Guest authors K. Werner, S. Theeuwen, J. de Boet, V. Bloem, W. Sneijers, H. Mollee, J-J Gommans from NXP Semiconductors present a white paper on the benefits of LDMOS (laterally diffused metal oxide semiconductor) for defence and avionics RF power applications up to 3.8GHz
High voltage LDMOS (laterally diffused metal oxide semiconductor) is the optimal technology choice for defence and avionics RF power applications up to 3.8GHz. It combines high power density and outstanding ruggedness with higher gain and efficiency than bipolar devices.
Furthermore, it’s based on high volume silicon manufacturing processes, with well-known and proven reliability behaviour. Intrinsic properties of LDMOS enable overdrive to +5dB without risk of failure, higher flexibility for different pulse formats and prevent thermal runaways, making the overall system design much simpler than for incumbent bipolar technologies.
LDMOS performance in general
Silicon LDMOS technology has been used in military and aerospace system designs for almost 10 years. The BLA1011-200 was the first LDMOS transistor, which was qualified for avionics applications in 2001.
In the meantime, LDMOS has made further inroads into this market space, supported by ever increasing performance of new process generations and recently also a high voltage (50V) technology to achieve power levels up to 600W per single device .
Ruggedness, or the ability to withstand “harsh” RF conditions in general, be it mismatch or extremely short pulse rise- and fall times, is of prime importance to reliable device performance.
The technologies have been hardened under the most stringent ruggedness tests during development, which is particularly true for the 50V technology.
Amongst other factors, the base resistance of the parasitic bipolar and the drain extension of the LDMOS device play key roles. Both parameters can be engineered to accommodate drain voltage peaks as high as 150V for the 50V technology node.
This ultimately results in devices, which can always withstand a VSWR of at least 10:1 at nominal load and all phase angles and performed flawlessly in the field.
Aluminum metalisation
Early LDMOS technologies worked with gold metalisation and bonding wires and hence had the built-in advantage of high electromigration resistance.
The development of more recent technology nodes in modern submicron fabs, triggered the use of aluminum-based metalisation and bond wires.
Of course, at that time the perception was that aluminum could not withstand the pulsed applications as reliably as gold can. However, extensive research and experience in the field have shown that aluminum is at least as good as gold in this respect, if not better in certain aspects.
Comparing technology generation 4 (Gen 4) with gold and Gen 5 with aluminum metalisation on device level, show identical mean time to failure results although the Gen 5 device has an even higher power density. Figure 5 depicts the results, in this case for GSM base station devices, which experience even higher operating stress (CW operation) than pulsed radar transistors.
Another point of contention was the use of aluminum bond wires. The worry here is the pulsed operation, which might lead to “movement” of the bond wires caused by Joule heating (per pulse) and may eventually lead to breakage or rupture by mechanical fatigue.
We devised run to failure experiments to calibrate models that allow lifetime predictions. It actually turned out, that the breakage of the bond wire would always occur at the heel of the first bond.
The breakage probability (TTF0.1%) also depends on the bond angle (the angle of the wire with the surface) and of course the applied current density.
Authors are K. Werner, S. Theeuwen, J. de Boet, V. Bloem, W. Sneijers, H. Mollee, J-J Gommans from NXP Semiconductors