At present, the RFMD/TRW collaboration and Rockwell are the only companies known to be selling HBT devices.
As for the MESFET vs. HBT issue: every type of transistor has its own set of strengths and weaknesses. Please see the following, which appeared in a recent issue of COMPOUND SEMICONDUCTOR Magazine. it is a summary of RFMD's position in favor of HBTs, plus a response from a company making MESFETs.
Why don't ANAD, TQNT and VTSS produce HBTs? The biggest reason has to do with process technology. HBTs are based on epitaxial structures, whereas these fellows have built their companies around non-epitaxial processes. It is no trivial thing to go set up a completely new process, and since these guys can sell every single MESFET they can make, they have no motivation to do so. And yes, there are intellectual property issues. RFMD is able to get off to a flying start only becasue they are building on the basis of a decade's worth of research into HBTs done at TRW; similarly, Rockwell has a substantial investment in HBT R&D.
begin
The Pros and Cons of HBTs
HBTs are most frequently compared to their GaAs cousins, MESFETs. Here's a summary of the most frequently discussed issues.
Advantages of HBTs:
Efficiency -- Efficiency is a measure of RF power output as a percentage of battery power consumed by the device. GaAs HBT power amplifiers are more efficient and use less power to transmit the same output power than comparable GaAs MESFET devices. Increased efficiency can translate into improved battery life and increased talk time.
Linearity -- Linearity is a measure of the distortion of a signal as a result of the amplification of the signal. GaAs HBT products have higher breakdown voltages and smaller and more constant base-collector capacitance than comparable GaAs MESFET devices, which results in better linearity. Improved linearity can translate into a higher quality signal, which is important for wireless networks using digital air interface standards such as TDMA and CDMA.
Complexity -- GaAs HBT transistors are bipolar devices that require only a single polarity power supply, while GaAs MESFET transistors generally require both a positive and negative power supply, which results in the need to include a negative voltage generator and other additional components or circuitry. As a result, GaAs HBT system architectures are simpler and easier to design, which can translate into reduced component costs and smaller equipment.
Die Size -- GaAs MESFET devices have a lateral or horizontal design structure while GaAs HBT devices have a more vertical structure. GaAs HBT circuits can be made smaller than GaAs MESFET devices, enabling more die per wafer, which can increase manufacturing yields and translate into reduced costs.
Feature Size - The critical geometries of an HBT are formed by epitaxial growth. As a result, the HBT circuits have large feature sizes, and can be fabricated using 1 - 2 micron linewidths. This is much more manufacturable than the 0.25 - 1.0 micron gate geometries typically required by GaAs MESFETs.
MESFET Manufacturers respond:
*HBTs, unlike MESFETs, are based on epitaxial layers. This added expense can raise the total cost of the end device by as much as 2x, removing any economic justification for their use.
*The advantage of the single supply can be easily overcome by an on chip charge pump that takes very little space.
*The supposed improved linearity is only a myth when HBTs are compared to state-of-the-art high volume MESFET process.
*Long term reliability is a chronic issue with HBTs. The performance of commercial HBT processes has been compromised so much to help reliability that any competitive performance advantages over high speed silicon and GaAs MESFETs are lost.
*Since HBTs have a very high turn on voltage (Vbe) the performance becomes severely impacted as power supplies move to ever lower voltages. At 3 V and below, HBTs can't match the high linearity, good gain and low noise characteristics of GaAs MESFETs. This is critical, because OEM manufacturers are relentlessly pushing to lower voltages.
Which side is correct? It's hard to say - this is one of those debates which is unlikely to be solved on paper. But, as the stories in this issue make clear, we won't have to wait much longer for a head-to-head match-up of these two technologies in the marketplace.
Even still, it is unlikely that we will ever be able to say the one of these technologies is demonstrably superior to the other. The more important question will be which company is capable of delivering inexpensive RFICs in a timely fashion.
end |
