Welcome to Laser Pointer Forums - discuss green laser pointers, blue laser pointers, and all types of lasers

LPF Donation via Stripe | LPF Donation - Other Methods

Links below open in new window

ArcticMyst Security by Avery

InGaN band structure question

dyeote

0
Joined
Feb 9, 2012
Messages
131
Points
18
I was thinking about N-level laser systems, and whether a diode laser is a '2-level' laser when a question occurred to me: Why must I pump my InGaN bluray diodes with over 4.5V (nearly 6V sometimes) to get a ~3eV photon (405nm)? is there a non-radiative transition in there? or is it a more complicated explanation necessitating a crystal band structure diagram?
Then I thought, even if there is a non-radiative transition in there (which I'm guessing there isn't), the wavelength variance of these diodes under different currents means that this is definitely a complicated band-structure matter :/

From various graphs found on the forum I saw that, with an increasing current, these diodes (PHR, BDR, etc...) increase their pump voltage (more energy per excitation) and increase their wavelength (less energy per emission). Am I correct to infer from this that their pumping band-gap increases and their lasing band-gap decreases ??
Does 'greater current' here directly correlate to a 'greater temperature' of the gain medium?

Responses shedding additional (coherent :D) light on the matter would be appreciated, and illustrative graphs doubly so.
 





Joined
Nov 7, 2008
Messages
5,725
Points
0
There's a member here that goes by pullbangdead.. his education directly involves GaN technology.. if there is someone here who can answer it's him. I haven't seen him on here in awhile, but if you were to shoot him a PM he might respond since it often sends out an email to alert the user when you send the message. Give him the link to this thread so he can respond here and we can all benefit from the answer..
 
Last edited:

dyeote

0
Joined
Feb 9, 2012
Messages
131
Points
18
PM'd him.

Meanwhile, I've been thinking, the pumping here is done by elastic collisions between electrons. In these collisions there's no necessity for all of the kinetic energy to be transferred, in fact it's statistical, right? So perhaps raising the voltage serves to increase the cross section of such collisions that do transfer sufficient energy to the medium to contribute to the population inversion? with the rest of the energy being lost to resistivity?? Gah, I wish remembered more of the solid state course...
 
Joined
Nov 7, 2008
Messages
5,725
Points
0
Well, it sounds to me like you've got more knowledge pertaining to the atomic level than many of the people here. What you say makes sense to me, but whether it's truly accurate or not I can't say for sure at this time. I'll see what I can dig up in terms of papers etc, and in the meantime maybe PBD will get back to us. This is a great topic, thanks for posting it!
 
Joined
Jul 4, 2008
Messages
2,499
Points
113
I was thinking about N-level laser systems, and whether a diode laser is a '2-level' laser when a question occurred to me: Why must I pump my InGaN bluray diodes with over 4.5V (nearly 6V sometimes) to get a ~3eV photon (405nm)? is there a non-radiative transition in there? or is it a more complicated explanation necessitating a crystal band structure diagram?
Then I thought, even if there is a non-radiative transition in there (which I'm guessing there isn't), the wavelength variance of these diodes under different currents means that this is definitely a complicated band-structure matter :/

From various graphs found on the forum I saw that, with an increasing current, these diodes (PHR, BDR, etc...) increase their pump voltage (more energy per excitation) and increase their wavelength (less energy per emission). Am I correct to infer from this that their pumping band-gap increases and their lasing band-gap decreases ??
Does 'greater current' here directly correlate to a 'greater temperature' of the gain medium?

Responses shedding additional (coherent :D) light on the matter would be appreciated, and illustrative graphs doubly so.

This is a VERY complex, question to answer.
I guess I will have a stab at it. One of my students at LG explained InGaN and LED materials
in regards to minimum emission voltage. Firstly, the semiconductor material determines the bandwidth gap possible, and when adjusted can give a range of outputs depending on the voltage potential. Were talking about electron hole recombination in the material and this is where electrons hit the active medium to create our "blue" photons. In the case of InN, the conductivity is quite low by comparison to say ,GaAs/AlGaAs thereby require a higher potential across the P/N junctions.
InGaN is a mixture of InN and GaN with a minute variation on amounts to give either a 470nm or a strong UV laser output more "Gallium Nitride". Highest output energy of InGaN is around 3.35eV Photon which is a little greater than 60% eff for a given 5.5V input.
normal conditions InGaN is around 50-55% efficient in respect to V to quantum energy (eV).

Then I thought, even if there is a non-radiative transition in there (which I'm guessing there isn't), the wavelength variance of these diodes under different currents means that this is definitely a complicated band-structure matter :/

Taken from str-soft Gallium Nitride materials and devices journal.

engineering approaches used for modeling III-nitride LEDs are based on a drift-diffusion model of carrier transport in heterostructures. This model has an intrinsic drawback: it predicts remarkably overestimated operation voltage of LEDs with a large number of deep InGaN QWs. The reason for that can be*
understood from Fig.8a where the band diagram is plotted for an MQW LED structure emitting light at*
515 nm. Because of large band offsets at the QW interfaces, every well is surrounded by potential barriers where the electron/hole concentration becomes extremely small,

*and lower, next to the*
barrier/QW interfaces. A low electrical conductivity of these regions results in a ladder-like conduction*
and valence band alignment shown in Fig.3a and, eventually, in unrealistically high operation voltage,
greater than 5 V at 30 mA (Fig.3c). Besides, the band alignment like that shown in Fig.8a implies a*
strong ballistic electron leakage to occur in such a structure [42]. *


First of all, some important materials properties of III-nitride compounds, like conduction and valence*
band offsets, spontaneous electric polarization, deformation potentials, ionization energies of donors and*
acceptors, hole effective masses, Auger recombination coefficients, etc., are still known with insufficient*
accuracy. In addition, the necessity of using advanced physical models requires knowing additional parameters related, for instance, to TDs or localized electron/hole states. The properties of accompanying*
materials used for LED fabrication also need a careful evaluation. For example, the optical properties of*
metals used as electrodes have a strong dispersion in the visible and UV spectral ranges, which should be*
accurately considered to predict properly LEE from an LED die. Another example is ITO having interrelated electrical and optical properties. Making simulations, one should specify the electrical conductivity*
of the ITO film, as well as the electron mobility, in order to estimate the materials optical characteristics*
depending on both parameters [30]. *
*Not only the reliable materials properties but also some additional information on the LED structure is*
frequently needed for accurate simulations. First, the LED band structure is quite sensitive to distribution*
of polarization charges depending on particular composition profile and strain. Because of indium surface segregation, the InGaN QW profile may by much different from the nominal one, providing considerable broadening and smoothing of the QW interfaces and penetration of indium in the QW barriers*
[45,46]. Second, the strain in the well may depend on both the composition profile and stress relaxation*
occurring in the layers preceding the QW. And, considerable Mg redistribution next to the active region*
during growth of various LED structures has been reported, eventually affecting the emission efficiency*
[47,48]. All the above effects are crucial for LED operation and can be allowed for in simulation on the*
basis of additional experimental information or advanced theoretical studies. *
4 Conclusion As discussed above, there is still a big room for improvement of the existing physical*
models and building up new models accounting for specific properties of III-nitride materials and devices. In addition, much effort should be made to understand better effects of III-nitride compound microstructure on the carrier transport and recombination processes. Such developments are expected to*
form the main stream in the future research. However, this will not provide any guideline for understanding the impact of technological factors and, first of all, of heterostructure growth conditions on the operation and characteristics of III-nitride LEDs, despite its evident importance. On the other hand, it is hard*
to expect a rapid progress in experimental studies of such issues, as strain relaxation and impurity redistribution in LED structures during growth, modification of the QW composition profile and instability of*
the QW thickness caused by indium surface segregation, as well as other technological factors. In this
respect, coupled modeling of heterostructure growth and device operation would be a solution to this*
problem. The analysis of indium incorporation during epitaxial growth and optical transitions in strained*
InGaN/GaN materials and QWs [49] provides an example of a small first step forward in this direction.*
Alternatively, detailed experimental information has to be invoked to highlight the contribution of technological factors to formation of the composition, *doping, and defect density profiles in LED heterostructures


Have a look at the link. Maybe it can explain to you what I can't. I don't have a background in quantum physics, sorry :(

http://www.waset.org/journals/waset/v55/v55-3.pdf
 
Last edited:
Joined
Aug 25, 2007
Messages
2,007
Points
63
I was thinking about N-level laser systems, and whether a diode laser is a '2-level' laser when a question occurred to me: Why must I pump my InGaN bluray diodes with over 4.5V (nearly 6V sometimes) to get a ~3eV photon (405nm)? is there a non-radiative transition in there? or is it a more complicated explanation necessitating a crystal band structure diagram?
Then I thought, even if there is a non-radiative transition in there (which I'm guessing there isn't), the wavelength variance of these diodes under different currents means that this is definitely a complicated band-structure matter :/

From various graphs found on the forum I saw that, with an increasing current, these diodes (PHR, BDR, etc...) increase their pump voltage (more energy per excitation) and increase their wavelength (less energy per emission). Am I correct to infer from this that their pumping band-gap increases and their lasing band-gap decreases ??
Does 'greater current' here directly correlate to a 'greater temperature' of the gain medium?

Responses shedding additional (coherent :D) light on the matter would be appreciated, and illustrative graphs doubly so.


So you're over-complicating it slightly, and there are some different things going on.

So I think the main question is this: 405nm is ~3.06eV, so why do you need more than 3.06V to generate light at 405nm?

Well first, you said part of the answer in your own question without realizing it:

Why must I pump my InGaN bluray diodes with over 4.5V (nearly 6V sometimes) to get a ~3eV photon (405nm)?

Do you need more than 3V to get A SINGLE 405nm photon out of a 405nm laser? Try it. Notice a SINGLE PHOTON. I think you may well find that you do get 405nm photons out at not much more than 3V, just not many of them.

Then lasing is more complicated: you don't need one photon from one e-h pair, you need a lot of photons. Which means you need a lot carriers. Which means you need a lot of current.

If you somehow had just an active region, just an InGaN quantum well existing in a vacuum and oyu could magically pump current into it, then sure, all that matters is the bandgap. But you don't have that: you have an InGaN active region inside a GaN p-n junction. You have waveguiding layers. You have signifcant layers of semiconductor above and below the active region. You have metal layers where you have to inject current into the quantum well.

So you don't just have an InGaN active layer to invert, you have a GaN diode to turn on. You have to pump a LOT of current into the device, and you have to pump that current through metal, from metal into semiconductor, and then through the semiconductor. All those things add voltage. Even if everything is perfect: your contacts are ohmic, your semiconductor is perfect, all that: you still have a series resistance. That series resistance still adds voltage, because you're not pumping in 1 e-h pair for 1 photon, you're pumping in a lot of current to get a lot of light. A lot of current means a lot of voltage, because you have a series resistance. With GaN, you often have a lot of series resistance, especially because p-GaN is quite resistive, and you have to have a lot of it in a laser.

Then you get into things not being perfect. Non-ideal diode. Non-ohmic contacts (which means a Schottky diode for a contact), not likely in a commercial diode but possible. Detrimental heterobarriers in the structure. All these things add voltage, because they all make it harder to inject current.

Then there are plenty of non-radiative recombination paths that eat current, requiring you to pump more current in, thus requiring higher voltage. The crystal isn't perfect, you have threading dislocations and other crystal defects, which act as nonradiative recombination centers. In GaN, you also have other mechanisms like Auger recombination, which seems to be a pretty big deal, especially at high current densities like in lasers.

So that's probably the biggest effect you're seeing as far as voltage.

Does that make sense?
 
Last edited:

dyeote

0
Joined
Feb 9, 2012
Messages
131
Points
18
Yeah, it does make sense now:
- All of the series resistance (as you said, a lot of the material in the die is far from being a perfect conductor) that comes before and after the inversion bandgap within the actual quantum well, will explain voltage rising with current (V=I*R).
- Non-ideal contacts and detrimental heterobarriers are like little Schottkys in the series and explain the extra bias voltage (Vf > Ephoton)
- And all the non-radiative recombination paths mean that I'm not even guaranteed one photon per e-h pair I force through, even if it did manage to contribute energy at the inversion bandgap. It's statistical, so to get more of my precious photons I'll need to give many more e-h pairs :)

Wow, super-awesome answer pullbangdead! :thanks:

The only point which you didn't address directly is the rising wavelength with rising current issue; is this purely due to temperature? Will I see this phenomenon disappear if I (somehow) perfectly regulate the die's temperature?
 
Joined
Aug 25, 2007
Messages
2,007
Points
63
Oh yeah, that part too.

The quick, hand-waving, unqualified answers are: blueshift is from band-filling, redshift is from bandtail states. But that's completely unqualified, that's just always the first answer that comes out, whether it's actually right or not.

For a more complete picture, you have to remember that GaN is polar (there is a built-in electric field in the crystal structure) and piezoelectric (there is a strain-induced electric field in the crystal). Commercial products are on c-plane, and the InGaN well is strained because of lattice constant vs GaN, which results in an electric field across the quantum well, which tilts the bands in the well. This results in QCSE, the quantum-confined Stark effect, which affects the emission wavelength, among other things. Start some reading about QCSE, and you'll start to get some picture about the complications.

And temperature matters too.

Suffice to say it's complicated, and there can be different regimes where one effect or another is more important, and I don't remember half of it off the top of my head.
 
Last edited:

benmwv

0
Joined
Sep 10, 2010
Messages
1,380
Points
48
Very interesting!

Just curious, where did/do you go to school?
 
Joined
Jul 4, 2008
Messages
2,499
Points
113
Oh yeah, that part too.

The quick, hand-waving, unqualified answers are: blueshift is from band-filling, redshift is from bandtail states. But that's completely unqualified, that's just always the first answer that comes out, whether it's actually right or not.

For a more complete picture, you have to remember that GaN is polar (there is a built-in electric field in the crystal structure) and piezoelectric (there is a strain-induced electric field in the crystal). Commercial products are on c-plane, and the InGaN well is strained because of lattice constant vs GaN, which results in an electric field across the quantum well, which tilts the bands in the well. This results in QCSE, the quantum-confined Stark effect, which affects the emission wavelength, among other things. Start some reading about QCSE, and you'll start to get some picture about the complications.

And temperature matters too.

Suffice to say it's complicated, and there can be different regimes where one effect or another is more important, and I don't remember half of it off the top of my head.

This what I was trying to get at with the articles. The GaN is polar, also varying the amounts of InN GaN causes the emission to vary in wavelength, and there is additional shift from temperature, though slight. More GaN shifts the photons into the higher energy eV (UV) and higher indium shifts the output towards 450nm range. Also, this chemistry now is being used to generate 515.9nm with a C plane configuration.
 




Top