Photoelectric Effect · Guide · glidepathscience.org
GuidePhotoelectric Effect
Contents
This Guide explains the ideas and equations behind every Walkthrough step, Concept Challenge question and Predict round, in the order a textbook would take them. Part 1 prepares the Intro track and Part 2 the Advanced track; each takes about fifteen minutes to read. Each section ends with a worked example that is not one of the sim's own problems.
Part 1 · The essentials
Light arrives in photons
Shine light on a clean metal in a vacuum and electrons can come off it. That much is no surprise: light carries energy, and the free electrons in a metal are only loosely held. The surprise is which light does it. Bright red light on zinc frees no electrons at all, however long it shines, while a faint ultraviolet lamp frees them at once. No picture of light as a smooth classical wave explains that, and the explanation Albert Einstein gave in 1905 helped start quantum physics. This is the photoelectric effect, and it is the subject of the sim.
Einstein proposed that light of frequency $f$ delivers its energy in separate lumps, later called photons. Each one carries
$E = hf = hc/\lambda$
where $h$ = 6.626 × 10⁻³⁴ J·s is Planck's constant, $c$ is the speed of light and $\lambda$ is the wavelength. One photon's energy is tiny in joules, so atomic physics measures it in electronvolts: 1 eV = 1.602 × 10⁻¹⁹ J, the energy an electron gains crossing a potential difference of one volt. In these units $hc$ = 1239.84 eV·nm, so a photon's energy in electronvolts is 1239.84 divided by its wavelength in nanometers. Shorter wavelengths, toward the blue and the ultraviolet, carry more energy per photon.
A metal holds its free electrons with a definite minimum binding. The work function $\phi$ is the least energy that will lift an electron out through the surface. It depends on the metal and on how clean its surface is; in this sim it runs from 2.14 eV for cesium to 5.64 eV for platinum. The photon picture adds one rule: a photon is absorbed whole, by a single electron. If its energy is less than $\phi$, that electron cannot leave, and it cannot store the energy while it waits for a second photon. So a photon frees at most one electron, and a photon below the work function frees none, however many arrive. There is no partial credit: a photon 0.03 eV short is as useless as one 3 eV short.
The usual mistake is to think that brighter light means more energetic photons. Brightness, the power the lamp delivers, sets how many photons arrive each second. The color alone sets the energy of each one, and so whether any electron can leave.
Every metal therefore has a dividing line between light that works and light that does not.
The threshold
Setting a photon's energy equal to the work function marks that line. Light at exactly the threshold frequency $f_0$ brings $hf_0 = \phi$, and any lower frequency frees nothing. In wavelengths the same line is the threshold wavelength $\lambda_0$, the longest wavelength that works:
$f_0 = \phi/h$ $\lambda_0 = hc/\phi$
The threshold is a longest wavelength, not a shortest, because longer waves carry less energy per photon. For most metals it lies in the ultraviolet, which is why ordinary room light does nothing to them and why the effect was first found with electric sparks. In 1887 Heinrich Hertz noticed that ultraviolet light from one spark helped a second spark gap to fire, and that a glass plate between them stopped the effect while a quartz plate did not. Glass absorbs ultraviolet shorter than about 300 nm, which for metals like zinc and copper is exactly the light whose photons carry enough energy; fused quartz passes ultraviolet down to about 160 nm, although oxygen in the air path absorbs below about 190 nm. The window of a phototube is part of the experiment.
In this picture the threshold is perfectly sharp, because every electron at the top of the metal needs exactly $\phi$. That is true for a metal at absolute zero. At room temperature a few electrons sit a little higher and need slightly less, so a weak current trickles out just past $\lambda_0$; Part 2 returns to this tail.
Most slips here are slips of units. With $h$ in J·s the work function must be in joules, and with $c$ in meters per second the wavelength comes out in meters. Working in electronvolts and nanometers, with $hc$ = 1239.84 eV·nm or $h$ = 4.1357 × 10⁻¹⁵ eV·s, avoids both.
Above the threshold, two things about the light can still change: how bright it is, and what color it is.
Brightness counts photons; color sets their energy
A lamp of power $P$ delivers its energy as a stream of photons, each carrying $hc/\lambda$. The number arriving each second is the power divided by one photon's energy:
$N = P\lambda/hc$
Each photon above threshold can free at most one electron, so the number of electrons leaving each second, $N_e$, rises in step with $N$, and the current they carry is $I = eN_e$, with $e$ = 1.602 × 10⁻¹⁹ C. The ratio $N_e/N$ is the quantum yield. For a clean metal just above threshold it is small, about one electron for every 10⁴ to 10⁵ photons: most photons are reflected, or absorbed by electrons too deep in the metal to escape. The sim uses one illustrative yield for every metal. Practical photocathodes, made from compounds of the alkali metals, reach yields of tens of percent.
Doubling a lamp's power at one wavelength therefore doubles the photons, the electrons and the current, but leaves each photon's energy, and so the fastest electron's energy, unchanged. Making the light bluer at the same power does something different: each photon carries more energy, so the electrons leave faster, and because each photon carries more, slightly fewer of them arrive each second, though each is a little more likely to free an electron. Brightness sets how many electrons leave; color sets how fast they go.
Watch the powers of ten in $N = P\lambda/hc$: a microwatt is 10⁻⁶ W, a nanometer is 10⁻⁹ m, and $hc$ = 1.986 × 10⁻²⁵ J·m. Dividing the power by the photon's energy in electronvolts instead of joules gives an answer about 10¹⁹ times too small.
That photons arrive one at a time matters most when the light is dim, which is where the wave picture fails most clearly.
What a wave would predict
In the classical picture light is a wave whose energy spreads evenly over its wavefront. An irradiance $I$, in watts per square meter, delivers $IA$ joules each second to a target of area $A$. An atom presents a target of only about 0.1 nm², $A$ ≈ 10⁻¹⁹ m², so before any of its electrons could gather the work function it would have to soak up energy for a time
$t = \phi/(IA)$
The wave picture makes three predictions: a delay under dim light, emission at any color given enough time, and faster electrons from brighter light. Experiment contradicts all three. Philipp Lenard found in 1902 that the electrons' energy does not depend on the light's intensity. Below the threshold color nothing happens at any intensity. And in 1928 Ernest Lawrence and Jesse Beams showed that emission begins within about 3 × 10⁻⁹ s of the light arriving, however dim it is. The photon picture predicts all of this: the first photon absorbed can free an electron at once, and dimmer light simply means fewer photons. The Wave or Photon? tab runs the two pictures side by side.
Two unit traps wait in $t = \phi/(IA)$. The work function must be in joules (1 eV = 1.602 × 10⁻¹⁹ J), and 0.1 nm² is 0.1 × (10⁻⁹ m)² = 10⁻¹⁹ m², because squaring the area squares the nanometer too.
The photon picture also says how fast the freed electrons move, and a voltage can measure it.
Stopping the fastest electron
An electron that absorbs a photon of energy $hf$ must pay the work function to leave, and keeps the rest as kinetic energy. The least-bound electrons, those at the very top of the metal's filled levels, pay exactly $\phi$ and leave fastest. Einstein's photoelectric equation gives their kinetic energy:
$K_{max} = hf - \phi$
The tube measures it with a voltage. Make the collector negative relative to the cathode, and the field between them pushes electrons back: at a retarding voltage $V$ an electron must climb $eV$ of potential energy to reach the collector. Raise the retarding voltage until even the fastest electron is turned back, and the current falls to zero. That voltage is the stopping voltage. For a collector made of the same metal as the cathode,
$eV_{stop} = K_{max}$
so the stopping voltage in volts is numerically equal to the fastest electron's energy in electronvolts.
The current does not drop to zero all at once. Electrons leave with a spread of energies, from zero up to $K_{max}$: those from deeper in the metal's filled levels have less to spare, and between parallel plates only the motion straight across the gap counts against the field, so a fast electron leaving at a steep angle is turned back early. As the retarding voltage rises it turns back the slowest electrons first, and the current fades smoothly to zero at $V_{stop}$. Brighter light makes the whole curve taller by the same factor and leaves the cutoff where it was; bluer light moves the cutoff to a larger retarding voltage.
Two mistakes are common. The collector's voltage at the cutoff is negative, but $V_{stop}$ is its size, a positive number. And forgetting the work function gives the photon's whole energy instead of $K_{max}$.
Plot the stopping voltage against frequency, and the photoelectric equation becomes a straight line.
Millikan's line: Planck's constant from a graph
Divide the photoelectric equation by $e$, and it predicts how the stopping voltage depends on frequency:
$V_{stop} = (h/e)f - \phi/e$
That is a straight line with slope $h/e$ and intercept $-\phi/e$. The slope is the same for every metal, 4.136 × 10⁻¹⁵ V·s; only the intercept moves with the work function, so the lines for different metals are parallel. Where a line crosses the frequency axis, $V_{stop}$ = 0 and the frequency is the threshold, $f_0 = \phi/h$. Extended back to $f$ = 0, the line meets the voltage axis at $-\phi/e$.
Robert Millikan set out to test this line. He doubted Einstein's light quanta, and he spent about a decade on the measurement, shaving fresh surfaces of alkali metals inside the vacuum to keep them clean. His 1916 data fell on straight lines whose slope gave $h$ within about 1 % of today's value, which has been exact by definition since 2019: $h$ = 6.626 070 15 × 10⁻³⁴ J·s.
Read the intercept with care: it is negative, and the work function is minus it, a positive number. And remember what a fit's uncertainty covers. The scatter of the points about the line gives the slope a standard error, but it says nothing about a mistake that shifts every point the same way, such as judging every zero a little early.
Part 2 opens up the model behind these lines: where the spread of energies comes from, what a collector really measures, and whose work function the intercept gives.
Part 2 · Going further
Inside the metal: the electron sea
Why do photoelectrons leave with a range of energies, and which ones carry $K_{max}$? To a first approximation a metal's conduction electrons behave like a gas of free particles. At absolute zero they fill every state up to the Fermi energy $E_F$. The work function is the climb from that Fermi level to the vacuum level, an electron at rest just outside the surface. An electron deeper in the sea has further to climb.
R. H. Fowler (1931) and Lee DuBridge (1933) added one assumption: an absorbed photon's energy goes into the electron's motion toward the surface, along the normal, and the electron escapes if that normal energy now clears the barrier. If $\varepsilon_\perp = p_\perp^2/2m$ is the electron's normal kinetic energy inside the metal, it escapes when
$\varepsilon_\perp + hf \ge E_F + \phi$
and its sideways, parallel energy rides along unchanged. Counting the escaping electrons of a free-electron gas gives a strikingly simple result. Outside the metal their flux is spread evenly over the normal kinetic energy $K_\perp$ and the parallel energy, up to a total of $K_{max} = hf - \phi$. At $T$ = 0 the total kinetic energy $K$ and its normal part are therefore distributed as
$N(K) \propto K$ $N(K_\perp) \propto K_{max} - K_\perp$
The fastest electrons, with the full $K_{max}$, are those that start at the Fermi level, leaving in any direction; only those that leave straight out carry all of it as normal energy. That difference between total and normal energy is what the next section's two collectors measure. Real metals add band structure and scattering, which Berglund and Spicer's three-step model of 1964 describes; the free-electron model keeps the essentials.
It is tempting to think that the deepest electrons leave fastest, since they absorb the same photon. They leave slowest, because they start furthest below the vacuum level.
Which energy a collector sees depends on its shape.
Plates or a sphere: what a collector measures
Between parallel plates the field is uniform and perpendicular to them, so it acts only on the velocity straight across the gap. An electron leaving at an angle $\theta$ from the normal with kinetic energy $K$ has only its normal part to spend against the field:
$K_\perp = K\cos^2\theta$
The square appears because kinetic energy goes as the square of the velocity, and the normal velocity is $v\cos\theta$. A retarding barrier $U$, which is $eV_{ret}$ for a collector of the cathode's own metal, turns back every electron with $K_\perp < U$. Integrating $N(K_\perp)$ from $U$ up to $K_{max}$ gives the fraction parallel plates collect at 0 K,
$I/I_{sat} = (1 - U/K_{max})^2$
which reaches zero tangentially. The curve flattens as it meets the axis, so its cutoff is hard to pin down by eye.
A small emitter at the center of a spherical collector changes that. The field is radial and the electrons fly straight out along it, so the barrier acts on their total kinetic energy. Integrating $N(K) \propto K$ from $U$ to $K_{max}$ gives
$I/I_{sat} = 1 - (U/K_{max})^2$
which reaches zero with a finite slope: a sharp cutoff at the same voltage. P. Lukirsky and S. Prilezhaev used a spherical collector for exactly this reason in 1928. The formula is the ideal limit of a point-like emitter; a real emitter of finite size gives the electrons some angular momentum about the center, and that part of their energy never counts against the field.
Three mistakes recur: using $\cos\theta$ instead of $\cos^2\theta$; measuring the angle from the surface instead of from the normal, which gives the parallel part, $K\sin^2\theta$; and using the plates' formula for a sphere.
So far the metal has been at absolute zero. Warming it blurs the threshold itself.
Warm metals: the Fermi tail and Fowler's law
At $T$ = 0 the electron sea has a sharp surface. At any real temperature its edge is smeared over a few $kT$, following the Fermi–Dirac distribution: some states just below $E_F$ empty and some just above it fill. At 300 K, $kT$ = 25.85 meV. Electrons above $E_F$ need less than $\phi$ to escape, so light just below the threshold still frees a few of them. Fowler worked out the yield $Y$, the electrons freed per photon, for the free-electron model, with $\mu = (hf - \phi_E)/kT$:
$Y \propto (kT)^2 f(\mu)$
Here $\phi_E$ is the work function of the emitter, the cathode, and Fowler's function is $f(\mu) = -\mathrm{Li}_2(-e^{\mu})$. Far below threshold it becomes an exponential, $f(\mu) \to e^{\mu}$: the yield falls by a factor of $e$ for every $kT$ that the photons fall short, the Fermi tail. Far above threshold $f \to \mu^2/2 + \pi^2/6$, and at $T$ = 0 the yield is simply
$Y \propto (hf - \phi_E)^2$
So near threshold at low temperature $\sqrt{Y}$ is a straight line in $hf$ that reaches zero at $\phi_E$. This is Fowler's method for the cathode's own work function. It needs no stopping voltage and does not depend on the collector's work function, only on saturated currents and photon counts.
At room temperature the points nearest threshold bend away from the straight line, so a line fitted through them crosses zero a little low; Fowler fitted his whole function instead. The square law holds within a few tenths of an electronvolt of threshold; the sim keeps it far above threshold too, which real metals do not. The usual slips are to forget that the threshold is sharp only at 0 K, and to compare photon energies where the law compares the excess energies $hf - \phi_E$.
The yield sees the cathode alone. The stopping voltage, it turns out, does not.
Whose work function? The contact potential
So far the collector has been made of the cathode's own metal. Real tubes rarely are. When the circuit is closed, electrons flow until the two metals' Fermi levels line up, and their vacuum levels then differ by the difference of their work functions. An electron that leaves the cathode with $K_{max} = hf - \phi_E$ must also climb this contact potential, $(\phi_C - \phi_E)/e$, to reach a collector whose work function is $\phi_C$. At the cutoff, then, $eV_{stop} = hf - \phi_E - (\phi_C - \phi_E)$, which is
$eV_{stop} = hf - \phi_C$
The cathode's own work function cancels. It still sets the threshold, whether electrons leave at all, but every stopping voltage, and so the intercept of a Millikan plot, belongs to the collector. Millikan had to measure and correct for this contact potential in 1916. The whole current–voltage curve shifts too: saturation begins at $V = (\phi_C - \phi_E)/e$ instead of at zero. The sign matters here. $V_{stop}$ is the retarding voltage at the cutoff, the cathode’s potential above the collector’s, so it is positive when the collector must be made negative. When $\phi_C$ exceeds $hf$ it is negative: the contact potential alone turns back every electron, and current flows only once the collector is made positive by more than $|V_{stop}|$. The Bench records it with this sign. The two methods therefore measure different things. Fowler's yield sees the cathode, and the stopping voltage sees the collector.
The mistakes are to use the cathode's work function in $eV_{stop} = hf - \phi$ when the collector differs, and to expect the two work functions to average.
A real measurement adds one more question: how do you know where the current reaches zero?
Where zero comes from
A real electrometer has noise, and a real cathode is warm, and both blur the meaning of "the current reaches zero". In practice the cutoff is read where the current sinks into the noise, for instance where it falls to three times the noise's standard deviation. That is the apparent stopping voltage. The current near the cutoff is proportional to the light, so a brighter lamp keeps the curve above that floor further out. At 300 K the Fermi tail also carries current past the ideal cutoff, falling there as $e^{-eV_{ret}/kT}$. The apparent stopping voltage therefore moves with brightness, and for bright light it lies past the ideal value. This is a systematic error: repeating the measurement does not reveal it, because every repeat shifts the same way.
On a logarithmic current scale the tail past the cutoff is a straight line that falls one decade for every
$(kT/e)\ln 10$ = 59.5 mV at 300 K
so the photocurrent doubles as a thermometer. DuBridge (1933) avoided the hunt for zero altogether by fitting whole retarding curves to Fowler's function, which uses every reading rather than the last few.
The mistake is to treat the apparent cutoff as the ideal one, and so to expect it to be independent of brightness, as the ideal one is.
A warm cathode does more than blur the cutoff: it can emit electrons with no light at all.
Electrons without light
The same tail of warm electrons that lets light just below threshold free electrons lets the hottest electrons cross the barrier on their own. This is thermionic emission, and the Richardson–Dushman law gives its current density:
$J = AT^2 e^{-\phi/kT}$
For a free-electron metal $A$ is about 120 A cm⁻² K⁻²; real surfaces give less, and the sim uses an illustrative 60.1. The exponential dominates. At one temperature the ratio of two metals' currents is $e^{\Delta\phi/kT}$ (taking the same $A$ for both, as the sim does), and for a low-work-function cathode a modest warming multiplies the dark current many times over. That is why photomultiplier tubes built for red light, whose photocathodes have work functions near 1 eV, are known for dark counts, and why serious photon counters cool them.
Two slips are common: giving the slowly changing $T^2$ factor the credit that belongs to the exponential, and reporting the exponent in powers of $e$ when the question asks for powers of ten. Divide by ln 10 = 2.303 to convert.
One rule from Part 1 remains to be stretched: one photon per electron.
More than one photon
"One photon, one electron" is a rule for ordinary light. A focused, pulsed laser can deliver photons so densely that an electron absorbs two at once, and then
$K_{max} = 2hf - \phi$
or $nhf - \phi$ for $n$ photons. Light below the one-photon threshold can then free electrons after all, at irradiances far beyond any lamp's. So the statement that brightness never changes $K_{max}$ is a low-intensity rule, true for every lamp in this sim but not for a femtosecond laser. Real photoemission is also richer than the free-electron model. Spicer's three-step model describes optical excitation between real energy bands, travel to the surface with scattering losses, and escape over the barrier. Angle-resolved photoemission (ARPES), the modern form of the experiment, measures each electron's energy and direction together and maps a crystal's bands.
The usual slip is to pay the work function twice: two photons free one electron, which pays $\phi$ once.
Reference
Common mistakes
- Brighter means more energetic photons. Brightness sets the number of photons; color sets the energy of each.
- Waiting for enough energy. Below threshold no electron leaves at any power or after any time (at 0 K).
- Mixing units. Use $hc$ = 1239.84 eV·nm, or convert electronvolts to joules before using $h$ in J·s.
- Forgetting the work function. $K_{max} = hf - \phi$, not $hf$.
- A negative stopping voltage, with matched metals. With a collector of the cathode’s own metal, the collector is negative at the cutoff and $V_{stop}$, the retarding voltage, is a positive number. With a collector of higher work function, $eV_{stop} = hf - \phi_C$ can be negative: the current starts only once the collector is made positive by $|V_{stop}|$.
- The intercept is the work function. The voltage intercept is $-\phi/e$; the frequency intercept is $f_0$.
- Cos for cos². Between plates the normal energy is $K\cos^2\theta$, with $\theta$ from the normal.
- The cathode sets the stopping voltage. With another collector, $eV_{stop} = hf - \phi_C$.
- Natural logs for decades. Divide an exponent in powers of $e$ by ln 10 to get powers of ten.
Equation sheet
| Part 1 | |
|---|---|
| photon | $E = hf = hc/\lambda$; $hc$ = 1239.84 eV·nm |
| Einstein | $K_{max} = hf - \phi$; $eV_{stop} = K_{max}$ (same metals) |
| threshold | $\lambda_0 = hc/\phi$, $f_0 = \phi/h$ |
| photons | $N = P\lambda/hc$ per second; $I = eN_e$, $N_e$ electrons per second |
| wave delay | $t = \phi/(IA)$, $A$ ≈ 10⁻¹⁹ m² |
| Millikan | $V_{stop} = (h/e)f - \phi/e$ |
| constants | $h$ = 6.626 × 10⁻³⁴ J·s = 4.1357 × 10⁻¹⁵ eV·s; 1 eV = 1.602 × 10⁻¹⁹ J |
| Part 2 | |
|---|---|
| plates | $K_\perp = K\cos^2\theta$; $I/I_{sat} = (1 - U/K_{max})^2$ |
| sphere | $I/I_{sat} = 1 - (U/K_{max})^2$ |
| contact | $eV_{stop} = hf - \phi_C$ |
| Fowler | 0 K: $Y \propto (hf - \phi_E)^2$; at $T$: $Y \propto (kT)^2 f(\mu)$, $f(\mu) \to e^{\mu}$ below threshold |
| tail | one decade per $(kT/e)\ln 10$; $kT$ = 25.85 meV at 300 K |
| thermionic | $J = AT^2 e^{-\phi/kT}$ |
| two photons | $K_{max} = 2hf - \phi$ |
Symbols: $f$ frequency · $\lambda$ wavelength (vacuum) · $\phi$ work function; $\phi_E$ the emitter's (cathode's), $\phi_C$ the collector's · $K_{max}$ the fastest electron's kinetic energy · $V_{stop}$ stopping voltage · $P$ power · $N$ photons per second · $N_e$ electrons per second · $Y$ yield, electrons per photon · $I$ current, or irradiance in $t = \phi/(IA)$ · $U$ retarding barrier (eV) · $\theta$ angle from the normal · $\mu = (hf - \phi_E)/kT$ · $T$ temperature · $J$ current density · $A$ atom's cross section, or the Richardson constant.
Further reading
OpenStax, University Physics Volume 3, section 6.2, "Photoelectric Effect", an open textbook that treats the Intro track's ideas; R. H. Fowler, Phys. Rev. 38, 45 (1931), and L. A. DuBridge, Phys. Rev. 43, 727 (1933), for Part 2. The Field Manual lists every source, the history and the model's limits.
Photoelectric Effect v1.5 · © 2026 Dr. David Meier · CC BY-NC-SA 4.0 · glidepathscience.org