SnO: First Stable P-Type 2D Semiconductor Discovered (phys.org)

New submitter Namarrgon writes: Transistors made with Ashutosh Tiwari's new semiconducting material could lead to computers and smartphones that are more than 100 times faster than regular devices. While researchers in this field have recently discovered new types of 2D material such as graphene, molybdenun disulfide and borophene, they have been materials that only allow the movement of N-type, or negative, electrons. In order to create an electronic device, however, you need semiconductor material that allows the movement of both negative electrons and positive charges known as "holes." The tin monoxide material discovered by Tiwari and his team at the University of Utah is the first stable P-type 2D semiconductor material ever in existence.

Comments

No, you don't

[Anonymous Coward]

Vacuum tubes work marvelously well with only electrons.

Re:No, you don't

[fyngyrz]

n-type (negative) electrons

ha

ha ha ha

blargh hah ha hah ha

N-type semiconductors... the materials have excess electrons, and leverage that.

P-type semiconductors... the materials have an electron deficit, creating "holes" in the structure, and the material leverages those deficits.

There are no "positive" electrons. Well, there are, sort of, but they have little to nothing to do with n-type and p-type materials. Unless physics has completely rewritten semiconductor theory while I wasn't looking, which I suppose is possible.

Re:No, you don't

[rubycodez]

more importantly you can get rectifier and transistor action with n-type and certain metals; don't need a p type at all

Re:No, you don't

[fyngyrz]

PS: New leadership: Would you *please* consider hiring some editors that are at least somewhat technically competent? It would also be nice if, you know, they could... edit the written word competently. The best that can be said of slashdot's "editors" to date is that they have been a constant source of amusement for some. Wouldn't it be amazing if TFS's that actually hit the site were edited into well written presentations? Well, it would be for me. Reading most of them so far has been like being poked in the eye with a sharp stick.

Re:No, you don't

[ananamouse]

They did, but that content is on the 'Paywalled' side of this site.

Re:No, you don't

[Anonymous Coward]

Electrons and an awful lot of heat. Enjoy watching the city's lights dim and getting a $1 million electric bill whenever you boot up your phone.

Re:No, you don't

[Locke2005]

Vacuum tubes? Yeah, my friend really enjoys waiting 15 minutes for his tube amp to warm up before he can listen to his electrostatic speakers... (although after the 500W amp is warmed up, it does sound really good.)

Re:No, you don't

[Anonymous Coward]

OK, well have fun getting an X-ray with a transistor x-ray source at the hospital. Information processing is one thing, but when it comes to insane power, frequency or current, tubes beat solid-state all the time. I mean how else can it be? An electron in a vacuum is much smaller than some silly rock stuck in plastic and trying to get its electrons to move.

Re: No, you don't

[Anonymous Coward]

Positron is a positive electron.

Re:No, you don't

[buchner.johannes]

Isn't it obvious that what was meant here was

allow the movement of N-type, or negative [charges], electrons.

Re:No, you don't

[Anonymous Coward]

No, it's not obvious, because N-type or P-type refers to a MATERIAL, not a charge or a particle.

Re:No, you don't

[Anonymous Coward]

Please. This is slashdot. Get real. Ask for something which has a chance of happening, or at least getting considered.

Re:No, you don't

[Anonymous Coward]

Very well: "In order to create an electronic device whose DC power consumption does not increase linearly with the number of elements, you also need a p-type semiconductor."

Silicon is intrinsically n-type, but it's possible to dope it heavily to turn it p-type. But you'll note that in any complementary pair of silicon transistors, the PNP or P-channel one is, without fail, inferior in every respect: Gain, standoff voltages, threshold frequency, max current... everything. But not profoundly inferior. However, other semiconductors have outstanding properties for one type of device (e.g., GaN bipolar transistors can run at 100s of GHz; Diamond npn devices can run at THz) but a complement that is basically worthless.

This is important because the only way to build logic circuits with a low DC power consumption is using complementary saturating transconductance amplifier elements (aka CMOS): Once FET gates are charged, they stop drawing power (except for leakage). It is of course possible to design logic circuits using only one polarity of transistor or using other active elements (RTL, TTL, ECL, tube logic, relay logic...), but they all have to continuously steer a set amount of current per device or gate. Every central processor IC ever made has used CMOS for this reason. Old emitter coupled logic chips were fabricated on GaAs wafers using only npn transistors. They might have achieved nanosecond propagation delays in the 1970s, but the Cray One's power substation and chiller plant were each larger than the Cray itself. A modern billion-logic-gate CPU built in ECL would consume 100 million watts of power!

Today, single-polarity circuitry is seen pretty much exclusively where the gate count is very very small and operating speeds exceed what CMOS is capable of. Example: take apart an old DISH receiver, and at the base of the microwave horn you'll find a handful of single-fet preamplifiers feeding Gilbert cells, both using (IIRC) GaAs, because the RF carrier frequency is extremely high. Including the LO (I think it's also an exotic-semiconductor VFO), total npn transistor count is in the vicinity of 12-24. The FET count in the Broadcom baseband processor IC it feeds is in the tens of millions. And yet the receiver horn draws a substantial fraction as much power as the Broadcom chip!

Another example - in older WiFi devices, you might trace the antenna jack through a privileged trace on the pcb to a single 4-terminal device before it goes to the processor chip. That's a single npn transistor or dual-gate fet for RF power gain. It probably eats as much current as the WiFi SOC does on standby.

The reason we haven't seen any diamond-substrate chips or processors is precisely because nobody can figure out how to make p-channel or pnp devices in diamond. Likewise, nobody can figure out how to make p-type graphene either. Having a p-type 2D semiconductor that can at least in principle be paired with graphene opens the door (in theory) to practical 2D CMOS circuitry.

Re:No, you don't

[Anonymous Coward]

Was there a contest to see how much wrong information you can put into a post? You win! You win! Hands down!

"Silicon is intrinsically n-type"

No, no it isn't, numbnuts. Go check your table of elements and check to see what column it's in... Gee...

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

Everything else you wrote is a headache-inducing mess of incoherent nonsense.

Re:No, you don't

[penguinoid]

Also, it's not really 2D since it's made of atoms and atoms are 3D. Do I win the pedantry contest or what?

Also, technically, it's not enough to have both P-type and N-type semi-conductors. You also need to be able to produce them cheaply and precisely and by the billions, both types next to each other. So, technically, don't expect to see this used for anything ever until they solve those problems.

Re:No, you don't

[Anonymous Coward]

You also need to be able to produce them cheaply and precisely and by the billions, both types next to each other.

There are semiconductor markets besides just integrated chips. The places I work purchases a lot of power semiconductors, where a production batch of a thousand units is a big deal, and at a price of a couple thousand each can integrate a lot of tech completely inappropriate to small, low power devices used in most consumer goods. There are other categories too, speciality semiconductors that get made for things like highly specialized light and radiation detectors used by various experiments, special low temperature devices, commercial devices intended for quantum mechanics researcher, etc.

Re: No, you don't

[Anonymous Coward]

"There are no "positive" electrons. Well, there are, sort of, but they have little to nothing to do with n-type and p-type materials."

He addressed that, pedant.

Re:No, you don't

[Bengie]

"2D" materials actually have 2 dimensional properties. In a mathematical 2D world, resistance is less because of fewer dimensions of potential movement. It seems that if you make really thin layers of stuff, the resistance works out to perfectly match that of these hypothetical 2D flat lands.

Sounds cheaper, too

[sanosuke001]

Tin + Oxygen sounds a lot cheaper (and more readily available) than those iridium, molybdenum, etc compounds, too

Re:Sounds cheaper, too

[Anonymous Coward]

In the quantities that would be needed for the average piece of electronics, cost is really not an issue. Availability could be though.

Ashutosh Tiwari?

[110010001000]

Is that the same Ashutosh Tiwari I did shots with at the Indian Institute of Technology?

Re:Ashutosh Tiwari?

[fyngyrz]

Perhaps you shouldn't have been carrying curry in your backpack, newb.

Fast?

[Anonymous Coward]

I gather that currently hole mobility in tin monoxide is about 1 cm^2 / V s. That leaves a long way to catch up to silicon...

Semi conductor

[Anonymous Coward]

Did you know that a semi conductor conducts semis?

Negative charges

[sjbe]

they have been materials that only allow the movement of N-type, or negative, electrons. In order to create an electronic device, however, you need semiconductor material that allows the movement of both negative electrons and positive charges known as "holes."

Captain pedantic here. Electron holes are not positive charges. They are the absence of an electron in a lattice where one could exist. This "hole" can be treated for convenience and practicality like a positively charged particle but that isn't technically the same thing.

Re:Negative charges

[bluefoxlucid]

As much as I remember 'bout PNP and NPN questions on my radio technician's licensing test from when I was 11, the whole description is a journalistic goofusism.

What I remember is the semiconductor substrate is doped to favor a particular charge--positive or negative--and so it acts according in an electrical circuit. If you have two N-type materials wired into a circuit with a P-type material separating them, the N-type material will resist electrical current flow because electrons want to move into a negative-charge area. Apply charge against the P-type material and it starts moving negative charge out of the N-type materials, and you wind up with a P-type channel between the two wires--a CONDUCTIVE P-type channel.

That's, of course, a messy description. The point is you're not moving holes through P-type material and electrons through N-type material; you're only moving electrons, and altering the charge of areas of the substrate. Semiconductor gates are highly-complex engineered structures, not cables that allow you to swap void with electron.

Re:Negative charges

[jouassou]

If you really want to be pedantic, then the negative charges aren't really electrons either. Both the positive and negative charges are quasiparticles, which are particle-like excitations of a large sea of actual electrons in the semiconductor. The collective behaviour of all these electrons then results in something that looks like a single electron with a different mass and sometimes the wrong charge. But it's usually easier to just call these quasiparticles "electrons" and "holes", because that's what they intuitively behave like.

Re:Negative charges

[Anonymous Coward]

11? Dear Sir, you are clearly very intelligent. Your apt description of N-type versus P-type materials from memory only seems to reinforce this observation. May I fellate your penis?

Re:Negative charges

[MattskEE]

Captain pedantic here. Electron holes [wikipedia.org] are not positive charges. This "hole" can be treated for convenience and practicality like a positively charged particle but that isn't technically the same thing.

True, but basically all semiconductor device engineers don't worry about this distinction. It's true but it doesn't matter.

This is similar to deciding whether to talk about the air or the water in two cases: A drop of water falling down in a container otherwise filled with air, or a bubble floating up in a container otherwise filled with water. In both cases you could look at either the movement of the air or the water. But in a container filled with air except for one drop of water it makes most sense to consider the drop of water rather than the air being displaced by the water. And for a bubble in a container otherwise filled with water it's much easier to keep track of the bubble rather than the water that is moving around the bubble.

Similarly in a semiconductor, in a valence band which is typically mostly filled with electrons we pay attention to the hole. And in a conduction band which is typically mostly empty of electrons we track the electrons . There is a real positive charge associated with a hole, coming from the uncompensated positive charge of an atomic nucleus. So is it a real particle? No. But when you actually look at it from a quantum perspective it is indistinguishable from one.

Re:Negative charges

[Anonymous Coward]

11? Dear Sir, you are clearly very intelligent. Your apt description of N-type versus P-type materials from memory only seems to reinforce this observation. May I fellate your penis?

As long as you are female, and maintain eye contact throughout.

Explain to me

[Anonymous Coward]

I was taught that a hole was the absence of an electron. As electrons move one way, the holes, in effect move the other.

Re:Explain to me

[rubycodez]

no an electron from any direction can fill the hole, only the holes move along current vector.

So finally we can say -

[blackmesadude]

Let it SnO, let it SnO...

Re:So finally we can say -

[rubycodez]

the Amazon planet has dioxide tech so they sing

let it SnOO-SnOO

Re:So finally we can say -

[Daetrin]

Death... by SnOO-SnOO!

Vacuous remarks

[fyngyrz]

Vacuum tubes work marvelously well with only electrons.

If by "marvelously well" you mean with high random noise levels, comparatively low current capacities, and comparatively huge volume requirements, sure.

And if by "only electrons" you mean "only electrons, neutrons, protons, electromagnetic fields and - of course - vacuum, sure.

Re:Vacuous remarks

[plover]

I have to agree with the OP, at least in a literal sense. Vacuum tubes were indeed "marvels", as people marveled at their function; so to call them marvelous is absolutely correct. You can also say they work marvelously well when compared to electrical devices such as relays. He didn't claim they were efficient, cool, small, low-voltage, short-lived, solid-state, distortion-free, or noise-rejecting. Doesn't mean they weren't marvels.

</nits_picked>

Re:Vacuous remarks

[Anonymous Coward]

Yes, you mean more current capability than any solid-state device

https://en.wikipedia.org/wiki/Mercury-arc_valve
https://en.wikipedia.org/wiki/Thyratron
https://en.wikipedia.org/wiki/Krytron

you mean as quiet as any solid-state device

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

with much higher frequency and power capabilities than solid-state devices

https://en.wikipedia.org/wiki/Backward-wave_oscillator

can generate x-rays and radiation for medical purposes

https://en.wikipedia.org/wiki/X-ray_tube

and you mean it can scale arbitrarily large, because what else is a particle accelerator but a large vacuum tube.

Then yes, absolutely, marvelously well. And I have literally *no idea* what "neutrons, protons, electromagnetic fields " is supposed to mean. As usual, your posts are filled with pig-ignorance and a kind of pride in your shallow knowledge and limited curiosity.

Re:Vacuous remarks

[Anonymous Coward]

Oh, and "short lived"? How is that data from Voyager 1 getting here?

http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19880010181.pdf

Or how about

http://www.radiomuseum.org/tubes/tube_175hq.html

You sure you want to compare that to a phone you throw out after a few years?

And what does "noise-rejecting" mean? Like somehow a transistor is smart and will actively reject noise? And what does "solid-state" mean? I\ve been told over and over that this means "no moving parts" now, and there are no moving parts in a tube. What does "distortion-free" mean? Are you claiming that a single-transistor circuit has no distortion? Are you clinically not all there or what?

Seriously.

Re:Vacuous remarks

[Anonymous Coward]

Solid state means the device is made of a solid material and the charge carriers are contained within that material.
Valves may not have moving parts, they the electrons aren't contained in a solid material.

Re:Vacuous remarks

[Anonymous Coward]

Yes, thank you, perhaps English is not your first language. Did you notice where I said "I've been told over and over that this means "no moving parts" now"?

Re:Vacuous remarks

[ChrisMaple]

Having actually studied the data sheets for Nuvistors, I can tell you that their noise performance is inferior to modern semiconductors designed for RF reception.

Nuvistors are thermionic electron tubes, and even with the smoothing effects of space charge their equivalent noise temperature is limited by using a hot cathode to emit electrons. Unlike semiconductors, they can't be cooled to reduce noise.

Re:Vacuous remarks

[Anonymous Coward]

That may well be, but I'm sure given half a century of research we'd have come up with something... Like lanthanum hexaboride cold cathodes, for example. Point is, tubes aren't these laughable things, they're as complex and wonderful as anything.

Devices 100 Times faster ? No

[Crashmarik]

Electron transit speed is not the limiting factor in device speed. Don't know who wrote the article but there is no way your Iphone is getting a 200 GHZ cpu from this.

Re:Devices 100 Times faster ? No

[Anonymous Coward]

Electron transit speed is not the limiting factor in device speed. Don't know who wrote the article but there is no way your Iphone is getting a 200 GHZ cpu from this.

Why? Because reasons? From TFA the electrons move faster and produce less heat soooo.

But of course I'd expect Intel to try to milk it for the next few Human generations by only increasing processor speeds by 1 GHz per cpu generation while selling the truly fast stuff to the Government black ops for outrageous sums.

Re:Devices 100 Times faster ? No

[WaffleMonster]

Electron transit speed is not the limiting factor in device speed.

Electrons move thru gates faster when there is less capacitance and less heat from reduction of resistance.

Don't know who wrote the article but there is no way your Iphone is getting a 200 GHZ cpu from this.

Actual text from TFA:

"Transistors made with Tiwari's semiconducting material could lead to computers and smartphones that are more than 100 times faster than regular devices."

Article is silent on the idea of 200 GHZ processors. There are many ways to get to 100 times faster.

Re:Devices 100 Times faster ? No

[Crashmarik]

Electrons move thru gates faster when there is less capacitance and less heat from reduction of resistance.
Article is silent on the idea of 200 GHZ processors. There are many ways to get to 100 times faster.

If you seriously believe that charge carrier speed in the substrate is limiting factor in device speed there is not much I can do for you except recommend a book

http://www.amazon.com/Semicond...

and maybe the following courses of study Electronic circuits I-IV or whatever they may be calling it these days.

Re:Devices 100 Times faster ? No

[Anonymous Coward]

Every article will take "theoretical threshold frequency of one transistor using this substrate" and convert it to breathless exaltations of "X tens/hundreds/bullshit times faster than your CPU" to impress you, while always ignoring the ever so inconvenient factor that actually limits CPU speeds: HEAT DISSIPATION.

Computers have a thermal brick wall problem: "Existing air-cooled CPU technology cannot realistically achieve a CPU-die-to-air thermal impedance below about .2 *C/Watt without unacceptable noise generation, and only ducted air cooling has the maintainence-free reliability typical users expect." And they drove into that brick wall ten years ago, which is why CPU power consumption still maxes around 150W and frequency (which is linearly related to power consumption) is still typically not much faster than a fast Pentium IV... and why the cooling fans on a K80 GPU sound like they're trying to breach the lightspeed barrier when it's at full throttle.

If you can remove a thousand watts of heat from a single square inch while keeping the die temperature below 60-70*C, sure it would be possible to push some descendant of an i7 to an operating frequency of 10GHz. Fix the thermal problem by replacing the cold side with a bottomless supply of liquid nitrogen and you can push a stock i7 to 7GHz today. Years ago Sandia Natl Lab supposedly came up with a better fan architecture, but so far apparent sales offerings have amounted to bupkis.

Heat dissipation isn't the only issue though. The farther you push into microwave frequencies, the worse your dielectrics behave. The worse the resistive losses get in your metal interconnects due to the skin effect, forcing you to spend even more power on negative impedance booster circuitry to maintain signal integrity... and of course, the faster your clock the more the causal domain associated with one clock cycle contracts.

Re:Devices 100 Times faster ? No

[WaffleMonster]

If you seriously believe that charge carrier speed in the substrate is limiting factor in device speed there is not much I can do for you except recommend a book

No of course not, my remarks refer to gate delay. When working with 2-D elements capacitance is much lower.

What the heck is "2D"?

[Bearhouse]

So after reading the fine article, it's apparently stuff that's only about one atom thick.
So, pedant maybe, but for me while that's pretty damn thin, it's still three-dimensional.
Blame the bullshit and sensationalism that seems to have to accompany even new announcement today.
In a scientific article, can we just have the facts without the crap?
That would be a good new direction for /. to take.

now get off my three-dimensional lawn!

Re:What the heck is "2D"?

[Anonymous Coward]

Yes it's physically 3D, but it's the same thing when you draw on a paper. The picture is 2D picture even though there's ink on top of the paper. Here the sheet of paper is the 2D circuit.

Re:What the heck is "2D"?

[Anonymous Coward]

If it were any thinner, it wouldn't exist.

Re:What the heck is "2D"?

[serviscope_minor]

What's wrong with calling it 2D? Electron motion is effectively limited to two dimensions, and it doesn't make much sense to talk about lateral movement through the degenerate dimension. And if you hate this you'll be even more angry that scientists often refer to quantum dots as zero dimensional.

Re:What the heck is "2D"?

[Anonymous Coward]

Outsid eof the *-ene compounds graphene and the like which are 2D. 3D electrical compounds like metals and the like have the effect that electron flow on the outside layer of the material, and tend to just bounce around on the interior. Yes, you can adjust the guage size, but you are really just adding surface area and making up for resistance in the material melting under load. So in a 2D material like the *-ene's the electrons are forced to move atom to atom in a general direction all the time, so everything is just more effecient and you waste less power.

Re:What the heck is "2D"?

[whit3]

So after reading the fine article, it's apparently stuff that's only about one atom thick. So, pedant maybe, but for me while that's pretty damn thin, it's still three-dimensional.

The band structure for a bulk material (full 3-d crystal structure) defines the behavior of electrons deep inside the material, not near a surface - and near-suface conditions are different. The permitted electron orbitals (and bonding, and atomic spacing...) in a very thin layer of SnO might be very different indeed (and have different bandgap, mobility, etc.) from the bulk material.

The fabrication and characterization of a material that is NOT similar to its parent 3-d lattice is what has been described here, and it is important to note the 2-d nature of this semiconductor, so that readers aren't confused by the chemical similarity to a bulk semiconductor with dissimilar behavior.

Other surface-dependent characteristics define the behavior of polysilicon (Ovonics), quantum dots, and some very useful low-noise HEMTs. Search on "2-d electron gas" ...

100 times faster

[Anonymous Coward]

Stopped reading. No. They might get the gates to switch faster but the reality is that before and after the gate electrons are on copper tracks and those are really the limiting factors in terms of maximum speed.

"Magic" nonsense is still nonsense

[gweihir]

No, they will not make anything "100 times faster". The limiter today is interconnect and that does not get any faster at all with this material.

Re: "Magic" nonsense is still nonsense

[Gefion]

I am not certain how to calculate the net performance benefit of the entire device, but if the core CPU/GPU on a die can perform 100x faster or more efficiently, certainly we could expect much better battery life and or other benefits that would be worthwhile. The equivalent conversation for the Overclockers out there is that more cooling allows for faster speeds... Why? Because of the resistance of the substrate and the faster the clock the more energy waste. If you have a lower resistance substrate, that would be pretty darn handy. Of course if your bridge/interconnect subsystems can't keep up, some parts of the performance improvement will be limited eventually, but you still might enjoy significant battery gain. Sadly who knows if this will ever make production.

Re: "Magic" nonsense is still nonsense

[gweihir]

The limiter is on-die interconnect. You may get individual transistors 100x faster, gates 30x faster and CPUs 1.1x faster (if that). Sorry.

Devices 100 Times faster ? Yeah, it's possible

[whit3]

Electron transit speed is not the limiting factor in device speed.

It's one of the limits; that speed goes along with a concept called 'mobility' which directly translates to better current-carrying capacity.

Higher mobility for p-type devices DEFINITELY would speed CMOS.

Since SnO is a p-type material, it could become half the circuitry of a CMOS IC, and because it is to be a layer atop (presumably silicon) other materials, it would make for lower silicon area for a given complexity. By using that third dimension, your interconnect wiring gets shorter and faster.

Re:Devices 100 Times faster ? no it's not

[Crashmarik]

It's one of the limits;

Your reply isn't even even logically sufficient.

Here let me give you a car analogy.

You have a junker Saturn and put in a Ferrari's engine, then take it out onto I-95 during rush hour. The engine was never the limiting factor, the tires transmission, steering, and the highway were all the much greater limiting factors.

2DEG?

[Anonymous Coward]

Are we talking about a 2D electron gas?
Those can already be made with compound semiconductors (GaN, GaAs), though it may be one or two layers thicker.
But it's very unclear how to integrate those materials with Silicon chips, so any new material is still intriguing.