Maxwell-Boltzmann Speed Distribution

If you could freeze time inside a container of nitrogen gas at room temperature and measure every single molecule's speed, you would not find them all moving at the same speed. Instead, you'd find a beautiful bell-shaped distribution: a few molecules crawling along at 100 m/s, many moving at a few hundred m/s, most clustered somewhere around 400–500 m/s, and a dwindling tail of extremely fast molecules at 1000+ m/s. The shape of this distribution is one of the most famous results in statistical physics. It's called the Maxwell-Boltzmann distribution, and it was derived by James Clerk Maxwell in 1860 (before anyone knew atoms were real, incidentally) and generalized by Ludwig Boltzmann in 1872.

The punchline: at any temperature above absolute zero, gas molecules have a distribution of speeds — not a single speed. And the distribution has a very specific mathematical shape that depends only on the temperature and the mass of the molecules. Nothing else.

Why molecules don't all have the same speed

Imagine shaking a box full of marbles. Some marbles bounce off each other, some ricochet off the walls, some gain energy in one collision and lose it in the next. At any given instant, a snapshot would show marbles at wildly different speeds. Only the average kinetic energy is constrained, and that's what we call temperature.

For gas molecules, the same picture applies, but with astronomical numbers of collisions. At room temperature and pressure, a single molecule collides about 7 billion times per second with other molecules. Each collision randomizes its speed and direction. Over many collisions, the speeds settle into a statistical equilibrium — the Maxwell-Boltzmann distribution — where some molecules happen to be fast, some slow, and the population distribution is stable even though individual molecules are constantly jumping around.

The distribution doesn't depend on what happened a second ago, or what the average pressure is, or how large the container is. It's a universal equilibrium distribution for any ideal gas, set by two things: the temperature and the mass of the particles.

The Maxwell-Boltzmann distribution formula

In three dimensions, the probability density that a molecule has speed $v$ is:

$f(v) = 4\pi \left(\dfrac{m}{2\pi k_{B} T}\right)^{3/2}\,v^{2}\,\exp\!\left(-\dfrac{m v^{2}}{2 k_{B} T}\right)$

where:

• $m$ is the mass of a single molecule (in kg)
• $k_{B} = 1.381 \times 10^{-23}\;\text{J/K}$ is Boltzmann's constant (the gas constant $R$ divided by Avogadro's number)
• $T$ is the absolute temperature in kelvin

The function $f(v)$ tells you the fraction of molecules whose speed falls in a tiny window around $v$. To get the fraction of molecules with speed between $v_{1}$ and $v_{2}$, you integrate: $\int_{v_{1}}^{v_{2}} f(v)\,dv$. The total integral from 0 to $\infty$ equals 1, which means the formula is correctly normalized (every molecule has some speed).

Let me unpack the formula piece by piece, because each part has a physical meaning.

The prefactor $4\pi (m/(2\pi k_{B}T))^{3/2}$ is just a normalization constant that makes $\int f(v)\,dv = 1$. You don't have to memorize this; it falls out of the math. The factor of $4\pi$ is a "3D area of a sphere of radius $v$" in velocity space — it accounts for the fact that molecules with speed $v$ can be going in any direction.

The $v^{2}$ factor is what makes the distribution rise from zero at low speeds. In 3D velocity space, there are more ways to have a high speed than a low speed (a sphere of radius $v$ has surface area $4\pi v^{2}$, growing with $v$). This geometric factor competes with the exponential drop and is what creates the characteristic bell shape.

The exponential $\exp(-mv^{2}/(2k_{B}T))$ is the Boltzmann factor. The exponent is exactly $-E_{\text{kinetic}}/(k_{B}T)$ — the ratio of a molecule's kinetic energy $\tfrac{1}{2}mv^{2}$ to the typical thermal energy scale $k_{B}T$. It says that high-kinetic-energy molecules are exponentially unlikely, with the decay rate set by temperature. Hotter gas → slower exponential decay → more high-speed molecules in the tail.

Multiplying them: at very low $v$, $f(v) \approx \text{const} \cdot v^{2}$ grows quadratically from zero. At very high $v$, $f(v) \approx \text{const} \cdot v^{2} e^{-v^{2}}$ dies off exponentially. Somewhere in between is the peak — and that's the "most probable speed."

The three characteristic speeds

Because the distribution isn't a single number, we describe it using three different "typical speeds," each with a clear physical meaning. All three are proportional to $\sqrt{T/m}$ — they differ only by constant factors.

Most probable speed $v_{p}$ — the peak of the curve, the speed at which the most molecules are found.

$v_{p} = \sqrt{\dfrac{2 k_{B} T}{m}}$

Mean speed $\bar v$ — the arithmetic average speed across all molecules.

$\bar v = \sqrt{\dfrac{8 k_{B} T}{\pi\,m}}$

Root-mean-square speed $v_{\text{rms}}$ — the square root of the average of $v^{2}$. This one appears whenever you care about kinetic energy, because $\tfrac{1}{2}m\langle v^{2}\rangle = \tfrac{3}{2}k_{B}T$.

$v_{\text{rms}} = \sqrt{\dfrac{3 k_{B} T}{m}}$

Their ratios are fixed and universal:

$v_{p} : \bar v : v_{\text{rms}} = 1 : \sqrt{\dfrac{4}{\pi}} : \sqrt{\dfrac{3}{2}} \approx 1 : 1.128 : 1.225$

The order is always $v_{p} < \bar v < v_{\text{rms}}$. The most probable speed is at the peak; the mean is slightly to the right of the peak (because the distribution has a longer tail on the right); the rms is slightly to the right of the mean (because squaring weights the high-speed tail even more).

Numerically, for $N_{2}$ (nitrogen, molar mass 28 g/mol, $m \approx 4.65 \times 10^{-26}\;\text{kg}$) at 300 K:

• $v_{p} \approx 422\;\text{m/s}$
• $\bar v \approx 476\;\text{m/s}$
• $v_{\text{rms}} \approx 517\;\text{m/s}$

All three are around 500 m/s — that's faster than the speed of sound (343 m/s in air). Which actually makes sense, because sound propagation in a gas depends on how fast molecular collisions can pass information along, which is tied to the molecular speed scale.

Method 1: Deriving $v_{\text{rms}}$ from kinetic theory

The equipartition theorem says each quadratic degree of freedom has an average energy of $\tfrac{1}{2} k_{B} T$. A gas molecule has three translational degrees of freedom (motion in $x$, $y$, $z$), so its average translational kinetic energy is:

$\left\langle \tfrac{1}{2}m v^{2}\right\rangle = \tfrac{3}{2} k_{B} T$

Solving for $\langle v^{2}\rangle$:

$\langle v^{2}\rangle = \dfrac{3 k_{B} T}{m}$

And $v_{\text{rms}}$ is defined as the square root of that:

$v_{\text{rms}} = \sqrt{\langle v^{2}\rangle} = \sqrt{\dfrac{3 k_{B} T}{m}}$

Done. No integration required. Kinetic theory + equipartition directly gives you $v_{\text{rms}}$ without even having to know the full shape of the distribution.

Method 2: Deriving $v_{p}$ from calculus

To find the peak of $f(v)$, take its derivative and set it to zero:

$\dfrac{df}{dv} = 0$

The derivative of the product $v^{2}\,e^{-\alpha v^{2}}$ (where $\alpha = m/(2 k_{B} T)$) is:

$\dfrac{d}{dv}\left[v^{2} e^{-\alpha v^{2}}\right] = 2v\,e^{-\alpha v^{2}} - 2\alpha v^{3}\,e^{-\alpha v^{2}} = 2v\,e^{-\alpha v^{2}}\,(1 - \alpha v^{2})$

Setting this equal to zero: either $v = 0$ (the boundary, not a maximum) or $1 - \alpha v^{2} = 0$. Solving the second:

$v^{2} = \dfrac{1}{\alpha} = \dfrac{2 k_{B} T}{m}$

$v_{p} = \sqrt{\dfrac{2 k_{B} T}{m}}$

Method 3: Deriving $\bar v$ from integration

The mean speed is $\bar v = \int_{0}^{\infty} v\,f(v)\,dv$. Plugging in the distribution:

$\bar v = \int_{0}^{\infty} 4\pi \left(\dfrac{m}{2\pi k_{B} T}\right)^{3/2} v^{3}\,e^{-\alpha v^{2}}\,dv$

The relevant integral is $\int_{0}^{\infty} v^{3} e^{-\alpha v^{2}}\,dv = \dfrac{1}{2 \alpha^{2}}$.

Plugging through:

$\bar v = 4\pi \left(\dfrac{m}{2\pi k_{B} T}\right)^{3/2} \cdot \dfrac{1}{2 \alpha^{2}} = \sqrt{\dfrac{8 k_{B} T}{\pi\,m}}$

The algebra is slightly tedious but the result is clean.

Temperature dependence

All three characteristic speeds scale as $\sqrt{T}$. Doubling the temperature (in kelvin!) increases every speed by a factor of $\sqrt{2} \approx 1.41$. Quadrupling $T$ doubles the speeds.

Visually, the whole distribution shifts to the right and flattens as $T$ increases. Toggle Show comparison temperature and pick a second temperature to see two curves overlaid. At higher $T$, the peak moves right (faster molecules) AND the peak height drops (the distribution spreads out more). The total area under each curve is always 1 — it's a probability distribution — so a wider curve has to be shorter.

Mass dependence and atmospheric escape

The speeds scale as $1/\sqrt{m}$. Lighter molecules are systematically faster at any given temperature.

This has a huge real-world consequence: light gases escape planetary atmospheres over long timescales. Earth's escape velocity is about 11.2 km/s (11,200 m/s). For a molecule to escape, it has to be moving faster than this at the top of the atmosphere where there's nothing above it to stop it.

Let's compute how many molecules have that escape speed at a typical upper-atmosphere temperature of 1000 K (exosphere is hot due to solar UV absorption):

For nitrogen ($m = 4.65 \times 10^{-26}$ kg): $v_{\text{rms}} \approx 940\;\text{m/s}$. Escape velocity is 11.2 km/s, which is about 12 standard deviations out in the Maxwell-Boltzmann tail. The fraction of N₂ molecules at that speed is roughly $e^{-144}$ — essentially zero. Nitrogen stays.

For helium ($m = 6.64 \times 10^{-27}$ kg): $v_{\text{rms}} \approx 2500\;\text{m/s}$. Escape velocity at 11,200 m/s is "only" about 4.5 standard deviations. A tiny but nonzero fraction of helium atoms exceed escape velocity — roughly $e^{-20} \approx 10^{-9}$ — and over geological timescales, Earth's helium slowly leaks away into space. This is why helium is so rare in Earth's atmosphere (5 parts per million) despite being one of the most common elements in the universe.

For hydrogen ($m = 3.32 \times 10^{-27}$ kg, even lighter than helium atom because $H_{2}$ dissociates in the exosphere): Hydrogen escape is orders of magnitude faster than helium. Earth has lost virtually all its original H₂ atmosphere over 4.5 billion years.

Jupiter has an escape velocity of ~60 km/s and keeps even hydrogen. The Moon has an escape velocity of only 2.4 km/s and can't hold any gas atmosphere — every molecule eventually boils away.

This is a direct consequence of the Maxwell-Boltzmann tail and nothing else. Temperature + mass + escape velocity → long-term atmospheric composition.

Speed of sound

The speed of sound in a gas is closely related to the Maxwell-Boltzmann speeds. For a diatomic ideal gas:

$v_{\text{sound}} = \sqrt{\dfrac{\gamma k_{B} T}{m}}$

where $\gamma = C_{P}/C_{V}$ is the adiabatic index (= 7/5 for diatomic). Compare to:

$v_{\text{rms}} = \sqrt{\dfrac{3 k_{B} T}{m}}$

The ratio $v_{\text{sound}}/v_{\text{rms}} = \sqrt{\gamma/3} \approx 0.68$ for diatomic gases. So sound travels at roughly 68% of $v_{\text{rms}}$ — a bit slower than the typical molecular speed because sound is a collective wave of density compression, not a single molecule flying across the room.

For air at 300 K: $v_{\text{rms}} \approx 517\;\text{m/s}$, $v_{\text{sound}} \approx 347\;\text{m/s}$ (matches the measured speed of sound in air, ~343 m/s). The Maxwell-Boltzmann distribution directly predicts the speed of sound once you put in the right constants.

Real-world applications

• Atmospheric chemistry. Reaction rates between gas molecules depend on how often they collide with enough energy to react. The high-speed tail of the Maxwell-Boltzmann distribution determines reaction rates via the Arrhenius equation. This is why a small temperature increase can dramatically speed up chemistry — it's the exponential Boltzmann factor at work in the reactive tail.
• Effusion (gas leaking through a small hole). The rate of molecular effusion is proportional to $\bar v/\sqrt{m}$, which means light gases effuse through a hole faster than heavy gases. This is called Graham's law of effusion and is the basis of uranium isotope separation via gas diffusion — U-235 and U-238 hexafluoride have slightly different masses, so they effuse at slightly different rates.
• Thermionic emission. Electrons in a hot metal also follow a Maxwell-Boltzmann-like distribution. Heat the metal enough and the high-energy tail of electrons can boil off the surface — the basis of vacuum tubes, old CRT TVs, and electron microscopes.
• Pressure broadening in spectroscopy. Atoms and molecules emit light at specific frequencies, but the Doppler shift of fast-moving molecules smears the emission lines into broader peaks. The line width is directly proportional to $v_{\text{rms}}$, and astronomers use it to measure the temperature of distant gas clouds.
• Rocket engine design. Exhaust velocity in a rocket nozzle is related to the thermal velocity of the gas at high temperature. Hotter combustion and lighter exhaust molecules both increase thrust, which is why hydrogen-oxygen rockets (light, hot exhaust) are more efficient than kerosene-oxygen rockets.
• Evaporation. Water evaporates because molecules in the high-speed tail of the distribution happen to be moving upward and have enough kinetic energy to escape the liquid's intermolecular bonds. The fraction of high-energy molecules grows exponentially with temperature — which is why hotter water evaporates faster.

Common mistakes

• Using Celsius for $T$. The formula uses absolute temperature, always. Plugging in 25°C instead of 298 K gives completely wrong answers.
• Confusing mean and most probable. Different quantities. The mean is to the right of the peak because the distribution has a right-skewed tail. The most probable is at the peak. Don't call them the same thing.
• Using molar mass instead of molecular mass. The formula wants $m$ in kilograms per molecule, not grams per mole. For N₂ (molar mass 28 g/mol), $m = 0.028 / (6.022 \times 10^{23}) \approx 4.65 \times 10^{-26}\;\text{kg}$. Forgetting the Avogadro division gives numbers off by 24 orders of magnitude.
• Forgetting the $4\pi v^{2}$ factor. Some formulations of Maxwell-Boltzmann look simpler because they give the distribution of velocity components (which doesn't have the $v^{2}$ factor), not the distribution of speeds (which does). These are different functions of different variables. Know which one you're looking at.
• Confusing the distribution with a single speed. Molecules in a gas never have a single speed — they always have a distribution. Asking "what's the speed of a nitrogen molecule at 300 K" is meaningless; the question is "what's $v_{\text{rms}}$ or $\bar v$ or $v_{p}$ at 300 K."
• Assuming the distribution is Gaussian. The Maxwell-Boltzmann is not a Gaussian/normal distribution — it starts at zero at $v = 0$ (because of the $v^{2}$ factor), peaks, then decays. A Gaussian would extend symmetrically to negative values, which doesn't make sense for a speed (which is always positive). MB is often called "Gaussian-like" but it's a different function.
• Thinking heavier molecules can't have high speeds. They can, it's just rarer. Xenon atoms at 300 K mostly move around 200 m/s, but a tiny fraction of them move at 500+ m/s. The tail is just suppressed by the $e^{-mv^2/(2k_BT)}$ factor.

Try it

• Slide the temperature from 100 K to 1500 K. Watch the whole distribution shift right and flatten as $T$ increases. The three marked speeds all move together, maintaining their fixed ratio of $1 : 1.128 : 1.225$.
• Switch gas types with the gas selector. Hydrogen's distribution is way out at high speeds (~2000 m/s peak at 300 K), while xenon's distribution is clustered near 200 m/s. The lighter the molecule, the faster it moves.
• Enable "Show comparison temperature" and pick a different $T$ for the overlay. Two bell curves appear on the same axes. The hotter one is lower and wider than the cooler one (because the total area under each is 1).
• Drag the $v_{\text{low}}$ and $v_{\text{high}}$ sliders to select a speed range. Enable "Show area" to shade the region between them — that shaded area is the probability that a random molecule has a speed in that range.
• Try temperatures up to 1500 K and see how the distribution tail extends. At 1500 K, a meaningful fraction of $N_{2}$ molecules have speeds above 1 km/s — which is why reaction rates at high temperature can be orders of magnitude faster than at room temperature.
• Compute $v_{\text{rms}}$ yourself using the formula $\sqrt{3k_{B}T/m}$ and check it against the HUD. At 300 K for $N_{2}$ you should get roughly 517 m/s.
• Try comparing the same gas at 300 K and 1200 K (4× higher). The $v_{\text{rms}}$ should double — you can see it in the HUD readouts.