Ostwald ripening as a check on the nucleation/growth trade-off. A precipitate of $\mathrm{BaSO_4}$ has, immediately after precipitation, a bimodal population: small crystals of average radius $r_1 = 0.20\ \mu\mathrm{m}$ and large crystals of average radius $r_2 = 2.0\ \mu\mathrm{m}$ , in contact with a solution held at the solubility limit for the large crystals (i.e., just saturated with respect to $r_2$ , but supersaturated with respect to the smaller, more soluble $r_1$ population, because small particles have higher local solubility). Using the Gibbs–Thomson (Ostwald–Freundlich) relation,

\ln\!\left(\frac{c(r)}{c_\infty}\right) = \frac{2\gamma V_m}{RTr}

where $c(r)$ is the solubility of a particle of radius $r$ and $c_\infty$ is the bulk (flat-surface) solubility, estimate the ratio $c(r_1)/c(r_2)$ at $T=353\ \mathrm{K}$ (a typical digestion temperature just below boiling), using $\gamma = 0.126\ \mathrm{J\,m^{-2}}$ and $V_m = 5.20\times10^{-5}\ \mathrm{m^3\,mol^{-1}}$ as before.

\ln\!\left(\frac{c(r_1)}{c_\infty}\right) = \frac{2(0.126)(5.20\times10^{-5})}{(8.314)(353)(0.20\times10^{-6})}

Numerator: $2 \times 0.126 \times 5.20\times10^{-5} = 1.310\times10^{-5}\ \mathrm{J\,m^2\,mol^{-1}}$ . Denominator: $8.314 \times 353 \times 0.20\times10^{-6} = 5.870\times10^{-4}\ \mathrm{J\,m\,mol^{-1}}$ .

\ln\!\left(\frac{c(r_1)}{c_\infty}\right) = \frac{1.310\times10^{-5}}{5.870\times10^{-4}} = 2.232\times10^{-2}

\frac{c(r_1)}{c_\infty} = e^{0.02232} = 1.0226

\ln\!\left(\frac{c(r_2)}{c_\infty}\right) = \frac{1.310\times10^{-5}}{(8.314)(353)(2.0\times10^{-6})} = \frac{1.310\times10^{-5}}{5.870\times10^{-3}} = 2.232\times10^{-3}

\frac{c(r_2)}{c_\infty} = e^{0.002232} = 1.00223

So:

\frac{c(r_1)}{c(r_2)} = \frac{1.0226}{1.00223} = 1.0203

The small crystals ( $r_1 = 0.20\ \mu\mathrm{m}$ ) are about $2.0\%$ more soluble than the large crystals under these conditions. That modest-looking 2% difference is exactly what drives Ostwald ripening: because the bath is saturated with respect to the large crystals but supersaturated with respect to the small ones, the small crystals slowly dissolve, feed dissolved $\mathrm{Ba^{2+}}$ and $\mathrm{SO_4^{2-}}$ back into solution, and that material redeposits onto the large crystals (which are undersaturated for them relative to the small-crystal-equilibrium solution). Over a digestion period of typically 30 minutes to an hour at near-boiling temperature, this slow net transfer coarsens the whole population, even though digestion happens entirely after the nucleation-versus-growth race that set the initial particle count. Digestion cannot manufacture large crystals from a solution that nucleated catastrophically into a billion sub-100 nm particles in the first place — the redistribution is a surface-tension-driven correction, not a substitute for controlling $S$ during precipitation.

Limiting-case check: as $r \to \infty$ for both crystal populations, $2\gamma V_m/(RTr) \to 0$ , so $c(r)/c_\infty \to 1$ for both, and the ripening driving force vanishes — large, well-formed crystals feel essentially no size-dependent solubility difference, exactly as expected, since Ostwald ripening is a small-particle phenomenon that becomes negligible once digestion has already coarsened the precipitate.

Limiting cases, scope & common traps

Regime	What happens	Practical consequence
$S \to 1^{+}$ ( $Q$ barely exceeds $K_{sp}$ )	$\ln S\to 0$ , nucleation barrier $\Delta G^{\ddagger}\to\infty$ , $J\to 0$	No new nuclei form; existing crystals grow slowly and cleanly — ideal but often too slow for lab timescales
Moderate $S$ (typical controlled precipitation, $S\sim 10$ – $50$ )	Barrier still substantial; a modest, controllable number of nuclei form early, then growth dominates	Textbook "digestion-ready" precipitate: coarse crystals, good for filtration
High $S$ (fast addition of concentrated reagent)	$\ln S$ large, but barrier collapses even faster relative to growth's mild response; explosive nucleation	Colloidal or gelatinous precipitate; high surface area; heavy coprecipitation risk
Heterogeneous nucleation present (dust, seed crystals, scratched glass)	Effective $\gamma$ and barrier are lowered at the foreign surface; nucleation proceeds well below the homogeneous threshold	Seeding deliberately controls where/when nucleation starts — used industrially and can be exploited analytically
After nucleation ends, during digestion	Ostwald ripening (small solubility differences from Gibbs–Thomson) slowly coarsens the population	Improves filterability and purity somewhat, but cannot fix a badly over-nucleated precipitate
$n\to$ large in growth law (strongly nonlinear, autocatalytic growth)	Growth itself accelerates faster with $S$ , narrowing (but not closing) the sensitivity gap with nucleation	Rare for simple ionic precipitates; more relevant to some organic/polymer crystallizations

Common trapMisconception: "digestion (heating and standing) fixes a poorly precipitated sample."

Digestion via Ostwald ripening genuinely improves crystal size and purity, and it is a standard, valuable step. But it operates on a population whose particle count was already fixed at the moment of nucleation. If a precipitation event created $10^{12}$ nuclei per liter instead of $10^{6}$ , digestion redistributes mass among the existing particles — it does not merge distant particles across the vessel or reduce the particle count by orders of magnitude in a practical digestion time. Controlling $S$ during addition (dilute reagents, slow addition, heat, stirring) is not optional groundwork that digestion can retroactively repair; it is the step that determines whether digestion has a reasonable population to work with at all.

The chemist behind it

The modern quantitative language of nucleation barriers traces to the joint efforts of statistical thermodynamics pioneers working on phase transitions in the early twentieth century, but the person whose name is most tightly bound to precipitate particle size control in analytical chemistry is Paul Peter von Weimarn, a Russo-German colloid chemist working in the 1900s–1920s. Von Weimarn systematically varied reagent concentration, temperature, and mixing order across dozens of precipitating systems — barium sulfate among them — and empirically established that particle size in a precipitate correlates with the relative supersaturation at the moment of mixing, long before the Gibbs–Thomson/classical-nucleation-theory machinery existed to explain why. His empirical rule is the direct qualitative ancestor of the quantitative treatment given here, and it is formalized fully in the companion treatment of Relative Supersaturation and von Weimarn.

Origins · 1 minPaul Peter von Weimarn and how this was discovered

Paul Peter von Weimarn

1879–1935 · Colloid chemistry, St. Petersburg and later Riga; systematic empirical studies of precipitation and crystallization from supersaturated solutions

Von Weimarn's central contribution was recognizing that the ratio of supersaturation to solubility — not the absolute concentration alone — governs whether a precipitating system yields crystalline solids or colloidal/amorphous ones. He demonstrated this across an enormous range of substances (including barium sulfate, silver halides, and various sulfides), establishing a generality that later theoretical treatments of nucleation kinetics would explain mechanistically.

History

- 1897 — Wilhelm Ostwald formulates the "step rule," and lays groundwork on supersaturation and metastable states relevant to nucleation.

- c. 1906–1925 — Paul von Weimarn publishes systematic studies correlating precipitate particle size with relative supersaturation across many chemical systems.

- 1926 — Marian Volmer and Alfred Weber (Volmer–Weber) develop early classical nucleation theory for vapor condensation, providing the surface/bulk free-energy framework.

- 1935–1939 — Richard Becker and Werner Döring, and independently Yakov Zeldovich, refine classical nucleation theory into the rate-equation form used across crystallization, condensation, and precipitation.

- Mid-20th century onward — analytical chemists (Kolthoff, Sandell, and successors) translate nucleation/growth kinetics into practical gravimetric technique: dilute reagents, slow addition, digestion, homogeneous precipitation.

Why does a high supersaturation ratio S produce a colloidal, hard-to-filter precipitate instead of coarse crystals?

Nucleation rate $J$ depends on $S$ through $\exp[-\text{const}/(\ln S)^2]$ , an extraordinarily steep function, while crystal growth rate depends on $S$ only through a low-order term like $(S-1)^n$ with $n\approx1$ – $2$ . At high $S$ , the nucleation barrier collapses almost completely, so huge numbers of new nuclei form throughout the solution nearly simultaneously. The finite pool of dissolved ions must then be shared among all those nuclei, and since total precipitated mass is fixed by stoichiometry, more nuclei necessarily means smaller average particle size. Below a certain particle size, precipitates behave as colloids: they resist settling, pass through filter pores, and carry large surface areas that adsorb impurities. This is precisely the mechanistic link between kinetics and the practical, well-known instruction to keep reagents dilute and add them slowly.

Does raising the temperature during precipitation help control particle size, and if so, how?

Yes, in two related ways. First, most sparingly soluble salts have solubility products that increase with temperature, so $K_{sp}$ rises and, for a fixed $Q$ , the supersaturation ratio $S = Q/K_{sp}$ falls — directly lowering the nucleation rate relative to growth. Second, higher temperature speeds up ion diffusion and the kinetics of Ostwald ripening during digestion, helping small crystals dissolve and redeposit onto larger ones faster. Both effects push the system toward fewer, larger, better-formed crystals, which is why gravimetric procedures routinely specify precipitating from hot solution and digesting near boiling rather than at room temperature.

What is the physical meaning of the pre-exponential factor A in the nucleation rate law, and why isn't it usually calculated precisely?

$A$ represents the frequency at which ion clusters near the critical size successfully cross the barrier, folding in the attachment rate of individual ions to a cluster surface, the density of potential nucleation sites, and vibrational/attempt frequencies at the cluster surface. In practice $A$ is difficult to compute from first principles for ionic crystals in aqueous solution — literature estimates for various systems range broadly, often $10^{25}$ to $10^{35}\ \mathrm{m^{-3}s^{-1}}$ — because it depends on molecular-scale attachment kinetics that are hard to measure independently of the nucleation rate itself. Analytical chemists generally do not need a numerically precise $A$ ; the qualitative and comparative behavior of $J$ (its extreme sensitivity to $S$ through the exponential term) is what drives practical technique, and that conclusion is robust to uncertainty in $A$ .

How does homogeneous precipitation exploit the nucleation/growth competition described here?

Homogeneous precipitation generates the precipitating ion in situ through a slow chemical reaction distributed throughout the entire solution volume, rather than by adding a concentrated reagent solution from outside. A classic example is generating hydroxide slowly via the controlled hydrolysis of urea, or generating sulfate slowly via hydrolysis of an organosulfate ester. Because the precipitating species appears gradually and uniformly everywhere in the solution rather than in a concentrated pulse near a burette tip, local supersaturation never spikes to the extreme values responsible for runaway nucleation. The bulk supersaturation ratio $S$ stays low and roughly uniform throughout the precipitation, so few nuclei form and those that do form are fed steadily by ongoing production of ions, yielding notably coarser, purer, and more easily filtered crystals than conventional direct addition.

Can seeding a supersaturated solution with pre-formed crystals bypass the nucleation barrier entirely?

Yes, and this is exactly why heterogeneous nucleation (on dust, container-wall imperfections, or deliberately added seed crystals) so often dominates over homogeneous nucleation in practice. Adding seed crystals provides ready-made surfaces onto which ions can deposit via the barrier-free growth pathway, without needing to assemble a brand-new nucleus from scratch. This is why some gravimetric and crystallization procedures include deliberate seeding, and why scrupulously clean glassware (removing dust and scratches that could otherwise nucleate unwanted fine particles) is part of good precipitation technique — the goal is to control where nucleation happens, not merely how much.

Is there a sharp, universal value of S below which nucleation simply does not happen?

Not a sharp universal value, because the barrier height $\Delta G^{\ddagger}$ depends on system-specific quantities — interfacial tension $\gamma$ , molar volume $v$ , and temperature $T$ — that vary from one precipitate to another. What is universal is the functional form: $\Delta G^{\ddagger} \propto \gamma^3 v^2/(T^2(\ln S)^2)$ , so systems with low interfacial tension (many soluble salts near their solubility limit) nucleate readily even at modest $S$ , while systems with high interfacial tension (many very insoluble salts, including several classic gravimetric precipitates) can sustain fairly high $S$ before homogeneous nucleation becomes significant — though heterogeneous nucleation on impurities usually intervenes well before the homogeneous threshold is reached in real glassware.

Why does this article emphasize supersaturation S rather than just absolute concentration?

Absolute concentration alone does not determine whether a system nucleates readily, because the relevant comparison is always to that system's own solubility. A precipitate with a very small $K_{sp}$ (like $\mathrm{BaSO_4}$ , $K_{sp}=1.1\times10^{-10}$ ) can reach enormous $S$ values even from dilute millimolar reagents, while a more soluble salt might need much higher absolute concentrations to reach the same $S$ . Because both the nucleation-rate law and the growth-rate law are naturally written in terms of the dimensionless ratio $S=Q/K_{sp}$ , comparing precipitation behavior across different chemical systems only makes sense on this normalized scale — a theme made fully quantitative in the treatment of relative supersaturation and the von Weimarn rule.

Nucleation and Particle Growth in Precipitates

Where this lives today

The physical picture — thorough prose

The core ideas — governing equations

Worked examples

Limiting cases, scope & common traps

The chemist behind it

Before this

Where this leads