Photothermal Detection and Sensing

Photothermal detection represents one of the most powerful and elegant techniques in modern optical sensing. While absorption spectroscopy measures light that is removed from a beam, photothermal sensing directly detects the heat generated when matter absorbs light. This seemingly simple idea opens an entirely new window into characterizing nanoparticles, single molecules, and biomolecules with unprecedented sensitivity and label-free contrast.

In this lecture, we connect fundamental principles of light-matter interaction, thermal diffusion, and optical interferometry to build a complete picture of photothermal microscopy and detection. By the end of this 90-minute session, you will understand:

This is a field at the intersection of condensed matter physics, biophysics, and optical engineering—with major applications from biomedicine to nanomaterial characterization.

1. Light Absorption and Heat Generation

Absorption Cross Section and Energy Conversion

When light propagates through an absorbing medium, the intensity decreases according to:

\[I(z) = I_0 e^{-\alpha z}\]

where \(\alpha\) is the absorption coefficient (units: cm\(^{-1}\)). For a nanoparticle much smaller than the wavelength, we define the absorption cross section \(\sigma_{\text{abs}}\) (units: nm\(^2\)), which represents the effective area over which the particle absorbs photons.

If a plane wave of intensity \(I_0\) illuminates the nanoparticle, the absorbed power is:

\[P_{\text{abs}} = \sigma_{\text{abs}} \cdot I_0\]

This absorbed energy must be dissipated as heat. In the steady-state regime (continuous-wave laser illumination), the thermal energy flows outward through conduction in the surrounding medium.

Temperature Profile Around a Heated Nanoparticle

Consider a small spherical heat source at temperature \(T_0\) dissipating power \(P_{\text{abs}}\) into an infinite medium with thermal conductivity \(\kappa\). The steady-state temperature profile in 3D is:

\[\Delta T(r) = T(r) - T_\infty = \frac{P_{\text{abs}}}{4\pi \kappa r}\]

This \(1/r\) decay is fundamental to heat diffusion. For a nanoparticle of radius \(a\) embedded in a medium:

  • At the surface \((r = a)\): \(\Delta T(a) = \frac{P_{\text{abs}}}{4\pi \kappa a}\)
  • Characteristic thermal length scale: \(\ell_T = \frac{\kappa}{|\text{absorption rate}|}\)

Key observations: - Smaller particles → larger temperature rise (for fixed power) - Higher thermal conductivity → faster heat dissipation - Modulation frequency → limited by thermal diffusion time \(\tau_T \sim a^2/D_T\), where \(D_T = \kappa/(\rho c)\) is thermal diffusivity

The thermal energy deposited at the particle scales with the absorption cross section, which for gold nanoparticles can be enormous—much larger than their geometric cross section!

2. The Photothermal Effect: Refractive Index Modulation

Temperature-Dependent Refractive Index

The key insight of photothermal detection is that most transparent media have a refractive index that changes with temperature:

\[\Delta n = \left(\frac{\mathrm{d}n}{\mathrm{d}T}\right) \Delta T = \beta \, \Delta T\]

where \(\beta = \mathrm{d}n/\mathrm{d}T\) is the thermo-optic coefficient (typically \(10^{-4}\) to \(10^{-3}\) K\(^{-1}\)).

Examples: - Water: \(\beta \approx -1.0 \times 10^{-4}\) K\(^{-1}\) (refractive index decreases with temperature) - Typical oils: \(\beta \approx -4 \times 10^{-4}\) K\(^{-1}\) - Fused silica: \(\beta \approx 1.0 \times 10^{-5}\) K\(^{-1}\)

Therefore, a heated absorber creates a spatial variation in refractive index:

\[n(\mathbf{r}) = n_0 + \beta \Delta T(\mathbf{r}) = n_0 + \beta \frac{P_{\text{abs}}}{4\pi \kappa |\mathbf{r}|}\]

Thermal Lensing Effect

When a probe beam passes through this refractive index gradient, it experiences a lens effect. The spatial phase modulation creates focusing (for \(\beta > 0\)) or defocusing (for \(\beta < 0\)).

The phase shift experienced by a probe beam passing at distance \(r_0\) from the heat source is approximately:

\[\Delta \phi(r_0) \approx \frac{2\pi}{\lambda} \int_{-\infty}^{\infty} \beta \frac{P_{\text{abs}}}{4\pi \kappa r(x,y)} \, \mathrm{d}z\]

For a probe beam co-aligned with the pump, the phase gradient creates an angular deflection and/or intensity redistribution—the thermal lensing effect.

This is not absorption of the probe: it is a phase modulation. A weak probe beam passing through the thermal lens experiences: - Spatial phase shift - Angular deflection (tilt) - Transverse intensity redistribution

These effects are the basis of all photothermal detection schemes.

3. Pump-Probe Detection Architecture

Time-Modulated Heating and Lock-in Detection

The strategy is simple but powerful: 1. Pump beam (at frequency \(\nu_0\)): absorbed → heats the sample 2. Probe beam (at frequency \(\nu_p\)): senses the refractive index change 3. Modulation: amplitude-modulate the pump at frequency \(f_m\) (typically 1–100 kHz) 4. Detection: use lock-in amplification at frequency \(f_m\) to extract the photothermal signal

Why Modulation?

  • The DC temperature and its associated refractive index change are usually slow or masked by other noise
  • By modulating the pump, we create a signal at frequency \(f_m\)
  • Lock-in detection (synchronous demodulation) rejects noise at other frequencies
  • This achieves background-free detection

Signal Model

If the pump intensity is modulated as: \[I_{\text{pump}}(t) = I_0 (1 + m \cos(2\pi f_m t))\]

then the absorbed power oscillates: \[P_{\text{abs}}(t) = \sigma_{\text{abs}} I_0 (1 + m \cos(2\pi f_m t))\]

This drives a temperature oscillation: \[\Delta T(t) = \Delta T_0 (1 + m \cos(2\pi f_m t - \phi))\]

where the phase lag \(\phi\) depends on the thermal diffusion time. The probe beam experiences a phase modulation, which creates an intensity modulation in the detection optics.

A lock-in detector multiplies the detected signal \(S(t)\) by a reference oscillation at \(f_m\) and low-pass filters:

\[S_{\text{LIA}} \propto \langle S(t) \cos(2\pi f_m t) \rangle_{\text{filtered}}\]

This rejects all noise at frequencies different from \(f_m\), achieving shot-noise-limited sensitivity.

Pump Modulation with Acousto-Optic Modulators

In practice, the pump beam intensity is often modulated using an acousto-optic modulator (AOM), a device that leverages the interaction between light and sound waves. An AOM consists of a transparent crystal (typically tellurium dioxide) with a piezoelectric transducer bonded to it. When a radiofrequency (RF) signal drives the transducer, it generates an acoustic wave (ultrasound) that propagates through the crystal, creating a periodic refractive index grating. Incident light diffracts off this acoustic grating according to the Bragg condition, and the first-order diffracted beam emerges with its frequency shifted by the acoustic frequency (Doppler shift) and with intensity directly modulated by the acoustic amplitude.

The key advantages of AOMs for photothermal microscopy are: (1) switching speed: typically ~100 ns, much faster than mechanical choppers; (2) modulation frequencies: 100 kHz to >1 MHz, well-suited for lock-in detection; (3) efficiency: 80–90% of light can be directed into the first-order beam; (4) programmability: RF frequency and amplitude can be dynamically controlled. In a typical photothermal setup, the AOM modulates the pump beam at 100 kHz–1 MHz, and the lock-in amplifier is tuned to detect the photothermal signal at that exact frequency. Beyond photothermal microscopy, AOMs are ubiquitous in confocal scanning, STED (stimulated emission depletion microscopy for depletion beam control), and optical trapping, where precise, fast modulation of laser beams is essential.

4. Photothermal Microscopy: Practical Implementation

Point-Scanning Configuration

A typical photothermal microscope operates as follows:

  1. Pump beam (high-power, typically 532 nm or 660 nm):
    • Focused to a diffraction-limited spot via high-NA objective
    • Positioned at sample plane
  2. Probe beam (low-power, often at different wavelength like 780 nm):
    • Co-aligned with pump (small offset possible)
    • Focused to same spot
    • Typically continuous-wave (CW)
  3. Detected signal:
    • Forward transmission: probe transmitted through sample
    • Backward (reflection): probe scattered by thermal lens
    • Usually prefer backward detection (higher signal for small scatterers)
  4. Scanning:
    • Raster-scan the objective (or sample)
    • Record lock-in amplitude \(A(x, y)\) at each pixel
    • Build a photothermal image

Advantages of Photothermal Microscopy

  • Label-free: no fluorescent dyes needed
  • Background-free: lock-in at modulation frequency eliminates autofluorescence, scattered light
  • Sensitive: single nanoparticles, eventually single molecules
  • Contrast mechanism directly related to absorption: measures \(\sigma_{\text{abs}}\) directly

5. Single-Particle and Single-Molecule Detection

Gold Nanoparticles

Gold nanoparticles (AuNPs) are ideal model systems for photothermal detection:

  • Plasmonic resonance: localized surface plasmon resonance (LSPR) near 520 nm
  • Large absorption cross section: \(\sigma_{\text{abs}} \approx 10\text{–}100\) nm\(^2\) for \(d \approx 20\text{–}100\) nm (much larger than geometric cross section \(\sim \pi (d/2)^2\))
  • Tunable absorption: resonance wavelength shifts with size and environment
  • Biocompatible: can be functionalized with antibodies, targeting ligands

For a 50 nm AuNP at resonance with 1 mW focused into a \(\sim 500\) nm spot: \[I_0 \sim 10^9 \text{ W/cm}^2 = 10^5 \text{ W/m}^2\] \[P_{\text{abs}} = \sigma_{\text{abs}} I_0 \approx (10 \text{ nm}^2) \cdot (10^5 \text{ W/m}^2) \sim 1\text{–}10 \text{ pW}\]

This tiny power still produces a measurable temperature rise and detectable phase shift!

Single-Molecule Photothermal Detection

Can we detect individual organic dye molecules or small proteins?

  • Absorption cross section of a dye: \(\sigma_{\text{abs}} \sim 10^{-3}\text{–}10^{-2}\) nm\(^2\) (much smaller than AuNP)
  • Required pump intensity: \(I_0 \sim 10^{10}\text{–}10^{11}\) W/cm\(^2\) (high photodamage risk)
  • Temperature rise around molecule: ΔT \(\sim\) 0.1–1 K
  • Signal: \(\Delta \phi \sim 10^{-5}\text{–}10^{-3}\) rad (still detectable with lock-in!)

Challenge: photobleaching at such high intensities. But under controlled conditions, single dyes and small proteins have been detected phototermally with good signal-to-noise ratio.

Key advantage: unlike fluorescence, no quantum yield limit—every absorbed photon counts, independent of emission efficiency.

6. Photothermal Heterodyne Imaging (PHI)

Heterodyne Detection Principle

A more sensitive variant of photothermal microscopy uses heterodyne detection. The idea:

  • Interfere the scattered probe field (from the thermal lens) with a reference field
  • This interference creates a signal proportional to the amplitude of the thermal lens effect
  • Similar to heterodyne interferometry in laser Doppler velocimetry

Mathematical Formulation

Let: - \(E_{\text{scat}}(t) = A_{\text{scat}} e^{i(\mathbf{k}_p \cdot \mathbf{r} + 2\pi f_p t)}\): scattered probe field due to thermal lens - \(E_{\text{ref}}(t) = A_{\text{ref}} e^{i(2\pi f_p t + \phi_0)}\): reference field - \(E_{\text{beat}}(t) = A_{\text{ref}} e^{i\phi_0}\): beating at the detector

The heterodyne gain comes from the term: \[I_{\text{het}} \propto 2 A_{\text{ref}} A_{\text{scat}} \cos(\phi_{\text{scat}} - \phi_0)\]

For weak scattering (\(A_{\text{scat}} \ll A_{\text{ref}}\)): \[I_{\text{het}} \approx 2 A_{\text{ref}} A_{\text{scat}}\]

Compared to homodyne (direct intensity) detection: \[I_{\text{hom}} \propto A_{\text{scat}}^2\]

Heterodyne amplifies the signal by the reference amplitude—a gain of \(A_{\text{ref}}/A_{\text{scat}} \gg 1\).

PHI Sensitivity Scaling

The photothermal heterodyne signal is proportional to: \[S_{\text{PHI}} \propto A_{\text{ref}} \cdot (\nabla n) \cdot P_{\text{abs}} \propto \sigma_{\text{abs}}\]

Direct measurement of absorption cross section!

This makes PHI particularly powerful for: - Comparing absorption spectra of single nanoparticles - Measuring small changes in absorption (e.g., conformational changes in proteins) - Wavelength-dependent single-particle spectroscopy

7. Applications: From Nanoscience to Biosensing

1. Label-Free Nanoparticle Imaging

Map the spatial distribution of absorbing nanoparticles (Au, Ag, carbon dots) without adding fluorescent labels.

Example: imaging plasmonic nanoparticles in cellular uptake studies - Detect size-dependent absorption changes - Map nanoparticle localization to subcellular organelles - No photobleaching, no fluorescence intermittency

2. Single-Molecule Absorption Spectroscopy

Measure absorption spectra of individual protein molecules, oligonucleotides, or chromophore complexes.

Key results: - Heterogeneous absorption profiles not visible in bulk measurements - Direct access to intrinsic absorption without fluorescence yield assumptions - Dynamic absorption changes during protein folding/unfolding

3. Biosensing and Molecular Diagnostics

Use photothermal microscopy to detect: - Biomarker binding: functionalize Au nanoparticles with antibodies; signal increases upon antigen binding - DNA hybridization: Au particles conjugated to complementary probes; detectable upon target DNA capture - Protein-protein interactions: monitor absorption changes at single-molecule level

Advantages: - No enzyme-linked assays needed - Label-free (or minimal labeling) - Real-time kinetics - Multiplexing possible (different wavelengths, sizes)

4. Nanoparticle Tracking Analysis (NTA)

Combine photothermal detection with particle tracking: - Track Brownian motion of gold nanoparticles through viscous fluid - Measure diffusion coefficient: \(D = k_B T / (6\pi \eta r)\) (Stokes relation) - Infer hydrodynamic size of viruses, exosomes, micelles - Distinguish aggregation states in real-time

Signal-to-noise is excellent because photothermal detection is background-free.

8. Connection to Photoacoustic Imaging

Pulsed Excitation and Acoustic Transduction

Photothermal detection operates in the thermal regime (modulation frequencies 1–100 kHz, thermal diffusion time \(\sim 1\) μs for micron-scale structures).

Photoacoustic imaging uses pulsed laser excitation (short pulses, sub-nanosecond timescale):

  1. Ultrashort pulse (\(\tau_p \ll \tau_T\)) deposits energy locally
  2. Rapid heating creates pressure transient: \(\Delta p \propto \beta \Delta T\)
  3. Acoustic wave propagation: ultrasonic frequency, ~1 MHz
  4. Detection: piezo transducers, microphone, or interferometric detection

Why Pulsed? Deeper Penetration

  • Photothermal imaging limited by thermal diffusion: penetration depth \(\sim \sqrt{D_T / f_m}\) (typically few hundred μm at 1 kHz modulation)
  • Photoacoustic imaging uses acoustic propagation (much longer range in tissue): penetration depth ~ few mm to cm
  • Acoustic scattering is much weaker than optical scattering
  • Perfect for biomedical imaging: can visualize hemoglobin, melanin in deep tissue layers

Unified Framework

Both techniques detect the same fundamental process: \[\text{Photon absorption} \to \text{Heat generation} \to \text{Refractive index/acoustic change}\]

  • Photothermal (CW, modulated): measures spatially-resolved temperature field via optical phase
  • Photoacoustic (pulsed): measures temporal pressure transient via acoustic wave

For molecules and nanoparticles, both measure the absorption cross section \(\sigma_{\text{abs}}\).

Code Example 1: Temperature Profile Around a Heated Nanoparticle

Code 
fig, (ax1, ax2) = plt.subplots(1, 2, figsize=get_size(15, 7))

# 1D radial temperature profile
r = np.linspace(0.01, 5, 1000)  # distance in micrometers
P_abs = 10e-12  # absorbed power in Watts (10 pW)
kappa = 0.6  # thermal conductivity of water, W/(m·K)
Delta_T_r = (P_abs / (4 * np.pi * kappa)) / (r * 1e-6)  # convert r to meters

ax1.loglog(r, Delta_T_r, 'b-', linewidth=2.5, label='Temperature rise')
ax1.axhline(y=0.1, color='r', linestyle='--', alpha=0.6, label='~0.1 K detection limit')
ax1.fill_between(r, 0.01, 0.1, alpha=0.2, color='gray', label='Detectable range')
ax1.set_xlabel('Distance (μm)')
ax1.set_ylabel('ΔT (K)')
ax1.legend(loc='upper right')
ax1.grid(True, which='both', alpha=0.3)
ax1.set_ylim([0.01, 100])

# 2D temperature profile (Gaussian pump)
x = np.linspace(-3, 3, 200)
y = np.linspace(-3, 3, 200)
X, Y = np.meshgrid(x, y)
R = np.sqrt(X**2 + Y**2) + 0.1  # distance, avoid singularity
# Pump with Gaussian profile (intensity)
w0 = 0.5  # beam waist in micrometers
I_pump = np.exp(-(X**2 + Y**2)/(w0**2))  # normalized Gaussian
# Temperature from local absorbed power
Delta_T_2D = (I_pump / (4 * np.pi * kappa * (R * 1e-6))) * (P_abs * 1e12) / 100

im = ax2.contourf(X, Y, Delta_T_2D, levels=20, cmap='hot')
ax2.contour(X, Y, Delta_T_2D, levels=10, colors='white', alpha=0.3, linewidths=0.5)
ax2.set_xlabel('x (μm)')
ax2.set_ylabel('y (μm)')
ax2.set_aspect('equal')

fig.set_layout_engine('constrained')
plt.savefig('temp_profile.png', dpi=150, bbox_inches='tight')
plt.show()

print(f"Temperature rise at r = 0.1 μm: ΔT ≈ {Delta_T_r[np.argmin(np.abs(r - 0.1))]:.2f} K")
print(f"Temperature rise at r = 1 μm: ΔT ≈ {Delta_T_r[np.argmin(np.abs(r - 1.0))]:.2f} K")

Temperature profile and 2D heat distribution (hot colormap: temperature rise \(\Delta T\) in K) around an absorbing nanoparticle (Gaussian pump profile).
Temperature rise at r = 0.1 μm: ΔT ≈ 0.00 K
Temperature rise at r = 1 μm: ΔT ≈ 0.00 K

Code Example 2: Thermal Lens Effect—Probe Beam Phase Shift

Code 
# Parameters
wavelength = 780e-9  # probe wavelength in meters (near-IR)
kappa = 0.6  # thermal conductivity of water, W/(m·K)
beta = -1e-4  # thermo-optic coefficient of water, K^{-1}
P_abs = 10e-12  # absorbed pump power, W

# Probe beam position relative to pump
r_probe = np.linspace(0.05, 5, 500)  # distance in micrometers

# Phase shift due to thermal lens
# Approximate: ∫ β ΔT(r) dz along probe path, assume 2D Gaussian profile
w0 = 0.5  # pump beam waist
# Integrated phase for probe at distance r from pump center
integral_Delta_n = beta * (P_abs / (4 * np.pi * kappa)) * np.log(r_probe / 0.1)  # approx integral

Delta_phi = (2 * np.pi / wavelength) * integral_Delta_n * (w0 * 1e-6) * 2  # path-integrated phase

ax = plt.figure(figsize=get_size(13, 7)).add_subplot(111)
ax.plot(r_probe, np.abs(Delta_phi) * 1e3, 'b-', linewidth=2.5, label='|Δφ|')
ax.axhline(y=0.001, color='g', linestyle='--', alpha=0.6, label='Quantum noise limit (~mrad)')
ax.fill_between(r_probe, 0, 0.001, alpha=0.1, color='green')
ax.fill_between(r_probe, 0.001, np.max(np.abs(Delta_phi)*1e3), alpha=0.15, color='blue', label='Detectable range')
ax.set_xlabel('Beam separation (μm)')
ax.set_ylabel('$|\\Delta\\phi|$ (mrad)')
ax.set_yscale('log')
ax.legend(loc='upper right')
ax.grid(True, which='both', alpha=0.3)
ax.set_ylim([1e-4, 10])

plt.tight_layout()
plt.savefig('thermal_lens_phase.png', dpi=150, bbox_inches='tight')
plt.show()

print(f"Maximum phase shift (co-aligned beams): {np.abs(Delta_phi[0]) * 1e3:.3f} mrad")
print(f"Phase shift at 1 μm separation: {np.abs(Delta_phi[np.argmin(np.abs(r_probe - 1.0))]) * 1e3:.4f} mrad")

Phase shift experienced by probe beam passing through thermal lens created by heated nanoparticle.
Maximum phase shift (co-aligned beams): 0.000 mrad
Phase shift at 1 μm separation: 0.0000 mrad

Code Example 3: Lock-in Detection Principle

Code 
# Time parameters
f_mod = 10e3  # modulation frequency, 10 kHz
t_total = 500e-6  # 500 microseconds
n_periods = int(f_mod * t_total)
t = np.linspace(0, t_total, 5000)

# Signal components
signal_amplitude = 1.0  # photothermal signal amplitude
noise_amplitude = 3.0  # broadband noise amplitude
noise_color = np.random.randn(len(t))

# Modulated photothermal signal
signal = signal_amplitude * np.cos(2 * np.pi * f_mod * t)

# Add broadband noise
noise = noise_amplitude * noise_color

# Total measured signal
measured = signal + noise

# Lock-in demodulation: multiply by reference and low-pass filter
reference = np.cos(2 * np.pi * f_mod * t)
demodulated = measured * reference

# Low-pass filter (simple moving average)
tau_filter = 50e-6  # filter time constant
n_filter = int(tau_filter * len(t) / t_total)
demodulated_filtered = np.convolve(demodulated, np.ones(n_filter)/n_filter, mode='same')

# Plot
fig, axes = plt.subplots(3, 1, figsize=get_size(15, 7))

# Raw measured signal
axes[0].plot(t*1e6, measured, 'k-', linewidth=0.8, alpha=0.7, label='Raw measured')
axes[0].plot(t*1e6, signal, 'b-', linewidth=2, alpha=0.5, label='True signal')
axes[0].set_ylabel('Signal (a.u.)')
axes[0].legend(loc='upper right', fontsize=8)
axes[0].grid(True, alpha=0.3)
axes[0].set_xlim([0, 200])

# Demodulated (before filtering)
axes[1].plot(t*1e6, demodulated, 'r-', linewidth=0.8, alpha=0.6, label='Demodulated × ref')
axes[1].set_ylabel('Demod. (a.u.)')
axes[1].legend(loc='upper right', fontsize=8)
axes[1].grid(True, alpha=0.3)
axes[1].set_xlim([0, 200])

# Filtered output
axes[2].plot(t*1e6, demodulated_filtered, 'g-', linewidth=2.5, label='After LPF')
axes[2].axhline(y=signal_amplitude/2, color='b', linestyle='--', alpha=0.6, label='Expected DC level')
axes[2].set_xlabel('Time (μs)')
axes[2].set_ylabel('Lock-in (a.u.)')
axes[2].legend(loc='upper right', fontsize=8)
axes[2].grid(True, alpha=0.3)
axes[2].set_xlim([0, 200])
axes[2].set_ylim([-0.5, 1.0])

plt.tight_layout()
plt.savefig('lockin_detection.png', dpi=150, bbox_inches='tight')
plt.show()

# Quantitative noise rejection
snr_before = np.var(signal) / np.var(noise)
snr_after = np.var(signal) / np.var(demodulated_filtered[int(0.1*len(t)):])
print(f"SNR before lock-in: {snr_before:.3f} ({10*np.log10(snr_before):.1f} dB)")
print(f"SNR after lock-in: {snr_after:.3f} ({10*np.log10(snr_after):.1f} dB)")
print(f"Noise rejection: {snr_after/snr_before:.1f}× improvement")

Lock-in detection principle: modulated signal buried in noise is recovered through synchronous demodulation and low-pass filtering.
SNR before lock-in: 0.054 (-12.6 dB)
SNR after lock-in: 63.282 (18.0 dB)
Noise rejection: 1163.4× improvement

Code Example 4: Photothermal Signal vs. Nanoparticle Size

Code 
# Gold nanoparticle absorption cross section (Mie theory, simplified)
# For Au nanoparticles in resonance (λ ≈ 520 nm for ~50 nm particles)
# Scaling: σ_abs ∝ r³ (in the dipole regime) plus resonance enhancement

# Particle radius in nm
radii = np.linspace(5, 100, 200)

# Simplified model: σ_abs = C * r^3 * f(resonance)
# where f(resonance) is a resonance factor peaking near 50 nm

def resonance_factor(r):
    """Approximate resonance factor for Au at λ = 532 nm"""
    r0 = 50  # resonance peak radius
    width = 15
    return 1.0 + 4.0 * np.exp(-(r - r0)**2 / (2 * width**2))

sigma_abs = (radii / 10)**3 * resonance_factor(radii)  # in units of 10 nm^2

# Photothermal signal: proportional to σ_abs
signal_strength = sigma_abs

# Temperature rise for fixed pump intensity
I_pump = 1e9  # W/cm^2 (typical focused laser)
P_abs = sigma_abs * 1e-18 * I_pump * 1e-4 * 1e12  # in pW (unit conversion)
Delta_T = P_abs / 10  # rough estimate (linear in power)

# Figure
fig, (ax1, ax2) = plt.subplots(1, 2, figsize=get_size(15, 7))

# Plot 1: Absorption cross section
ax1.plot(radii, sigma_abs, 'b-', linewidth=2.5)
ax1.axvline(x=50, color='r', linestyle='--', alpha=0.6, label='Resonance peak')
ax1.fill_between(radii, 0, sigma_abs, alpha=0.2, color='blue')
ax1.set_xlabel('Radius (nm)')
ax1.set_ylabel('$\\sigma_a$ (×10 nm²)')
ax1.legend()
ax1.grid(True, alpha=0.3)
ax1.set_yscale('log')

# Plot 2: Photothermal signal (relative)
# Signal ∝ σ_abs for heterodyne detection
ax2.semilogy(radii, sigma_abs, 'g-', linewidth=2.5, label='PHI signal ∝ σ_abs')
ax2.semilogy(radii, radii**3/1000, 'r--', linewidth=2, alpha=0.7, label='r³ scaling (dipole)')
ax2.axvline(x=50, color='orange', linestyle='--', alpha=0.6, label='Opt. size: ~50 nm')
ax2.set_xlabel('Radius (nm)')
ax2.set_ylabel('Signal (a.u.)')
ax2.legend(fontsize=8)
ax2.grid(True, alpha=0.3, which='both')

plt.tight_layout()
plt.savefig('signal_vs_size.png', dpi=150, bbox_inches='tight')
plt.show()

# Print some values
idx_50 = np.argmin(np.abs(radii - 50))
idx_20 = np.argmin(np.abs(radii - 20))
idx_80 = np.argmin(np.abs(radii - 80))

print("Absorption cross section (relative units):")
print(f"  r = 20 nm: σ_abs = {sigma_abs[idx_20]:.1f}")
print(f"  r = 50 nm: σ_abs = {sigma_abs[idx_50]:.1f} (resonance peak)")
print(f"  r = 80 nm: σ_abs = {sigma_abs[idx_80]:.1f}")
print(f"\nPhotothermal signal enhancement from 20→50 nm: {sigma_abs[idx_50]/sigma_abs[idx_20]:.1f}×")

Photothermal signal scales with absorption cross section, which varies strongly with nanoparticle size and material.
Absorption cross section (relative units):
  r = 20 nm: σ_abs = 11.9
  r = 50 nm: σ_abs = 620.3 (resonance peak)
  r = 80 nm: σ_abs = 789.5

Photothermal signal enhancement from 20→50 nm: 52.3×

Summary and Key Takeaways

Photothermal detection is a powerful, label-free technique that exploits a fundamental connection:

\[\text{Absorbed light} \longrightarrow \text{Heat} \longrightarrow \text{Refractive index change} \longrightarrow \text{Phase/intensity modulation}\]

Key concepts: 1. Temperature gradients scale as \(\Delta T(r) = P_{\text{abs}} / (4\pi \kappa r)\) in 3D 2. Thermal lensing creates phase shifts detectable with a probe beam 3. Lock-in detection at the pump modulation frequency provides background-free signal 4. Absorption cross section is directly measured—especially large for plasmonic nanoparticles 5. Single-particle and single-molecule sensitivity is achievable with appropriate configurations 6. PHI (photothermal heterodyne imaging) uses interference for enhanced sensitivity 7. Applications span nanoscience (nanoparticle tracking), biosensing (molecular binding), and deep-tissue imaging (photoacoustic variant)

This technique beautifully demonstrates how fundamental physics principles—thermal conduction, optics, and signal processing—combine to create modern analytical tools used daily in research labs worldwide.

Experimental Connections

Photothermal detection combines optics, thermodynamics, and signal processing — a rich playground for experiments:

Thermal lens effect in absorbing liquids Focus a laser (even a laser pointer) into a cuvette of dilute ink or coffee. Place a screen behind the cuvette and observe the far-field beam profile changing over seconds — the thermal lens causes the beam to diverge. Block the beam and watch it relax. The timescale is set by thermal diffusion.

Lock-in detection principle Modulate a light source (LED with a function generator) at a known frequency. Detect with a photodiode and an oscilloscope — the signal is buried in noise. Now use a lock-in amplifier (or a software lock-in: multiply by the reference and low-pass filter). The signal-to-noise improvement is dramatic and teaches why modulation is essential for sensitive detection.

Photothermal contrast of metal nanoparticles Gold nanoparticles (50–100 nm) absorb strongly at 532 nm. Deposit them on a coverslip, illuminate with a modulated green laser (heating beam), and probe the refractive index change with a red probe beam. Even without a commercial photothermal microscope, the principle can be demonstrated with a simple two-beam setup and a lock-in amplifier.

Absorption vs. scattering: comparing nanoparticle signals Compare darkfield (scattering) and photothermal (absorption) images of the same gold nanoparticle sample. Small particles (< 40 nm) are nearly invisible in darkfield but clearly detected in photothermal imaging — this teaches the \(a^3\) vs. \(a^6\) scaling of absorption vs. scattering cross-sections.

Thermal diffusion timescale measurement Vary the modulation frequency of the heating beam and measure the photothermal signal amplitude. The roll-off frequency corresponds to \(f_c \sim D_\text{th} / w_0^2\), where \(D_\text{th}\) is the thermal diffusivity and \(w_0\) the beam waist. This connects the signal to the heat equation solved in the lecture.

Further Reading

The following references are linked to the central Resources & Recommended Reading page:

  • Bialkowski (2019) — Photothermal Spectroscopy Methods. Comprehensive on thermal lensing, deflection, and interferometric detection.
  • Berciaud et al. (2004) — Seminal paper on photothermal heterodyne imaging of single nanoparticles.
  • Novotny & Hecht (2012) — Principles of Nano-Optics. Single-molecule and single-particle detection theory.
  • Bohren & Huffman (2004) — Absorption and Scattering of Light by Small Particles. Cross-section calculations for understanding signal strength.
  • Saleh & Teich, Ch. 22 — Acousto-optics and modulation techniques (relevant to lock-in methods).

Lecture 13: Photothermal Detection & Sensing — Introduction to Photonics