Special Report: R&D '04

Corporate spending on research and development remains the driving force behind innovation. Indeed, the five corporations with the largest R&D budgets alone spent $33.6 billion last year, more than the U.S. government spent on R&D conducted by federal agencies. But a troubling trend that began in 2001 continues: corporate R&D spending is on the decline. Last year’s decrease of .6 percent is slight, but it follows sharper declines in 2001 and 2002, after more than a half-decade of robust growth. Perhaps most worrisome, the declines were not limited to a few obviously troubled sectors, such as telecommunications, but affected a cross section of industries and included some of the world’s top spenders on R&D. In fact, three of the five largest corporate R&D spenders showed significant decreases in their 2003 budgets: the top spender, Ford Motor, cut its budget by $200 million, and Siemens, a long-time powerhouse in research and development, decreased spending by $900 million.

The picture, however, is far from being universally bleak. Technology Review’s innovation index, which is calculated from four key measures, shows healthy – and growing – spending by leading pharmaceutical, biotech, and computer companies. Familiar names like Pfizer, Amgen, and Nokia lead the list. But the innovation index also contains some surprises. High on the list were BMC Software and Swiss biotech firm Serono. And Volkswagen and Nissan Motor ranked, unlike most of their competitors in the auto industry, right up with Merck and Intel.

Predictably, pharmaceutical and electronics companies feature prominently among the 150 top spenders, with some 28 pharma and medical-device companies and 19 electronics firms showing up. But the continued consolidation of research within some sectors is also notable. For example, Intel spent $4.36 billion last year, while its nearest rival in the semiconductor business had an R&D budget of only $1.72 billion. Even in the technologically competitive biopharmaceutical industry, one company, Pfizer, dominates, with a $7.13 billion budget; the second-ranked drug corporation spends nearly $2.2 billion less.

Of course, R&D budgets only partially reflect the state of innovation. In the following pages, Technology Review also profiles three corporate research projects; each tells a very different story about the challenges of commercializing radically new technologies. As you read in the profiles, business, legal, and financial factors all play key roles in determining how – and whether – new technologies move out of the lab. Together these profiles present a peek into the workings of today’s corporate research labs. -David Rotman

• Philips: Polymers to Pixels
• Nokia: Peer-to-Peer Phones
• Texas Instruments: Single-Electron Transistors
• The Top 15: Companies Spending the Most on R&D
• The Innovation Index
• The Business of Innovation

Polymers to Pixels
Company: Philips Electronics
Benefit: Cheap, high-quality flat-screen TVs

By Jessica Gorman

Electronics superstores like Best Buy are bustling right now, as holiday shoppers search for the perfect gift. Some of the big-ticket items are the most mesmerizing – row after row of 50-inch flat-screen TVs costing thousands of dollars. But in a few years, if ongoing research at Philips Electronics succeeds, stores will offer a new, cheaper option: a flat-screen TV that brings images to life with a layer of light-emitting diodes that use novel organic molecules.

Philips, one of the world’s largest makers of flat-screen plasma and liquid-crystal displays (LCDs), has spent more than a decade at its labs in Eindhoven, the Netherlands, working to perfect the polymer-based screens. The technology, says Nijs van der Vaart, the leader of the project at Eindhoven, “has so many advantages.” For one thing, black is “very black,” while LCD screens allow light to leak through blacked-out pixels. And unlike LCDs, the new technology allows viewing from any angle; it also eliminates the shadows that follow fast-moving objects like soccer balls. But the advantage that consumers might find most eye-catching is a price that’s potentially lower than either plasma or LCD televisions’; in theory, at least, manufacturing displays out of light-emitting plastics will be far cheaper.

Philips’s new type of display relies on a fundamental breakthrough in materials science. Normally, polymers don’t emit light, but in 1989 physicists at the University of Cambridge developed a new type of plastic that shone brightly when sandwiched between electrodes. Philips started research on the new technology soon after. For a display manufacturer like Philips, the implications were obvious: selectively addressing small areas of the polymer layer with electricity would get those areas to selectively emit light. In other words, you could make pixels.

Philips is not without competition in trying to come up with a better flat-screen display, but it hopes to gain an edge by leveraging its expertise in chemistry and materials science. Van der Vaart and his research colleagues are looking for further ways to improve the light-emitting properties of the polymers. They have also built a novel ink-jet printer that uses separate print heads to deposit the polymers as red, green, and blue pixels. Philips believes this printing system is potentially a cheap and versatile way to make the supersize television screens.

One problem that Philips’s researchers continue to wrestle with is the light-emitting lifetime of the polymers. In particular, Philips’s blue polymer fades after far less use than the roughly 20,000 hours a television needs to be able to endure. Another challenging problem is that some parts of the polymer-LED screen fade before others. And to make a 30-inch TV screen, Philips’s researchers must improve the display’s efficiency, so that smaller currents will elicit the needed light from the polymers.

But after more than a decade of work, the research is beginning to yield results. In 2002, the company introduced the technology in a display on an electric shaver. In 2004, Philips released a cell phone with a small polymer-LED display. Larger, higher-resolution displays will probably appear in cell phones around 2005, says van der Vaart, who is optimistic that a 30-inch or larger TV could reach stores by 2008.

In the meantime, holiday shoppers who want cheap flat-screen televisions based on brightly glowing polymers will have to wait. Perhaps an electric shaver would make a nice gift instead; you could even get one with a polymer display.

Peer-to-Peer Phones
Company: Nokia
Benefit: Image sharing on cell phones

By Patric Hadenius

Peer-to-peer computing has become an enormously popular way to share digital data, enabling, among other applications, music downloads via the now defunct sharing service Napster. Now Nokia, the Finnish telecom giant, is working to bring it to multimedia mobile phones.

Why is the world’s largest maker of cell phones suddenly interested in the technology that made Napster a household name and an industry villain? The sale of more and more smart phones, camera phones, and game phones is driving a constant search for new applications to support them. Indeed, the killer application for all those camera phones has yet to be found – but peer-to-peer could help. With file-swapping technology, you and your friends could easily share photos you take with your phones. Or when you have a colleague on the phone, you could share a document and even edit it at the same time.

At least, that’s what Nokia is betting on. With one-third of the mobile-phone market, $36.2 billion in net sales, and $6.3 billion in profit last year, Nokia is indisputably the world’s leader in mobile communication. But it’s an increasingly competitive business. So the pressure is on Nokia’s research departments to devise cell-phone improvements that will distinguish their products from competitors’.

To help reinvigorate the company’s technology, Jukka Nurminen at Nokia’s research center in Helsinki, Finland, recruited Balázs Bakos at its site in Budapest to start investigating the possibilities of moving peer-to-peer from the wired world to the mobile world. Any peer-to-peer system sends information between computers with a minimum of hierarchy, using few, if any, servers and databases. So at first glance, the technology might seem a natural for the mobile world: isn’t talking on the phone already a kind of peer-to-peer networking? In fact, there are plenty of servers and directories needed to connect two phones and keep a line open between them. And mobile phones have other limitations that typical wired devices don’t, including far less processing power and memory, a short battery life, and limited bandwidth.

One of the first questions Nurminen and his Hungarian partner addressed was whether a typical mobile network could support a widely used file-sharing protocol called Gnutella. The experiment proved a dead end. “I think of it as an important finding,” argues Nurminen, “but we saw that it didn’t scale above 10,000 phones.” The problem was that peer-to-peer applications use a lot of bandwidth as they hunt for information. And the demand for bandwidth multiplies rapidly as more computers join the network. On the Internet this is less of a problem, since there is plenty of bandwidth, and most service providers charge their users flat rates, regardless of the number of bits they send. But on the mobile networks, where bandwidth is limited, and the pricing is per connection and per bit sent, users need a protocol that skyrocketing traffic won’t overwhelm.

So Nurminen and Bakos focused on reducing bandwidth requirements. The first step was to restrict search traffic by dividing the whole network into smaller clusters. Each phone in the network keeps a list of the images and other files stored within its cluster and can respond to queries from outside on behalf of the whole cluster. In a simulated mobile network, this approach proved ideal, enabling fast searching without sacrificing network resilience.

Having solved a key technical challenge, the researchers took their work to the company’s business units. But here the project ran into a hitch: concerns about digital rights management. With the fate of Napster still fresh in everyone’s mind, the business side didn’t want to start promoting technology that could facilitate the exchange of copyright-protected material. Erich Hugo, Nokia’s technology marketing manager, says, “The technology is still in development.”

Maybe so, but if the Napster phenomenon is any indication, once potential users understand the possibilities of peer-to-peer cell phones, it might be next to impossible to go back. Peer-to-peer technology, after all, has always distinguished itself by a very strong reluctance to be controlled.

Single-Electron Transistors
Company: Texas Instruments
Benefit: Ultrasmall integrated circuits
By Peter Fairley

CEOs at technology firms like to boast that the intuitions of individual researchers and engineers are their companies’ greatest assets. A string of recent patent filings authored by Texas Instruments electronics researcher Christoph Wasshuber shows that in some cases, at least, there’s truth to that claim. By giving the 36-year-old Austrian-born engineer the flexibility to follow his instincts in designing a novel type of single-electron transistor, Texas Instruments has secured a toehold in the development of a technology that could transform semiconductor microchips in the decades to come.

In many ways, a single-electron transistor, which is turned on and off by the addition or subtraction of a lone electron, is the ultimate in semiconductor miniaturization. Not only could it allow the manufacture of powerful, ultrasmall electronic devices, but it could also slash power consumption. While exotic versions of these highly sensitive electronic switches have been around since the late 1980s, research on them has stalled because of severe problems in making them robust enough. The same property that makes them attractive, their ultrasensitivity, also makes it difficult to get them to work effectively in the real world. In particular, single-electron transistors are easily overwhelmed by background noise or signals from neighboring circuits. But Wasshuber and his collaborators at the Swiss Federal Institute of Technology in Lausanne have designed a single-electron transistor that, incorporated into standard silicon circuitry, is immune to interference.

If the innovative design works, suggests Wasshuber, it could result in ultrafast single-electron processors. What’s more, Wasshuber’s transistors should be compatible with standard semiconductor fabrication processes, enabling manufacturers to push beyond conventional microchip technology without abandoning their multibillion-dollar investments in production capacity. The first uses of the single-electron transistors will likely be in memory chips and ultrasensitive electrometers for testing electrical circuits. Konstantin Likharev, a physicist at New York’s Stony Brook University, estimates that a memory chip with single-electron transistors could store a terabit of data in a square centimeter of silicon, a data density about 100 times greater than that of today’s best memory.

But even more important to the future of microelectronics, single-electron transistors could solve one of the gravest problems facing conventional chip technology; as more and more transistors are packed together, heat becomes harder to dissipate. Hundreds of thousands of electrons flow through a conventional transistor, and as a result, switching it on and off usually takes at least one volt. Over the next decade, chips will be jammed with billions of transistors, and the power required to switch them could literally cook the circuits. In contrast, a single-electron transistor, turned on or off by just one electron, runs cool and consumes one-tenth as much power. If you look ahead to the end of the industry’s technology road map, we’re going to have 30 to 50 billion transistors on a chip, says Dennis Buss, Texas Instruments’ vice president of silicon technology development. The thought of operating those at one volt is unthinkable.

Texas Instruments hired Wasshuber in 1998 for his computer modeling expertise and set him to work on a series of near-term projects. But at night the young physicist cranked out designs of single-electron transistors like those he had worked on in graduate school. When a Texas Instruments task force scouting future technologies vital to the semiconductor industry selected single-electron transistors as one idea warranting a closer look, Wasshuber was in luck. Finally, he had the green light to pursue his hobby in the daylight hours at the lab.

Despite single-electron transistors’ broad implications, however, experts who have been working on them for more than a decade caution against overenthusiasm. Likharev calls Wasshuber’s ideas clever but wants to see proof that the new designs will work in real circuits. Then there’s the challenge of manufacturing actual devices based on the designs. A single-electron transistor that operates at room temperature will require features as small as one to two nanometers across. That’s the size of molecules, notes Greg Snider, an electrical engineer at the University of Notre Dame and an expert in single-electron transistors. The semiconductor industry is quite a ways away from doing that controllably.

Wasshuber agrees that plenty of work remains before devices using single-electron transistors show up in cell phones and desktops. And a large-scale research and development effort on the new chip technology is far more than Texas Instruments can justify funding today, given its expectation that it can continue to miniaturize conventional transistors through 2015. For the moment, Texas Instruments is salting away its patents while keeping a close  eye on the emerging field. But whether Wasshuber’s design for single-electron transistors proves practical or not, the company’s opportunity to explore the future of microelectronics is worth the investment in turning his after-work hobby into part of his day job.

The Top 15: Automotive, computer, and pharmaceutical companies dominate the list of the companies with the highest R&D expenditures.
Company	Country	R&D 2003 ($Mil)	R&D Percent change	Absolute change in R&D ($Mil)
FORD MOTOR	United States	$7,500	-3%	($200)
PFIZER	United States	$7,131	38%	$1,955
DAIMLERCHRYSLER	Germany	$6,689	-8%	($600)
TOYOTA MOTOR	Japan	$6,210	2%	$97
SIEMENS	Germany	$6,084	-13%	($903)
GENERAL MOTORS	United States	$5,700	-2%	($100)
MATSUSHITA ELECTRIC	Japan	$5,272	5%	$257
IBM	United States	$5,068	7%	$318
GLAXOSMITHKLINE	United Kingdom	$4,910	-4%	($192)
JOHNSON and JOHNSON	United States	$4,684	18%	$727
SONY	Japan	$4,683	16%	$649
MICROSOFT	United States	$4,659	8%	$352
NOKIA	Finland	$4,514	23%	$850
INTEL	United States	$4,360	8%	$326
VOLKSWAGEN	Germany	$4,233	22%	$762

The Innovation Index, as calculated by Technology Review, takes into account not only the size of a company’s R&D budget but also its increase in spending and the amount spent in relation to sales. Nonetheless, the second-largest R&D spender, Pfizer, comes out at the head of the pack.
Rank by Innovation Index	Company	Innovation Index	R&D 2003 ($Mil)	R&D percent change	Absolute change in R&D spending ($Mil)	R&D as a percent of sales
1	PFIZER	$178	$7,131	38%	$1,955	16%
2	AMGEN	$149	$1,655	48%	$539	20%
3	NOKIA	$146	$4,514	23%	$850	13%
4	JOHNSON and JOHNSON	$141	$4,684	18%	$727	11%
5	BMC SOFTWARE	$138	$586	20%	$96	41%
6	VOLKSWAGEN	$138	$4,233	22%	$762	4%
7	SONY	$136	$4,683	16%	$649	7%
8	MERCK (U.S.)	$135	$3,178	19%	$501	14%
9	SERONO	$134	$468	31%	$110	25%
10	ASTRAZENECA	$134	$3,451	12%	$382	18%
11	MICROSOFT	$133	$4,659	8%	$352	14%
12	ROCHE	$133	$3,694	12%	$396	15%
13	NOVARTIS	$133	$3,756	12%	$394	15%
14	INTEL	$132	$4,360	8%	$326	14%
15	NISSAN MOTOR	$129	$3,225	18%	$491	5%
The Innovation Index is calculated by indexing each of the four R&D measures shown to the highest-performing company and then creating a combined, unweighted average. Source: Standard and Poor’s, Technology Review

The Business of Innovation: When factors like R&D growth are taken into account, the biotech, pharmaceutical, and technology sectors rank highly. But when it comes to pure budget size, industrial conglomerates and transportation companies are tops.
Industry sector	Average rank by Innovation Index	Number of companies in the top 2003 R&D spenders	Average R&D 2003 ($Mil)	Average R&D percent change	Average absolute change in R&D spending ($Mil)	Average R&D as a percent of sales
Biotechnology	22	$3	$917	28%	$225	21%
Computer software	41	$7	$1,341	11%	$111	18%
Pharmaceuticals/ medical devices	48	$28	$2,045	9%	$176	14%
Semiconductors	50	$10	$1,318	1%	$41	22%
Transportation	72	$26	$2,273	5%	$72	5%
Computer hardware	81	$9	$2,251	0%	$4	7%
Heavy machinery	84	$5	$663	9%	$58	5%
Electronics/ electrical	90	$19	$1,616	0%	$11	6%
Aerospace and defense	93	$8	$1,369	1%	$29	6%
Telecom- munications	100	$13	$2,156	-10%	($269)	11%
Chemicals	104	$10	$1,083	-1%	($19)	6%
Consumer products	108	$5	$1,043	2%	$6	2%
Industrial conglomerates	109	$4	$2,490	-3%	($241)	4%
Energy	112	$3	$586	2%	$2	2%
Source: Standard & Poor’s, Technology Review