Multi-Fibre Laser

[Enthalpy]

Hello everyone and everybody!

Fibre lasers produce a high quality beam up to very few kW output and with excellent power efficiency. Nice to cut steel sheets for instance. But 2kW in D=100µm and e_r=2 make already 7MV/m, and the material's breakdown doesn't allow much more.

To obtain more power from several fibres and keep the phase coherence among them, I propose to synchronize the phase by evanescent waves. With a cladding thinner than usual, fibres side-by-side share some light, and if done often enough while the laser light is produced, I hope the bundle emits coherent light.

This shall be easier than a single oscillator driving independent lasers amplifiers before merging the beams, which needs difficult precautions to limit the phase drift over the long amplifiers.

The fibres alternate a certain number of amplifying sections, with colour centres and dissipation, and synchronizing sections where the evanescent wave drops slowly enough to couple the fibres. Short distances between synchronization avoid excessive phase discrepancy.

If the coolant has low optical losses and an index close to the fibres' one, it too can couple the fibres by evanescent wave. Then the synchronization sections can become mere space holders. Cladding at the amplifying sections reduces the light amplitude in the coolant and at the interface, if this is necessary against losses.

If the synchronization sections or space holders are very short, dissipation there can become acceptable, and the fibres can contain colour centres there too.

Pumping light can arrive from any end, or both, or from the sides. The coolant too can flow axially, though radial flow makes a greater flux or a lower speed, especially with four poles or more as in electric machines. Thin fibres are easily cooled.

The multifibre laser can oscillate or amplify. The pilot light might sometimes be fed in a single fibre, but I expect it will be spread most often. Mirrors at none, one or both ends are commonly Bragg gratings, more easily because synchronization is done elsewhere.

The multifibre laser needs not be straight, as long as the fibre lengths are identical enough. It can build a ring for instance, optionally with several turns.

I believe to have described this already on the Net. I someone knows where, please tell!
Marc Schaefer, aka Enthalpy

[swansont]

Evanescent coupling to do this is not a new idea.

[Enthalpy]

A first pumping scheme uses one fibre laser to pump each fibre of the multifibre laser.

The pumping light arrives in the narrow multiple lasing fibres, with single cladding or no cladding. Coupling by the evanescent wave is easier, and cooling the multifibre laser has no additional constraint by the pump.

Fibre lasers are already common. Just build many independent ones and cool them.

The multifibre laser receives pumping light, for instance at 1µm, that is not coherent from one fibre to the other, and makes light, for instance at 1.3µm, that is coherent across the fibres and can be well focussed.

A good power efficiency for this step resembles the wavelength ratio. The pumping fibre lasers had already converted light from laser diodes which themselves received electric power.

Two pumping lasers per fibre, one per side, make sense too. Splitting or merging the pumping light is also possible if somehow advantageous.

Marc Schaefer, aka Enthalpy

19 hours ago, swansont said:

Evanescent coupling to do this is not a new idea.

Thanks Swansont! Source?

[Enthalpy]

This illustrates coolant flowing transversally among the fibres:

With two "poles" (similarly to an electric machine) and pumping illumination from the sides, or six coolant poles and pumping unspecified. Notice the lack of fibres at the centre where the coolant stagnates.

A powerful laser needs much coolant throughput, transverse flow eases that.

==========

An evanescent wave through the coolant may be possible but is not easy.

Imagine the evanescent wave at vacuum 1330nm shall decrease by exp(6)~400 over transverse 120µm in the coolant. That needs transverse k=j314krad/m in the coolant. Propagating in silica at k=6.850Mrad/m from N=1.450, it needs 6.843Mrad/m in the coolant, or N=1.448.

Coolants with a matching index exist (N=1.449 for CCL4, mix it) and some will be transparent enough, but the index drifts with temperature in silica and in liquids. The close match must be kept despite heating.

Surrounding a fibre core with a material of similar composition, like a bit of fluorosilicates added to silica, keeps more naturally the tiny index difference over temperature.

Marc Schaefer, aka Enthalpy

[Endy0816]

Quote

 Splitting or merging the pumping light is also possible if somehow advantageous.

Think this is actually where I've heard of this from before, in the context of tapping fiberoptic cables.

[swansont]

4 hours ago, Enthalpy said:

Thanks Swansont! Source?

Don't have a specific cite. Probably a DAMOP talk, several years ago. 

[Enthalpy]

Oops : e^-6~1/400 coupling by the evanescent wave through 120µm coolant needs transverse j50krad/m that demand 1.45-1.449 96 difference between the refractive indices, impossible to stabilize.

========== Cooling figures

The following figures aren't optimum nor well chosen. They shall only shaw that some set of parameters looks feasible. I use data from the silica family, but fibre lasers prefer aluminosilicates. Gaseous nitrogen seems easy, but I compute with gaseous hydrogen.

• Colour centres can work under the synchronization sections too, as the overtemperature is small. This eases the manufacture.
• I compute with an even distribution through the core's depth, but concentrating them near the cladding as depicted would even the core's temperature hence light intensity and is made by just one deposition step more.
• The sync sections can share the cladding's material. Or maybe use the core's material, but then phase shifts happen.
• I suppose the fibre can receive the thicker cladding everywhere and be etched thinner outside the sync sections. A smooth surface matters: heat again?
• Sources don't tell it, but I guess some lossy material is needed around the cladding to force the single mode. Here it needs an intermediate low-index material in between.
• The low-index and lossy materials must be absent from the sync sections. Remove them mechanically? That's easier if they are organic. Or deposit them after the sync sections are assembled?
• I hope the sync sections can be sintered together, or maybe glued. Electric sealing would need Na⁺ ions to my knowledge.

Gaseous hydrogen at 300K 300bar has compressibility Z=1.19, 10110mol/m³, 20.4kg/m³, capacity 28.8J/mol/K 14290J/kg/K 291kJ/m³/K, viscosity 9.7µPa*s and 0.48mm²/s.

420 fibres, 2m long, shall deliver 420*140=58,8kW light and produce 420*70=29,4kW heat. The sides of a hexagonal pattern are 8+1 fibres wide (err, no, it's more), or 1.6mm.

54dm³/s hydrogen exit 2K warmer, so silica fibres whose index drifts by 8.7*10^-6/K propagate 1300nm light over 700µm with +-2.5° discrepancy to be evened out by the sync sections.

Hydrogen flows at 5.6m/s at the 3*2m*1.6mm ports and some 13m/s between the fibers. If the fibres didn't interact, then Reynolds ~1600 makes the flow transitional and C_d=1.0. Tripling the drag without reason to account interactions, the flow pushes 0.14N/m on the fibres. Stiff supports with 700µm period leave 5.5nN*m bending moment, so R=70µm and E=90GPa give 0.4nm deformation and 20kPa stress. The experimental 0.32 gauge factor over the volume changes the refractive index by <70ppb only.

1.4W/m/K silica conduction from r=0 to R=50µm and uniform 35W/m per fibre make the centre 2.0K warmer than the interface, so the index 17*10^-6 bigger amplifies the field by <2.0 over the outer 20µm radius. The algebraic solution would be better, and colour centres only near the interface would give uniform temperature and field. From r=50µm to R=70µm, the temperature drops by 1.3K, the index by 12*10^-6 and the field by <1.4 more than what both materials do.

The coolant resides some 8µs around each fibre so heat diffuses only 3µm deep, and turbulence shouldn't reduce that further. This coolant layer warms by mean 3K and this won't differ much among the fibres.

Arbitrary exp(6)~400 field attenuation over 2*40µm thick sync sections needs a transverse wave number j75krad/m. Indices of 1.45 et 1.450 080 achieve that, needing for instance 0.05mol% GeO₂ doping at the core, and possibly some compensation of the colour centres.

Marc Schaefer, aka Enthalpy

[Enthalpy]

420 fibres fit on a hexagonal pattern 13+1 fibres wide, not 8+1. The coolant poles are 2.3mm wide, so if keeping 5.6m/s there, 75dm³/s flow through the bundle.

300bar gaseous hydrogen exits 1.6K warmer at the pole centres so the phase mismatch between the fibers is smaller. The bundle could be upscaled, for instance by *3² for 530kW light at identical hydrogen speed, and more with faster hydrogen, as the fibers accept a stronger force.

==========

I use data from these sources:
nvlpubs.nist.gov silica's n versus T and P
springer.com silica's n versus doping
brennen.caltech.edu aerodynamic drag

Edited October 26, 2018 by Enthalpy