Boston University

Optical Characterization and Nanophotonics Laboratory








High Resolution Raman Microscopy

Members: Anna K. Swan, M. Selim Ünlü


Alumni: Bennett B. Goldberg, Mark James Lande


Hot Topic-Cool Science

Semi-conductor technology has progressed very rapidly since the solid state transistor was first invented, exemplified by the stunning increase in number of processors/area over time. This progress is not at all matched in heat management, which still remains decidedly primitive, with chips mounted on heatsinks with fans to cool them.

We are working in a collaboration with Jet propulsion laboratory, UCLA and University of Huston, where the overall goal is to develop pulsed mid-infrared (IR) semiconductor lasers with effective heat managment. Since the band gaps of 2-5 micrometer IR lasers are so small, elevated temperatures are detrimental. By integrating miniturized versions of thermo-electric (TE) cooling directly with the laser, the small active region of the laser can be spot cooled.

At Boston University, we will determine the time evolution of the local temperature as the TE cooling and laser are pulsed. This is necessary inorder to understand the complicated issues of heat transport through the layered material the laser is composed of. Below, you will find how we are doing it.

Raman microscopy is based on interaction of laser light with lattice vibrations (phonons). The light can either gain a small extra energy from absorbing a phonon or loose some energy by creating a phonon. The population of phonons depends on the temperature, so the temperature is measured by the ratio between the Raman peaks at the phonon loss and gain sides. Since visible light is used, a spatial resolution below 1 micron can readily be achieved using a conventional microscope, with a temperature resolution of a few degrees. Semiconductors have very short optical penetration depth since visible light has energy above the material electronic bandgap. Hence, Raman scattering provides information from the near surface region. The micro-Raman probe we are using comes from Renishaw Inc. We are adding time resolution capabilities to this instrument, so check back with us later. In the meantime, you can have a look at some of our current results.

Measuring temperature with scattered light-Raman spectroscopy

Inelastic light scattering

In the first quarter of the 20th century, C.V. Raman pursued the idea of inelastic light scattering. He was searching for a change of photon frequency for light interacting with a material due to emission or absorbtion of quantized excitations. The discovery of so called Raman lines provided another convincing piece in the puzzle to support the quantum theory of light. Raman was awarded the Nobel Prize in physics in 1930 for his discoveries. Currently, Raman scattering is presently used in a multidude of different fields.

Micro-Raman spectroscopy

Type II superlattice mid IR lasers, consisting of InAs/GaSb/GaInSb/GaSb superlattices with InAs/AlSb superlattice cladding is grown on GaSb substrates. By monitoring the Raman lines of GaSb, we can determine the temperature in the area the laser light is illuminating. Below, we show an example of temperature Raman measurements on GaSb(100). The green light (514.5nm) we are using to probe the sample is well above the bandgap (1.8 mm), so most of the light gets absorbed in the material and only a small fraction of all impinging photons interact with the material to produce a ramanshift. Hence, almost all photon energy is deposited in the material. In order not to heat up the volume we sample, care has to be taken to use sufficiently low power density. This poblem gets worse for low tempertures, where the heat capacity and heat-conductivity drops significantly.

This figure shows the Raman signal from the transverse optical phonon in GaSb at 3 different temperatures. The small peak is on the high energy side where a phonon has been absorbed (called anti-Stokes) and the larger peak (Stokes) is due to creation of phonons. The phonon population follows Bose-Einstein statistics, so that the higher the phonon energy is, or the lower the temperature, the fewer phonons are thermally exited. The ratio R between Stokes and anti-Stokes peaks can be used as a measure of the temperature since R=A exp (hbarW/kT). Here hbarW is the phonon energy, k is the Boltzmann constant and T is the absolute temperature. A is a constant that includes several effects such as the efficiency of the detection system at the two different wavelengths and the possibility of enhancement of one of the Raman peaks closer to an electronic resonance.

Here you can see how the ratio of the raman peaks changes with T, ln(R) vs 1/T for GaSb where the temperature is known and has been varied in a cryostat. The value of the phonon energy is also measured very accurately, so the data points are only fitted for the constant A.

This figure shows the small but dicernable variation of the Raman shift due to temperature.

Inelastic light scattering

We have explored and tested many techniques in thermal imaging including Atomic Force Microscopy and Near-Field Scanning Optical Microscopy, but our research primarily focused on Raman Microscopy and Blackbody Thermal Imaging.

1. AFM based scanning thermal microscopy. The first technique is the Atomic Force Microscope based thermal imager. A cantilevered tip thermal tip is used to replace the Si3N4 cantilever and tip. The tip is composed of the junction of two dissimilar forming a thermocouple. While keeping another junction at a reference temperature the thermoelectric potential is measured and the temperature information is extracted. It is important to note that in the presence of an external bias on a sample, there could be an attractive electrostatic force between the thermocouple tip and the sample. The spatial resolution of such instruments is sub-micrometer and the temperature accuracy is within 0.5 oC [1]. The drawbacks of this instrument are that it remains in contact with the sample causing both thermal and electrical surface alteration. It is a 2-D imaging system and cannot examine any sub-surface thermal distribution. Its application to Si circuits, which are typically covered with overglass may be difficult or impossible. A similar technique may be employed by utilizing the temperature dependence of a Wollaston process wire. The temperature of the probe is monitored by measuring its resistance [2]. Thermal measurements can also performed utilizing differential thermal expansion of a composite cantilever probe on AFM. The composite cantilever probe is made of a thin metal film(aluminum or gold) deposited on a regular silicon nitride AFM probe. During tip-surface contact, heat flow through the tip changes the cantilever temperature which bends the cantilever due to differential thermal expansion of the two probe materials. An AC measurement is used to separate cantilever bending due to temperature and topography. To eliminate image distortion due to air heat conduction, thermal images of a biased resistor were obtained in vacuum(10-5 Torr) [3]. This technique can provide sub-micron spatial resolution with relative temperature accuracy of better than 0.5 oC. However, the measurement is strongly influenced by local thermal conductivity.

In summary, AFM based thermal imaging systems provide very high-resolutions but the samples that can be characterized are very limited. Our laboratory in the Photonics Center at Boston University acquired a thermal AFM system. This system will be used to benchmark and verify resolution for our optical imaging system on samples designed to allow for thermal imaging by AFM.

2. Thermal Imaging by Near-Field Scanning Microscopy (NSOM) Near field scanning optical microscopy and spectroscopy (NSOM) is a recent technique where a tapered optical fiber probe is placed within a fraction of a wavelength of a sample and scanned over the surface. The tapered single-mode optical fiber provides a tiny aperture through which the light is coupled. Because both the tip--to--sample separation and the tip aperture are a small fraction of the wavelength, the spatial resolution is given approximately by the tip diameter typically much better than diffraction limit. Unlu and Goldberg have extensive experience in NSOM specifically applied to semiconductors [9,10]. We have developed and successfully demonstrated the high-resolution capability of NSOM for thermal imaging [11]. Figure 1 shows a IR-NSOM image of a high power Heterojunction Bipolar Transistor (HBT) from by Texas Instruments. These devices are developed for high-speed high-power applications for satellite communications. The uniform temperature distribution is a critical parameter for the optimized performance of these devices. For the relevant device dimensions (for example, the specific example has 2.5 um emitter finger width) conventional thermal microscopy can not provide useful information. Our results demonstrate the first and to the best of our knowledge the only high-resolution (~ 1 um) thermal imaging using a direct optical method. FIGURE 1. Thermal NSOM image of a high power AlGaAs/GaAs HBT. This image shows the non-uniform temperature distribution across the device. Note that only one of the collector contacts (top) is biased, resulting in an asymmetrical current and thus temperature distribution.

The difficulties associated with thermal NSOM are mostly associated with the scarcity and prohibitive cost of single mode fibers at mid-IR wavelengths. In our experiments we have used multimode chalcogenide glass fibers which lack the mechanical robustness required for a reliable tip making process as well as a scanning instrument. Since the IR signal collected through the small aperture is very weak the background radiation from the metal coating on the fiber tip may dominate the signal. Our results were obtained using a lock-in technique. The small throughput of NSOM also prevents two-color imaging.


C. Feng, M. S. Ünlü, B. B. Goldberg, and W. D. Herzog, "Thermal Imaging by Infrared Near-field Microscopy," Proceedings of IEEE Lasers and Electro-Optics Society 1996 Annual Meeting, Vol. 1, November 1996, pp. 249-250


Collaboration partners include Jet propulsion laboratory, UCLA and University of Houston


Support for this project is provided by DARPA HERETIC

© 2007 Trustees of Boston University. All rights reserved.  |  Last modified April 16, 2007 at 12:00 AM EDT