The Evolution of Silicon Pixel: How Tiny Chips Transformed Particle Physics

Silicon pixel sensors revolutionized particle detection with microscopic tracking precision, enabling huge physics discoveries.

Silicon pixel sensors have revolutionized particle detection in cutting-edge physics experiments. These tiny silicon chips can pinpoint the exact location where a subatomic particle passed through with incredible precision. In my science blog, I want to explain the evolution of this technology and why it’s so important.

The Origins of Silicon Pixel Sensors

In the old days of particle physics, detectors were much cruder. They could tell that a particle hit somewhere in a general area but not exactly where. Imagine trying to study a complex particle collision with a blurry camera—you would miss a lot of important details.

That’s why silicon pixel sensors were such a huge leap forward. They work kind of like a digital camera sensor. An array of miniature pixels turns the energy from an incoming particle into an electrical signal. By seeing which pixels got activated, you can reconstruct the particle’s path with high resolution.

The first silicon pixel sensors were “hybrid” devices. They combined a pixelated silicon sensor wafer with a separate silicon chip containing the readout electronics. This worked well but wasn’t ideal. The readout chips were built with a process optimized for circuits, while the sensors used a specialized process to get high-purity silicon. Mating these together was a hassle.

Scientists realized it would be easier to put both the sensing pixels and the electronics on a single silicon chip using “monolithic” fabrication. However, this was tricky with the microchip manufacturing processes available in the 1980’s and 90’s.

The move to monolithic silicon pixel sensors was motivated by several factors. Cost is a major one. Building hybrid detectors requires meticulously aligning and connecting two pieces. Monolithic fabrication on one wafer is simpler and less expensive. It also enables smaller, lighter, and lower power designs.

These advantages are desirable for high-energy physics. Reduced material means fewer unwanted interactions with particles. Experiments need vast amounts of sensing coverage, so cost matters.

Early Innovations to Enable Monolithic Silicon Pixel Sensors

The first hurdle was integrating the sensor and electronics into the high-purity silicon needed for particle detection. Regular microchips use lower-grade silicon wafers seeded with impurities. This is great for making transistors but lousy for sensors.

Researchers in the 90’s developed ways to fabricate chips on the fancy silicon and still get working transistors. One idea from Lawrence Berkeley Lab involved “gettering” to remove impurities from the chip’s active areas. Another technique developed at Stanford used super high-temperature processing to minimize contaminants.

This paved the way for real monolithic devices. One milestone was MIMOSA-1 in 1999, the first monolithic chip made using fairly standard manufacturing techniques. While far from today’s standards, it proved the concept was viable.

Photograph of MIMOSA 1
Photograph of MIMOSA 1

But MIMOSA-1 had limitations. Its pixel array contained only simple components and couldn’t handle complex logic. In 2008, engineers in the UK developed “inMAPS” which added extra transistor layers so more advanced circuitry could be integrated right into the pixel matrix.

) Schematic cross-section of the INMAPS process [38][38] with an additional deep p-well implant in the pixel to prevent the n-wells with PMOS transistors from collecting
Schematic cross-section of the INMAPS process with an additional deep p-well implant in the pixel to prevent the n-wells with PMOS transistors from collecting
signal charge so that only the n-well collection electrode collects the signal charge
Schematic cross-section of the INMAPS process with an additional deep p-well implant in the pixel to prevent the n-wells with PMOS transistors from collecting signal charge so that only the n-well collection electrode collects the signal charge

This innovation enabled all kinds of improvements to silicon pixel sensors. With full circuits on each pixel, you can add intelligence like time-stamping detected particles to reconstruct events in 4D. You can also use advanced “sparse” readout architectures. This means only reading out the pixels that detected something, reducing data volume.

Another clever technique called “AC coupling” was introduced in some silicon pixel sensors. This uses capacitors to block the DC baseline voltage of the sensor from reaching the electronics. AC coupling simplifies the circuit design since it removes the need to handle the large sensor voltage.

Some sensors also began incorporating a “reverse bias” on the sensor substrate. Applying a voltage gradient depletes the silicon of charged carriers, creating an active sensing region. This provides higher sensitivity and faster collection of the charge generated when a particle passes through.

Making Silicon Pixel Sensors More Radiation Hardened

Another big advance was making the sensors more radiation hard. Particle physics experiments bombard chips with intense energy. Early monolithic detectors could withstand maybe 10,000 radiation hits per square cm. But by using specialized transistor layouts and optimizing the pixel sensors to operate with higher voltage bias, modern chips can endure over 1,000,000 times more radiation while still working.

The key to radiation tolerance is accelerating the charge collection. When high-energy particles pass through the silicon, they generate electron-hole pairs. But radiation also gradually damages the crystal structure, creating defects that can trap these charges before they reach the pixel electrode. A higher electric field provides the “oomph” to sweep the electrons and holes to the collection point faster.

Various techniques are used to maximize the field strength. Special pixel layouts with a small collection of electrodes reduce their capacitance. Added sensor implants alter the internal electric potential. Together these allow the application of much higher voltage across the sensor region, up to hundreds of volts, to achieve extreme radiation hardness.

Schematic view of two different sensor designs for increased radiation tolerance: (a) a large collection electrode with circuitry embedded and (b) a small collection electrode
surrounded by wells containing the readout circuitry. The low dose n-type implant is added to move the junction away from the collection electrode and allow full depletion of
both the very lowly doped P-type epitaxial layer (P=) and the n-type implanted region.
Schematic view of two different sensor designs for increased radiation tolerance: (a) a large collection electrode with circuitry embedded and (b) a small collection electrode surrounded by wells containing the readout circuitry. The low dose n-type implant is added to move the junction away from the collection electrode and allow full depletion of both the very lowly doped P-type epitaxial layer (P=) and the n-type implanted region.

Pushing Silicon Pixel Sensors to Ever Smaller Feature Sizes

To pack more functionality into each pixel, engineers are constantly shrinking component sizes. This follows the whole semiconductor industry trend towards smaller transistors. Moving to finer lithography also reduces capacitance, improving radiation tolerance and analogue performance.

Early monolithic sensors used process nodes around 0.5-0.6 microns—huge by modern standards! Now they are transitioning to 65nm and below. For comparison, leading-edge microprocessors are at 3-5nm. So there is room for silicon pixel sensors to keep scaling down.

Of course, tiny transistors are more vulnerable to radiation damage effects. But specialized hardening techniques like redundant vias and circuits allow highly tolerant designs, even in very thin geometries.

Pixel size and line width of CMOS imaging technologies and DRAM technologies. Pixel imaging technologies are about 10 years behind DRAM technologies
with pixels containing only few transistors at a pitch of about 20 times the minimum
technology linewidth
Pixel size and line width of CMOS imaging technologies and DRAM technologies. Pixel imaging technologies are about 10 years behind DRAM technologies with pixels containing only few transistors at a pitch of about 20 times the minimum technology linewidth

Continued Innovation for Large Scale Silicon Pixel Sensors

Engineers keep innovating on monolithic fabrication to make larger and more complex pixel chips. Stitched sensors can be made bigger than a single wafer by carefully aligning multiple reticles on the silicon. Wafer stacking techniques allow integrating chips from multiple wafers into a single 3D component. This opens the door for detectors with over 10 billion pixels across ten square meters!

Manufacturing large stitched silicon pixel sensors requires very careful design to ensure acceptable yields. Entire wafers have to work reliably even if small areas are defective. Power delivery is also a major challenge. Resistive losses can cause voltage drops across large chips. Clever segmented power distribution and regulation helps address this.

3D integration via wafer stacking provides another path to large devices. This blurs the lines between “monolithic” and “hybrid” sensors. Stacking can combine optimized sensing layers with separate tiers containing complex drive electronics. The main obstacles are aligning and interconnecting stacks reliably while keeping heat dissipation manageable.

The Impact of Silicon Pixel Sensors on Fundamental Physics Research

The ability to track particles with near microscopic accuracy has been absolutely crucial for today’s groundbreaking discoveries in physics. As silicon pixel sensors continue improving, scientists can probe deeper into the quantum foundations of our universe.

One major application is detecting Higgs bosons, elusive particles that explain how others acquire mass. Finding the Higgs at the Large Hadron Collider relied on tracking subtle particle jets with ultra-fine precision only possible with pixel vertex detectors. This confirmed a 50 year old theory and led to the 2013 Nobel Prize.

Looking ahead, more advanced silicon pixel sensors will be vital to push the energy frontiers with higher collision rates and intensities. For example, the proposed International Linear Collider would smash electrons and positrons at unprecedented energies. Instrumenting the miles-long accelerator tunnels to analyze events will require ultra-fast, ultra-robust pixel sensors covering hundreds of square meters.

Applications of Silicon Pixel Sensors Beyond Physics Research

While primarily used in physics, these ultra-advanced imaging chips also have applications in medicine, space science, and beyond. For example, silicon pixel sensors enable proton therapy machines to precisely target tumors without damaging healthy tissue.

Time-of-flight positron emission tomography (TOF-PET) is another medical use. Photons from radioactive tracer molecules hitting the pixel array provide both spatial and timing information to reconstruct activity inside the body. TOF-PET promises substantial improvements in image quality and scanning efficiency.

Silicon pixel sensors are also transforming X-ray scanning for dentistry and mammography with higher resolution and lower dose. Lighter and more flexible CMOS detectors simplify these imaging procedures. X-ray astronomy missions also benefit from large area, highly efficient pixelated sensors.

Conclusion

In summary, the evolution of monolithic silicon pixel sensors has enabled tremendous advances in high energy physics experiments. Clever techniques to integrate high purity sensors with complex circuitry in a radiation-hardened way have led to devices unimaginable just decades ago. The future looks bright for silicon pixel sensors to enable ever more detailed particle tracking and unravel even deeper mysteries of our universe.

Reference

Stefano Meroli
Stefano Meroli

CERN scientist, history lover.
PhD in Nuclear Physics and counting.

Articles: 12

Leave a Reply

Your email address will not be published. Required fields are marked *