The maximum continuous current that the LED may draw is 100 mA.
If you use a linear power supply from the USB circuit you will not need more than 100 mA. At low duty cycles you will need proportionately less.
Operation at 100 mA continuous is liable to be risky.
If pulsing to 1A duty cycle of <= 1% is required = mean of 10 mA.
From data sheet:
The rated LED continuous current is 20 mA
The absolute maximum continuous LED current is 100 mA.
Lifetime can be expected to suffer at this level - plus see below re thermal ratings.
The maximum pulsed current is 1A but
this is at 1% duty cycle and with not more than 100 uS oulse.
As 1A x 1% = 10 mA mean the max continuous rating is higher.
Note that the seem to be somewhat serious about the absolute maximum continuous rating.
ie some data sheet writers put figures like that BUT they are inconsisent with other figures in the datasheet.
In this case the consistency of the other data suggests they mean what they say OR that they have cut and pasted the datasheet from elsewhere and changed some figures. Less likely here. The datasheet IS cut and pasted between their other IR LED products, but that's legitimate **. There is however some overlap in thinking between various parameters suggesting that walking to near the data sheet limits will get you (or the LED burnt. (eg see below - dissipation at 100 mA is for 1% duty cycle but 100 mA is abs max continuous current.)
Absolute max current is at 25C ambient.
This derates at about 27 C/Watt (BOTE calculation from fig 1.)
Power dissipation in free air <= 25C is 150 mW.
Max Vf at 100 mA = 1.4V typical and 1.8V worst case.
Dissipation will be 140 mW and 180 mW in each case *.
This is around the 150 mW figure.
The 100 mA dissipation figure is given for 1% pulsed output suggesting that they expect die heating to alter this adversely.
If the data sheet can be trusted, you may be able to operate this LED at say 80 mA continuous at 25C ambient in free air.
Thou may be able to pulse it at up to 1A at <= 1% duty cycle at <= 100 uS on time per pulse.
(* Ignoring the 1 mW or so that escapes as IR light)
Fig 8 - The word "relative" on the Y axis should be removed.
Fi6 6 - The radiant output / angle graph has been hand drawn and is not reliable. Nothing really looks like that. They were trying to fit the +/- 10 degrees half power points and then hand draw the rest.
** Cut and pastes:
This appears to be a cut and paste of Everlight ds..