21 Jun 2007
GaN chip developers look set to benefit from an in situ pyrometer that can measure wafer temperatures with a precision of ±0.1°C. Richard Stevenson speaks to Thomas Zettler of optical sensor specialist LayTec to find out more about its ultraviolet pyrometer.
Profitable chip making demands low development costs and high production yields. To cut initial expense, growth recipes should be established using the minimal number of runs, since this optimization process can consume a large proportion of the development budget. To do this, process engineers must know as much as possible about the reactor's local environment, including the wafer temperature – a primary driver of epilayer growth rates and compositions.
Pyrometry is the standard and established method for measuring the wafer temperature within a reactor. The technique involves measuring the intensity of thermal radiation emitted by the wafers over a narrow wavelength band using a photodetector, and then correlating this intensity to a temperature.
The temperature of wafers based on InP and GaAs material systems can be measured with a pyrometer operating at 950 nm. However, this spectral region is less suitable for nitrides, because they do not produce any radiation at this wavelength, so only the susceptor surface below the wafer is measured.
To address this deficiency, pyrometers have been built specifically for nitride growth that operate at 400 nm. These devices measure the temperature of GaN directly, and more precisely, than their longer wavelength equivalents. The first of these was constructed by J Randall Creighton and co-workers from Sandia National Laboratories, US. Then at the end of 2006, optical sensor expert LayTec of Germany (see tinted box) introduced a commercial version of this tool, the Pyro 400, at the MRS Fall meeting in Boston, US. This instrument is primarily designed for Aixtron multi-wafer reactors, but could also be adapted for Thomas Swan tools.
Developing a pyrometer that operates in the near ultraviolet is a significant challenge, according to LayTec's president Thomas Zettler. "In a reactor at 1000 °C there is very intense radiation at all wavelengths, and it's a huge problem to measure 400 nm emission accurately," explained Zettler.
Emission at 400 nm can be many orders of magnitude weaker than that at longer infrared wavelengths, and the careful selection of several filters is required to block out this unwanted radiation. Operating in this wavelength range enables LayTec's instrument, which consists of collection optics, filters and a detector, to measure the surface temperature of multiple wafers with a precision of ±0.1 °C or less. This precision exceeds that of the company's EpiTT instrument and of all other commercial tools, claims Zettler.
Although the Pyro 400 can measure temperatures with a very high precision, it does not feature emissivity correction. This means that the instrument ignores the variations in emissivity between different objects, and that the growth of an antireflection film would produce a change in the value of the recorded temperature. However, Zettler believes that this weakness is not an issue. According to him, there is only a small difference between the emissivities of GaN and AlN layers and that the product is designed as a calibration tool to check the bare wafer surface temperature.
To drive sales, LayTec will have to convince its customers to add the Pyro 400 to other in situ instruments already installed in reactors, such as reflectance tools to monitor wafer bowing. These reflectance monitors can ensure wafer flatness, but they cannot reveal whether temperature variations occurred across the substrate during the growth. These differences can exist even before growth begins, says Zettler, as they come from slight variations in the susceptor's geometry.
Zettler believes that the Pyro 400 can deliver the greatest benefit during the development of new growth recipes for future products. These new recipes must produce wafers with uniform, high-quality active regions and a low degree of bowing. The number of runs required to fine-tune this growth can be reduced with the new 400 nm pyrometer, says Zettler. He adds that the in-house expertise generated from the development of the 400 nm instrument has also enabled the company to show its customers how to get more out of their wafer-bowing equipment.
At the MRS meeting Zettler's colleague Elizabeth Steimetz presented a poster that detailed the capabilities of the Pyro 400, which was co-authored by researchers from the Ferdinand-Braun-Institute for Very High Frequency Technology (FBH) in Berlin. The 400 nm pyrometer was fitted to an Aixtron AIX2600HT planetary reactor equipped with LayTec's EpiCurve sensor, and mapped the temperature of a platter containing 11 different 2-inch substrates and epiwafers with a precision of ±0.1 °C and a ±3 mm spatial resolution.
These trials were conducted at 1100 °C and a platter rotation speed of six revolutions per minute. The tests showed that convex bowing results from the growth of GaN on sapphire and produces a 4 °C temperature variation across the wafer, which can be measured with a resolution of 0.1 °C. Switching to a SiC platform replaces a convex bow with a concave one, and variations in wafer temperature increase to 12 °C. According to the researchers, lowering the reactor's temperature to 800 °C dramatically reduced the bowing of both of these epiwafers, leading to growth of quantum wells with excellent uniformity.
• This article originally appeared in the June 2007 issue of Optics & Laser Europe magazine.Optics & Laser Europe magazine – subscribe here