+44 (0)24 7671 8970
More publications     •     Advertise with us     •     Contact us
News Article

Components and Subsystems

Modelling down LED production costs
M. Dauelsberg and D. Brien of AIXTRON AG discuss how companies can reduce the cost of ownership of LED production equipment by modelling and simulating thin film deposition processes with ESI group’s CFD-ACE+ multiphysics CFD software.

The lighting industry is experiencing a fundamental change as revolutionary as the introduction of digital cameras has been to photography. Light Emitting Diodes (LEDs), either based on compound semiconductors or novel organic polymers – referred to as OLEDs, Organic Light Emitting Diodes – are substantially more energy efficient in converting electricity into light, last longer, and are more versatile due to their small size than the traditional incandescent light bulb. In addition, LED lighting contributes to energy conservation and reduces CO2 emission.

According to the market research firm Strategies Unlimited, the High Brightness LED market was $4.2 billion in 2006, reflecting an impressive 25-30% average annual growth rate from 2001 to 2006. This growth rate is expected to continue, with a market projection of $9 billion by 2011. Yet despite their already wide spread use in numerous every day applications such as red, amber and green traffic signals, architectural lighting (figure 1), indicators, brake and head lights in cars, backlighting units in large LCD flat screens or even UV disinfection lamps, LEDs still appear to be a few years away from general use for indoor office illumination. In spite of recent productivity advances, the major roadblock to their widespread adoption remains the high manufacturing cost of LEDs.


LEDs based on compound semiconductors are by far the most common type today, and are usually composed of multiple thin layers of so called III-V semiconductors. The name alludes to the fact that these compounds consist of elements from the third and fifth columns of the periodic table, like GaAs, InP, GaN, and multinary compounds thereof, e. g. InGaN, that emit in the blue-green to near UV range of wavelengths. Their properties, like the colour of light emission, are controlled by the thickness and composition of each individual layer in a multilayer stack. The method of choice for making LEDs is Metal Organic Vapour Phase Epitaxy (MOVPE), a deposition technique using gaseous metalorganic and hydride species. The precursors are supplied to the MOVPE reaction chamber with a flow of carrier gas at a process pressure below atmosphere, and the monocrystalline solid thin film is formed on heated substrates at temperatures from 600°C to 1400°C, depending on material and application.

AIXTRON AG develops, manufactures and markets chemical reactors and basic deposition process technology for MOVPE. The accurate control of layer composition and thickness, which can be as low as a few nanometers, with uniformity requirements on multiple substrates of less then +/- 1% deviation from the mean value, poses a formidable challenge to reaction chamber and process design. Simply put, MOVPE is a highly precise chemical process used for the production of optoelectronic devices.

Modelling Deposition
Multiphysics modelling and simulation of semiconductor deposition processes has become an indispensable tool to meet accuracy requirements of the sensitive MOVPE process. AIXTRON has been using ESI Group’s CFD-ACE+ multiphysics CFD software for nearly ten years to provide guidance during conceptualisation, hardware development and process tuning. Today’s industrial requirements to reduce the cost of LED manufacture call for production scale MOVPE reactors with lower cost of ownership. This goal is accomplished by larger and larger reaction chambers accommodating an increased number of wafers, reduced gas consumption per wafer area and increased through-put at enhanced process performance. Modelling and simulation play a key role in reducing the technological risk and the time-to-market for product innovations. The impressive increase in reactor wafer load capacity over the past few years speaks for itself [1]. Figure 2 shows AIXTRON’s flagship MOVPE reactors for the production of GaN based green, blue and white LEDs, the Planetary Reactor for 42x2” wafers (or 11x4” and 5x6” equivalently) and the Close Coupled Showerhead system for 30x2”.


The modelling approach is based on the numerical computation of mixed convective laminar gas flow coupled with heat transfer and multi-component gas species transport. For the growth of group-III Nitrides this basic approach is supplemented by advanced mechanisms including complex gas phase reactions and even gas phase nucleation, i.e. the formation of nano-size particles by interaction between low volatile reaction byproducts, and the deposition of Nitride semiconductor layers on the substrate surface [2,3]. The models have been developed in a joint collaboration with STR Inc. [4] and incorporated into CFD-ACE+ via user subroutines and user defined scalars. While the above mentioned phenomena do not normally occur at lower process pressures below 200 mbar, they may more or less deplete the gas phase from reactants at higher pressures.

The resulting deposition rate fall-off with increasing pressure is related to longer residence time and higher intermolecular collision rates in the gas phase. On the other hand, at certain stages of the LED manufacturing sequence, higher process pressures up to one atmosphere are preferred for device quality reasons.

Figure 3 illustrates how the pressure dependence of GaN growth efficiency could be substantially improved in a Close Coupled Showerhead reactor by a reduced chamber size and modified thermal ambient. Model prediction and measured pressure dependence of GaN growth rates are in very good agreement, considering the complex physical and chemical processes going on in the reaction chamber during the MOVPE process, which must be taken into account to capture the pressure dependence of growth efficiency. Using a quantitative model to understand these phenomena is key to designing larger chambers without compromising process performance and device quality.


The gas inlet of a MOVPE reactor chamber is one of the key components and essentially determines process performance and versatility. Primarily using a rigorous CFD based multiphysics modelling approach, a novel gas injector (figure 4) was developed for a new generation AIXTRON Planetary Reactor with 42x2” wafers [5], which is currently the world’s largest production scale reactor for processing GaN based LEDs. The gas injector features multiple gas inlets for advanced deposition uniformity control, symmetric and uniform gas flow to guarantee wafer-to-wafer reproducibility, and thermal control of the injector to prevent premature gas phase reactions.

In summary, advanced multiphysics modelling helps evaluate the impact of key process and reactor design parameters on performance criteria like the uniformity of layer thickness and composition on wafer and from wafer to wafer, the growth efficiency and the robustness of the process. At AIXTRON, upfront modelling analysis has become an integral part of any development project that involves the reaction chamber of an MOVPE reactor.



[1] D. Brien et al. J. Crystal Growth 303 (2007) 330.
[2] M. Dauelsberg et al. J. Crystal Growth 298 (2007) 418.
[3] J. R. Creighton, G. T. Wang, W. G. Breiland, M. E. Coltrin, J. Crystal Growth 261 (2004) 204.
[4] Nitride CVD module, STR Inc., www.semitech.us.
[5] C. Martin et al. J. Crystal Growth 303 (2007) 318.


Photonis announces agreement to acquire Xenics, a leader in Infra-Red imaging solutions
Phlux Technology Secures £4m in Seed Funding
Start-ups: build up your ecosystem
Trends in the semiconductor industry
Advantest Unveils E5620 DR-SEM for Review and Classification of Ultra-Small Photomask Defects
Xenics introduces Wildcat+ 640
Advantest Introduces New inteXcell Series
Sony Semiconductor Israel Redefines IoT Connectivity Platform
Sensor Sales Stay Strong Due to Big ASP Rise
Advantest Launches ACS Solution Store to Enable Real-Time Data Analytics Solutions
Lam Research Acquires SEMSYSCO to Advance Chip Packaging
Infineon planning a major investment in a new factory in Dresden
Particle Measuring Systems introduces 10 nm aerosol particle counter
EBARA presents DYNOX Gas Abatement System
TSMC Launches OIP 3DFabric Alliance
DuPont reshapes IC materials manufacturing for post-COVID success
Evolving subfab vacuum challenges demand collaborative solutions
Next-gen low-k films can address present, future fabrication challenges
Critical Manufacturing redefines semiconductor MES for greater agility, success
Brewer Science bonding & dielectric materials deliver packaging solutions for 5G, IoT devices
Cutting FOPLP pattern distortion
EV Group advances leadership in optical lithography
ASM joins semiconductor climate consortium
KLA Launches New Double-Sided Direct Imaging Platform
Micron ships advanced DRAM technology with 1β node
Nor-Cal will do business as Pfeiffer Vacuum
ClassOne Equipment Releases Major New Lot of Used Semiconductor Tools for Resale
Park Systems Celebrates the Merger and Acquisition of Accurion GmbH
Amkor to support European semiconductor ecosystem
Nanometre components require sub-nanometre precision
ALD TechDay for More-than-Moore devices
Cadence Expands Collaboration with Samsung Foundry

Search the news archive

To close this popup you can press escape or click the close icon.
Register - Step 1

You may choose to subscribe to the Silicon Semiconductor Magazine, the Silicon Semiconductor Newsletter, or both. You may also request additional information if required, before submitting your application.

Please subscribe me to:


You chose the industry type of "Other"

Please enter the industry that you work in:
Please enter the industry that you work in: