How do they work? Femto.dc Current amplifiers (TIA, photo diode amplifier)

[bnevins]

I am wondering how they meet the noise specification.
Equ. Input Noise Current 65 fA/√Hz (@ 10 kHz)

Judging by other specs, they are using 10Meg feed back resistor.
So the 65 fA is expected, but 10 kHz seems quite an achievement.
Assuming Cin=1nF, the RC gives you ~ 16 hz (100 rad/s), yet the amp keeps the
noise of the R to 10kHz?

1) Could they do it with a Brisebois boot strap?
https://www.analog.com/media/en/reference-design-documentation/design-notes/dn399f.pdf
2) the quoted input noise 5nV/rt hz, is possible using an opamp.

3) They don't show a noise spectrum, and only spec the noise at 10kHz. 
|I have written them asking about the noise at 200KHz, so far no response.

4) The circuits shown in the data sheet don't seem to be consistent with bootstrap, since one terminal of the
diode is connected to shield/ground.

Any insights most welcome!

[David Hess]

I am wondering how they meet the noise specification.
Equ. Input Noise Current 65 fA/√Hz (@ 10 kHz)

Judging by other specs, they are using 10Meg feed back resistor.
So the 65 fA is expected, but 10 kHz seems quite an achievement.
Assuming Cin=1nF, the RC gives you ~ 16 hz (100 rad/s), yet the amp keeps the
noise of the R to 10kHz?

The input capacitance combined with the feedback resistance does not make a low pass filter because feedback reduces the input impedance to essentially zero.

4) The circuits shown in the data sheet don't seem to be consistent with bootstrap, since one terminal of the
diode is connected to shield/ground.

400 kHz of bandwidth is consistent with a simple transimpedance amplifier design.

[ejeffrey]

Transimpedance amplifiers gain bandwidth is not really affected by capacitance.  The input is a virtual ground, so the voltage on the device capacitance is essentially constant.  No change in voltage == no current flow.

However, the device capacitance does affect the noise gain when combined with the feedback resistor.   The voltage noise gain is just the standard formula for the non-inverting opamp: |1 + (Zf / Zg)|.  Here, Zg is the complex impedance of the input capacitance.  And in fact the data sheet you linked shows that the noise goes up with frequency.  At high frequency trans-impedance amplifiers are almost always dominated by voltage noise not current noise.