The flourishing semiconductor microelectronics equipment industry can lend its practices to the semiconductor laser industry to engender cost savings and efficiencies. The fast-growing but still relatively small semiconductor laser market relies on a lot of the same tools as the microelectronics equipment industry, such as PVD equipment (physical vapor deposition), but at a smaller scale. In 2017 for instance, the laser market investment in these tools stood at $200 million against a $3 billion spending in the larger industry.
Semiconductor lasers are more efficient than conventional lasers, requiring less power and being smaller in size. They are used in fiber optics, as signal lasers and pumps for fiber lasers in welding and cutting applications of metal. They also see applications in optical amplifiers in telecommunications. The market makes up nearly half of the $11 billion laser market and is expected to reach $8.8 billion by 2022.
The growth strategies of the two semiconductor industries are different. Production technologies in microelectronics rely on a widely adopted roadmap. On the other hand, players in the semiconductor laser industry keep process information close to the chest.
Despite the differences, semiconductor laser process flow can benefit from adopting relevant practices used in the microelectronics industry. Many leading semiconductor laser vendors are adopting these practices to great benefits to the factory cost of ownership and process equipment. These vendors tend to be medium-sized, able to offer customers need-based platform customization along with process and equipment support, control systems and proprietary platforms.
Thin films process flow in semiconductor laser
The quality and design of the passive thin films that coat the active layer stack are essential in determining the efficiency of the internal diode and the performance of the device it will power.
Thin films include metallization layers, coatings for reflectivity, dielectric coatings etc.
Various PVD techniques are available for passive layer thin-film fabrication.
- Thermal resistive evaporation: This is the most affordable process with good directionality, at temperatures of 50 to 100 degrees C. The process has poor uniformity, though excellent with planetary and masks. It is used typically for metals with a low melting points. The deposition rate is typically less than 50 A/s and the film quality is poor, though it can be improved with ion assist and moderate stress. Scalability is limited at a reduced deposition rate and utilization.
- E-beam evaporation: This process is of moderate complexity and cost, with good directionality, also at temperatures of 50 to 100 degrees C. The process has low impurities and poor uniformity, though planetary and masks can vastly improve uniformity. It can be used for metals and dielectrics. The deposition rate is typically less than 100 A/s and the film quality is poor, though it can be improved with ion assist and moderate stress. Scalability is limited at reduced deposition rate and utilization.
- Magnetron sputtering: This costly process of moderate complexity is highly scalable with automation. It can be applied to metals and dielectrics. It produces very good quality film withstanding moderate to high stress. Impurities are low, and uniformity is a good improvement over the earlier processes. The deposition rate for dielectrics lies between 1 and 10 A/s. The process is carried out at 200 degrees C and requires cooling. Directionality is low and can be improved with system geometry.
- Ion beam sputtering: This is the most expensive and complex process, producing an excellent quality film that can withstand high stress. Very low levels of impurities are present. Deposition rates are between 1 to 2 A/s. The process is carried out at low temperatures and directionality is excellent. Scalability is, however, low.
The choice of deposition technology will be a tradeoff between film quality, expense, throughput, and yield. Film quality and productivity requirements will be the major drivers of the choice.
It can help, when making a choice, to consider the intrinsic properties and the extrinsic properties. The former depends on layer design, material selection, the precision of process control and choice of process, and the integrity of the deposition environment. The latter depends on the process choice along with deposition system quality and system design.
If we were to look at the example of reflective facet coatings in this context, we can compare the AR (anti-reflective) and HR (high reflectivity) coatings.
- AR coatings are meant to reduce reflectivity (3 to 10%) of the faces where the light hits as it enters and exits the laser cavity.
- HR coatings stop light from escaping the rear face of the diode. Instead, most of the light is reflected back in (90 to 99.9%.)
Consider the wavelength vs reflectance for AR and HR in low power and high power laser bars as an example.
Since film quality requirements aren’t so stringent in low-power laser bars, low cost and high throughput are the basic requirements. To select a production process, we’ll identify the intrinsic and extrinsic requirements. The former includes minimal oxidation and good adhesion. The latter includes a low cost.
Now we can match these requirements with one of the processes we have access to. For oxidation and adhesion requirements, precleaning with an in situ source and a flip-fixture will help. E-beam process will help keep costs low. These considerations will drive choices.
In high-power laser bars, high-stress tolerance and precise optical properties are necessary. High film density, low refractive index shift in air, and low defect density will be additional necessities in this application. Extrinsic properties will include high uniformity, a high-count film stack with multiple layers and repeatability of each layer.
Magnetron sputtering suits the needs of the scenario: it provides an ion source for pre-cleaning and assisted deposition. Evaporation doesn’t work for high-count multilayer film stacks.
High precision and high-cost monitoring system like optical monitoring systems will help. It will enable them to achieve the precision and complexity needed for the high repeatability of layers.
Process & equipment development plan
Closely guarded, company-specific
Capital spend in PVD
Customer-defined hardware to match process specs
Vendor defines hardware to meet process specifications
Variable response time; non-local service engineers common
Quick response time; dedicated service engineers on-site
Purchasing process equipment
Comparing the microelectronics and laser industries in the table below we can see the former would have been in the place of laser industries a few decades ago.
A growing market would have rendered independent and in-house R&D expensive. The industry was forced to adopt a unified roadmap as a result. The laser market is in a similar place of transition today. To scale up into large-volume applications, production costs have to drop by at least 25%.
Equipment will usually be purchased with hardware in mind, and the lowest price from a selection of vendors will be considered. This process may not be suitable for key process steps, as it may end in hardware that’s functional but ineffective.
For key process steps, lessons can be learned from the microelectronics industry. Film requirements can define the choice, and collaborations can take place with vendors. The end result will be equipment that works precisely for the application.
The factory cost of ownership can also benefit from a similar approach. Working with a number of multiple vendors, a different one for each process step, can lead to a lack of cohesive understanding among them. The customer will be held responsible for the workflow.
A fewer number of vendors will enhance understanding of the requirements and improve service time. In other words, it’s important to choose the process technology that fits in with the requirements of the production process.