This is gonna sound like babel, Seriously: begin nanosized, quick, dissertation. This will take 20 minutes to type and a year to edit, and I dont have a year.
One thing you need to know, lasers are supposed to be monochromatic, but that is just a concept. They often have a broad range of gain bandwith, and diodes have perhaps the most of all, except for organic dyes. So in a two mirror cavity, there are N possible gain path lengths described as multiples of N* C/2L where C is speed of light, L is the length of the cavity and N is a integer. Each one corresponds to a slightly different frequency. There are various means for narrowing this down to one longtudinal mode in a cavity and thus true single frequency operation.
A free running diode is anything but single longitudinal mode. 10 of them running in parallel is a mess on the spectrum analyser several nanometers wide and hopping around like crazy.
So he's got a spread spectrum source. This is bad.
Ok, the 514.5 line of the argon laser (roughly 600 terahertz) has a gain bandwidth of say 10 gigahertz and I could lock it with a temperature tuned etalon element intracavity to less then 1 mhz, and lock it to a adsorbtion line in iodine vapor. That doubles easily, for it is very much single frequency.. His diode has a gain bandwith of upwards of a nanometer or two , and nothing to stabilize it. It has a coherence length, ie length over which interference can occur, of maybe 1 to 2 cm when unstabilized. My argon when stabilized had a coherence length of 10s of kilometers, a free running argon is typically 30 cM.
UH, um, Type I and Type II phase matching for SHG. Where to begin.
Required Course Materials:
Download PSST! from St Andrews and Download SNLO from LLNL.
Book is Quantum Electronics, 3rd Edition by Amnon
Yariv.
PSST is worth having anyways as it models many laser parameters. SNLO is a LSD TRIP through ABCD matricies and nonlinear mixing and heterodyning of laser light. YARIV is incomprehensible math.
Ok, we've established that a big diode has lousy beam quality and very, very poor frequency stability. It has a nasty coherence length.
To get enough energy density for SHG, unless you have a Qswitched or other ultrashort pulse laser (think of it as dumping a fast cap) , you need to be intracavity to get enough energy to drive SHG forward. SHG probably only existed in the stars until lasers came along.
So if he just focuses the diode array in the doubling crystal, he might get 10-20 mW of violet TOPS for 60 watts of pump. Its going to be unstable, because diodes frequency hop as much as .1 nm/'C, and back reflections cause heating of the diode.
Also your type I phase matched doubling crystal only works well over a few microradian of input angle and that is selected by wavelength. And in type one, the Poynting Vector is off the doubling input angle by a few milliradian, the doubled light comes off at a odd angle to the pump light. This causes issues when you try to extract it. So you have huge mechanical stabilty issues, most of which cancel out intracavity, but cant be cancled extracavity very easily. Type II is more forgiving, but requires more power, but it doesnt have the walkoff. Type I crystals are within budget, type II are very expensive.
Of course the diode cavity is too small for him to access, and there is a array of diodes there, each at their own temperature and own wavelength, though sometimes they do lock up. Since each is at a slightly different wavelength, each wavelength starts its own doubling process to a degree, and since doubling basicly occurs because of writing a pattern in the electron orbits in the lattice 1/2 wavelength long, having multiple wavelengths fuzzes up the pattern and effiency goes to hell in a bandbasket, if doubling occurs at all.
If you have a single frequency laser, that pattern in the electrons is rock stable, and doubling occurs easily. If you have a mess of wavelengths, they fight each other and cancel out.
If he couples the diode array into a 4 mirror bow tie resonant cavity, and puts the crystal at the bow tie crossover , he can start to get enough energy to really double. But due to the wavelength spread, its damn unstable and will fluctuate like mad. And he gets noise, but no power.
This is usually solved in the lab by ordering a diode missing one outer faucet and using a grating as a littrow reflector to tune the diode. Heck, you can reflect the order from a grating into one of these 660 CD player reds and tune it about 2.5 nm from center, its done all the time for spectroscopy purposes in labs
Pound Drever Hall is a scheme to lock the diode wavelength by locking the ideal diode wavelength to a external block of quartz that acts a Febry Perot cavity, and thus basicly the optical equvalent of a FM discriminator for frequency purposes. This gives him a voltage proportional to wavelength of the diode. Since he can't change the diode cavity length of the tiny diode chip, he can change the bow tie cavity length with a Piezo to match the aggregate diode wavelength. Its called a "Tweeter" mirror. Call it inverse pound drver hall locking. In other words you scan the
the position of one mirror to lock the cavity resonance to the mess of stuff coming out of the diode. Then he gets good conditions for doubling.
So he needs to achive resonance to gain power and he needs to track the resonace to the unstable diode. This is backwards to the normal way of tuning a single diode to match the crystal,
See
http://www.toptica.com/products/item...56/bow-tie.jpg
http://www.photonics.com/Content/Rea...ArticleID=7599
http://members.misty.com/don/verdi1.gif
END INCOHERENT TRAIN OF THOUGHT BABEL.
Steve