Raman spectroscopy-based measurements of thin film thickness

Raman spectroscopy-based measurements of thin film thickness

Silicon-on-sapphire (SOS) devices are used in a range of defense and aerospace radiation-hardened communications applications. Metrology of the Si film thickness is needed during the fabrication of these devices for quality and process control purposes. However, due to the nature of the sapphire substrate, conventional thin metrology tools such as ellipsometry are challenging to use.

In a recent research study funded by the Defense Microelectronics Activity (DMEA) (Contract No. HQ072718C008), Advanced Cooling Technologies, Inc. (ACT), in collaboration with Iowa State University, has developed a new thin-film metrology tool based on the principles of Raman spectroscopy.

What is Raman Spectroscopy?

Raman spectroscopy is a measure of the inelastic scattering of light, in which incident photons exchange energy with vibrational modes of the material. When light is incident on a crystalline material, it sends the vibrational modes of the material into a higher energy “virtual state” before being reemitted, i.e., scattered. The majority of photons scatter at the same frequency as the incident light (Rayleigh scattering). However, around one in a million photons scatter at a different frequency as the molecules settle into a different vibrational state. This is known as Raman scattering (Figure 1)

Figure 1: Energy level diagram illustrating transitions involved in Rayleigh (elastic) and Raman (inelastic) scattering

Calculating the Relative Raman Intensity

Raman spectroscopy is commonly used for identifying materials or probing their molecular structures. In our research, we have shown that if the experimental conditions are adequately controlled, the intensity of the Raman signal from a thin film material can be correlated to its thickness. However, due to interference by the multiple reflections of both the incident and Raman scattered light, the relation between Raman signal intensity and thin film thickness is complicated.

A theoretical model was developed for calculating the relative Raman intensity as a function of film thickness [link to paper]. To account for the uncertainty associated with the complicated Raman signal intensity trend with thickness, a dual-laser approach was developed. A dual-laser Raman spectrometer, equipped with 532 nm and 785 nm lasers, along with associated LabVIEW-based controls, was setup to demonstrate the technique (Figure 2).

Figure 2: Dual-laser Raman spectrometer setup, highlighting key components

The developed model was verified with Raman measurements of SOS thin films with thicknesses from 80-500 nm (Figure 3). For comparison of the measurements to the model, an instrument-specific scaling factor is determined using a least-squares fitting of the measured intensities (Figure 4).

Based on these results, ACT has successfully demonstrated the determination of SOS thin film thicknesses using this dual-laser Raman intensity approach. This Raman-based technique may also be extended to other semiconductor materials and thin-film systems, as well as next-generation two-dimensional atomic layer materials. This technology may be used for semiconductor device process quality control where alternative metrology techniques such as ellipsometry may not be feasible.

Figure 3: Raman spectra obtained from SOS thin film samples using 532 nm laser. Spectra offset on y-axis for clarity. Notice the non-monotonic trend of peak intensity with thickness

Figure 4: Comparison of predicted intensity to measured intensity of the 521 cm-1 Si Raman peak for SOS thin films, taking into account instrument-specific scaling factor.

Rigorous Prediction of Raman Intensity Whitepaper       CONTACT US TO DISCUSS YOUR APPLICATION

THERMAL OPTICAL TEST CART FOR LASER DIODE THERMAL MANAGEMENT

Advanced Cooling Technologies, Inc. (ACT) has developed a Thermal and Optical Test Cart to evaluate the performance of microchannel coolers used in high-power laser diode thermal management. Copper microchannel coolers (MCCs) used in these applications are prone to degradation due to erosion-assisted corrosion. To provide protection from this degradation mechanism, ACT developed the Applied Nanoscale Corrosion Erosion Resistant (ANCER™) coating which can be applied uniformly to the high aspect ratio heat transfer features inside the MCCs. The Thermal and Optical Test Cart was developed to evaluate the effectiveness of this coating to prevent erosion-assisted corrosion in MCCs by performing extended life tests of ANCER™-coated and baseline-uncoated copper MCCs.

Figure 1: ACT’s Thermal and Optical Test Cart for performing life test evaluations of microchannel coolers used in laser diode thermal management.

The Thermal and Optical Test System consists of a pumped loop for de-ionized water (DIW) coolant, which is required in laser diode cooling applications to prevent leakage currents. The test cart has DIW conditioning and controls to monitor and maintain DIW quality parameters such as resistivity, pH, and dissolved oxygen content. Laser diode microchannel coolers are typically assembled into “stacks” of several MCCs that are manifolded in parallel. The test cart can accommodate several MCC stacks and provides measurements of DIW pressure differential and temperature increase for each stack. Additionally, the Thermal and Optical Test System is equipped with the capability to periodically monitor the optical performance of the light emitted by the laser diodes or other light-emitting devices such as LEDs. By collecting the light into an integrating sphere, the total optical power output and wavelength distribution can be measured. The Thermal and Optical Test System is shown in Figure 1.

ACT has used the Thermal and Optical Test System to evaluate two sets of commercially available copper MCCs. The first set uses resistive heaters to simulate the laser diode waste heat load. Temperature measurements on the surface of these heaters allow a calculation of the thermal resistance of the MCCs. A 5200 hour life test of ANCER™-coated and baseline-uncoated MCCs has shown that the thermal resistance of the uncoated coolers has increased by over 30%, while the thermal resistance of the ANCER™ coated coolers has remained stable. These results are shown in Figure 2.

Figure 3: Spectral distribution of ANCER™ coated and baseline uncoated laser diode MCC stacks at the beginning of an extended life test, as measured in the Thermal and Optical Test Cart.

The second set of commercially available copper MCCs evaluated by ACT was assembled in stacks that have actual laser diodes bonded to the coolers. Figure 3 shows the spectral distribution measured for the ANCER™ coated and baseline uncoated laser diode stacks at the beginning of an extended life test, demonstrating the capability of the Thermal and Optical Test System.

For more information on testing microchannel coolers or cold plates, or developing a custom test apparatus to evaluate the thermal and hydraulic performance of your thermal management devices, contact ACT at (717) 295-6061.

See more Custom Thermal and Mechanical Systems

C.T.E. Matched Vapor Chambers

Figure 1. C.T.E. matched vapor chamber allows direct bonding, eliminating a thermal interface. The vapor chamber acts as a thermal transformer, spreading the heat so that it can be removed by air cooling.

There are a growing number of high power  applications across industries that require highly efficient cooling solutions. Some of these products have very high localized heat fluxes, greater than 300 W/cm², but must maintain a tight temperature range so as not to disrupt sensors or optical devices. Direct die attachment can lead to mechanical stresses at the interface if the coefficient of thermal expansion (C.T.E.) is mismatched between the die and the substrate.  The standard method to overcome this is to add an interface material, such as thermal gap pads or thermal pastes, to accommodate the mismatch.  Unfortunately the presence of this thermal interface layer increases the thermal resistance and likewise increases the temperature on the device itself.

Vapor Chambers  are an important tool in the thermal management toolbox, since they act as flux transformers, spreading the high input heat flux over the entire surface of the vapor chamber.  This allows the heat to be removed from the vapor chamber by conventional cooling methods.  Most vapor chambers are limited to an input heat flux of about 75 W/cm², however, ACT has developed a C.T.E matched vapor chamber that allows for direct bonding of the heat source; see Figure 1.  This unique device has been demonstrated to dissipate heat fluxes as high as 700 W/cm² and 2kW overall.  The evaporator thermal resistance of vapor chambers with this wick design is only 0.05 K-cm²/W.

The overall envelope structure is aluminum nitride with a direct bond copper exterior.  The copper on the inside of the vapor chamber ensures that the well-known water/copper performance is maintained.  In areas where the heat source is to be attached, the copper layer is removed exposing Aluminum nitride.  Aluminum nitride has a CTE of ~5.5 ppm/⁰C, which is close to many common semiconductor materials.  The devices can be directly attached to the vapor chamber, eliminating the need for a thermal interface layer.

Return to Vapor Chamber Assemblies…

PCB Level Heat Spreading

At the printed circuit board (PCB) level, the heat flux is the highest in the system.  While it is advantageous to dissipate heat as close to the source as possible, this can be difficult to do while simultaneously satisfying the electrical isolation requirements.  Unfortunately, in many cases electrical isolation is only achieved using materials that are thermally insulating, such as with FR4 boards.  Recent work at ACT has explored adding heat pipes to the structure of Metal Core Printed Circuit boards to help spread heat right at the source.  If the ratio of the heat source to circuit board area is sufficient, this can be an effective way to improve heat spreading at the board level while requiring minimal design impacts to the system.

Figure 1. Photograph of metal core printed circuit board with heat pipes soldered into the metal core of the circuit board.

Figure 1 shows an example of heat pipes embedded into a Metal Core Printed Circuit Board (MCPCB).  Heat pipes are seen on the left, while the reverse ‘circuit side’ is seen on the right.

In Figure 2, a 3 LED structure is shown on the right, while a thermal image of that structure is seen on its left.  Measured results show that embedded heat pipes can reduce the heat spreading resistance by 45% over the standard aluminum MCPCB and even 15% over a copper MCPCB.  This is a valuable improvement, particularly so close to the heat source.

Figure 2. IR images and photographs of heat pipe embedded circuit board during LED operation. The scale (58°C to 68°C) has been set to emphasize the thermal spreading in the circuit board. The heat spreading resistance is reduced by 45% over the standard aluminum MCPCB.

Learn more about Heat Pipes…

HiK™ Plates to Improve Size, Weight, and Power (SWaP)

Aluminum HiK™ Plate

High Conductivity (HiK™) heat sinks can also improve the Size, Weight, and Power (SWAP) compared to standard heat sinks. Placing a discreet heat source on a large metal heat sink will produce large thermal gradients as the heat slowly conducts through the aluminum to the fins.  Embedding heat pipes in a HiK™ heat sink can increase the thermal conductivity from around 180 W/m K to 500-1,200 W/m K, providing an opportunity to reduce heat sink plate thickness and fin area. This approach has been proven in a variety of weight/volume sensitive applications including: Ruggedized Electronics, UAVs, Handheld/Portable Devices, LEDs and optical devices.

Figure 1. A HiK™ natural convection heat sink reduces the mass by over 34% when compared with an all-aluminum heat sink with the same thermal performance.

Embedded heat pipes can improve performance and reduce the mass of forced and natural convection heat sinks.  ACT fabricated a HiK™ heat sink and an all-aluminum heat sink with the same performance; see Figure 1.  The total heat dissipation is 150W in both cases.   The conventional aluminum heat sink is 12 inches (30.5 cm) long, weighs 9.6 lbs. (4.4 kg) and has a base thickness of 0.6 inch (1.5 cm).  Introduction of 5 heat pipes, 3 in close proximity to the heat source and another two a little further out for improved spreading reduced the length to 10 inches (25.4 cm), reduced the thickness to 0.28 in (0.7 cm), and reduced the mass to 6.3 lbs. (2.9 kg) for an overall material reduction of over 34%. Thermal images that demonstrate the improvement are shown in Figure 2.  The Hi-K heat sink seen on the right maintains the same source temperature, even though the heat sink is shorter, lighter, and thinner. The improvement is directly attributable to the addition of heat pipes which can be seen as red lines in the picture on the right.

LED Extrusion Isothermalization

Heat pipes have an effective thermal conductivity of 10,000 to 100,000 W/m K, compared with aluminum’s thermal conductivity of about 180 W/m K.  Therefore, the performance of large aluminum heat sinks can be improved with embedded heat pipes.  The high effective conductivity allows the heat pipes to spread the heat throughout the heat sink.  This heat spreading reduces the thermal gradient and likewise reduces the max temperature at the LED source.  Benefits include:

• Improved lifetime and reliability by operating with the same power at a lower temperature
• Increased optical output by operating at the same temperature with higher powers
• Reduced heat sink size and weight

Our testing and analysis has confirmed that the longer the extrusion, the more heat pipes improve performance.  As seen in Figure 1, the percentage improvement in thermal resistance with heat pipes increases approximately linearly with increasing heat sink length. For example one can expect to see a 5% improvement in thermal resistance for a 5 cm long heat sink, increasing to 30% for a 30 cm long heat sink.  Please note that one can expect the benefit will be more noticeable in natural convection heat sinks, as fan operation plays a major role in forced convection performance.

Figure 2. IR image and photograph of a heat pipe embedded heat sink dissipating 100 W.

Figure 2 shows a photograph and an infrared (IR) image of a heat sink with embedded heat pipes.  In this case heat pipes were embedded in a 200mm long radial heat sink that that was dissipating 100 watts of heat.  The heat pipe improves the efficiency of the heat sink, by transferring heat with a very low ΔT from the LED at the bottom to the entire length of the heat sink.

Figure 3 shows calculated temperatures for identical heat sinks, with and without embedded heat pipes.  The heat pipes decreased the LED maximum temperature by 10⁰C, which will help achieve long, reliable operation.

Figure 3. Comparison of identical heat sinks with (B) and without (A) embedded heat pipes, dissipating 100 W. The heat pipes reduce the LED temperature by 10°C, helping to increase life and reliability.

Learn More about Photonics Thermal Management…

LED Case Study – The Remote Sink

In many lighting applications the LED device must fit in a fixed space to accommodate a variety of customer requirements, which often do not account for thermal management considerations.  A common example is luminaire design, where the ceiling or wall fixtures are based on pre-existing designs using non-LED technologies.  These designs commonly have both restricted space for heat dissipation through conduction, and limited air flow to remove heat via convection.   In cases where there is space to remotely dissipate the heat, heat pipes can be used to transport the heat from the device to a heat sink located elsewhere.  This is called the remote sink.

Figure 1. Heat pipes transfer heat from the LED to a remote sink, with very small temperature drops.

The remote sink solution has a heat pipe in direct contact with the LED device at one end, which serves as the evaporator. At the other end the heat pipe is connected to the heat sink, the condenser.  A sketch of a conceptual design can be seen in Figure 1.  Here two heat pipes are in direct contact with both the LED at the bottom and heat dissipating fins at the top. A wall or other enclosure can be placed in between the LED and heat sink to separate the two.

Aluminum has a thermal conductivity of about 180 W/m K, while the thermal conductivity of copper is only 400 W/m K.  In contrast, the effective conductivity of a heat pipe can range from 10,000 to 100,000 W/m K.   This high effective thermal conductivity allows the heat sinks to be located remotely from the LED.