New wavelengths expand cryptography's appeal

"Integrity of the data was the issue," Leonard Widmer, the company's vice-president of sales, told OLE. "It was vital that the data could not be modified or interfered with during the communication between the place where the ballots were counted and registered to where they were disclosed."

According to Widmer, the introduction of the technology did not change the way people voted. "People in Switzerland typically vote by post so the voters did not have a different experience – they still used paper ballots," he explained. "This is not e-voting over the internet. We simply secured the integrity of the data when it was transmitted."

id400

One of the benefits of this system is that it operates over an existing optical telecommunication fibre. The company simply used a free channel in a singlemode dark fibre to produce a secure point-to-point link between two buildings roughly 5 km apart. Each building housed one of id Quantique's Cerberis quantum key distribution (QKD) systems to encrypt and decode the data.

"Cerberis is essentially a hybrid system," explained Widmer. "It integrates a QKD apparatus and a classical encryptor running at 1 Gbit/s. The encryption key used by the encryptor is provided and secured by the QKD and can be changed up to four times per second. The entire system is fully automated and transparent from the user's point of view."

This successful demonstration has certainly motivated the firm to continue developing its QKD technology. "I would say that the QKD system is likely to be used again in future elections, especially in Geneva," commented Widmer.

Counting single photons at 1064 nm
QKD systems, such as Cerberis, require single-photon transmission and detection. While in Cerberis these functions are performed by a highly attenuated laser and an InGaAs single-photon detector both optimized for 1550 nm, moving to wavelengths such as 1064 nm presents a problem – the lack of available high-quality single-photon detectors. It is a challenge worth pursuing however, as this wavelength could open doors to free-space quantum cryptography.

"The best approach today at 1064 nm is to use a single-photon detector based on a silicon avalanche photodiode," Alexis Rochas, R&D manager of the firm's instrumentation business unit, told OLE. "There are two significant problems with this. Firstly, silicon's bandgap does not permit absorption above 1.1 µm. This means that at 1064 nm you typically have a single-photon detection probability of 2%. The second problem is the timing resolution, which is typically larger than 300 ps."

To overcome these problems, the company has teamed up with US firm Princeton Lightwave (PLI) and plans to launch a single-photon counting module optimized for 1064 nm at Photonics West. The product, the id400, will combine an avalanche photodiode with integrated biasing and quenching electronics.

As Rochas explains, this brings the expertise of both companies together into a single product. The end result is an InGaAsP/InP avalanche photodiode optimized for single-photon counting at 1064 nm. "The single-photon detection probability of the id400 will be 30% at 1064 nm," he commented. "We expect a timing resolution of less than 100 ps."

Fabrication

In simple terms, the id400 comprises an InGaAsP absorption layer and an InP multiplication layer. The first step is to absorb the photons to give a primary electron-hole pair. The next job is to multiply this carrier and create an avalanche, which is done in the InP layer.

Two other significant factors to consider are the dark current and afterpulsing, both of which degrade the performance of the device. The dark current is created by processes other than photoexcitation, such as thermal excitation of free carriers. Afterpulsing occurs when carriers created during an avalanche are trapped by defects in the multiplication layer, and freed later through processes such as thermal emission.

Development at PLI has centred on the InGaAsP absorber optimized for 1064 nm. "PLI has previously produced InGaAs/InP diodes for 1550 nm. They reduced the band-gap by introducing phosphide, which improved detection at 1064 nm," explained Rochas. "Their knowledge also lies in fabrication techniques that avoid defects and impurities."

PLI is now working to increase the diameter of its diodes. "PLI's diodes are typically 40 µm in diameter," said Rochas. "The problem is that when you increase the diameter, you have a larger impurity concentration, and in turn greater dark current and afterpulsing."

Circuit design
To operate in single-photon counting mode, a circuit must bias the diode to above its breakdown voltage (to start the avalanche) and then immediately stop the avalanche to limit the number of carriers flowing in the device and reduce the number of trapped carriers. This is particularly important for reducing afterpulsing.

"We have developed a fully integrated 1 mm² circuit that can be placed within 5 mm of the avalanche photodiode with a very limited connection in between," said Rochas. "The quenching time with standard electronics is often larger than 10 ns. With the id400, it is lower than 5 ns. This value is obtained thanks to our fast electronic circuit and the short connection length."

Single-photon counters must also be compatible with low operating temperatures in order to reduce the dark current. "The id400 has a three-stage thermoelectric cooler and the single-photon chip in a T08 package," said Rochas. "We cool the diode to –50 °C to limit the thermal generation and the dark current."

Users of the id400 will also be able to change the single-photon detection probability. The maximum probability will be 30%, although Rochas says that this could be reduced to 10% for customers wanting to limit the dark current.

It is worth noting that there is a significant trade-off between the single-photon detection efficiency and the dark current. The larger the bias above the breakdown voltage, the higher the detection probability and the better the timing resolution will become – but the dark current and afterpulsing will also increase.

Although reluctant to predict which applications will adopt the technology first, Rochas says that target applications for the id400 include free-space optical communications for space missions, satellite laser ranging, laser range finding for military applications and free-space quantum cryptography.

Another benefit of id Quantique's circuit is its ability to run in both a free-running and a gated mode. "The gate width can be set between 1 ns and 1 ms, and the gate frequency up to 20 MHz," said Rochas. "Free-running is useful when the user cannot predict the arrival of the photons. If the incoming photons arrive at a given frequency, gated mode is preferred."

Quantum key distribution systems
In quantum key distribution systems, each data bit is represented by a single photon which is assigned a variable such as polarization. In the example opposite, this allows Alice and Bob to communicate by choosing a particular value for the polarization of each bit, using values taken from one of two randomly chosen quantum mechanical "bases", represented by horizontal/vertical polarizations and 45°/–45° polarizations. In each basis, one polarization direction corresponds to a digital "1" and the other to a "0".

Alice, the sender, prepares a photon with the correct polarization for her first bit value (1 or 0). If Bob's randomly chosen basis matches Alice's, he measures the correct bit value. However, if he chooses the wrong basis, quantum mechanics dictates that there is a 50% chance that he will measure the intended bit value and a 50% chance that he will measure the wrong value. After receiving all of the information, Bob informs Alice via an insecure channel which basis he has used to measure each photon. Alice responds by telling Bob for which bits his basis choice was correct and these bits are used as the encryption key.

This method is 100% secure because it is impossible for an eavesdropper, known as Eve, to intercept the key. To try and break the code she has to randomly choose a polarization basis for each photon measurement, just like Bob, but even if she monitors Alice and Bob's public discussion concerning their choice of basis, her random basis choices differ from Bob's.

Eve's situation is further disadvantaged because she cannot intercept photons during the key distribution process without being detected. Even if she attempts to replace intercepted single photons, her incorrect polarization bases will mean that her replacement photons are not equivalent to Alice's. Eve's presence is exposed when Alice and Bob check the validity of their transmitted key through error-checking procedures, and the corrupted key is discarded.

• This article originally appeared in the January 2008 issue of Optics & Laser Europe magazine.