Here we present a protocol to build a rapid Brillouin spectrometer. Cascading virtually imaged phase array (VIPA) etalons achieve a measurement speed more than 1,000 times faster than traditional scanning Fabry-Perot spectrometers. This improvement provides the means for Brillouin analysis of tissue and biomaterials at low power levels in vivo.
The goal of this protocol is to build a parallel high-extinction and high-resolution optical Brillouin spectrometer. Brillouin spectroscopy is a non-contact measurement method that can be used to obtain direct readouts of viscoelastic material properties. It has been a useful tool in material characterization, structural monitoring and environmental sensing. In the past, Brillouin spectroscopy has usually employed scanning Fabry-Perot etalons to perform spectral analysis. This process requires high illumination power and long acquisition times, making the technique unsuitable for biomedical applications. A recently introduced novel spectrometer overcomes this challenge by employing two VIPAs in a cross-axis configuration. This innovation enables sub-Gigahertz (GHz) resolution spectral analysis with sub-second acquisition time and illumination power within the safety limits of biological tissue. The multiple new applications facilitated by this improvement are currently being explored in biological research and clinical application.
Brillouin scattering, first described by Leon Brillouin1 in 1922, is the inelastic scattering of light from the thermal acoustic modes in a solid and from the thermal density fluctuations in a liquid or gas. The spectral shift of the scattered light, usually in the sub GHz-range, provides information about the interaction between the incident light and the acoustic phonons in the sample. As a result, it can provide useful information regarding the viscoelastic properties of the examined material.
In its spontaneous version, Brillouin scattering generally has cross-sections in the order of Raman scattering, resulting in a very weak signal. Additionally, Brillouin frequency shifts are orders of magnitude smaller than Raman shifts. As a consequence, elastically scattered light (from Rayleigh or Mie scattering), stray light, and back-reflections off of the sample can all easily overshadow the Brillouin spectral signature. Hence, a Brillouin spectrometer needs to not only achieve sub-GHz spectral resolution but also high spectral contrast or extinction.
In traditional Brillouin spectrometers these requirements are met by scanning-grating monochromators, optical beating methods, and, most popularly, multiple-pass scanning Fabry-Perot interferometers2. These methods measure each spectral component sequentially. This approach leads to acquisition times for a single Brillouin spectrum ranging from a few minutes to several hours, depending on the instrument and on the sample. The two-stage VIPA spectrometer, built using this protocol, has the ability to collect all of the spectral components within less than a second while providing sufficient extinction (>60 dB) to effectively suppress other spurious signals2.
The integration of the VIPA etalons is the key element of this spectrometer. A VIPA is a solid etalon with three different coating areas: in the front surface, a narrow anti-reflection coating strip allows the light to enter the VIPA, while the rest of the surface features a highly reflective (HR) coating; in the back surface, a partially reflective coating allows a small portion (~5%) of the light to be transmitted. When focused onto the narrow entrance of the slightly tilted VIPA, the light beam gets reflected into sub-components with fixed phase difference within the VIPA2. Interference between the sub components achieves the aspired high spectral dispersion. Aligning two VIPAs sequentially in cross-axis configuration introduces spectral dispersion in orthogonal directions3. The spectral dispersion in orthogonal directions spatially separates the Brillouin peaks from unwanted crosstalk, which makes it possible to pick up only the Brillouin signal. Figure 1 displays a schematic of the two stage VIPA spectrometer. The arrows below the optical elements indicate the degree of freedom in which the translational stages should be oriented.
Figure 1. Instrumental setup. An optical fiber delivers the Brillouin scattering into the spectrometer. A cylindrical lens C1 (f=200 mm) focuses the light into the entrance of the first VIPA (VIPA1). Another cylindrical lens C2 (f=200 mm) maps the spectral angular dispersion into a spatial separation in the focal plane of C2. In this plane, a vertical mask is used to select the desired portion of the spectrum. An analogous configuration follows, tilted at 90 degrees. The beam passes through a spherical lens S1 (f=200 mm) and is focused into the entrance slit of the second VIPA (VIPA2). A spherical lens S2 (f=200 mm) creates the two-dimensional spectrally separated pattern in its focal plane, where another horizontal mask is placed. The horizontal mask is imaged onto the EMCCD camera using an achromatic lens pair. Please click here to view a larger version of this figure.
An undergraduate student with some optics coursework and basic aligning experience should be able to build and use this two-stage spectrometer. The spectrometer has recently been shown to be compatible with a variety of standard optical probes3,4,5 (e.g., confocal microscope, endoscope, slit-lamp ophthalmoscope). Here, the spectrometer is connected to a confocal microscope. The laser light is aligned into a standard research inverted system microscope after integrating a 90:10 beam splitter. The backscattering light from the sample is coupled into a single mode fiber, making the microscope confocal.
Note: Brillouin spectral analysis requires a single-longitudinal mode laser (~10 mW at the sample). For aligning purposes, use a strongly attenuated portion of this laser beam (<0.1 mW).
1. Initial Setup of Fiber and the EMCCD (Electron Multiplied Charge Coupled Device) Camera
2. Horizontal Stage of Spectrometer
3. Vertical Stage of Spectrometer
4. Combination of the Two Stages and Final Alignment
5. Measuring the Brillouin Shift
6. Calibration and Analysis of Brillouin Spectrum
Figure 2. Spectrometer calibration. (A) EMCCD camera frame obtained from calibration sample. (B) Lorentzian curve fit (red) to the measured data (blue). Please click here to view a larger version of this figure.
Figure 3 shows representative Brillouin spectra and their fits for different materials. The VIPAs both have a thickness of 5 mm which results in a FSR of approximately 20 GHz. The integration time for these measurements was 100 msec. 100 measurements were taken and averaged. One calibration measurement was taken prior to acquiring the spectra.
Figure 3. Brillouin Spectra of different materials. Lorentzian curve fit (red) to the measured data (blue). (a) Brillouin spectrum of Methanol. The measured Brillouin shift is 5.59 GHz. (b) Brillouin spectrum of Ethanol. The measured Brillouin shift is 5.85 GHz. (c) Brillouin spectrum of Polystyrene. The measured Brillouin shift is 14.12 GHz. Please click here to view a larger version of this figure.
The obtained Brillouin shifts agree with previously published data3,6,7. To determine if the alignment of the spectrometer is optimal, many spectral measurements of the same material can be taken sequentially, and the standard deviation of the peak positions can be calculated. Figure 4A shows a time-trace of 250 Brillouin measurements of methanol taken sequentially; a histogram of the evaluated Brillouin shifts is shown in Figure 4B. A well-aligned spectrometer with 5 mW of light at the sample and an integration time of 100 msec will have a standard deviation of σ ~ 10 MHz. Changes in Brillouin shift within corneal and lens tissue have been measured to be on the order of 1 GHz9,10,11. Therefore, Brillouin shift readings with variability of ≤10 MHz will enable the measurement of relevant mechanical variations in tissue.
Figure 4. Deviation in Brillouin shift over 250 methanol measurements. (A) Time-trace of 250 Brillouin measurements of methanol. (B) Histogram of Brillouin shift deviation over 250 measurements. Please click here to view a larger version of this figure.
A key design feature of this spectrometer configuration is that the two stages can be aligned independently. When a VIPA etalon is slid out of the optical path, the remaining lenses of the spectrometer stage form a 1:1 imaging system, so that the spectral pattern from each stage is imaged onto the CCD camera. Therefore, it is straightforward to go back to either one of the stages to improve its performance without affecting the alignment of the other stage. The set of translational stages and degrees of freedom suggested in the protocol follow this philosophy of maintaining the ability to independently optimize both stages of the spectrometer.
It is difficult to mount the VIPA etalons such that the tilt degree of freedom rotates around the input window axis. Subsequently, when operating with commercially available opto-mechanical components, tilting the etalon inevitably introduces a small translation of the input window. The translation along the optical beam axis should not induce negative effects because of the large Rayleigh range of the input lens (~3 mm). On the other hand, the translation on the axis orthogonal to the beam propagation may significantly decrease the throughput of the spectrometer. This is why the translational and the tilt degrees of freedom on the VIPA holder have to be operated in tandem to maximize the finesse and the throughput. It is suggested to start the alignment procedure with relatively high tilt angle (~3-5 degrees) so that coupling light into the VIPA etalon is nearly lossless. As the alignment improves, the tilt angle can be decreased to improve the finesse. The optimization of the finesse and the throughput is an important part of the alignment protocol, especially for applications in biological materials where integration time and laser power must be kept as low as possible.
In the protocol, it has been suggested to extract the Brillouin shift from the analysis of two spectral peaks, namely the Stokes and Anti-Stokes peaks from two adjacent diffraction orders. This is an intrinsically robust procedure that minimizes artifacts due to laser frequency drifts, or etalon temperature changes. However, measuring peaks with significantly different spectral shift may have drawbacks. In fact, the provided procedure for spectral calibration is based on the assumption of constant spectral dispersion across the spectral pattern. In reality, the spectral dispersion increases in the lower orders of diffraction. As a result, the regions of the spectrum that are analyzed (i.e., the Stokes and Anti-Stokes peaks from two adjacent diffraction orders) may have different spectral dispersions. In this case, the spectral calibration procedure written here will provide inaccurate Brillouin shifts. It is suggested that more materials of known Brillouin shift should be used in this situation to build a spectral calibration curve with polynomial fit. This is particularly important if the absolute value of the Brillouin shift has to be compared rather than the relative change in Brillouin shift between two conditions.
Typically, the finesse of free-space etalons, including the VIPA described here, does not surpass ~50. As a consequence, there will be a trade-off to consider between high spectral resolution and free spectral range. In this protocol a free spectral range of 20 GHz is suggested for green (532 nm) laser operation since most biomaterials have Brillouin shifts of less than 10 GHz. Only half of the FSR is available for Brillouin analysis because the Stokes and anti-Stokes frequency shifts larger than half FSR will appear aliased in the spectrum.
If difficulties are encountered in observing the Brillouin signal, it is important to recognize whether the issues are related to an excessive amount of stray light, or to a low number of Brillouin photons. Excessive stray light should be effectively eliminated. Ensure that the black box enclosure is light-tight. Without sample, turning on room lights or turning off the laser should not significantly change the EMCCD camera background photon count. To eliminate laser light reflected off of the sample’s surface, slightly tilt the sample, and start the observation with spatial masks closed as much as possible without blocking the signal. These two procedures will enable increasing the integration time to allow observation of a very dim Brillouin signal. If still there is no signal, it is probable that the Brillouin signal is too weak. Use a different sample with strong Brillouin signal such as Methanol, or check the alignment of the collection optics in the confocal microscope. After successfully observing a signal, optimize it further by following step 5.4 described in the protocol.
Loss of incident light due to absorption or scattering will increase the acquisition time required for sample analysis. As a result, best results are usually obtained in transparent or thin materials. A well aligned spectrometer should be able to get a high photon count (>1,000 at the peak) with 5 mW of light at the sample and 100ms of integration time for liquid materials or clear plastic samples. This is significantly faster than traditional scanning spectrometers. Due to its low acquisition time and illumination power, such a spectrometer has enabled using Brillouin spectroscopy for in vivo-mechanical imaging3,10,11,12. Using this type of spectrometer, several groups have demonstrated a variety of applications such as measuring the rheological properties of the eye lens13, detecting bacterial meningitis in spinal fluid4, and analyzing the corneal elastic modulus14.
Further improvements to the spectrometer are expected in the near future, especially if low-loss ultra-narrow band pass and/or notch filter can be incorporated to relax the extinction requirements on the VIPA spectral dispersion. Since the spectrometer can be used in combination with a variety of standard optical probes, for example confocal microscopes3,4,5, endoscopes, and slit-lamp ophthalmoscopes, the VIPA spectrometer could be an instrumental component in several biomedical applications.
The authors have nothing to disclose.
This work was supported in part by the National Institutes of Health (P41-EB015903, R21EY023043, K25EB015885), National Science of Foundation (CBET-0853773) and Human Frontier Science Program (Young Investigator Grant).
OPTICS: | |||
VIPA (virtual image phase array) | LIGH MACHINERY | Quantity: 2 | |
Bundle of Three 423 Linear Stages with SM-25 Micrometers | NEWPORT | 423-MIC | Quantity: 1 |
SS Crossed-Roller Bearing Translation Stage, 0.5 in., 8-32, 1/4-20 | NEWPORT | 9066-X | Quantity: 1 |
Vernier Micrometer, 13 mm Travel, 9 lb Load Capacity, 50.8 TPI | NEWPORT | SM-13 | Quantity: 1 |
Adjustable Width Slit | NEWPORT | SV-0.5 | Quantity: 2 |
Compact Dovetail Linear Stage, 0.20 in. Z Travel, 1.57×1.57×1.38 in. | NEWPORT | DS40-Z | Quantity: 2 |
Slotted Base Plate, 25 or 40mm to 65mm Stage, 1.1 in. Range | NEWPORT | B-2B | Quantity: 2 |
Ø1/2" Optical Post, 8-32 Setscrew, 1/4"-20 Tap, L = 2", 5 Pack | THORLABS | TR2-P5 | Quantity: 2 |
Ø1/2" Post Holders, Spring-Loaded Hex-Locking Thumbscrews, L = 2", 5 Pack | THORLABS | PH2-P5 | Quantity: 1 |
Ø1/2" Post Holders, Spring-Loaded Hex-Locking Thumbscrew, L = 3", 5 Pack | THORLABS | PH3-P5 | Quantity: 1 |
Imperial Lens Mount For 2" Optics, 8-32 Tap | THORLABS | LMR2 | Quantity: 2 |
f=200.0 mm, Ø2" Achromatic Doublet, ARC: 400-700 nm | THORLABS | AC254-200-A | Quantity: 2 |
Kinematic Mount for up to 1.3" (33 mm) Tall Rectangular Optics, Right Handed | THORLABS | KM100C | Quantity: 2 |
Fixed Cylindrical Lens Mount, Max Optic Height: 1.60" (40.6 mm) | THORLABS | CH1A | Quantity: 2 |
f = 200.00 mm, H = 30.00 mm, L = 32.0 mm, N-BK7 Plano-Convex Cylindrical Lens, Antireflection Coating: 350-700 nm | THORLABS | L1653L1-A | Quantity: 2 |
Right-Angle Post Clamp, Fixed 90° Adapter | THORLABS | RA90 | Quantity: 1 |
Adapter with External C-Mount Threads and Internal SM1 Threads | THORLABS | SM1A9 | Quantity: 1 |
Studded Pedestal Base Adapter, 1/4"-20 Thread | THORLABS | PB4 | Quantity: 2 |
Spacer, 2" x 3", 1.000" Thick | THORLABS | Ba2S7 | Quantity: 2 |
543 nm, f=15.01 mm, NA=0.17 FC/APC Fiber Collimation Pkg. | THORLABS | F260APC-A | Quantity: 1 |
SM1-Threaded Adapter for Ø11 mm collimators | THORLABS | Ad11F | Quantity: 1 |
Translating Lens Mount for Ø1" Optics, 1 Retaining Ring Included | THORLABS | LM1XY | Quantity: 1 |
Single Mode Patch Cable, 450 – 600 nm, FC/APC, 2 m Long | THORLABS | P3-460B-FC-2 | Quantity: 1 |
1:1 Matched Achr. Pair, f1=30 mm, f2=30 mm, BBAR 400-700 nm | THORLABS | MAP103030-A | Quantity: 1 |
SM1 Lens Tube…length to adjust depend on CCD, we have 3.5 inches | THORLABS | SM1LXX | Quantity: 1 |
Base Adapters for Ø1/2" Post Holders and Ø1" Posts | THORLABS | BE1 | Quantity: 8 |
Clamping Forks for Ø1/2" Post Holders and Ø1" Posts | THORLABS | CF125 | Quantity: 8 |
HW-KIT5 – 4-40 Cap Screw and Hardware Kit for Mini-Series | THORLABS | HW-KIT5 | Quantity: 1 |
D20S – Standard Iris, Ø20.0 mm Max Aperture | THORLABS | D20S | Quantity: 2 |
FOR ENCLOSURE | |||
25 mm Construction Rail, L = 21" | THORLABS | XE25L21 | Quantity: 6 |
1" Construction Cube with Three 1/4" (M6) Counterbored Holes | THORLABS | RM1G | Quantity: 8 |
Right-Angle Bracket for 25 mm Rails | THORLABS | XE25A90 | Quantity: 12 |
25 mm Construction Rail, L = 15" | THORLABS | XE25L15 | Quantity: 4 |
25 mm Construction Rail, L = 9" | THORLABS | XE25L09 | Quantity: 8 |
High Performance Black Masking Tape, 2" x 60 yds. (50 mm x 55 m) Roll | THORLABS | T743-2.0 | Quantity: 1 |
Low-Profile T-Nut, 1/4"-20 Tapped Hole, Qty: 10 | THORLABS | XE25T3 | Quantity: 1 |
1/4"-20 Low-Profile Channel Screws (100 Screws/Box) | THORLABS | SH25LP38 | Quantity: 1 |
60" (W) x 3 yds. (L) x 0.005" (T) (1.5 m x 2.7 m x 0.12 mm) Blackout Fabric | THORLABS | BK5 | Quantity: 1 |
CAMERA, LASER and MICROSCOPE | |||
EMCCD camera | ANDOR | iXon Ultra 897 | Quantity: 1 |
400 mW single mode green laser | LASER QUANTUM | torus 532 | Quantity: 1 |
Research Inverted System Microscope | OLYMPUS | IX71 | Quantity: 1 |