Can you make a diode just by connecting a piece of N-type material like phosphorus to a P-type material like gallium? Why are they added to silicon?

  • 18
    \$\begingroup\$ They are not n and p type materials, they are just metal/non-metal dopants added to Silicon to create n and p type semiconductors. Si is the actual semiconductor here. \$\endgroup\$
    – Mitu Raj
    Oct 30, 2021 at 18:23
  • 13
    \$\begingroup\$ My OC72s say it isn't. \$\endgroup\$ Oct 30, 2021 at 18:23
  • 5
    \$\begingroup\$ 774400, Silicon and Germanium are metalloids. See here. Note: there is no accepted standard definition nor agreement on what is and isn't a metalloid. But it is agreed that Si and Ge are metalloids. Phosphorus is not a metalloid. The solid state physics for silicon (2-8-4) and germanium (2-8-18-4) crystals are appropriate (tin is similar with 2-8-18-18-4 but is overly conductive at room temperatures.) And silicon purification is well understood, relatively safe, and its boules (with or without dopants) are readily pulled from a melt. \$\endgroup\$
    – jonk
    Oct 30, 2021 at 18:38
  • 3
    \$\begingroup\$ @774400:What is "N-type" and "P-type" material? \$\endgroup\$
    – Curd
    Oct 30, 2021 at 19:18
  • 14
    \$\begingroup\$ @774400: In the title you are asking why Si is needed to make semiconductors. (BTW you don't need it to make a semiconductor; it is a semiconductor). In the body of your posting you are asking what it takes to make a diode. Those are different things. \$\endgroup\$
    – Curd
    Oct 30, 2021 at 19:23

5 Answers 5


First, silicon is not necessary for making semiconductors, nor even working semiconductor devices. The first commercially successful transistors were Germanium, and other semiconductor materials (e.g. selenium, copper oxide, etc.) were used to make rectifiers before transistors came along. Beyond that, there's a large number of other semiconducting materials (such as gallium arsenide) that work -- some even commercially.

Second, there are specific constraints on what makes a semiconductor work in a semiconductor device. A full treatment gets down into the quantum physics of it, which I'm neither inclined, nor entirely qualified, nor have the space to fully describe here. Basically, however, semiconductor devices work because the available states for electrons to exist in within the material are separated. This leads to the terms "valence band", "conduction band", and "band gap". You need a material that supports distinct valence and conduction bands, and that can be tuned to be P-type or N-type. All of the semiconductor materials do this.

Elemental gallium has overlapping conduction and valence bands, so it's a metal. It can be part of a semiconductor, but it's not a semiconductor. Elemental phosphorus has conduction and valence bands that are so far apart that it's an insulator. It can be part of a semiconductor, but it's not a semiconductor.

  • 2
    \$\begingroup\$ I don't think there is a limit to the number of characters for an answer, so technically you do have the space ;) \$\endgroup\$
    – DonQuiKong
    Oct 31, 2021 at 15:15
  • 3
    \$\begingroup\$ @DonQuiKong meta.stackexchange.com/a/176447/155659 \$\endgroup\$
    – OrangeDog
    Oct 31, 2021 at 15:45
  • \$\begingroup\$ I'm not so sure elemental phosphorus is always an insulator. White phosphorus I'm pretty sure is, but I wouldn't be surprised if red or black phosphorus turned out to have some conductive or semiconductive properties. \$\endgroup\$
    – Hearth
    Oct 31, 2021 at 18:24
  • 2
    \$\begingroup\$ @Hearth I hear you, but for the purposes of my answer I'm going to pretend I didn't. Phosphorus actually has a fairly modest band gap (that changes with the allotrope). But I'm going to keep my fingers seated firmly in my ears even so. \$\endgroup\$
    – TimWescott
    Oct 31, 2021 at 20:34
  • 1
    \$\begingroup\$ @DonQuiKong Having the space does not mean the space should be used! \$\endgroup\$ Nov 1, 2021 at 17:37

A PN junction needs a semiconductor. This is a material that is 'halfway' between a conductor and an insulator. It's not just that it conducts less well than a metal; it's that carriers (electrons and holes) in the material are constrained to different energy levels (think of floors in a building). You also need two sets of distinctly different levels for a P-N junction.

In addition, it's not 'just connecting' that is needed. The crystal structure needs to be continuous across the junction or else the energy levels will 'spread out', and the desired behaviour (conduction in only one direction) won't occur.

Note that not only silicon is a practical semiconductor -- germanium is. Semiconductors are also made of gallium arsenide (as in LEDs); gallium nitride (GaN -- used in high frequency RF circuits like 5G cellphone transmitters) silicon carbide (SiC - often used in electric vehicles to drive the motor) and even diamond for extremely high temperature circuits.

Fundamentally you need a crystalline non-metal (which therefore has distinct energy levels), a separation between these levels which is not too large (so that practical levels of current can flow); not too small (or too much current will flow -- germanium is in this range); and is practical for doping (which changes the energy levels). Silicon is extremely suitable for this -- it is plentiful (so low cost); easily purified; robust (strong); straightforward to dope, yet chemically quite inert.

  • 2
    \$\begingroup\$ Silicon also happens to have other desirable properties - like the fact that silicon dioxide, which is easily produced on a wafer, is a pretty good insulator. \$\endgroup\$
    – Luaan
    Oct 31, 2021 at 22:05
  • 1
    \$\begingroup\$ Silicon is also not a direct bandgap semiconductor. Direct bandgap is pretty awesome for an LED device, since radiative recombination is so easy. Pew-pew ooh pretty! It is not as cool to have your processor unit blinking away all your 1's due to radiative recombination. pew-pew whoops, blue screen. Thus, some momentum separation of minima and maxima is for the better. \$\endgroup\$ Nov 1, 2021 at 14:07
  • \$\begingroup\$ MOSFETs would not blink much even if made of direct-bandgap material. \$\endgroup\$
    – fraxinus
    Nov 1, 2021 at 16:04
  • \$\begingroup\$ @StianYttervik I think the electrical circuits that lead to 0s and 1s being stored are one or two layers above the properties of the electrons in the semiconductor material. We don't represent a 1 by the presence of electrons in the conduction band. Rather it's represented by a high voltage over here and a low voltage over there, and the voltage difference comes from the design of the overall electrical circuit and not the individual pieces of semiconductor material. \$\endgroup\$
    – user253751
    Nov 2, 2021 at 10:08
  • 1
    \$\begingroup\$ @user253751 True, I was being hyperbolic. Recombination is nonetheless an important factor of a semiconductor's performance which promotes the use of an indirect band gap. You'd need stronger fields across the junctions to ensure recombination isn't an issue, which in turn leads to ohmic losses and possibly a larger minimum thickness of the different doped zones. \$\endgroup\$ Nov 2, 2021 at 10:21

Semiconductor rectifiers have been used since the 1920s. Typically using Copper Oxide or Selenium

Metal rectifiers consist of washer-like discs of different metals, either copper (with an oxide layer to provide the rectification) or steel or aluminium, plated with selenium. The discs are often separated by spacer sleeves to provide cooling.


It is not. And phosphorus, in its allotropic form called black phosphorus, is actually a semiconductor and could be used; the problem is that it is expensive.

Black phosphorus (BP), a novel two-dimensional (2D) layered semiconductor material, has attracted tremendous attention since 2014 due to its prominent carrier mobility, thickness-dependent direct bandgap and in-plane anisotropic physical properties. BP has been considered as a promising material for many applications, such as in transistors, photonics, optoelectronics, sensors, batteries and catalysis. However, the development of BP was hampered by its instability under ambient conditions, as well as by the lack of methods to synthesize large-area and high quality 2D nanofilms.


Gallium Nitride is a III/V (three-five) semiconductor. It is best know as the Blue LED that is the basis for making the billions and billions of white LEDs that are on the market. The .white LEDs are GalliumNitride (GaN) diodes with some phosphors to make the green and red emissions to make a "white" LED.

There is growing interest in GaN semiconductors for transistors and companies like CUI are using commercially made GAN transistors for high frequency power supplies (smaller indicators, less ripple, less heat loss, ...). Also, GaN can be used to make circuits efficiently run at frequencies well beyond where silicon starts to get inefficient (very inefficient). Microwave frequencies in excess of 70GHz used for automotive radar is not a problem for GaN.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.