Wave-Particle Duality and the Double-Slit Experiment
Wave-Particle Duality and the Double-Slit Experiment
The concept of wave-particle duality sits at the heart of quantum mechanics and represents one of the most profound departures from classical intuition in the history of science. In classical physics, particles and waves are categorically distinct: particles are localized, discrete, and follow deterministic trajectories; waves are extended, continuous, and superpose. The experiments of the early twentieth century demolished this clean separation. Electromagnetic radiation — long understood as a wave — behaved as if composed of discrete quanta (photons) in the photoelectric effect and Compton scattering. Conversely, electrons — paradigmatic point particles — produced interference fringes in double-slit experiments, a behavior exclusive to waves. The synthesis of these observations into a coherent framework required entirely new mathematics: the quantum-mechanical wave function and its probabilistic interpretation.
Louis de Broglie's 1924 hypothesis generalized the wave-matter connection to all particles. If photons of energy $E = hf$ and momentum $p = h/\lambda$ exhibit wave-like behavior, then by relativistic symmetry, any particle with momentum $p$ should have an associated wavelength $\lambda = h/p$. For macroscopic objects, this de Broglie wavelength is fantastically small — a 1 kg ball moving at 1 m/s has $\lambda \approx 6.6 \times 10^{-34}$ m, far smaller than any resolvable scale. But for electrons with kinetic energies of tens of eV, $\lambda$ is on the order of atomic spacings (Ångströms), making diffraction from crystal lattices observable. The 1927 Davisson-Germer experiment confirmed this directly: electrons scattered from a nickel crystal exhibited diffraction maxima consistent with de Broglie's formula to high precision.
The double-slit experiment, famously described by Feynman as containing "the only mystery" of quantum mechanics, distills wave-particle duality to its purest form. A beam of electrons (or photons, atoms, even large molecules such as C₆₀ fullerenes) is directed at a barrier with two narrow slits. The intensity pattern recorded on a distant screen is not the sum of two single-slit patterns, as classical particle mechanics would predict. Instead, it is an interference pattern: $I(\theta) \propto \cos^2(\pi d\sin\theta/\lambda)$, characterized by alternating bright and dark fringes. Each electron arrives at the screen as a discrete dot — particle-like — yet the statistical distribution of many such dots builds up the wave-like interference pattern. The wave function, not the electron itself, propagates through both slits simultaneously.
The role of information is crucial. If one attempts to determine through which slit each electron passes — by placing a detector at one slit, coupling the electron to a "which-path" photon, or any other means — the interference pattern disappears and the distribution becomes the classical sum of two single-slit patterns. This is Bohr's principle of complementarity: the experimental arrangement determines whether wave-like or particle-like behavior is observed, and no experiment can reveal both simultaneously. The quantum eraser goes further: if the which-path information is erased after the electron has hit the screen (but before the experimenter looks at the data), the interference pattern is restored — demonstrating that it is the availability of information, not the physical disturbance, that controls the outcome.
Feynman's path integral formulation provides the deepest theoretical foundation for these phenomena. In this framework, a quantum particle simultaneously takes all possible paths from source to detector, each path contributing an amplitude $e^{iS[\text{path}]/\hbar}$ where $S$ is the classical action along that path. The total amplitude is the sum (integral) over all paths, and the probability is the squared modulus of this sum. Interference arises from the relative phases of different path amplitudes. In the double-slit experiment, paths through slit 1 and paths through slit 2 acquire different phases, and their superposition produces the observed interference pattern. The classical limit emerges when $S \gg \hbar$: adjacent paths have rapidly varying phases that cancel, leaving only the contribution from paths near the stationary-action (classical) trajectory — Fermat's principle and Hamilton's principle emerge naturally.
Wave packets — superpositions of plane waves — provide the reconciliation between localized particles and extended waves. A particle localized near position $x_0$ with approximate momentum $p_0$ is described by a wave packet $\psi(x,t) = \int \tilde{\psi}(k)e^{i(kx-\omega(k)t)}dk/(\sqrt{2\pi})$, where $\tilde{\psi}(k)$ is peaked near $k_0 = p_0/\hbar$. The group velocity $v_g = d\omega/dk|_{k_0} = p_0/m$ matches the classical particle velocity. The phase velocity $v_\phi = \omega/k$ is generally different and not directly observable for massive particles. Dispersion — the dependence of $\omega$ on $k$ — causes the wave packet to spread over time: $\Delta x(t) \approx \Delta x_0\sqrt{1 + (\hbar t/2m\Delta x_0^2)^2}$, reflecting the spreading of a quantum probability distribution.
The Fourier decomposition of wave packets is not merely a mathematical tool but encodes fundamental physics. The width of the wave packet in position space $\Delta x$ and the width of its Fourier transform in momentum space $\Delta k$ are related by $\Delta x \cdot \Delta k \geq 1/2$ (a purely classical Fourier inequality). Upon multiplying by $\hbar$: $\Delta x \cdot \Delta p \geq \hbar/2$ — the Heisenberg uncertainty principle emerges as a consequence of wave mechanics. A Gaussian wave packet $\tilde{\psi}(k) \propto e^{-(k-k_0)^2/(4\sigma_k^2)}$ saturates this inequality and represents the minimum-uncertainty state — the archetype of a coherent quantum state.
The coherence length of the source determines whether interference can be observed. For a source with spectral width $\Delta\lambda$, the coherence length is $L_c = \lambda^2/\Delta\lambda$. Only if the path length difference between the two slits is less than $L_c$ will the interference terms survive ensemble averaging. This explains why a distant star produces a nearly monochromatic (long coherence length) beam that can produce interference fringes, while a hot incandescent source requires spectral filtering. In modern quantum information, coherence is the resource underlying quantum computation and communication, and understanding its origin in wave-packet physics is essential.
For any particle with relativistic energy $E$ and momentum $\mathbf{p}$, the associated matter wave has frequency $f = E/h$ and wavelength $\lambda = h/|\mathbf{p}|$, equivalently $\mathbf{p} = \hbar\mathbf{k}$ where $\mathbf{k}$ is the wave vector. For a non-relativistic particle of mass $m$ and kinetic energy $E_k$: $$\lambda = \frac{h}{\sqrt{2mE_k}} = \frac{h}{mv}.$$ For an electron accelerated through potential $V$: $\lambda = h/\sqrt{2meV}$. At $V = 54$ V (Davisson-Germer conditions): $\lambda \approx 1.67$ Å, matching the nickel lattice spacing.
For two slits of width $a$ separated by center-to-center distance $d$, illuminated by coherent radiation of wavelength $\lambda$ at screen distance $L \gg d$, the intensity pattern is: $$I(y) = I_0 \left(\frac{\sin(\pi ay/\lambda L)}{\pi ay/\lambda L}\right)^2 \cos^2\left(\frac{\pi dy}{\lambda L}\right),$$ where $y$ is the transverse screen coordinate. The fringe spacing (distance between adjacent maxima) is $\Delta y = \lambda L/d$. The angular position of the $m$-th maximum satisfies $d\sin\theta_m = m\lambda$. The pattern is modulated by the single-slit diffraction envelope: minima occur when $a\sin\theta = n\lambda$ for integer $n$.
A one-dimensional wave packet is constructed by superposing plane waves: $$\psi(x,t) = \frac{1}{\sqrt{2\pi}}\int_{-\infty}^{\infty}\tilde{\psi}(k)e^{i(kx-\omega(k)t)}dk.$$ Expanding $\omega(k)$ around $k_0$: $\omega(k) \approx \omega_0 + (k-k_0)v_g + \frac{1}{2}(k-k_0)^2\omega''$, the packet moves at the group velocity $v_g = (d\omega/dk)|_{k_0}$, which for a massive particle gives $v_g = \hbar k_0/m = p_0/m$ (classical velocity). The quadratic term $\omega'' = \hbar/m$ causes dispersion: the packet spreads as $\sigma_x(t) = \sigma_x(0)\sqrt{1+(\hbar t/2m\sigma_x^2(0))^2}$. The phase velocity $v_\phi = \omega_0/k_0 = E/p = v/2$ (non-relativistic) is half the group velocity and carries no information.
Wave behavior and particle behavior are mutually exclusive yet jointly necessary (complementary) descriptions of quantum systems. Let $\mathcal{V}$ denote fringe visibility (wave-like measure) and $\mathcal{D}$ denote path distinguishability (particle-like measure). The quantitative complementarity relation states: $$\mathcal{D}^2 + \mathcal{V}^2 \leq 1.$$ Here $\mathcal{V} = (I_{\max}-I_{\min})/(I_{\max}+I_{\min})$ and $\mathcal{D} = |p_1 - p_2|$ where $p_1, p_2$ are the probabilities of finding the particle in paths 1 and 2 as determined by a which-path measurement. The inequality becomes equality for pure states. Full which-path information ($\mathcal{D}=1$) implies zero visibility ($\mathcal{V}=0$) and vice versa. This inequality follows from the Schmidt decomposition of the joint particle-detector state.
The amplitude for a particle to propagate from $(x_i, t_i)$ to $(x_f, t_f)$ is: $$K(x_f,t_f;x_i,t_i) = \int \mathcal{D}[x(t)]\, e^{iS[x(t)]/\hbar},$$ where the integral is over all continuous paths $x(t)$ with $x(t_i)=x_i$ and $x(t_f)=x_f$, and $S[x(t)] = \int_{t_i}^{t_f}L(x,\dot{x},t)\,dt$ is the classical action. The propagator $K$ is the kernel of the time-evolution operator: $\psi(x_f,t_f) = \int K(x_f,t_f;x_i,t_i)\psi(x_i,t_i)\,dx_i$. For a free particle, $K = \sqrt{m/(2\pi i\hbar T)}\exp(im(x_f-x_i)^2/(2\hbar T))$ where $T = t_f-t_i$. In the double-slit context, $K = K_1 + K_2$ (sum over paths through each slit), and $|K|^2 = |K_1|^2 + |K_2|^2 + 2\text{Re}(K_1^*K_2)$ — the interference term.
Calculate the de Broglie wavelength of: (a) an electron with kinetic energy 50 eV; (b) a thermal neutron at room temperature ($T = 300$ K); (c) a carbon-60 fullerene molecule ($m = 1.2 \times 10^{-24}$ kg) moving at $v = 100$ m/s. Compare with relevant length scales.
(a) $E_k = 50\text{ eV} = 50 \times 1.6\times10^{-19} = 8\times10^{-18}$ J. $\lambda = h/\sqrt{2m_eE_k} = 6.626\times10^{-34}/\sqrt{2\times9.109\times10^{-31}\times8\times10^{-18}} = 6.626\times10^{-34}/(1.208\times10^{-24}) \approx 1.73$ Å. Comparable to bond lengths ($\sim1-2$ Å) — ideal for electron diffraction crystallography. (b) Mean thermal kinetic energy: $E_k = \frac{3}{2}k_BT = \frac{3}{2}\times1.38\times10^{-23}\times300 = 6.21\times10^{-21}$ J. $\lambda = h/\sqrt{2m_nE_k} = 6.626\times10^{-34}/\sqrt{2\times1.675\times10^{-27}\times6.21\times10^{-21}} \approx 1.46$ Å. Also near atomic spacings — basis for neutron diffraction. (c) $\lambda = h/(mv) = 6.626\times10^{-34}/(1.2\times10^{-24}\times100) \approx 5.5\times10^{-12}$ m $= 0.055$ Å. Far below atomic scales, yet interference of C₆₀ has been observed experimentally (Arndt et al. 1999), confirming matter waves for complex molecules.
In a double-slit experiment with electrons, the slit separation is $d = 0.5$ μm, the screen is at $L = 1$ m, and the accelerating voltage is $V = 100$ V. Find: (a) the de Broglie wavelength, (b) the fringe spacing, (c) the angular position of the third dark fringe, (d) the coherence length required for 50% visibility.
(a) $\lambda = h/\sqrt{2m_eV} = 6.626\times10^{-34}/\sqrt{2\times9.109\times10^{-31}\times1.6\times10^{-19}\times100}$. Numerator: $6.626\times10^{-34}$. Denominator: $\sqrt{2.918\times10^{-47}} = 1.708\times10^{-24}\cdot\sqrt{100} \approx 1.708\times10^{-24}/\sqrt{1} $. Let me compute: $2m_eV = 2\times9.109\times10^{-31}\times1.6\times10^{-17} = 2.915\times10^{-47}$, $\sqrt{2.915\times10^{-47}} = 1.707\times10^{-24}$ kg·m/s. $\lambda = 6.626\times10^{-34}/1.707\times10^{-24} = 3.88\times10^{-10}$ m $\approx 0.388$ nm. (b) Fringe spacing: $\Delta y = \lambda L/d = 0.388\times10^{-9}\times1/(0.5\times10^{-6}) = 7.76\times10^{-4}$ m $\approx 0.776$ mm. (c) Dark fringes at $d\sin\theta = (m+\frac{1}{2})\lambda$. Third dark fringe: $m=2$, $\sin\theta = 2.5\lambda/d = 2.5\times0.388/500 = 1.94\times10^{-3}$, $\theta \approx 0.111°$. (d) For visibility $\mathcal{V} = 50\%$, path difference must be within $L_c/2$ of zero. At the edge of the central peak ($\Delta = \lambda L/2d \approx 0.388$ mm), coherence length $L_c > \Delta_{\max}$ needed. For $\mathcal{V}=0.5$, $L_c \approx 2\Delta_{\max} \approx 2\times d\sin\theta_{1st\,dark} \approx 0.5\lambda/d \cdot d = 0.5\lambda$ gives a rough estimate $L_c \gtrsim \lambda/2 \approx 0.19$ nm. In practice $L_c = \lambda^2/\Delta\lambda$ sets the spectral purity requirement.
Derive the time-evolution of a Gaussian wave packet $\psi(x,0) = (2\pi\sigma_0^2)^{-1/4}\exp(ip_0x/\hbar)\exp(-x^2/(4\sigma_0^2))$ for a free particle ($V=0$). Find $|\psi(x,t)|^2$ and identify the spreading rate.
The momentum-space wave function is obtained by Fourier transform: $\tilde{\psi}(k) = (2\pi\sigma_0^2)^{-1/4}(2\pi)^{-1/2}\cdot\sqrt{2\pi}\cdot(2\sigma_0^2)^{1/4}e^{-(k-k_0)^2\sigma_0^2}\cdot(2\pi)^{1/4}$. More directly: $$\tilde{\psi}(k) = \left(\frac{2\sigma_0^2}{\pi}\right)^{1/4}\exp(-(k-k_0)^2\sigma_0^2),$$ a Gaussian centered at $k_0 = p_0/\hbar$ with width $\sigma_k = 1/(2\sigma_0)$. For a free particle $\omega(k) = \hbar k^2/(2m)$, so expanding around $k_0$: $\omega(k) = \omega_0 + v_g(k-k_0) + \frac{\hbar}{2m}(k-k_0)^2$ with $v_g = \hbar k_0/m$. After integration: $$|\psi(x,t)|^2 = \frac{1}{\sqrt{2\pi}\sigma(t)}\exp\left(-\frac{(x-v_gt)^2}{2\sigma(t)^2}\right),$$ where $\sigma(t) = \sigma_0\sqrt{1+(t/\tau)^2}$ with $\tau = 2m\sigma_0^2/\hbar$ (spreading time). The packet moves at $v_g = p_0/m$ and spreads, with the probability density remaining Gaussian at all times. For an electron with $\sigma_0 = 1$ nm: $\tau = 2\times9.109\times10^{-31}\times(10^{-9})^2/(1.055\times10^{-34}) \approx 17$ fs — extremely rapid spreading on atomic scales.
Analyze the quantum eraser. Two slits are labeled by orthogonal photon polarizations: slit 1 imparts horizontal polarization $|H\rangle$, slit 2 imparts vertical polarization $|V\rangle$. Show that which-path information destroys interference, and that placing a 45° polarizer before the detector restores it. Compute the visibility in each case.
The joint electron-photon state after the slits is: $|\Psi\rangle = \frac{1}{\sqrt{2}}(|\psi_1\rangle|H\rangle + |\psi_2\rangle|V\rangle)$, where $|\psi_j\rangle$ is the wave function for path $j$ at the screen, containing the phase information $\psi_j(y) \propto e^{i\phi_j(y)}$. The electron's reduced density matrix is $\rho_{\rm elec} = \text{Tr}_{\rm photon}|\Psi\rangle\langle\Psi| = \frac{1}{2}(|\psi_1\rangle\langle\psi_1| + |\psi_2\rangle\langle\psi_2|)$ — a mixed state. The intensity $I(y) = \langle y|\rho_{\rm elec}|y\rangle = \frac{1}{2}(|\psi_1(y)|^2 + |\psi_2(y)|^2)$, with no cross terms: $\mathcal{V} = 0$. Now insert a 45° linear polarizer, projecting onto $|D\rangle = (|H\rangle+|V\rangle)/\sqrt{2}$. The post-selected state is $\langle D|\Psi\rangle = \frac{1}{2}(|\psi_1\rangle + |\psi_2\rangle)$ (up to normalization). The intensity $I(y) \propto |\psi_1(y)+\psi_2(y)|^2 = |\psi_1|^2+|\psi_2|^2+2\text{Re}(\psi_1^*\psi_2)$ — full interference, $\mathcal{V} = 1$. The 45° polarizer erases the which-path information, restoring the cross term. Projecting onto $|A\rangle = (|H\rangle-|V\rangle)/\sqrt{2}$ gives a shifted fringe pattern with $\mathcal{V}=1$ but phase shifted by $\pi$. The two patterns are complementary — they sum to the no-interference distribution, consistent with the complementarity inequality $\mathcal{D}^2+\mathcal{V}^2 = 0+1 = 1$.
Practice Problems
Show Answer Key
1. $E_k = 25\text{ keV} = 4.0\times10^{-15}$ J. $m_ec^2 = 8.187\times10^{-14}$ J. $\lambda = hc/\sqrt{E_k(E_k+2m_ec^2)} = (6.626\times10^{-34}\times3\times10^8)/\sqrt{4\times10^{-15}\times1.678\times10^{-13}} \approx 7.7\times10^{-12}$ m $= 0.077$ Å. Relativistic correction is significant at this voltage.
2. $\lambda = h/\sqrt{2m_eV} \approx 1.67$ Å. Bragg: $\sin\theta = \lambda/(2d) = 1.67/(2\times2.15) = 0.388$, $\theta \approx 22.8°$. Davisson and Germer observed maximum at $50°$ from normal (i.e., $40°$ grazing), consistent with this calculation for the appropriate lattice planes.
3. Path lengths from slits 1,2 to screen point $y$: $r_{1,2} = \sqrt{L^2+(y\mp d/2)^2} \approx L + (y\mp d/2)^2/(2L)$. Path difference: $\Delta r = r_2 - r_1 \approx -yd/L$. Constructive interference when $\Delta r = m\lambda$: $y_m = m\lambda L/d$. Fringe spacing $\Delta y = y_{m+1}-y_m = \lambda L/d$. Approximation valid when $d \ll L$.
4. Spreading time to double: $\sigma(t) = 2\sigma_0$ gives $t = \sqrt{3}\tau$ where $\tau = 2m\sigma_0^2/\hbar$. (a) Electron: $\tau = 2\times9.109\times10^{-31}\times(10^{-10})^2/(1.055\times10^{-34}) = 1.73\times10^{-16}$ s, $t = \sqrt{3}\times1.73\times10^{-16} \approx 3.0\times10^{-16}$ s. (b) Proton: $m_p/m_e \approx 1836$ times larger, $t \approx 5.5\times10^{-13}$ s.
5. Path difference $\Delta = yd/L = 0.25\times10^{-3}\times10^{-6}/1 = 2.5\times10^{-10}$ m. Phase: $\delta = 2\pi\Delta/\lambda = 2\pi\times2.5\times10^{-10}/(5\times10^{-7}) = 2\pi\times5\times10^{-4} \approx 3.14\times10^{-3}$ rad. This is near zero phase — constructive fringe (very close to central maximum).
6. Amplitude through slit at $x_s$: $K(X,T;x_s,0)\cdot K(x_s,0;0,0)$. Total: $A = K_1 + K_2$ where $K_j = C\exp(im[(x_s^{(j)})^2+X^2-...]/2\hbar T)$. Intensity $|A|^2 = |K_1|^2+|K_2|^2+2\text{Re}(K_1^*K_2)$, the last term being the interference contribution with phase $\propto a\cdot X/T\cdot m/\hbar$ — giving fringes in $X$.
7. (a) $\sigma_k = 1/(2\sigma_0) = 1/(2\times10^{-9}) = 5\times10^8$ m$^{-1}$. (b) $\Delta p = \hbar\sigma_k = 1.055\times10^{-34}\times5\times10^8 = 5.28\times10^{-26}$ kg·m/s. $\Delta E = p_0\Delta p/m_e$ where $p_0 = \sqrt{2m_eE_0} = 1.71\times10^{-24}$ kg·m/s. $\Delta E = 1.71\times10^{-24}\times5.28\times10^{-26}/(9.109\times10^{-31}) \approx 9.9\times10^{-20}$ J $\approx 0.62$ eV. (c) $\tau_c = \hbar/\Delta E \approx 1.055\times10^{-34}/(9.9\times10^{-20}) \approx 1.1\times10^{-15}$ s $= 1.1$ fs.
8. Schmidt decompose $|\Psi\rangle = \sum_j \sqrt{\lambda_j}|\alpha_j\rangle|\beta_j\rangle$. Visibility $\mathcal{V} = 2|\langle\psi_1|\psi_2\rangle|/(|\psi_1|^2+|\psi_2|^2)$ and distinguishability $\mathcal{D}$ are bounded by the Schmidt coefficients. For a two-path system with detector state entangled, $\mathcal{D}^2+\mathcal{V}^2 \leq 1$ follows from the Cauchy-Schwarz inequality on the overlap of detector states.
9. $E_k = 25\text{ meV}$. $p = \sqrt{2m_nE_k}$, $\lambda = h/p = h/\sqrt{2m_n\times25\times1.6\times10^{-22}} \approx 1.81$ Å. Phase difference: $\Delta\phi = 2\pi\Delta L/\lambda = 2\pi\times10^{-5}/(1.81\times10^{-10}) = 3.47\times10^5$ rad. Modulo $2\pi$: $3.47\times10^5/(2\pi) \approx 5.52\times10^4$ full cycles — need to know fractional part to determine constructive or destructive.
10. (a) $E = \sqrt{p^2c^2+m_0^2c^4}$, $\omega = E/\hbar$, $k=p/\hbar$. $v_g = d\omega/dk = c^2p/E = c^2\cdot\gamma m_0v/(\gamma m_0c^2) = v$. ✓ (b) In dispersive medium, $k = n(\omega)\omega/c$, $v_g = (dk/d\omega)^{-1} = c/(n+\omega dn/d\omega)$, the group velocity accounting for dispersion.
11. $\lambda = h/mv = 6.626\times10^{-34}/(1.2\times10^{-24}\times100) \approx 5.5\times10^{-12}$ m $= 0.0055$ nm. Fringe spacing: $\Delta y = \lambda L/d = 5.5\times10^{-12}\times1.25/10^{-7} = 6.9\times10^{-5}$ m $= 69$ μm. Observed and matches prediction. Significance: C₆₀ has 3N=180 vibrational modes, yet its center-of-mass wavefunction exhibits coherent interference — demonstrating wave-particle duality extends to complex many-body objects, provided decoherence from environment is avoided.
12. The interference pattern in the signal photon alone is absent regardless of the idler measurement timing — only the full joint coincidence data (conditioned on specific idler outcomes) reveals fringes. Future events cannot influence past records; the apparent retrocausality is illusory. The coincidence condition selects a subensemble, and the complementary subensembles sum to a featureless distribution. No information can be sent backward in time: the marginal signal-photon distribution is always uniform.