Why is silicon the preferred material for wafers in integrated circuits?

  • Thread starter Thread starter fog37
  • Start date Start date
  • Tags Tags
    Silicon
Click For Summary

Discussion Overview

The discussion centers around the reasons for silicon being the preferred material for wafers in integrated circuits (ICs), exploring its properties, the manufacturing process, and comparisons with other semiconductor materials. Participants delve into the relationship between ICs and silicon wafers, including the processes of doping and the creation of electronic components.

Discussion Character

  • Exploratory
  • Technical explanation
  • Conceptual clarification
  • Debate/contested

Main Points Raised

  • Some participants note that silicon is commonly used because it is an easy semiconductor material to work with, being a pure crystal that allows for effective doping to alter electrical properties.
  • Others highlight that silicon is cheap, stable across a wide temperature range, and the processes for modifying its properties are repeatable and easy to control.
  • There is mention of other semiconductor materials like gallium arsenide and germanium, but participants agree that silicon is the most prevalent due to its abundance.
  • Participants discuss the process of doping, where different substances are introduced to create various electronic components embedded in the silicon wafer.
  • Some contributions clarify that while some connections between components can be made through doping, metal traces are often added later to facilitate connections.
  • Questions are raised about the necessity of both ICs and discrete components on circuit boards, with some suggesting that cost and practicality influence this design choice.
  • Participants explore the manufacturing processes, including photomasking and photolithography, used to define component locations on the silicon substrate before doping occurs.
  • There is curiosity about the current state of optical computing and its feasibility, with participants noting that it remains an active field of research.

Areas of Agreement / Disagreement

Participants generally agree on the advantages of silicon as a semiconductor material, but there are multiple competing views regarding the necessity of discrete components alongside ICs and the feasibility of optical computing. The discussion remains unresolved on some technical aspects and the future of optical chips.

Contextual Notes

Limitations include the complexity of semiconductor manufacturing processes and the varying definitions of terms like "doping" and "photolithography," which may depend on specific contexts or applications. The discussion does not resolve the nuances of these processes.

fog37
Messages
1,566
Reaction score
108
Hello Forum,

I would like to understand why silicon, specifically, is used to make wafers (substrates) to make integrated circuits.
ICs are small and dense electric circuits. Are the various gates, resistors, capacitors, transistors and other components made by doping different regions of the semiconductor silicon wafer? Or electronic components installed on top of the silicon wafer? I am trying to understand how the relation between IC and the silicon wafer...

Metal traces are still need to create connections between all those components, correct?

thanks,
fog37
 
Engineering news on Phys.org
Hello Fog .. there is too much to cover in a forum here -- There are many videos : http://www.computerhistory.org/revolution/digital-logic/12/288/2220

As for the reason Silicon is used (most common) is that it is an easy semiconductor material to work with. The wafer is a slice of a pure crystal - meaning the SI atoms are all set in a fixed pattern. By adding (doping) - individual atoms are replaced with a different one - typically having one more ( - doped) or one less (+ doped) election - but then other doping atoms will yield other characteristics.

Keep in mind the topic from material science and crystal structure, to basic semiconductor to complete IC deign and fabrication could be from 3 to 6 classes - in succession in a Bachelors degree program.
 
Top reasons: It's cheap and stable across a large range of temperatures, it's very easy to change the electrical properties from good conductor to good insulator by doping/oxidation and the results are very repeatable using easy to control processes.

Did I say cheap already?

A very simple example of the process.
 
Last edited:
fog37 said:
I am trying to understand how the relation between IC and the silicon wafer...
As Windadct said,
Keep in mind the topic from material science and crystal structure, to basic semiconductor to complete IC deign and fabrication could be from 3 to 6 classes - in succession in a Bachelors degree program.
Take a look at this page and click on some of the links to see what he means.

http://hyperphysics.phy-astr.gsu.edu/hbase/solids/sselcn.html#c1
 
Silicon crystalline structure allows dopants to alter the semiconductor electrical characteristics. There are other crystals that provide semiconductor characteristics (gallium arsenide, for example can be used for IC's)

In addition to doping, there are many layers of silicon dioxide (basically glass) and metal (generally copper or aluminum) to provide multilayer wiring, as well as polysilicon resistors.

The difference between the silicon wafer and the IC is "a pile of" masks and manufacturing steps.

http://wiki.usgroup.eu/wiki/public/tutorials/cmosfabricationprocess

http://bwrcs.eecs.berkeley.edu/Classes/icdesign/ee141_s02/Lectures/Lecture5-Manufacturing.pdf <--- many neat pictures
 
Last edited by a moderator:
Thanks everyone.

So, monocrystalline silicon is used because it is cheap, easier make and to control, etc.

By doping (altering) different regions of the blank silicon wafer at different level and with different substances we can create different electronic components (like resistors, capacitors, transistors, diodes) that are embedded into the material itself, i.e. from the silicon itself. Are the connecting wires between those created components made the same way (by doping) or are metal traces added later?

Most circuit boards have both IC and the standard lumped components (capacitors, resistors, etc.). Why? Why not make everything in the IC form?

Is the process of photomasking and photolithography used to first map where the various components are supposed to be on the silicon substrate, before the doping step and before the components are actually created?

thanks,
fog37
 
fog37 said:
Thanks everyone.

So, monocrystalline silicon is used because it is cheap, easier make and to control, etc.

Silicon is the most common semiconductor material used, but is not the only one. Gallium and germanium are also used in some specialized semiconductors, but neither is as plentiful as plain old silicon.

By doping (altering) different regions of the blank silicon wafer at different level and with different substances we can create different electronic components (like resistors, capacitors, transistors, diodes) that are embedded into the material itself, i.e. from the silicon itself. Are the connecting wires between those created components made the same way (by doping) or are metal traces added later?

Some components are connected by conducting channels created on the wafer itself; others are connected by special metal layers which are deposited on the surface of the chip. Some processor chips contain millions of transistors and other components, which would be too costly to wire up as if they were discrete components. ICs were invented to prevent making all these connections separately.

Most circuit boards have both IC and the standard lumped components (capacitors, resistors, etc.). Why? Why not make everything in the IC form?
Because it would be too expensive, and for some components, impractical or impossible. For example, inductors still have to be made by coiling a wire around something; no practical semiconductor analog exists. To an extent, the same is true for capacitors; if you want a certain size capacitor, a discrete component is used.

It takes roughly a billion dollars to set up a production line for a complex semiconductor, like a processor, exclusive of the cost of developing the design. There are some "computers on a chip" out there, but for maximum flexibility in configuring a computer system, it's still easier and cheaper to use ICs to make only a portion of the entire system.

Is the process of photomasking and photolithography used to first map where the various components are supposed to be on the silicon substrate, before the doping step and before the components are actually created?

thanks,
fog37
The various masks are created from the design of the chip. The circuit logic (i.e., the various components and pathways connecting them) is created and checked by a chip designer, and the mask to make that particular arrangement of circuitry is then produced from this design.

An un-doped wafer is coated with a light-sensitive material, and this coated wafer is exposed to light. The portions of the coating exposed to light change chemically, and are washed away by treating them with a special cleaner. The masked portions of the wafer are protected from absorbing the doping material during that part of the manufacturing process.

This article discusses some of the IC manufacturing aspects:

https://en.wikipedia.org/wiki/Semiconductor_device_fabrication
 
Thank you!
Everything is much clear now. It is interesting that inductors still have to be made by coiling a wire while other components are mimicked by doping silicon...

On a different note, there is buzz about optical computers. Are optical chips already a reality?

fog37
 
fog37 said:
Thank you!
Everything is much clear now. It is interesting that inductors still have to be made by coiling a wire while other components are mimicked by doping silicon...

On a different note, there is buzz about optical computers. Are optical chips already a reality?

fog37

It's still an active field of research, but there is no indication that an optical computer will be sitting on your desk anytime soon:

https://en.wikipedia.org/wiki/Optical_computing

Ditto for quantum computers:

https://en.wikipedia.org/wiki/Quantum_computing
 
  • #10
SteamKing said:
It's still an active field of research, but there is no indication that an optical computer will be sitting on your desk anytime soon: ...
The research was going on during my college years '60s and '70s. The term "photonics" was used after semiconductor light sources (LEDs) were first made. Funny thing:
https://en.wikipedia.org/wiki/Photonics#Photonic_integrated_circuits said:
The most widely deployed PICs are based on Indium phosphide material system. Silicon photonics is an active area of research.
bold by me
 
  • #11
fog37 said:
Thank you!
Everything is much clear now. It is interesting that inductors still have to be made by coiling a wire while other components are mimicked by doping silicon...

We can emulate some properties (mainly in filters) of inductors on the IC substrate by using circuits like the Gyrator at low frequencies and powers.
 

Similar threads

Replies
5
Views
1K
  • · Replies 13 ·
Replies
13
Views
2K
  • · Replies 1 ·
Replies
1
Views
6K
  • · Replies 6 ·
Replies
6
Views
4K
  • · Replies 3 ·
Replies
3
Views
3K
Replies
68
Views
7K
  • · Replies 2 ·
Replies
2
Views
2K
  • · Replies 6 ·
Replies
6
Views
2K
  • · Replies 2 ·
Replies
2
Views
1K
  • · Replies 0 ·
Replies
0
Views
749