The Gold Standard of Vacuum Science | Why a Freely Effusing Reservoir Can Cool as It Empties

Imagine a reservoir containing a dilute gas. One observation selects a particle uniformly from all particles inside at an instant. Another selects a particle that crosses a small opening during a short interval. Those selections need not produce the same speed or energy distribution.

A faster particle can reach an opening more often. Direction matters too: only velocity components pointing through the opening contribute to outward crossings. The escaping sample is therefore weighted by crossing rate, rather than simply copied from an inventory list.

This is the first distinction to stabilise before introducing any formula. A population and a flow drawn from it can have different averages even when no special gate explicitly measures and selects particle energy. The physical crossing process itself supplies a selection mechanism.

Keep the measurement question visible. We will investigate the reservoir’s changing particle number, temperature and pressure. The energy of its outgoing particles is an intermediate quantity needed to explain that change, not a replacement definition of the reservoir temperature.

Contents · Next chapter

Use a rigid reservoir of fixed volume V containing a classical ideal monatomic gas. Assume its interior stays sufficiently close to a Maxwellian distribution with a well-defined instantaneous temperature T and particle number N. Particle escape is slow relative to the internal redistribution needed for that description.

The population is assumed macroscopic, so continuous number and temperature balances are meaningful rather than a description of a last few individual particles. The opening is small enough for effusive transport, with negligible collisions inside the opening and negligible incoming gas from outside. Its effective transmission is represented by a fixed area A. We neglect gas release from surfaces, leaks elsewhere, changes in volume and energy supplied through the walls during the model interval.

These conditions are a linked contract. A hole can be small relative to the vessel while still requiring a transport check relative to molecular collision lengths. A rarefied opening does not imply that the whole interior must be collisionless.

Real reservoir walls commonly exchange energy with gas. We temporarily exclude that exchange to make the outgoing-energy selection visible. Later chapters restore a heat term and show why a familiar constant-temperature result can then return.

Contents · Previous chapter · Next chapter

Choose the z direction normal to the opening, positive outwards. In the Maxwellian model, each Cartesian velocity component has a centred Gaussian distribution. The components are independent, with mean square kBT/m, where m is particle mass and kB is Boltzmann’s constant.

The average translational kinetic energy inside is therefore one half m times the sum of three component mean squares, giving 3kBT/2. Each direction contributes kBT/2. This is an inventory average over particles inside the reservoir.

For crossings, weight the distribution by positive vz. A particle travelling sideways with vz near zero contributes little to outward flux, even if its sideways speed is large. A particle with negative vz is moving away from the opening and does not contribute to the outward crossing count at that moment.

The weighting changes the normal component’s distribution. It does not change the independent sideways component distributions in this ideal factorised model. That separation provides a particularly clear route to the escaping mean energy.

Contents · Previous chapter · Next chapter

Write a = m/(2kBT). For the positive normal velocities, the crossing-weighted mean square is the ratio of the integrals of vz³ exp(−avz²) and vz exp(−avz²), each integrated from zero to infinity.

The denominator is 1/(2a), and the numerator is 1/(2a²). Their ratio is 1/a = 2kBT/m. Thus the normal kinetic-energy contribution among escaping particles is kBT, twice the inventory contribution kBT/2.

Each sideways contribution remains kBT/2 because its distribution is independent of the normal crossing weight. Adding the three contributions gives escaping mean translational energy 2kBT.

Notice what has and has not doubled. The normal energy contribution doubles; the total escaping mean energy is four thirds of the interior mean, not twice it. Writing the three contributions separately avoids that common inference error.

This derivation uses a distribution of particle velocities, not an assertion that every escaping particle has exactly the mean energy. Individual crossing energies still vary.

Contents · Previous chapter · Next chapter

Integrating the positive normal velocity gives outward number flux per area J = n√(kBT/(2πm)), where n = N/V is number density. Multiplying by A gives the number escaping per time.

Consequently dN/dt = −(A/V)N√(kBT/(2πm)). Define K = (A/V)√(kB/(2πm)), so the compact equation is dN/dt = −KN√T.

The negative sign describes loss from the reservoir. The coefficient K has units of inverse time per square root kelvin, so K√T has units of inverse time. It is not itself a fixed number of particles removed each second.

At a fixed temperature, the model removes a fixed fraction per time. When temperature changes, that fractional loss rate changes too. This is why inserting a constant rate into the entire cooling trajectory would silently remove one of the relationships we are trying to follow.

A quick limiting check is useful: setting A to zero gives no outward loss. Doubling V at fixed N and T halves number density and therefore halves the initial crossing rate through the unchanged opening.

Contents · Previous chapter · Next chapter

The reservoir’s internal translational energy is U = 3NkBT/2. During a short loss interval dN is negative. Each departing particle carries mean energy 2kBT evaluated at the reservoir’s current temperature.

With no heat input or other included energy transfer, dU = 2kBT dN. Both sides are negative during escape. Using the current temperature matters because the outgoing mean energy changes as the reservoir cools.

Differentiate U: dU = (3kB/2)(N dT + T dN). Equating the two expressions and cancelling kB gives (3/2)N dT = (1/2)T dN.

Thus dT/T = dN/(3N). Since dN is negative, dT is negative too. The sign follows from the energy selection rather than from a slogan that losing gas must always cool a container.

A different escape process or substantial heat transfer could give another relation. The balance equation is useful because it makes exactly which energy accompanies the lost particles visible.

Contents · Previous chapter · Next chapter

Integrating dT/T = dN/(3N) from the initial state gives ln(T/T₀) = (1/3)ln(N/N₀). Hence T/T₀ = (N/N₀)^(1/3).

Let f = N/N₀ denote the remaining particle fraction. Then T/T₀ = f^(1/3). The rigid-volume ideal-gas law pV = NkBT gives p/p₀ = f^(4/3). Internal energy has the same ratio f^(4/3), since both pV and U are proportional to NT in this monatomic model.

These exponents have different jobs. A particle fraction is not a pressure fraction, and temperature does not fall in direct proportion to N. Naming the ratios before substituting helps prevent a correct exponent from being attached to the wrong quantity.

There is no arbitrary choice of the temperature scale’s zero here. T and T₀ are absolute temperatures in kelvin. Using Celsius values in these ratios would not preserve the kinetic-energy relationship from which the equation was derived.

Contents · Previous chapter · Next chapter

Suppose the original gas temperature is 320 K and exactly one eighth of the original particle population remains within the ideal model. The cube root of one eighth is one half, so the new temperature is 160 K.

The pressure fraction is (1/8)^(4/3) = 1/16. A starting pressure of 0.032 Pa would therefore correspond to 0.002 Pa. These are invented numbers illustrating the model, not proposed apparatus conditions.

If the original population were 8 × 10¹² particles, the remaining population would be 10¹². The population fell eightfold, temperature twofold and pressure sixteenfold. Reconstruct the pressure ratio by multiplying the first two fractions: (1/8)(1/2) = 1/16.

At fixed volume, internal energy also falls to one sixteenth. The escaped particles have removed fifteen sixteenths of the original internal energy while accounting for seven eighths of the original population. That difference is the accumulated consequence of energy-biased escape, not a violation of conservation.

Contents · Previous chapter · Next chapter

If half the particles remain, f = 1/2. Temperature is approximately 0.793701 T₀ and pressure approximately 0.396850 p₀. A half-population condition is therefore already below half the original pressure in this cooling model.

If instead pressure is observed to be half its original value, solve f^(4/3) = 1/2. This gives f = 2^(−3/4), approximately 0.594604. Temperature is then 2^(−1/4), approximately 0.840896 T₀.

The two observations answer different inverse questions. A pressure trace alone cannot be converted to particle fraction by assuming unchanged temperature when the model explicitly predicts cooling.

For an independent check, raise the inferred temperature ratio to the third power and recover the inferred particle fraction. Then multiply those two ratios to recover one half. This checks the relationship in both directions rather than merely repeating the rearrangement that produced it.

The guide on temperature change and gas accumulation develops the wider measurement distinction without assuming this particular escape mechanism.

Contents · Previous chapter · Next chapter

Substitute T = T₀f^(1/3) into the number balance. It becomes df/dt = −αf^(7/6), where α = K√T₀ is the initial fractional particle-loss rate.

Separate variables and use f(0) = 1. The result is f^(−1/6) = 1 + αt/6. Define τ = 6/α, giving N/N₀ = (1 + t/τ)^(−6).

The temperature and pressure ratios follow without solving another differential equation: T/T₀ = (1 + t/τ)^(−2), and p/p₀ = (1 + t/τ)^(−8).

These are power-law expressions in the shifted time coordinate, rather than exponentials. Their logarithmic slopes change with time because cooling reduces the speed scale feeding the opening.

Check the initial derivative. Differentiating the number ratio at zero gives −6/τ = −α, agreeing with the original flux equation at the initial temperature. This local check catches a missing factor of six that would otherwise affect every predicted time while leaving the fraction-to-temperature relation unchanged.

For a further timing check, the pressure half-time is τ[2^(1/8) − 1], approximately 0.090508τ. The particle half-time is τ[2^(1/6) − 1], approximately 0.122462τ. They differ because pressure also tracks the falling temperature.

Neither interval equals τ. The symbol τ was defined as a convenient time scale in a power-law solution, not as the time for a fixed exponential fraction to remain. Defining a symbol is part of the model; a familiar letter does not bring every property it had in another equation.

A learner should obtain each time by setting the relevant ratio equal to one half, solving for 1 + t/τ and then subtracting one. Reusing the particle exponent for pressure would preserve plausible-looking arithmetic while answering the wrong endpoint question.

Contents · Previous chapter · Next chapter

Use y = 1 + t/τ to keep the arithmetic compact. Particle, temperature and pressure ratios are y^(−6), y^(−2) and y^(−8) respectively. The identity p/p₀ = (N/N₀)(T/T₀) holds at every listed state.

The last rows are mathematical consequences of continuing the ideal assumptions. They are not evidence that a real reservoir can follow the same model indefinitely. Thermal contact, changing collision conditions, surface gas and instrument limitations may become important well before such large reductions.

Read the table by relationships rather than memorising its entries. Squaring y gives the inverse temperature ratio; cubing that temperature ratio gives the particle fraction. The combined relationships provide several independent arithmetic checks.

Contents · Previous chapter · Next chapter

Now explicitly maintain the gas at T₀ by supplying the required heat. The number balance becomes dN/dt = −αN, with solution N/N₀ = exp(−αt). Pressure has the same ratio because temperature is fixed.

At the same initial state and opening coefficient, both models have the same initial particle-loss rate. Later, the maintained-temperature reservoir empties more rapidly in particle-number terms because it does not lose the molecular speed scale through cooling.

Pressure can nevertheless fall differently from particle number in the cooling reservoir. Comparing pressure curves without tracking both variables can therefore create a misleading intuition about which reservoir has removed more particles.

For a concrete comparison, at t = τ the cooling model has particle fraction 1/64 and pressure fraction 1/256. The maintained-temperature model has exp(−6), approximately 0.002479, for both fractions. Its remaining population is smaller than 1/64, although that conclusion should be derived from N rather than guessed from a temperature label.

Contents · Previous chapter · Next chapter

Restore a heat-input rate H, defined positive into the gas. The balances become dU/dt = H + 2kBT dN/dt and U = 3NkBT/2.

Differentiation gives (3kB/2)N dT/dt = H + (kBT/2)dN/dt. Since dN/dt is negative, the escape term cools the reservoir when H is zero.

To hold T constant, set the left side to zero. The required heat rate is H = −(kBT/2)dN/dt, a positive quantity. Each lost particle needs an additional half kBT of heat input to offset the difference between its escaping mean energy and the interior energy per particle.

This is an energy-accounting result under the stated model, not a heater specification. Real thermal contact involves surfaces, geometry and transport. The value here identifies what the simplified maintained-temperature assumption asks the environment to provide.

The exercise also reveals why an apparently innocuous phrase such as constant temperature carries physical content. It may require an energy pathway that a closed mathematical equation has left implicit.

Contents · Previous chapter · Next chapter

A gas expanding into another initially empty part of an isolated rigid container is a different system boundary. If no particles leave the combined container and no energy crosses its exterior, the ideal-gas total internal energy stays fixed. That familiar free-expansion case need not have the cooling law derived here.

Our reservoir is an open subsystem. Particles and their kinetic energy cross its boundary, and the escaping sample is energy-weighted. We are following what remains inside rather than the combined original gas wherever it has travelled.

Changing only the word vacuum does not make those processes equivalent. Draw the boundary, identify whether N changes and write the energy transfer before choosing a temperature relation.

For a learner, compare three paper sketches: a partition removed inside a sealed vessel, an effusing reservoir losing particles outside its boundary, and a constant-temperature reservoir exchanging heat. Ask which terms appear in each balance. This method resolves the apparent contradiction more reliably than memorising that expansion either does or does not cool gas.

Contents · Previous chapter · Next chapter

The existing guide on temperature-driven gas pumping examines opposing transfers between regions held at different temperatures. Its zero-net-flow condition and reservoir particle balance address that linked two-region problem.

The present model instead follows outward loss into negligible back pressure while the remaining reservoir temperature evolves from its own energy balance. We do not impose a hot-side/cold-side pressure ratio or maintain two reservoir temperatures.

Both problems use crossing-weighted molecular statistics. Their boundary conditions and questions differ. Reusing one result without its conditions can turn a correct formula into an incorrect interpretation.

For a transfer check, ask what would change if an external reservoir supplied an appreciable incoming flux. The net particle balance would need both directions, and the energy balance would need the energy carried by each direction. The one-way cooling law could no longer be assumed unchanged merely because the opening is still small.

Contents · Previous chapter · Next chapter

Our derivation uses one monatomic species with translational internal energy 3NkBT/2. It does not automatically apply unchanged to molecules with active rotational or vibrational energy storage, mixtures undergoing fractionation, adsorption or chemical reactions.

Even when a common temperature can be assigned, different species have different crossing speed scales through their particle masses. Their relative populations may change during escape. A single mean particle mass inserted once at the start can conceal that changing composition.

The guide on residual gas fractions and incoming gas fractions develops a different species-dependent removal problem. Use it to strengthen the distinction between an inventory fraction and a flow fraction.

For this article’s calculations, keep the single-species contract. Extending a model responsibly means adding the relevant balances and evidence, not simply replacing one symbol with a more complicated name. The original result remains useful precisely because its assumptions can be stated and checked.

Contents · Previous chapter · Next chapter

The temperature law assumes the remaining gas can be described by one instantaneous Maxwellian distribution. If escape strongly distorts that distribution before collisions redistribute energy, a single temperature may no longer capture the state needed for the flux calculation.

The opening’s effusive condition and the reservoir’s equilibration condition concern different length and time scales. Collisions can be rare inside a tiny opening yet frequent enough over the larger reservoir’s paths to support the interior approximation.

A useful paper investigation compares a characteristic internal redistribution time with a characteristic fractional-loss time. The latter is of order 1/(K√T) in our number balance. We have not supplied enough physical information to certify either timescale for actual apparatus.

As N and T change, those scale comparisons can change too. A model valid near its initial state may not remain valid throughout an arbitrarily long ideal trajectory. State the interval of intended application rather than treating the analytic solution as evidence that its premises sustain themselves forever.

Contents · Previous chapter · Next chapter

From p/p₀ = (1 + t/τ)^(−8), the logarithmic pressure slope is d(ln p)/dt = −8/(τ + t). Its magnitude decreases with time.

If someone fits a single exponential over a short interval, it may approximate part of this curve. The fitted decay coefficient then describes that interval’s approximation, not a constant physical coefficient established for the whole record.

The particle logarithmic slope is −6/(τ + t). Their ratio is four thirds, matching the pressure–particle relation. Confusing them would overstate the fractional particle-loss rate inferred from a pressure slope in this cooling model.

A local tangent and a global model are different objects. Before interpreting a straight segment on a semilog plot, ask which state variables were held fixed, which changed and whether the model supplies a constant logarithmic derivative. The guide on two connected pump-down volumes explores another reason a pressure trace can depart from one simple exponential.

An additional model check is available without fitting an exponential. Taking the inverse square root of the temperature ratio gives (T/T₀)^(−1/2) = 1 + t/τ. The transformed temperature is linear in time under the ideal contract. The inverse sixth root of the particle ratio and inverse eighth root of the pressure ratio give the same line.

Agreement among these transformations is stronger internal consistency evidence than a visual impression that one trace falls smoothly. It still does not uniquely establish the physical mechanism: other processes or correlated measurement errors could produce a similar finite record. The transformed comparison is a prediction to test alongside the assumptions, not an automatic certification of them.

Contents · Previous chapter · Next chapter

At one time, a pressure ratio of one sixteenth could arise in the cooling example with particle fraction one eighth and temperature fraction one half. It could also arise in an isothermal ideal-gas state with particle fraction one sixteenth.

These observations share a pressure ratio and a fixed geometric volume, yet imply different inventories. The pressure number alone does not select the thermal history or escape mechanism.

Independent temperature evidence can help reconstruct the inventory under an appropriate ideal-gas interpretation. Independent knowledge of opening transport, thermal coupling and gas sources can help test the proposed dynamical explanation. Fitting the pressure curve alone does not make each assumption independently observed.

The wider lesson is to distinguish a forward calculation from an inverse identification. Given our full contract, the pressure trajectory follows. Given a pressure trajectory, several contracts may still be possible. A useful report names what was stipulated, what was measured and what alternative explanation remains compatible with the record.

Contents · Previous chapter · Next chapter

The model predicts a defined reservoir pressure. A real gauge responds through its own physical mechanism, location, calibration and response time. Some gauge responses also depend on gas species or temperature in ways needing a separate interpretation.

An indicated pressure and the model’s p should not be declared identical merely because they share a unit. The evidence needs to show why that measurement represents the intended reservoir quantity over the relevant range and time interval.

The guide on gauge offset and gas load develops one concrete ambiguity. The guide on connection conductance from two readings keeps measurement locations explicit.

For our educational exercise, use supplied ideal state variables. Do not claim that the exact mathematical cooling law has been confirmed by a display unless the measurement pathway and competing effects have actually been examined. More displayed digits do not remove an unidentified physical relationship.

Contents · Previous chapter · Next chapter

At remaining fraction f, internal energy is U₀f^(4/3). With no heat input, cumulative escaped energy is U₀[1 − f^(4/3)]. This gives an independent global conservation check on the differential balance.

For f = 1/8, cumulative escaped energy is 15U₀/16. Divide by the escaped particle number 7N₀/8. Since U₀/N₀ = 3kBT₀/2, the mean energy per particle over the entire escaped population is 45kBT₀/28, approximately 1.60714kBT₀.

That average is below the initial instantaneous escape mean 2kBT₀ because later particles leave a cooler reservoir. It is above the initial inventory mean 1.5kBT₀ for this particular cumulative interval. These are three different averages with three different selection rules.

The distinction prevents another tempting shortcut: multiplying every escaped particle by the initial 2kBT₀ over the whole process. That would ignore the falling instantaneous temperature and remove too much energy from the account.

Contents · Previous chapter · Next chapter

Double the reservoir volume and initial particle number together while holding initial temperature, opening area and initial number density fixed. The initial absolute escape rate is unchanged because J depends on n and T, which are unchanged.

But the initial fractional-loss rate α is halved because the same outward count is now drawn from twice the inventory. Thus τ doubles. At equal scaled time t/τ the two models share the same particle, temperature and pressure fractions.

At the same physical time, they generally do not share those fractions. The larger reservoir has progressed less far along the depletion trajectory. This is a useful separation between absolute flow, fractional depletion and geometric scale.

The result assumes the same valid interior equilibration and thermal-isolation conditions for both hypothetical reservoirs. Scaling volume may change real thermal and collision behaviour. The numerical relationship follows from the specified model; it is not a general design claim that doubling a vessel automatically doubles every useful operating time.

Contents · Previous chapter · Next chapter

With all other model quantities fixed, doubling A doubles K and α, halving τ. The fraction-to-temperature law T/T₀ = f^(1/3) remains unchanged. Geometry alters the speed of movement along that relationship rather than its exponent.

If the initial temperature is quadrupled while N₀, V, A and particle mass are unchanged, α doubles through √T₀. Initial pressure quadruples through the ideal-gas law. One change therefore affects both the initial state scale and the time scale.

Ask the learner to state both effects before calculating. Saying the process becomes twice as fast misses the fourfold pressure scale; saying pressure quadruples misses the changing initial crossing rate.

A compact nondimensional account keeps these roles separate: initial p₀ and T₀ set state scales, while τ sets the time scale. The analytic ratio curves can then be compared without pretending that equal dimensionless coordinates imply equal absolute pressures or temperatures.

Contents · Previous chapter · Next chapter

Start a fictional monatomic reservoir at 405 K. Suppose a later state contains 8/27 of its original particle number under the same ideal effusive-cooling contract. Find its temperature ratio, pressure ratio and remaining internal-energy ratio.

The cube root of 8/27 is 2/3. Temperature is therefore 270 K. The pressure ratio is (8/27)(2/3) = 16/81, and internal energy has the same ratio.

Now reverse the question. A pressure ratio of 16/81 in this model implies particle fraction (16/81)^(3/4), which is 8/27, and temperature fraction (16/81)^(1/4), which is 2/3. Reconstructing both directions checks the inverse exponents.

Finally remove the effusive-cooling assumption while retaining only that pressure ratio and fixed volume. The unique particle answer disappears unless temperature or another relation is supplied. A learner demonstrates transfer when they can explain why the same numerical pressure statement supports a unique answer in one contract and not in the other.

Contents · Previous chapter · Next chapter

If a learner treats escaping particles as a fair sample, return to positive normal velocity and crossing weights. If they confuse pressure with inventory, reconstruct pV = NkBT with a changed temperature. If they lose the sign in the energy balance, write a short interval in which a known number of particles leaves.

Stabilise the repaired idea using a nearby exact fraction such as one eighth. Then remove a cue: provide pressure instead of particle fraction, or ask for cumulative escaped energy rather than remaining energy.

For extension, restore a heat-input term or compare two reservoir sizes. Require an explanation of which equation changes and which relationship remains valid. The aim is selective model use rather than a long list of formulas remembered without their conditions.

Parents and teachers can use the paper cases without equipment. A strong explanation names the population, crossing selection, boundary and energy pathway. Those habits transfer into measurement and open-system reasoning well beyond this one ideal gas.

Contents · Previous chapter · Next chapter

Does every gas leak cool its container by this law?

No. The result belongs to the stated effusive, monatomic, well-equilibrated and negligible-heat-input model. Other transport regimes, gas sources or thermal coupling require different balances.

Is the escaping gas simply at a higher equilibrium temperature?

The escaping sample has a crossing-weighted energy distribution. Its mean energy can be reported, but it is not automatically an isotropic equilibrium reservoir described by the same temperature definition.

Does zero heat input mean zero energy transfer?

No. Escaping particles carry energy across the reservoir boundary. Heat input and energy transported with matter are different terms.

Can a pressure curve alone prove effusive cooling?

It can be compared with a prediction, but identifying the mechanism needs the relevant transport, temperature, source and measurement evidence. A fitted curve does not independently verify every premise.

Contents · Previous chapter · Next chapter

A reproducible paper calculation records initial N, V and T or an equivalent complete state, the opening coefficient, the gas assumption and the chosen thermal boundary. It also states whether the result concerns instantaneous flow, remaining inventory or cumulative escaped energy.

For an actual scientific comparison, measurements would need appropriate calibration, temporal alignment and a clear relationship to those model quantities. The evidence should distinguish cooling from gas removal, additional sources and gauge response, with qualified apparatus procedures outside this teaching exercise.

Save alternatives as well as the favoured fit. A maintained-temperature prediction, a heat-coupled model and the ideal cooling prediction answer different conditional questions. Their disagreement can identify which additional observation would be useful.

The smallest meaningful next step is often a better-defined quantity rather than a more elaborate numerical solver. Decide what the pressure represents and which energy transfers belong in the boundary before increasing the computational detail.

Contents · Previous chapter · Next chapter

Rebuild the vacuum quantities

Return to Vacuum Science for pressure, particles, surfaces and transport. Its temperature, composition, gauge-offset, conductance and connected-volume branches linked above develop different parts of the measurement account.

Trace the specialist foundation

Missouri University of Science and Technology’s January 2023 qualifying examination, problem B6 provides a primary academic route into ideal-gas effusion and its energy balance. The numerical examples and changed-context checks here are original teaching calculations. Source checked on 25 October 2026.

Transfer the reasoning

Continue through Gold Standard Education and How X Works. In a new open-system problem, ask what crosses the boundary, how that crossing selects the sample and which conserved account links the flow to what remains.

Contents · Previous chapter · Return to Vacuum Science