Streak camera

A streak camera measures how light intensity changes in time by turning time into distance. Light enters through a narrow slit and is imaged onto a photocathode, which emits electrons in proportion to the instantaneous intensity. The electrons are accelerated and pass between deflection plates carrying a fast voltage ramp, so electrons emitted later are deflected further. They strike a phosphor screen, usually after amplification in a microchannel plate, and a camera records the resulting streak: position along the sweep is time, and position along the slit is a second coordinate, often wavelength. Time resolution is typically a few picoseconds in general-purpose instruments and below 1 ps in the fastest designs, with time windows from about 100 ps to milliseconds.

Sweep speed and time resolution

The time resolution is set mainly by two terms. The first is the width of the slit image on the screen divided by the sweep speed. A sweep that spreads a 1 ns window across a 10 mm screen moves at 10 µm/ps, so a 50 µm slit image corresponds to 5 ps; spreading a 100 ps window across the same screen gives 100 µm/ps and 0.5 ps. The second is the spread of photoelectron transit times, which arises because electrons leave the photocathode with a range of initial energies $\Delta\varepsilon$. In an extraction field $E$, an order-of-magnitude estimate is

$$\Delta t \approx \frac{\sqrt{2 m_e\,\Delta\varepsilon}}{e\,E}.$$

With $E$ = 2 kV/mm, an energy spread of 0.1 eV gives 0.53 ps and 0.5 eV gives 1.2 ps. Sweep jitter relative to the light pulse adds a third term when many shots are averaged.

Single-shot and synchroscan operation

In single-shot mode a trigger starts one fast ramp and the camera records one event, such as a single amplifier pulse. The dynamic range is limited by space charge at the photocathode at high intensity, which broadens the streak, and by noise at low intensity.

In synchroscan mode the deflection voltage is a sine wave locked to the repetition rate of a mode-locked laser, typically near 80 MHz, a period of 12.5 ns. Only the steep central part of each cycle is used, and successive sweeps fall on the same place on the screen, so the signal from millions of pulses accumulates. Timing jitter between the laser and the sweep drive then limits the resolution. Synchroscan cameras are used for weak time-resolved emission, such as semiconductor photoluminescence and molecular fluorescence, where a spectrograph disperses the light along the slit and one image gives intensity versus wavelength and time.

Comparison with other time-resolved methods

Time-correlated single-photon counting times individual photons with a fast detector and electronics; its instrument response is usually tens of picoseconds, set by detector timing jitter, and it records one spectral channel at a time unless a detector array is used. A streak camera with a spectrograph records all wavelengths at once at higher time resolution, at the cost of higher instrument price and lower dynamic range per shot. At the lowest light levels a streak camera can also be read out in a photon-counting mode, in which individual photoelectron spots on the screen are located and accumulated.

Pump-probe spectroscopy reaches femtosecond resolution, limited by the pulse durations, and also detects states that do not emit, but it builds the time axis point by point with a delay stage. An intensity autocorrelation measures pulses shorter than a streak camera can resolve, but gives only a symmetric trace that requires an assumed pulse shape. For pulses longer than a few tens of picoseconds, a fast photodiode and a sampling oscilloscope are simpler.

Where it is used

Streak cameras measure the pulse duration and shape of picosecond lasers, gain-switched diodes and Q-switched pulses. With X-ray-sensitive photocathodes they record the emission of laser-produced plasmas and inertial-confinement fusion targets. In range-resolved lidar and fluorescence imaging the slit axis carries a spatial coordinate instead of wavelength.

Pitfalls

The sweep is not perfectly linear, so the time axis must be calibrated, commonly with a pulse train passed through an etalon or delay lines of known spacing. Space-charge broadening makes intense pulses appear longer than they are; a series at decreasing intensity shows whether the measured width is still changing. Photocathode response varies along the slit and with wavelength and needs flat-field correction. Lifetimes approaching the resolution must be extracted by deconvolution with the instrument response.

Common questions

What is the time resolution of a streak camera?

A few picoseconds for general-purpose synchroscan and single-shot instruments, and a few hundred femtoseconds in the fastest designs, set by sweep speed, slit width and photoelectron energy spread.

Streak camera or TCSPC for fluorescence lifetimes?

For lifetimes of hundreds of picoseconds or longer at a single wavelength, TCSPC is usually sufficient and more economical. For lifetimes of a few picoseconds, or when the full spectrum must be followed in time, a streak camera is the usual choice.

References: J.-C. Diels and W. Rudolph, Ultrashort Laser Pulse Phenomena, 2nd ed. (Academic Press, 2006); J. R. Lakowicz, Principles of Fluorescence Spectroscopy, 3rd ed. (Springer, 2006).