Fibre optics illuminate the path to success

Anette Hegner, IP manager for NKT Research and Innovation: "Patents are extremely important for start-up companies!"

Anders Bjarklev was one of the pioneers of a technological blockbuster – the optical fibre amplifier used to transfer high-speed data over the Internet. Unfortunately, the professor from the Technical University of Denmark never earned a cent from it. Why not? Because he never patented his early ideas. He learned a valuable lesson though, and in the late 1990s, when he and two of his PhD students came up with another bright idea concerning optical fibre technology, he made sure he patented it. Today, Mr Bjarklev and one of those former students, Jes Broeng, are at the helm of Crystal Fibre, a Danish company with 22 employees and customers in the US, Germany, UK and Japan.


Patents protecting a unique technology


Crystal Fibre supplies photonic crystal fibres (also known as microstructured or “holey” fibres), which are highperformance optical fibres used in lasers, telecommunications, defence, metal machining and entertainment. The company is a small niche player in a massive market dominated by giant companies such as the US firm Corning and the Japanese-owned company OFS Fitel. Crystal Fibre’s main advantages in this huge and competitive industry are its unique technology and the patents that protect it.The Technical University of Denmark, where Professor Bjarklev worked, helped pay for the first patent application. That patent not only protected the basic technology; it also formed the basis for founding Crystal Fibre and raising capital from NKT, a Danish industrial group.“ Patents are extremely important for start-up companies,” Anette Hegner, IP manager for NKT Research and Innovation, explains. One year after filing the first patent application, NKT made a full commitment to the start-up company, and NKT now owns around 90%. Professor Bjarklev, Dr Broeng and the Technical University of Denmark, which continues to be a partner in research and development, share the rest.

Important lessons in patent management

Crystal Fibre is more fortunate than most start-ups. Unlike other companies of its size, which tend to hope for the best when it comes to patent strategy or rely on the help of an external patent attorney, this little company has access to a team of lawyers and patent specialists from NKT, who also serve 16 other companies including 7 NKT subsidiaries. With the help of this team, the management of patents has played a key part in Crystal Fibre’s business strategy. In 2004, for example, Crystal Fibre bought the assets of BlazePhotonics, a UK competitor. Before doing so, Crystal Fibre opposed two of BlazePhotonics’ important European patents to put BlazePhotonics under pressure, since both companies were creating crucial IP in the same area. Dr Broeng explains that the UK competitor had filed patents on basic production processes that his company had also invented at an early stage but chosen to protect by keeping them secret rather than patenting them. If Crystal Fibre had not bought BlazePhotonics, it risked being prevented from using these processes, which are keys to production.Other important lessons included filing patents too broadly in markets it was not necessarily competing in. Such decisions are difficult early on when it is not clear how important a patent will become, much less which markets a company will ultimately focus on.“ In the beginning, we were in a race to protect everything we could,” says Dr Broeng. “Faced with the cost incurred by this early patenting-everything patent strategy, we learned to select our patent investment more from a business point of view than from a technological point of view.”

An effective filing procedure

Today, Crystal Fibre has 45 patent families in a variety of countries. Its filing procedure is generally well thought out. Normally, Crystal Fibre files the basic or the first application in Denmark and/or US. Only if the technology is truly promising, however, will the company make a broad PCT filing. In general, the minimum criterion is filing in the main market and the country of a main competitor.Although a subsidiary of a large company, budgets are still tight and the company exercises discipline when it comes to managing these patents, especially those that do not contribute to the bottom line. Patents are periodically eliminated. Non-core technologies, for example, may be licensed to other companies.
Only patents that can boost the bottom line are prosecuted. The guiding principle is that the potential market must cover the expense of filing and maintaining protection. When a patent would only protect incremental additions to a pre-existing patent, other means of protection are considered. The company aims to have many applications in a few key countries rather than a few applications in many countries. Ultimately, it is not about the number of patents, but about their impact. Small companies have little choice. Patenting everything can leave a company broke. “You’ll soon discover that your bank account is empty,” says Dr Broeng.

Scanning files for new ideas

Dr Broeng spends a lot of time scanning the Delphion patent database, searching by keyword and industry. This exercise helps him keep an eye out for filings by competitors, aggressive moves that might infringe upon Crystal Fibre’s business. Every now and then such a search also inspires a new idea. For example, Dr Broeng once spotted a US filing by a competitor for a new technological procedure related to high-powered fibres for lasers. It was a market in which Crystal Fibre saw great potential. After further research, the company filed an improvement to the patent and then, with the leverage of the further patent, acquired a licence for the US company’s underlying technology. Today, high-power laser fibre is Crystal Fibre’s primary market and only it and its licensing partner have the technology.

Individual industry business centres would be useful

Dr Broeng says that small companies are hampered by patent attorneys’ sometimes limited knowledge of specific industries. Building a patent strategy early on that matches a company’s business goals is difficult. He suggests that business centres for individual industries would be useful. Over time, they could provide the kind of expertise small companies need. He also says that the high translation costs paid by small companies to obtain patents are a serious burden. Finally, large companies filing and obtaining broad, unclear patents concern him. Large firms can use such patents to dissuade smaller companies (that cannot afford to litigate the patent) from entering a market.


Crystal Fibre’s main advantages are its unique technology and he patents that protect it.

Product facts

Crystal Fibre designs, manufactures and markets photonic crystal fibres (PCFs). Photonic crystal fibres are a new range of optical fibers offering significant new possibilities and functionality within a broad range of applications. Crystal Fibre launched its first product in the summer of 2000 and has since then expanded its product range considerably.
The company is today the biggest commercial supplier of photonic crystal fibre solutions and offers a diversity of specialty fibers within nonlinear fibers, large mode area fibres as well as high NA and UV optimized fibers

Optical Fibres for Extremely Fast Communication

Optical fibers can be used to transmit light and thus information over long distances. Fiber-based systems have largely replaced radio transmitter systems for long-haul optical data transmission. They are widely used for telephony, but also for Internet traffic, long high-speed local area networks (LANs), cable TV, and increasingly also for shorter distances. In most cases, silica fibers are used, except for very short distances, where plastic optical fibers can be advantageous.
Compared with systems based on electrical cables, the approach of optical fiber communications (lightwave communications) has advantages, the most important of which are:
The capacity of fibers for data transmission is huge: a single silica fiber can carry hundreds of thousands of telephone channels, utilizing only a small part of the theoretical capacity. In the last 30 years, the progress concerning transmission capacities of fiber links has been significantly faster than e.g. the progress in the speed or storage capacity of computers.
The losses for light propagating in fibers are amazingly small: ∼ 0.2 dB/km for modern single-mode silica fibers, so that many tens of kilometers can be bridged without amplifying the signals.
A large number of channels can be reamplified in a single fiber amplifier, if required for very large transmission distances.
Due to the huge transmission rate achievable, the cost per transported bit can be extremely low.
Compared with electrical cables, fiber-optic cables are very lightweight, so that the cost of laying a fiber-optic cable can be lower.
Fiber-optic cables are immune to problems that arise with electrical cables, such as ground loops or electromagnetic interference (EMI).
However, fiber systems are more sophisticated to install and operate, so that they tend to be less economical if their full transmission capacity is not required. Therefore, the “last mile” (the connection to the homes and offices) and is usually still bridged with electrical cables, whereas fiber-based communications do the bulk of the long-haul transmission. Gradually, however, fiber communications are being used within metropolitan areas (metro fiber links), and even fiber to the home (FTTH) is being developed – particularly in Japan, where private Internet users can already obtain affordable Internet connections with data rates of 100 Mbit/s – well above the performance of current ADSL systems, which use electrical telephone lines.
Telecom Windows
Optical fiber communications typically operate in a wavelength region corresponding to one of the following “telecom windows”:
The first window at 800–900 nm was originally used. GaAs/AlGaAs-based laser diodes and light-emitting diodes (LEDs) served as transmitters, and silicon photodiodes were suitable for the receivers. However, the fiber losses are relatively high in this region, and fiber amplifiers are not well developed for this spectral region. Therefore, the first telecom window is suitable only for short-distance transmission.
The second telecom window utilizes wavelengths around 1.3 μm, where the loss of silica fibers is much lower and the fibers' chromatic dispersion is very weak, so that dispersive broadening is minimized. This window was originally used for long-haul transmission. However, fiber amplifiers for 1.3 μm (based on, e.g. on praseodymium-doped glass) are not as good as their 1.5-μm counterparts based on erbium. Also, low dispersion is not necessarily ideal for long-haul transmission, as it can increase the effect of optical nonlinearities.
The third telecom window, which is now very widely used, utilizes wavelengths around 1.5 μm. The losses of silica fibers are lowest in this region, and erbium-doped fiber amplifiers are available which offer very high performance. Fiber dispersion is usually anomalous but can be tailored with great flexibility (→ dispersion-shifted fibers).
The second and third telecom windows are further subdivided into the following wavelength bands:

Band Description Wavelength range
O band original 1260–1360 nm
E band extended 1360–1460 nm
S band short wavelengths 1460–1530 nm
C band conventional (“erbium window”) 1530–1565 nm
L band long wavelengths 1565–1625 nm
U band ultralong wavelengths 1625–1675 nm

The second and third telecom windows were originally separated by a pronounced loss peak around 1.4 μm, but they can effectively be joined with advanced fibers with low OH content which do not exhibit this peak.
System Design
The simplest type of fiber-optic communication system is a fiber-optic link providing a point-to-point connection with a single data channel. Such a link essentially contains a transmitter for sending the information optically, a transmission fiber for transmitting the light over some distance, and a receiver. The transmission fiber may be equipped with additional components such as fiber amplifiers for regenerating the optical power or dispersion compensators for counteracting the effects of chromatic dispersion. The article on fiber-optic links gives more details.
A typical channel capacity for long-haul transmission is nowadays 2.5 or 10 Gbit/s; 40, 100 or even 160 Gbit/s may be used in the future. More advanced systems increase the transmission capacity by simultaneously using several or even many different wavelength channels (coarse or dense wavelength division multiplexing). The main challenges are to suppress channel cross-talk via nonlinearities, to balance the channel powers (e.g. with gain-flattened fiber amplifiers), and to simplify the systems. Another approach is time division multiplexing, where several input channels are combined by nesting in the time domain, and solitons are often used to ensure that the sent ultrashort pulses stay cleanly separated even at small pulse-to-pulse spacings.
Another important development is that of systems which link many different stations with a sophisticated fiber-optic network. This approach can be very flexible and powerful, but also raises a number of non-trivial technical issues, such as the need for adding or dropping wavelength channels, ideally in a fully reconfigurable manner, or to constantly readjust the connection topology so as to obtain optimum performance, or to properly handle faults so as to minimize their impact on the overall system performance. As many different concepts (e.g. concerning topologies, modulation formats, dispersion management, nonlinear management, and software) and new types of devices (senders, receivers, fibers, fiber components, electronic circuits) are constantly being developed, it is not clear so far which kind of system will dominate the future of optical fiber communications.
For a discussion of aspects such as bit error rates and power penalties, see the article on optical data transmission.
Transmission Capacity of Optical Fibers
Within the last 30 years, the transmission capacity of optical fibers has been increased enormously. The rise in available transmission bandwidth per fiber is even significantly faster than e.g. the increase in storage capacity of electronic memory chips, or in the increase in computation power of microprocessors.
The transmission capacity of a fiber depends on the fiber length. The longer a fiber is, the more detrimental certain effects such intermodal or chromatic dispersion are, and the lower is the achievable transmission rate.
For short distances of a few hundred meters or less (e.g. within storage area networks), it is often more convenient to utilize multimode fibers, as these are cheaper to install (for example, due to their large core areas, they are easier to splice). Depending on the transmitter technology and fiber length, they achieve data rates between a few hundred Mbit/s and ∼ 10 Gbit/s.
Single-mode fibers are typically used for longer distances of a few kilometers or more. Current commercial telecom systems typically transmit 2.5 or 10 Gbit/s per data channel over distances of ten kilometers or more. Future systems may use higher data rates per channel of 40 or even 160 Gbit/s, but currently the required total capacity is usually obtained by transmitting many channels with slightly different wavelengths through fibers; this is called wavelength division multiplexing (WDM). Total data rates can be several terabits per second, sufficient for transmitting many millions of telephone channels simultaneously. Even this capacity does not reach by far the physical limit of an optical fiber. In addition, note that a fiber-optic cable can contain multiple fibers.
In conclusion, there should be no concern that technical limitations to fiber-optic data transmission could become severe in the foreseeable future. On the contrary, the fact that data transmission capacities can evolve faster than e.g. data storage and computational power, has inspired some people to predict that any transmission limitations will soon become obsolete, and large computation and storage facilities within high-capacity data networks will be extensively used, in a similar way as it has become common to use electrical power from many power stations within a large power grid. Such developments may be more severely limited by software and security issues than by the limitations of data transmission.
Key Components for Optical Fiber Communications
Optical fiber communication systems rely on a number of key components:
optical transmitters, based mostly on semiconductor lasers (often VCSELs), fiber lasers, and optical modulators
optical receivers, mostly based on photodiodes (often avalanche photodiodes)
optical fibers with optimized properties concerning losses, guiding properties, dispersion, and nonlinearities
dispersion-compensating modules
semiconductor and fiber amplifiers (mostly erbium-doped fiber amplifiers, sometimes Raman amplifiers) for maintaining sufficient signal powers over long lengths of fibers, or as preamplifiers before signal detection
optical filters (e.g. based on fiber Bragg gratings) and couplers
optical switches and multiplexers (e.g. based on arrayed waveguide gratings); for example, optical add/drop multiplexers (OADMs) allow wavelength channels to be added or dropped in a WDM system
electrically controlled optical switches
devices for signal regeneration (electronic or optical regenerators), clock recovery and the like
various kinds of electronics e.g. for signal processing and monitoring
computers and software to control the system operation
In many cases, optical and electronic components for fiber communications are combined on photonic integrated circuits. Further progress in this technological area will help optical fiber communications to be extended to private households (→ fiber to the home) and small offices.