Last week the new custom developed photodiodes that are used in the Lens R&D fine Sunsensors (BiSon and MAUS) have been tested at the Delft Reactor Institute (DRI) to levels beyond what would be required for any known mission.
Given the high radiation levels to be tested for the devices it had become impractical to perform regular Co60 tests for Total Ionizing Dose (TID) and Proton tests for displacement damage (TNID). Following discussions with ESA on how to best test the diodes it was decided to use 1MeV electrons instead. These tests are also used for solarcells which means that a significant amount of background information is available. Construction wise, Sunsensors are not unlike Solarcells in the sense that they use a silicon photodiode (like solarcells in the past) and a membrane (similar to a cover glass but much thicker).
One main advantage for Lens R&D is the vicinity of the facility. For the rest, the advantages and disadvantages of using this test method lay in exactly the same properties.
-high dose rate so short test duration
-tests both TID and TNID at the same time
The high dose rate is to the advantage of the test campaign as it means a short test time. It also means though that enhanced low dose rate sensitivity (ELDRS) is not tested
Testing both TID and TNID in one test means that the effects caused cannot be separated.
As per ESA standards one has to test at least two devices from at least three wafers, Which means we prepared six samples in total by building some new diodes in old BiSon64-ET-RH housings.These samples were then taken to DRI for irradiation.
As the 1MeV electrons generate atomic oxygen which is highly corrosive, the units are protected by a Kapton tent which is flushed with Nitrogen. In order to compensate for the energy loss, the beam dose rate is calibrated underneath this Kapton tent before the irradiation.
A previous batch of diodes had been tested up to 4.10^14 1MeV electrons with success, it has been decided to split the batch in two sub-batches of 3
3 diodes have therefore been irradiated up to 4.10^14 1MeV electrons, the other 3 to 8.10^14 1MeV electrons.
As a 1MeV electron lead to a ionizing dose deposit of 24.10^-9 rad/1MeV electron and a Non-Ionising Energy Loss (NIEL) of 3.14.10^-5 Mev.cm²/g this can be re-calculated to a TID of 19.2Mrad and TNID of 2.5.10^10 MeV.cm²/g.
Taking the 1mm Al equivalent shielding as provided by the window into account (0.65mm sapphire) this level of tolerance means that the sensors will build with these detectors will be able to survive any known mission with margin and can even be used without modification for long duration missions at 1100 or 1200km.
Although the TID and TNID effects cannot be separated, it is known that TID mainly leads to dark-current increase and TNID to displacement damage. The latter will cause a change in the quantum efficiency profile of the detectors and will lead to less current generation with increasing dose.
A dark current test has been performed on all six devices before and after the first irradiation as well as on the three devices after the second irradiation. These tests showed that the dark current directly after irradiation was still a factor three or more above the threshold value to be achieved after annealing. Therefore it can be confidently stated that the limit of the diodes has not yet been reached with these levels of radiation.
The effect of the displacement damage can be readily seen when comparing the quantum efficiency profiles before and after irradiation.
After the first irradiation it is obvious that the quantum efficiency has degraded especially in the near infra red region (NIR). The light blue line measured before irradiation shows a significant decrease for wavelengths longer than 900nm. The cut-off wavelength has shifted from approximately 1040nm to 940nm.
After the second irradiation the cut-off wavelength shifted down even a bit further to some 900nm. From these shifts it can be concluded that the shift is not linear with dose which is confirmed by previous measurements when the sensors were irradiated with 0.5/1/2 and 4 .10^14 1MeV electrons.
The loss in QE may seem significant, but it has to be born in mind that the sensors will be Sun illuminated which means that the higher energy densities are at the wavelengths which are least affected by the irradiation. The actual loss in generated current will be reported on at a later stage after the entire radiation qualification is finished (at this moment the sensors are in the prescribed high temperature annealing stage and further QE and dark current profiles will be taken to determine the recovery after annealing).
Although the diodes survived, the tests were not without further detrimental effects.
Testing up to 8.10^14 1MeV electrons has left it’s mark on the sensors.
The quantum efficiency profiles have changed, the dark current has gone up (both as expected) but also the Al2O3 ceramic substrate has been scourged by the radiation.
Where Alumina substrates are normally bright white, the radiation has given them a tan that is remarkably equal to a Sun tan. This despite the fact that Alumina is generally considered to be very inert. In actual practice the substrates will not see this level of radiation, but it goes to show that the irradiation had quite an impact and the test was quite severe.
Having shown their radiation tolerance and equipped with gold plated front and back contacts, the diode as produced by our supplier Micron Semiconductor Ltd in the UK are ready for space applications. As a result there is now a new verified source for radiation hard photodiodes in Europe. This claim can be made despite the fact that the diodes are still annealing. The annealing will only lead to an improved behaviour and can be used to judge if the extreme high dose rate has led to a worse performance. Furthermore, the annealing is expected to only improve upon the dark-current which is already well above threshold. Annealing generally doesn’t affect TNID related QE profile variations so isn’t expected to change much in that respect.
This leaves the question about low dose rate sensitivity. ELDRS testing is normally performed at 360 rad/h and testing up to 19Mrad would have taken >50.000 hours and is therefore not feasible. The very high margin shown as to the project requirements for various projects however lead to the conclusion that even some ELDRS will not lead to any issues during any application and ELDR testing will not add any confidence in the application.
Dark-current measurements are a means of assessing the quality of the diodes, but it should be remembered that during normal use the diodes are in 0 bias. Therefore the dark current has no effect on the operation but for a negligible thermal noise increase. Furthermore being used as fine Sunsensors means that the common mode QE decrease will be automatically cancelled by the angle calculating algorithm which takes the ratio of the generated currents. As a results, fine Sunsensors based on these diodes can be expected to provide a very high quality Sun angle output over a very long period of time.
As the same diodes are used for the BiSon as well as the MAUS Sunsensors, the MAUS is unbeaten as for radiation tolerance by any other Sunsensor for cubesat applications.