A fast multispectral light synthesiser based on LEDs and a diffraction grating

Abstract

Optical experiments often require fast-switching light sources with adjustable bandwidths and intensities. We constructed a wavelength combiner based on a reflective planar diffraction grating and light emitting diodes with emission peaks from 350 to 630 nm that were positioned at the angles corresponding to the first diffraction order of the reversed beam. The combined output beam was launched into a fibre. The spacing between 22 equally wide spectral bands was about 15 nm. The time resolution of the pulse-width modulation drivers was 1 ms. The source was validated with a fast intracellular measurement of the spectral sensitivity of blowfly photoreceptors. In hyperspectral imaging of Xenopus skin circulation, the wavelength resolution was adequate to resolve haemoglobin absorption spectra. The device contains no moving parts, has low stray light and is intrinsically capable of multi-band output. Possible applications include visual physiology, biomedical optics, microscopy and spectroscopy.

Figure 1. The optical table layout and principle.

(a) Top-view of the optical bench. The LEDs with collimators are mounted on a profile (left side). G: Grating; L: Launching lens, F: fibre; scale bar: 20 cm. (b) The principle of wavelength combining with a grating under a constant exit angle. An LED, incident angle α, produces a specular reflection at the angle β₀ = −α, and a positive first order diffraction at the angle β₊₁. The negative first order angle β₋₁ may exist for shorter wavelengths. The dashed arc represents the approximate input angle range for wavelengths 300 to 700 nm for grating density G = 1850 mm⁻¹, and the dotted arc indicates the negative first order beams (UV to blue). UV_m, B_m, R_m indicate the positions of diffracted orders m, (c) Input angles for a grating with G = 1850 mm⁻¹, for exit angles β₊₁ between 30° and 90°. Solid lines represent input angles α. Dashed lines represent negative order angles β₋₁. Thick lines represent the configuration with exit angle β₊₁ = 50°, used in the prototype. Input angle α (dotted), specular reflection β₀ (dot-dashed) and negative first order β₋₁ (dashed). Shaded area shows the design constraint for small exit angles (β₊₁ <50°), where the input angles may overlap with the exit angle (α = β₊₁) in the wavelength range 500 nm to 700 nm.

Figure 2. Bandwidths and efficiency of the wavelength combiner.

(a) Spectral irradiances measured at the tip of the LEDs. Dotted and dashed line represent the white and green phosphorescent LEDs, respectively. (b) Spectral irradiances of the individual LEDs at the fibre output, adjusted for equal photon flux via PWM. Thick and thin lines show the spectra of consecutive LEDs. (c) The data of panels (a,b) compared in a semi-logarithmic plot. The dotted and dashed lines represent the irradiance spectra of a white LED and green phosphorescent LED, respectively. The arrow shows the spectral narrowing of the white LED. The crosses mark peak irradiances of the LEDs. (d) A semi-logarithmic plot of the overall coupling efficiency of individual LEDs (squares), range 0.1~1.6%, and the PWM duty cycle used to achieve equal photon fluxes at the fibre output (2% for LED 420 nm, 100% for the white LED) (diamonds). The scale on the right side shows the PWM quantisation levels.

Figure 3. Intracellular spectral sensitivity measurements from R1-6 photoreceptors of a blowfly.

(a) Membrane potential trace of an R1-6 blowfly photoreceptor upon stimulation with a graded series of 25 ms pulses with 474 nm light. The pulse intensity was graded in a geometric series (PWM 1, 2, 4, …, 2¹²). The decadic logarithm of the light intensity is indicated on the top axis. The last response (dot) was evoked with a 350 mA pulse from a power LED peaking at 525 nm, directed into the zero-order beam. (b) Membrane potential trace of an R1-6 blowfly photoreceptor upon isoquantal spectral stimulation with 25 ms pulses from the LED synthesiser. (c) Spectral sensitivity obtained from an average of ten sweeps with the LED synthesiser (solid circles). For comparison, the spectral sensitivity obtained from a single spectral sweep with a classical photostimulator is shown (open circles).

Figure 4. Imaging spectroscopy of lysed human blood and of Xenopus skin in situ.

(a) Comparison of tabulated molar extinction spectra of human oxyhaemoglobin (HbO₂, dotted line) and deoxyhaemoglobin (Hb, dashed line) with scaled absorbance measurements of diluted fresh human blood. The spectrophotometric measurement (solid line) closely follows the HbO₂ spectrum above 440 nm; at wavelengths below 440 nm, the measurements are unreliable (shaded area). The absorbance spectrum obtained with imaging spectroscopy using the LED synthesiser (solid circles) closely follows the HbO₂ spectrum above 420 nm. The absorbance spectrum obtained with bare LEDs (open circles) fails to reproduce the fine structure of the HbO₂ absorbance spectrum at 530–580 nm. (b) Absorbance spectra from the artery (A), vein (V), capillary (C), skin background and bright spots. (c) Images of the frog skin obtained with the LED synthesiser. The thin capillaries are best seen in the images between 390 and 430 nm. At 625 nm, the thick vessels show a bright lumen and are surrounded by dark pigmented cells. (d) Images obtained by unmixing with six absorbance components: HbO₂ and Hb above (1, 2) and below 455 nm (3, 6), a spectrally neutral (4) and a red component (5). The LW Hb component discriminates the vein from the artery. The SW Hb component shows an image of the capillaries. (e) False-color image of frog skin. Components 1–3 from (d) were allocated to RGB channels (Red, LW HbO2; Green, SW Hb; Blue: LW Hb). Inset: colour photo of the analysed patch. Scale bar in (d): 1 mm.