Reaction. N₂O₄(g) ⇌ 2 NO₂(g), ΔH° = +57.2 kJ/mol (endothermic), ΔS° = +175.9 J/(mol·K). The equilibrium constant follows from the free energy: Kp = exp(−(ΔH − TΔS)/RT) = 0.147 bar at 298 K, and d ln K/d(1/T) = −ΔH/R (van 't Hoff). Kc = Kp/(R T) in mol/L. ΔH and ΔS are treated as temperature independent.
Why a "stress" shifts the balance. Compare the reaction quotient Q = p(NO₂)²/(p(N₂O₄)·p°) with Kp. If Q < K the reaction runs forward (more brown NO₂), if Q > K backward. Heating raises K, so the equilibrium moves right. Squeezing the piston raises all partial pressures but NO₂ enters squared, so Q rises above K and the system moves left (fewer molecules). Adding Ar at constant pressure expands the gas, lowers partial pressures and moves it right; at constant volume the concentrations are unchanged and nothing moves. Adding NO₂ at constant volume raises Q, so some of it recombines.
Dynamics. Each particle stands for 0.5 mmol. Dissociation is first order, rate kf; recombination is second order, kr = kf/Kc, so the two rates balance exactly at K. Reactions are drawn as random events (the Gillespie algorithm), so the counts fluctuate around the equilibrium as in a real nanoscale sample. kf = 2×10⁴ s⁻¹ at 298 K is an order-of-magnitude value with Ea ≈ ΔH (barrierless recombination); time is shown in slow motion (a screen second is about 1.5 µs at 1×).
Colour. The tint of the gas follows Beer–Lambert transmission 10−εcl through a 1.6 cm path with illustrative absorption coefficients (blue 130, green 40, red 3 L/mol/cm): NO₂ absorbs blue, so more NO₂ looks browner; N₂O₄ is colourless.
Simplifications: ideal gases, molecules do not collide with each other (reactions are random events, not collisions), piston height is drawn on a compressed scale (V is printed exactly), no heat of reaction effect on the bath (the jacket is a perfect thermostat), and the pressure comes from the piston load, not from drawn weights (the discs are a visual cue).
Running numerical checks...