Explosions and Branching Chains

Mix hydrogen and oxygen at low pressure and gently warm the flask. Nothing happens for a while, then the whole vessel detonates within microseconds. This dramatic threshold behavior is the signature of a branched-chain explosion, and it cannot be explained by simple Arrhenius heating alone. Chain reactions were introduced as sequences of initiation, propagation, and termination steps carrying a single reactive intermediate through the mechanism. Explosions push this idea further: some propagation steps produce more radicals than they consume. That branching multiplies chain carriers exponentially, and the reaction rate can outrun any mechanism for removing heat or radicals. This article separates thermal explosions, driven by a runaway feedback between temperature and rate, from branched-chain explosions, driven by a runaway feedback in radical population. You will see how the same H₂ + O₂ system displays three distinct explosion limits as pressure changes. Then you will derive the critical condition for branching explosion. Finally you will work two numerical examples showing how close a mixture sits to detonation.

Key takeaways

A thermal explosion occurs when heat release outpaces heat loss, causing temperature and rate to escalate together.
A branched-chain explosion occurs when chain-branching steps multiply radical concentration faster than termination removes it.
The critical branching condition is $\delta = k_b - k_t = 0$ ; positive $\delta$ gives exponential radical growth.
The H₂ + O₂ system shows three pressure-dependent explosion limits, each governed by a different mechanism.
Explosion limits are boundaries in pressure–temperature space, not fixed rate constants; small changes in vessel geometry shift them.
Key idea: explosions arise from positive feedback, either in heat (thermal) or in radical number (chain branching), that overwhelms the corresponding loss process.

The core insight

An explosion is not simply "a fast reaction" — it is the macroscopic signature of positive feedback, where the rate of a process increases its own driving force faster than any competing loss process can remove it, so a small perturbation grows without bound until the system is limited by something else entirely, typically fuel depletion or vessel destruction.

The physical picture

In an ordinary chain reaction, each propagation step consumes one radical and produces one radical, so the population of chain carriers stays roughly steady after initiation. The reaction proceeds at a rate set by that steady radical concentration until termination steps (radical–radical recombination or wall collisions) balance production. A branching step breaks this symmetry: one radical enters, and two or more emerge. The classic example is the hydrogen–oxygen chain,

\text{H} + \text{O}_2 \rightarrow \text{OH} + \text{O}

which consumes one H atom but produces an OH radical and an O atom — two reactive species where there was one. The O atom then attacks H₂ to give OH and H back, and the OH attacks H₂ to regenerate H, so each branching event seeds two new chains. If branching outpaces the destruction of radicals, the number of chain carriers grows geometrically with time, and so does the rate, even at constant temperature. This is fundamentally different from a thermal explosion, where the feedback loop is thermal: reaction releases heat, heat raises temperature, and (through the Arrhenius factor) higher temperature accelerates the reaction further. Both mechanisms produce the same experimental signature — an abrupt transition from slow, controlled reaction to violent, essentially instantaneous combustion — but the underlying feedback variable differs: temperature for thermal explosions, radical concentration for branched-chain explosions. Real explosive systems, including H₂ + O₂ and hydrocarbon oxidation, often involve both mechanisms operating together, which is why explosion boundaries in these systems have a rich, non-monotonic shape.

Figure

Feedback loops: thermal explosion versus branched-chain explosion

The two feedback loops are formally analogous — both are described by an equation of the form $dx/dt = (\text{gain} - \text{loss})\,x$ — but $x$ means something completely different in each case: temperature deviation for thermal runaway, radical concentration for chain-branching runaway.

The core relations

\frac{d[\text{R}]}{dt} = r_i + (k_b - k_t)[\text{R}]

where: $r_i$ = rate of initiation, $k_b$ = effective first-order branching rate constant, $k_t$ = effective first-order termination rate constant, $[\text{R}]$ = radical concentration.

\delta \equiv k_b - k_t

where: $\delta$ is the net branching factor; the sign of $\delta$ decides steady state versus explosion.

[\text{R}](t) = \frac{r_i}{\delta}\left(e^{\delta t} - 1\right)

where: valid for $\delta \neq 0$ ; for $\delta<0$ this saturates at a steady state, for $\delta>0$ it diverges.

[\text{R}]_{ss} = \frac{r_i}{-\delta} = \frac{r_i}{k_t - k_b} \qquad (\delta < 0)

where: steady-state radical concentration below the explosion limit.

\frac{d\Delta T}{dt} = \frac{q_{rxn}}{C_v}\,k(T) - \frac{h\,A_s}{C_v}\,\Delta T

where: $q_{rxn}$ = heat released per reaction event, $h$ = heat-transfer coefficient, $A_s$ = vessel surface area, $\Delta T = T - T_{surroundings}$ , used in the Semenov thermal-explosion model.

p_2^{\text{(1st limit)}} \propto k_t^{\text{wall}},\qquad p_2^{\text{(2nd limit)}}:\ k_b(p) = k_t^{\text{gas}}(p),\qquad p_3^{\text{(3rd limit)}}:\ \text{thermal runaway dominates}

where: the three empirical explosion limits of H₂ + O₂ correspond to different dominant loss mechanisms at different total pressures.

Derivation AThe critical branching condition

Goal. Find the condition under which radical concentration — and hence reaction rate — grows without bound, defining the branched-chain explosion limit.

Step 1. Write the radical balance. Assume a small, constant rate of initiation $r_i$ (e.g., from a slow bimolecular step or a wall-catalyzed process) continually seeds new chains. Radicals are removed by termination (first order in $[\text{R}]$ , representing wall destruction or three-body gas-phase recombination) at rate $k_t[\text{R}]$ , and are multiplied by branching at rate $k_b[\text{R}]$ , since each branching event consumes one radical but the net production of new radical equivalents scales linearly with how many radicals are currently present to branch. The balance is

\frac{d[\text{R}]}{dt} = r_i + k_b[\text{R}] - k_t[\text{R}] = r_i + \delta[\text{R}]

with $\delta = k_b - k_t$ .

Step 2. Solve the linear ODE. This is a first-order linear differential equation with constant coefficients. With initial condition $[\text{R}](0) = 0$ , the solution is

[\text{R}](t) = \frac{r_i}{\delta}\left(e^{\delta t} - 1\right).

Step 3. Examine the two regimes. If $\delta < 0$ (termination faster than branching), then as $t \to \infty$ , $e^{\delta t} \to 0$ , and

[\text{R}](t) \to \frac{r_i}{-\delta} = \frac{r_i}{k_t - k_b}

a finite, bounded steady state. The system behaves as an ordinary chain reaction with a well-defined stationary rate. If $\delta > 0$ (branching faster than termination), the exponential term grows without bound: $[\text{R}](t) \to \infty$ as $t$ increases, in principle in a finite but very short time once $[\text{R}]$ becomes large enough that side reactions (like thermal effects or fuel depletion) intervene. Physically, this divergence is the explosion: the radical population, and with it the reaction rate $\propto [\text{R}]$ , increases without limit until the fuel is consumed or the vessel fails.

Step 4. Identify the critical condition. The boundary between bounded and explosive behavior is

\delta = k_b - k_t = 0 \quad\Longleftrightarrow\quad k_b = k_t.

Since $k_b$ and $k_t$ generally depend on pressure, temperature, and vessel geometry (gas-phase termination is second or third order in concentration, i.e., pressure-dependent, while wall termination is first order but depends on diffusion to the walls, i.e., inversely pressure-dependent), this equality traces out a curve in $(p, T)$ space: the explosion limit.

\boxed{\delta = k_b(p,T) - k_t(p,T) = 0 \ \text{defines the explosion-limit boundary; } \delta>0 \text{ inside it, explosion occurs.}}

Worked examples

Example 1 — Steady-state radical concentration and how close a mixture sits to explosion. A chain-branching mixture has an initiation rate $r_i = 1.0 \times 10^{10}\ \text{molecule cm}^{-3}\text{s}^{-1}$ , a termination rate constant $k_t = 5.0\times 10^{3}\ \text{s}^{-1}$ , and a branching rate constant $k_b = 4.0\times 10^{3}\ \text{s}^{-1}$ at the operating pressure. Step 1 — net branching factor.

\delta = k_b - k_t = 4.0\times10^3 - 5.0\times10^3 = -1.0\times10^3\ \text{s}^{-1}

Since $\delta<0$ , the system sits below the explosion limit and reaches a steady state. Step 2 — steady-state radical concentration.

[\text{R}]_{ss} = \frac{r_i}{-\delta} = \frac{1.0\times10^{10}}{1.0\times10^3}\ \text{cm}^{-3} = 1.0\times10^7\ \text{molecule cm}^{-3}

Step 3 — sanity check via the time to approach steady state. The characteristic relaxation time is

\tau = \frac{1}{|\delta|} = \frac{1}{1.0\times10^3\ \text{s}^{-1}} = 1.0\times10^{-3}\ \text{s} = 1.0\ \text{ms}

a physically reasonable induction time for gas-phase radical build-up, and it confirms that after a few milliseconds the mixture settles into steady, non-explosive combustion at $[\text{R}]_{ss}=1.0\times10^7\ \text{cm}^{-3}$ . Step 4 — how close to the limit? The mixture is close to explosion: $k_b/k_t = 0.80$ , i.e., branching supplies 80% of what would be needed to match termination. A 25% increase in pressure that raised $k_b$ (a gas-phase, pressure-favored process) to $5.0\times10^3\ \text{s}^{-1}$ while leaving $k_t$ unchanged would push $\delta$ to zero — the explosion limit — illustrating why H₂/O₂ mixtures can detonate from a seemingly modest pressure increase.

Example 2 — Locating the second explosion limit of H₂ + O₂ (order-of-magnitude). The second explosion limit of a stoichiometric H₂ + O₂ mixture at 500 °C occurs near $p_2 \approx 2\ \text{Torr}$ (real literature value, depending on vessel). Below $p_2$ , gas-phase termination (the three-body step $\text{H} + \text{O}_2 + \text{M} \rightarrow \text{HO}_2 + \text{M}$ ) is too slow, because it is third order and scales with $[\text{M}]^2$ -like dependence through pressure, so branching by $\text{H} + \text{O}_2 \rightarrow \text{OH} + \text{O}$ (first order in pressure) dominates and the system explodes. Above $p_2$ , raising the pressure accelerates the termination step faster than the branching step, and the mixture becomes stable again. Step 1 — express both rate constants in terms of pressure. Approximate at fixed temperature:

k_b = k_b^{\circ}\,p, \qquad k_t = k_t^{\circ}\,p^2

(branching is bimolecular, effectively first order in the bath pressure since $[\text{O}_2]\propto p$ ; three-body termination is effectively second order in $p$ since it needs both a collision partner $M$ and $\text{O}_2$ -derived radical density, each scaling with $p$ ). Step 2 — apply the critical condition $k_b=k_t$ .

k_b^{\circ}\,p_2 = k_t^{\circ}\,p_2^2 \quad\Rightarrow\quad p_2 = \frac{k_b^{\circ}}{k_t^{\circ}}

Step 3 — plug in typical literature magnitudes at 800 K. Using $k_b^{\circ} \approx 5\times10^{2}\ \text{s}^{-1}\,\text{Torr}^{-1}$ and $k_t^{\circ}\approx 2.5\times10^{2}\ \text{s}^{-1}\,\text{Torr}^{-2}$ (order-of-magnitude values consistent with the observed limit),

p_2 = \frac{5\times10^{2}}{2.5\times10^{2}}\ \text{Torr} = 2.0\ \text{Torr}

matching the experimentally observed second limit of order a few Torr. This confirms the qualitative picture: doubling total pressure from 1 Torr to 4 Torr moves the system from inside the explosive peninsula (at 1 Torr, $k_b=5\times10^2$ s⁻¹ > $k_t=2.5\times10^2$ s⁻¹, so $\delta>0$ ) to outside it (at 4 Torr, $k_t=4.0\times10^3$ s⁻¹ now exceeds $k_b=2.0\times10^3$ s⁻¹, so $\delta<0$ ).

Limiting cases & common traps

Situation	Behavior	Common pitfall
$\delta < 0$ , small $r_i$	Steady, slow chain reaction	Assuming any chain mechanism must eventually explode
$\delta > 0$	Branched-chain explosion, isothermal in principle	Confusing this with a thermal explosion; branching can occur near ambient $T$
Low pressure, first limit	Radicals escape to walls before reacting; increasing $p$ suppresses wall loss and can trigger explosion	Expecting explosion only at high pressure — the first limit is explosive below a pressure threshold
Intermediate pressure, second limit	Gas-phase termination increases with $p^2$ , faster than branching's $p^1$ , stabilizing the mixture	Forgetting that raising pressure can stop an explosion here
High pressure, third limit	Exothermicity accumulates faster than heat loss; thermal explosion mechanism takes over	Treating the third limit as more branching when it is actually thermal runaway

The chemist behind it

The quantitative theory of branched-chain explosions was developed largely by Nikolay Semenov and Cyril Hinshelwood, who shared the 1956 Nobel Prize in Chemistry for their independent work on chain reaction mechanisms in the 1920s–1930s. Semenov built the mathematical theory of thermal explosion and, together with his collaborators, mapped the branching-chain kinetics of hydrogen–oxygen mixtures. Hinshelwood's Oxford group performed the painstaking pressure–temperature mapping of the explosion peninsula in H₂ + O₂, discovering the counter-intuitive result that raising pressure could extinguish an explosive mixture — direct experimental evidence for the branching–termination competition captured in $\delta = k_b - k_t$ .

Origins · 1 minNikolay Nikolaevich Semenov and how this was discovered

Nikolay Nikolaevich Semenov

1896–1986 · Soviet physical chemist who founded the quantitative theory of chain and thermal explosions; shared the 1956 Nobel Prize with Hinshelwood

Ask the tutor: How does adding an inert third-body gas shift the second explosion limit of a hydrogen–oxygen mixture?

Why does H2 + O2 have three separate explosion limits instead of one?

Each limit reflects a different dominant loss process for chain carriers. The first (lowest-pressure) limit occurs because at very low pressure radicals diffuse to the vessel walls and are destroyed before they can branch; raising pressure suppresses this diffusive loss and can trigger explosion. The second limit occurs at intermediate pressure, where gas-phase three-body termination, which scales roughly as pressure squared, overtakes branching, which scales roughly as pressure to the first power, stabilizing the mixture. The third limit at higher pressure and temperature reflects a switch to genuine thermal-explosion behavior, where accumulated heat cannot escape fast enough. All three are governed by the same competition between chain multiplication and chain destruction, just with different physical loss channels dominating.

Can a mixture be both a thermal explosion and a branched-chain explosion at once?

Yes, and this is common in real combustion systems. Chain branching multiplies radicals, which accelerates heat release; the released heat raises temperature, which in turn accelerates both branching and propagation steps through their Arrhenius factors. In hydrocarbon oxidation, degenerate branching from intermediates like peroxides couples tightly with thermal feedback, producing complex phenomena such as cool flames and two-stage ignition. The Semenov and branching pictures are limiting idealizations; most practical explosions sit somewhere between them, with both feedback loops reinforcing each other.

How is the branching factor δ related to the explosion limit boundary in a p–T diagram?

The explosion limit is the locus of points where δ equals zero, that is, where the branching rate constant exactly equals the termination rate constant. Because both rate constants depend on pressure, temperature, and the third-body concentration, this locus traces a curve rather than a single value. Inside the curve, where δ is positive, any small radical population grows without bound and the mixture explodes. Outside the curve, where δ is negative, the mixture reaches a bounded steady-state radical concentration and burns steadily or not at all.

Does the branching explosion require any minimum temperature, unlike a thermal explosion?

Not fundamentally. Chain-branching explosions are, in the idealized linear model, isothermal — they depend on the algebraic sign of δ, which can in principle be positive even near room temperature if the branching step has a low enough activation energy relative to termination. In practice, however, branching rate constants like that for H + O2 → OH + O do have appreciable activation energies, so raising temperature increases kb and can push a mixture across the explosion limit even without invoking thermal runaway explicitly.

What role does the vessel wall play in the first explosion limit?

At low total pressure, mean free paths are long and radicals reach the container walls quickly, where they can recombine or be destroyed catalytically, acting as an efficient first-order termination channel. This wall termination rate depends on the diffusion coefficient, which increases as pressure decreases, so wall loss is most effective at low pressure. Raising the pressure slows diffusion to the walls, reducing this termination pathway relative to gas-phase branching, and can push the mixture across the first limit into explosion territory.