Conduction is drift between collisions
Guide one gave us resistivity as an intrinsic property — a number that sorts materials across more than twenty orders of magnitude — and guide two explained why metals have carriers at all: their conduction band overlaps the valence band, so a sea of electrons is free to move the instant you apply a field. Now we answer the question that trips up almost everyone: if the electrons are free, why do they get stopped? Why does a copper wire have any resistance at all, and why does that resistance climb when the wire gets hot or dirty? The whole answer is one shift in mental picture.
Do not picture a free electron flying down an open highway. Picture a pinball table. Switch on the voltage and the electron does start to accelerate along the field — but almost immediately it slams into something and its motion is randomized, then it accelerates again, slams again, over and over. What actually reaches the far end of the wire is not a fast bullet but a slow, jostling drift on top of frantic random bouncing. The drift is astonishingly gentle: in a normal wire carrying a real current the electrons drift only a fraction of a millimetre per second, even though each electron between collisions is moving roughly a million metres per second. Resistance is simply the tax those collisions charge on the drift.
This turns conductivity into a tidy product of three things: sigma = n x e x mu. Here n is how many carriers you have per cubic metre, e is the charge each one carries, and mu is the mobility — how freely each carrier drifts for a given push, which is set entirely by how often it collides. In a metal n is fixed and gigantic (around 10^28 to 10^29 electrons per cubic metre, and it barely changes when you heat the metal). So for a metal the charge count is off the table: everything about its resistivity rides on the mobility — that is, on the collision rate. Make the electrons collide more often and resistivity rises; let them coast farther between hits and it falls. Keep that one lever in mind, because the rest of the guide just pulls it two ways: with heat, and with dirt.
Why heat slows a metal down
So what does an electron collide with? Not, as you might guess, with the atoms sitting neatly in their lattice sites — a perfectly periodic crystal is, quantum-mechanically, transparent to an electron wave; it would glide through a flawless, motionless lattice forever. What actually scatters the electron is anything that breaks that perfect periodicity. And heat breaks it constantly. Raise the temperature and every atom vibrates harder about its site; those thermal vibrations are quantized ripples in the lattice called phonons, the very same lattice waves that carry heat through an insulator. A hotter metal is a lattice sloshing with more, bigger phonons — more wobbling targets for the drifting electron to hit.
More phonons means shorter flights between collisions, lower mobility, and so higher resistivity. Above roughly room temperature the number of phonons grows in proportion to temperature, so a metal's resistivity rises very nearly linearly with T — a clean, almost straight line. Engineers capture this with a temperature coefficient of resistance: for copper it is about 0.0039 per degree C. Put in numbers: warm a copper wire from 20 to 120 degrees C, a rise of 100 degrees, and its resistivity climbs by about 0.0039 x 100, roughly 39 percent. This is not a footnote — it is why the same motor winding draws a different current cold than hot, and why resistance-based thermometers work at all.
Why dirt slows it too — and the two effects simply add
Heat is not the only way to break the lattice's perfect periodicity. Any defect does it too — and every real crystal is full of them. A foreign atom dissolved into the lattice, a vacancy, a grain boundary where two crystal patches meet at mismatched angles like floor tiles laid crooked, a tangle of dislocations left by cold working — each one is a local disruption the drifting electron can scatter off. Crucially, these defects sit there whether the metal is hot or cold, so the resistance they add is roughly independent of temperature. Even at absolute zero, where the phonons have all frozen out, this defect scattering remains: it is called the residual resistivity, the floor a metal's resistance can never drop below.
The beautiful simplification is that the two contributions just add up: total resistivity equals the temperature-dependent phonon part plus the temperature-independent defect part. That is Matthiessen's rule, rho = rho_thermal(T) + rho_residual. It says the two causes barely talk to each other — a hot dirty wire is a hot clean wire's resistivity plus a fixed dirt penalty stacked on top. It is only an approximation (the mechanisms are not perfectly independent), but it is a superb one, and it is how you should reason about metal conductors: heat sets the sloped part, purity sets the height of the floor.
Now the honest engineering sting: this is exactly why the best conductors are pure metals, and why alloying — which we prize for making metals strong — almost always makes them worse at carrying current. Dissolve about 30 percent zinc into copper to make brass and every zinc atom is a scattering centre; brass conducts only around a quarter as well as pure copper, its resistivity roughly three to four times higher. So there is a real, unavoidable tradeoff: brass is stronger and machines beautifully, but you would never make a power line out of it — transmission lines are near-pure aluminium or copper precisely to keep that residual floor as low as possible. Strength and conductivity pull against each other, because the same dissolved atoms that pin dislocations also scatter electrons.
RESISTIVITY vs TEMPERATURE (why the two material classes go opposite ways)
rho rho
^ METAL (rises, ~linear) ^ SEMICONDUCTOR (falls, steeply)
| / | \
| / <- thermal (phonon) | \
| / part grows with T | \___
| / | \_____
|----/------------------------ | \________
|___/ <- residual (impurity) floor | \____
+-------------------------> T +-------------------------> T
0K heating --> cold heating -->
Matthiessen: rho_metal = rho_thermal(T) + rho_residual(impurities,defects)
metal: n fixed -> hotter = more phonons = more scattering = higher rho
semi: n ~ exp(-Eg/2kT) -> hotter = MANY more carriers, swamps everything = lower rhoThe semiconductor does the exact opposite
Here is the twist that catches people out, and it falls straight out of guide two's band picture. In a metal, heat can only hurt conduction because the carrier count n is already fixed and enormous — heat has nothing to add there, so all it can do is scatter. A semiconductor is the mirror image. Its valence and conduction bands are separated by a band gap — the step an electron must jump before it can conduct — so at room temperature almost no electrons have made the jump and n is tiny (in pure silicon only about 10^16 carriers per cubic metre, roughly a trillion times fewer than copper's). That microscopic n, not the mobility, is what makes an intrinsic semiconductor so resistive.
Now heat it. Thermal energy is exactly what lifts electrons across the gap, and the number that make it across grows exponentially, roughly as exp(-Eg / 2kT). This is a ferociously steep dependence: warming silicon a little kicks a flood of new electrons up into the conduction band (and leaves an equal flood of holes behind). The mobility still sags with temperature, just as in a metal — but that gentle sag is utterly swamped by an exponential explosion in the number of carriers. So a semiconductor's resistivity falls, and falls steeply, as it heats. Same heat, opposite outcome, for one reason: in a metal you already had all your carriers, while in a semiconductor heat manufactures them.
Reading it off: sensors, heaters, and one honest limit
This one framework quietly designs a lot of hardware. Because a metal's resistance rises so cleanly and linearly with temperature, you can run the logic backwards and measure temperature by measuring resistance: that is a platinum RTD, the workhorse precision thermometer in industry. Because a semiconductor's resistance falls so steeply with temperature, you get a thermistor — far more sensitive over a narrow range, the little bead that reads your engine coolant or a lithium battery pack. Two sensors, opposite slopes, both straight off this guide.
The same 'purity flattens, dirt raises the floor' idea explains a second family of parts. Precision resistors and resistance standards are made from special alloys like manganin and constantan, tuned so their two temperature effects nearly cancel and the resistance barely drifts with heat. Heating elements go the other way: nichrome is deliberately a dirty, high-resistivity alloy, around sixty times more resistive than copper, so it dissipates a lot of power and glows without melting. And there is a lovely bonus — because in a metal the same free electrons carry both charge and heat, good electrical conductors are also good thermal conductors, a proportionality captured by the Wiedemann-Franz law. Copper is prized for pans and heat sinks for the very reason it is prized for wire.
Be honest about where this scattering picture stops. Everything above says a metal's resistivity heads toward its residual floor as you cool it but can never reach zero, because some defects always remain. That is true — except that certain metals, cooled below a critical temperature, drop abruptly to exactly zero resistivity and stay there. That is superconductivity, and it is not a cleaner version of low scattering; it is a different quantum mechanism (electrons pairing up and moving as one condensate) that this drift-and-collision model simply does not contain. The pinball picture is powerful and right for ordinary conductors and semiconductors — just know its edge, and do not stretch it past the cliff where a whole new physics takes over.