Strong growth means TriQuint VP has good reason to be cheerful

"2006 was an important year: TriQuint emerged as a strong player in mobile handsets." So says Tim Dunn, vice-president of the handset business unit of the Hillsboro, OR, company. And it is hard to argue with the bare figures. Thanks to design wins in nearly 100 new phones last year, TriQuint Semiconductor sold 43 million transmit modules – up from only 5 million the year before.

Dunn believes that this strong momentum will continue, and expects to ship 100 million transmit modules in 2007. "This forecast is based on existing design-ins for phones just hitting their stride," said the upbeat VP at the 3GSM World Congress held in Barcelona, Spain, in mid-February.

Key design wins in what have turned out to be some of the world s most popular GSM phones, such as Motorola s RAZR platform, are the major reason behind the upturn. Dunn admits that luck has played some part in this success, saying, "Fortunately, we got into some hot phones last year." Samsung, the world s third-biggest mobile phone maker after Nokia and Motorola, was TriQuint s top customer in 2006. The chip manufacturer has benefited from the relationship, as a wide range of Samsung s dual-band and quad-band phones became big sellers.

Measured specifically in terms of the global consumption of GaAs components, Dunn now estimates that TriQuint has a market share of around 13%. Additionally, he is hopeful of seeing that proportion grow as the wireless communications business enters the broadband era.

The 3GSM event, which switched from Cannes to Barcelona in 2006 because it became far too large for the original French location to accommodate, is where critical trends in this industry tend to emerge. Dunn identified two key drivers that are now shaping the handset business: the need for increased RF semiconductor content to support communication in multiple frequency bands; and the emergence of ultra-low-cost phones.

With up to 11 frequency bands now requiring support, plus the increasing need for WiFi-enabled handsets, and the emergence of wideband-CDMA and EDGE as dominant protocols, demands on RF components are tougher than ever. As a result, Dunn says that TriQuint is now less concerned about the challenge to GaAs from all-CMOS power amplifiers (PAs) than it has been for many years.

At the other end of the mobile handset business, phone makers are aiming to sell their products for just $22 in developing countries. They need the front-end of the phone to represent no more than 5% of that sale price. Although this might suggest a route for all-CMOS PAs to get to market, Dunn believes that it will also become a profitable area for GaAs specialists.

Its recent success in the handset business means that TriQuint has had to cope with a rapid increase in wafer volumes through its 6 inch GaAs fabrication facility in Oregon. "Wafer volumes have more than doubled over the past two years," Dunn said. "Revenue per wafer is also going up, because of the value that our modules add."

Coupled with a rebounding market for GaAs-based driver components that are used in optical communications – of which Dunn estimates that TriQuint has a 60% market share – and strong demand for front-end modules in broadband and WiFi applications, the company s GaAs fabs are under pressure to deliver more volume than ever before.

As a result, TriQuint is adding capacity in some innovative ways. While 6 inch manufacturing has expanded at the high-volume Oregon fab, the company s 4 inch GaAs fab in Richardson, TX, is also being upgraded to the larger format. However, this is not quite as simple a move as it first appears. That is because TriQuint s surface and bulk acoustic wave (SAW/BAW) filter factory in Orlando, FL, is approaching its capacity limit.

Though not based on GaAs, the BAW fabrication process also employs a 6 inch wafer platform, and this will be switched to Richardson, where a multi-functional manufacturing line is being established to help free up extra capacity.

Innovation of a different kind is at the heart of TriQuint s latest PA technology, which is the company s response to industry demands for better front-end components to support phones with very small transceiver and baseband systems.

This has led TriQuint to develop its so-called CuFlip assembly technology, which replaces a number of wire bonds with a flip-chip process. One of TriQuint s key products – the TQM7M4006 quad-band PA module – is made using this process. The module includes two InGaP PA chips, as well as a passive output matching die (also GaAs) and a CMOS controller component. All of these die are flip-chip mounted onto a laminate substrate.

This is a bit of a divergence from the usual approach, because GaAs die assembly has typically required wire-bond interconnects that are also used to fine-tune RF circuits. While flipping the die and attaching to laminate has obvious advantages in reducing the wire-bond count and leading to better product yields, the process does require more design passes at the outset of product development, as it is much less of a "black art" than the traditional approach. As a result, TriQuint has greatly increased its use of simulation tools at the design stage.

Another popular theme among the estimated 55,000 3GSM delegates was WiMAX, the emerging broadband wireless technology that some – though by no means all – see as a rival to third-generation cellular services. With the mobile industry divided from the very top over the merits of WiMAX, this is a tricky area for PA component manufacturers. Nobody will be keen to invest heavily in a technology that may become a white elephant. "The jury is still out," said Dunn of WiMAX s likely success, although he does believe that WiMAX applications will find a niche at the very least. For example, the broadband system could be deployed rapidly and layered over an existing network, perhaps temporarily for disaster relief, or where ultra-high demand on cellular bandwidth is expected, such as the 2008 Summer Olympics in Beijing, China.

But the uncertainty within the industry over WiMAX is of no great concern to Dunn – for now, the volume business is in cellular transmit modules and PAs, and the former Intel man is confident that the progress made in 2006 will continue thanks to some new products and a new strategy for mobile communications – both of which were launched in Barcelona. Announcing this new handset strategy, Dunn highlighted the "customer-first", responsive approach that TriQuint is taking, rather than trying to second-guess which services and technologies will turn out to be the dominant ones of the future.

For some, says Dunn, the wireless future means high-speed handsets operating over wideband–CDMA–EDGE (or "WEDGE") networks. Others believe that the key trend will be basic communications services rolling out around the world. Others still regard widespread inclusion of short-range connectivity like Bluetooth and WiFi as the next logical step.

The TriQuint strategy is to meet any and all of these demands from a wide range of customers. New products launched for mobile handset applications are now grouped into a set of product "families", the names of which are physics-inspired, emphasizing the focus on the fundamental building blocks with which products such as mobile phones are built.

The "Hadron" family includes industry-standard PA modules; "Tritium" includes an integrated duplexer and filter for CDMA and wideband-CDMA applications; and the "Quantum" family consists of full transmit modules that combine a GaAs PHEMT switch, a low-pass filter and a control circuit with the PA function.

Curious though these product families may sound, it is the understanding of semiconductor device physics that underpins all these GaAs-based components. In case you were wondering, a hadron is defined as a subatomic particle that experiences the nuclear force. According to Wikipedia (your correspondent s grasp of particle physics is tenuous at the best of times), hadrons can disappear altogether under certain conditions. Thankfully, there appears to be no danger of that happening to the market for GaAs chips in mobile handsets for the foreseeable future.