Supports commercial testing and assembly of silicon photonics
Release time:2020-04-10   Browse:

Silicon photon (SiP) is the world's surging demand for data that promises new advantages in bandwidth, scalability and energy efficiency.The shift in data transmission from electrons to photons (and the computing process to come) is comparable to the development of integrated circuits 50 years ago: a fundamental shift that marks a watershed in the history of technology.However, many challenges still need to be overcome to achieve practical, profitable manufacturing and testing of silicon photonic devices.The growing need - combined with the fundamental need for nanoscale precise mutual orientation of tiny photonic elements - requires powerful, rapid automation solutions.

In silicon photonics, optical components are made together with electronic microcircuits onto conventional silicon wafers.These components can include complex lasers, complex waveguide structures, detectors, modulators, delay and multiplexing/demultiplexing structures, and many other integrated photon components.The chips are then integrated into a package that can include other chips and active components, lenses and other tiny optical devices, as well as optical fibers, fiber arrays and electrical I/O.

Cascade Microtech is the first to enable the CM300xi photonic engineering wafer probe platform integrated with PI's fast multi-channel photon alignment system for high-throughput, wafer safe, nanoscale optical detection of silicon photonic devices on wafers.Image courtesy of FormFactor company Cascade Microtech.

To do this, and to test the chip to ensure its functionality before the packaging process even begins, the photonic elements must be precisely aligned.Both testing and encapsulation require coupling of optical inputs and outputs to sources and detectors, and all elements must be tested and aligned to each other across the optical path to ensure effective coupling.For today's SiP devices, horizontal alignment tolerances of less than 50 nm are increasingly common.For devices with multichannel photon inputs or outputs, additional precision optimizations around Theta-Z are required to effectively couple all inputs and outputs of the array.In general, the Theta-X and Theta-Y directions also need to be optimized.

Challenges and solutions

These specialized, highly precise positioning tasks, often with multiple degrees of freedom, are "repeated many times from chip to final package," meaning that engineers face vexing geometry.For example, if the rotation center position is not ideal (as it always is), adjusting the Angle will translate the coupling in the X and Y directions.In addition, the increasingly common short SiP waveguide shows a steering effect, so the optimization on the input side will generate a deviation on the output side.This means that the overall best alignment becomes a "moving target".Therefore, in the past, an iterative approach has been necessary to solve this problem and reach an acceptable consensus.Most importantly, the process is time-consuming.

PI overcomes these challenges and integrates new firmware based algorithms into its industrial optical fiber alignment subsystem to achieve fast multi-channel/multi-freedom photonic alignment.These automation subsystems are designed for integration into production tools from wafer detection to chip testing to final packaging, and can perform multiple linear and angular digital gradient search optimizations simultaneously.

The resulting parallelism eliminates the previously required iterative approach, thereby increasing process throughput by more than two orders of magnitude.Clearly, the implications for the economics of production and competitiveness are profound.Furthermore, the inherent parallelism of this closed-loop all-digital technique means that the total alignment time in multi-channel/multi-degree of freedom applications is only weakly related to the number of individual allocations performed.For example, in the case of a wafer probe operation, waveguide I/O coupling can typically be optimized in less than 500 milliseconds, regardless of the number of inputs or outputs to the SiP chip.In the subsequent chip testing and packaging process, the same impressive throughput was achieved.

Designed for a variety of applications

Because different devices and production applications have different requirements, there are several fast positioning system variants that can be used for alignment of single channel or array devices with or without Angle optimization requirements.All systems are based on very strict Settings.If angular optimizations are not required, the most popular configurations start with three phases of precise linear motion of the stack.

Triaxial alignment system for applications that do not require Angle optimization.(credit: PI)

These provide a 25mm range for rough positioning and initial exploration.The fast NanoCube locator is installed at the end of this long-travel platform stack to provide fast surface scanning motion for pattern positioning and fast transverse and z-gradient search capabilities to dynamically compensate for drift effects, all with nanoscally resolution.Flexible guides and all-ceramic insulated piezoelectric actuators ensure long life in all-weather industrial deployments.Position sensors on all drives provide microsecond position determinations, thereby enabling soft confinement that protects mechanical and expensive devices, such as fully patterned wafers.

If Angle optimization is required, such as when aligning a photon array device, a parallel-motion hexapod can be used to provide long stroke, six-degree-of-freedom positioning and initial light seek, as well as rapid automatic Angle optimization.These hexapods provide freely defined coordinate systems and pivot points that can be rotated around optical optima such as the waist of the beam and the center of the physical channel.The NanoCube locator was again deployed to achieve fine horizontal and z-axis alignment, as well as dynamic compensation for drift effects and residual geometric errors in parallel with the Angle optimization process.

The hexapod based system has six degrees of freedom for fast and accurate alignment of optical fibers, optical fiber arrays and optical components.(credit: PI)

The combination of hexapod and Nanocube laid the foundation for the pioneering parallel alignment capability of the PI FMPA system.For example, to align a linear or two-dimensional fiber array, the NanoCube performs a horizontal gradient search by tracking to keep the first channel of the array aligned, while the hexapode performs a parallel theta-Z gradient search on the NTH channel of the fiber.The array.The freely defined center of rotation of the hexapod has ensured that the optical axis of the first array of elements can be located near the optical axis of the first channel.Then, any small residual geometric errors can be compensated by tracing the NanoCube.The parallelism of overall alignment means that the entire process is 10-100 times faster than the previous iteration approach.

Consider the economics of production

These systems are available in standard configuration for one - or two-sided alignment tasks with or without Angle optimization.In addition, the modular architecture allows for additional alignment of robots to meet virtually any application requirements.The supply range includes high-performance digital controllers (e-712), firmware routines with algorithms for rapid positioning, and a wide range of software packages covering all aspects of the application, from easy to start to easy to control.The system USES a graphical user interface to integrate it quickly and managably into external programs.

For systems requiring Angle optimization, the hexapod controller (c-887) performs the same scanning and alignment algorithm.These features include built-in sinusoidal and helical scanning, which can be automatically modeled to achieve accurate peak positioning even without fast sparse scanning.Automatic centroid determination in top hat and other challenging coupling;And the unique parallel gradient search process, can be carried out across the input single step optimization, output and degree of freedom.

Together, these novel subsystems help ensure the cost-effective testing and manufacturing of today's and future silicon photonic devices.

0

Prev:Silicon photonics technology was developed by using wafer level test stations
Next:最后一页