Atomically precise surface engineering of silicon CCDs for enhanced UV quantum efficiency
(Color online) Measured and absolute quantum efficiency for a back-illuminated, delta-doped silicon CCD. This CCD was not AR coated. Note that the data, once corrected for quantum yield, lies along the silicon transmittance curve. This indicates that the CCD is exhibiting reflection-limited response and 100% internal quantum efficiency. Data for an unmodified, front-illuminated CCD is shown for comparison purposes to illustrate the improvement in UV sensitivity that is achieved by the delta-doping and back-illumination processes.
(Color online) Cassini CCD that has been frame thinned to enable back-illumination for proof-of-concept demonstration of ultrahigh absolute UV quantum efficiency. Its frontside detection electronics are face down as pictured. This CCD is mounted in a picture frame package with a window cut to enable the backside illumination. The clover leaf pattern evident in the thin membrane is due to compressive stresses present in the processed silicon. It should be noted that an alternative thinning approach, currently under development at JPL, involves the use of a handle wafer to support the imaging membrane. This approach eliminates the deformation and is appropriate for high volume production of scientific grade CCDs.
(Color online) Demonstration of shadow masking approached used in AR coating experiments. A cleaved silicon wafer was placed in direct contact with the backside of a thin imaging membrane to block the deposition of the AR coating from a portion of the device. This shadow masking is used to ensure that coated and uncoated references were present within the same device.
Flat field images produced by delta-doped back-illuminated silicon CCDs when illuminated with 300 nm light. The top of each image comes from the half of the CCD that was coated with a sputtered film. (a) CCD half coated with sputtered MgO film. (b) CCD half coated with sputtered HfO2 film. The brightness of the image is proportional to the QE. Note that in each case, the side of the CCD that is sputter coated demonstrates lower quantum efficiency than the uncoated side.
(Color online) Quantum efficiency for the AR coated and uncoated sides of delta-doped CCDs. The AR coatings were applied by sputter deposition [(a): HfO2, (b): MgO]. Note that the quantum efficiency of the coated side of the CCD has worse performance, contrary to the calculations made using TF-calcTM that had indicated the utility of HfO2 and MgO at those thicknesses.
Flat field image taken under 270 nm illumination. The dark circles at the top and bottom of the image are regions that were coated with HfO2 using electron beam evaporation. It is believed that x-ray damage during the evaporation degraded the CCDs imaging capability in these regions. The bright rectangular regions are coated with HfO2 using thermal evaporation. The differences in the brightness of the two regions are due to different thicknesses of AR coating applied to each (25 and 40 nm, respectively). Changing AR coating thickness enables one to maximize performance at different target wavelength ranges (see Fig. 14 ).
(Color online) Flat field image produced by delta-doped AR coated back-illuminated silicon CCD when illuminated with 183.2 nm light. The top of this image comes from the portion of the CCD that is coated with a PE-ALD Al2O3 film. The brightness of the image is proportional to the QE. Note that the side of the CCD that is AR coated has higher quantum efficiency than the uncoated side. The shadow masking used is imperfect for this device due to the highly conformal nature of ALD deposition processes, which allows the coating to sneak under the shadow mask (see Figs. 2 and 3 ).
(Color online) Demonstration of the repeatability and control of the ALD technique. (a) Quantum efficiency for two individual delta-doped-CCDs coated on one half by 16.5 nm thick ALD Al2O3 AR coating. Device #1 was a conventional CCD, while device #2 is an electron multiplied CCD with gain. Note that there is no measurable quantitative or qualitative difference between the two devices when the measurement error (+/−5%) is taken into account. The depositions were carried out using the same deposition recipe but were separated by a month between coatings. (b) Comparison of modeled and measured performance of a 23 nm thick ALD Al2O3 AR coated CCD. Note that the qualitative QE behavior predicted by our model matches the measured performance.
(Color online) TEM image of PEALD HfO2 grown on a blank delta-doped silicon wafer. Note that the PEALD process forms an interfacial layer despite the presence of the native oxide that forms after the delta-doping process. The prevention of the formation of this silicate layer provides the motivation for the ultrathin (2 nm) aluminum oxide barrier layer grown for the HfO2 AR coatings in this study.
TEM images of a PEALD HfO2/Al2O3 bilayer grown on a blank delta-doped silicon wafer. Note that the presence of the thin (2 nm) aluminum oxide barrier layer prevents the formation of the interfacial layer observed Fig. 9 .
(Color online) High resolution hafnium 4f XPS scans for three hafnium oxide AR coatings. The samples were prepared to target HfO2 film thicknesses of ∼1.2 nm in all cases to enable the XPS to characterize the interface properties. The sputtered HfO2 deposited on silicon and the PEALD HfO2 grown on the 2 nm barrier layer of Al2O3 show Hf peaks consistent with the data observed for much thicker films. However, the presence of a low binding energy tail for the PEALD HfO2 grown directly on silicon suggests that a chemical interaction has occurred between of the HfO2 and the underlying silicon (see TEM in Fig. 9 ).
TEM images of a sputtered HfO2 thin film on a blank delta-doped silicon wafer. Unlike the case of PEALD HfO2 deposited directly on silicon, there is no evidence of a large silicate interfacial layer. The lighter layer at the boundary is likely the native oxide that forms after the delta-doping process. The sputtered HfO2 film appears less dense, amorphous, and rough as compared to the corresponding PEALD HfO2 bilayer (see Fig. 10 ).
(Color online) Optical properties of ALD and sputter deposited AR coatings. (a) Index of refraction of HfO2 and HfO2/Al2O3 bilayer coatings. Note that the ALD coatings have higher index of refraction over the NUV band (above 235 nm), which is indicative of a higher quality thin film as it is closer to the value for bulk HfO2. (b) Absorption coefficient, as modeled by UV spectroscopic ellipsometry for HfO2 and HfO2/Al2O3 bilayer AR coatings. Note that the ALD bilayer shows the lowest absorption over the wavelength range of 265–300 nm.
(Color online) Comparison of HfO2 AR coated delta-doped CCDs. The ALD bilayer outperforms the thermally evaporated HfO2 coatings at shorter wavelengths. This may be due to superior film quality, or the slight difference in thickness between the 23 nm ALD and 25 nm evaporated films. The sputter deposited HfO2 data are not shown here, due to its negligible UV response.
(Color online) Quantum efficiency for AR coated delta doped-CCDs compared against that of the uncoated portion of a delta doped-CCD and the multichannel plate onboard GALEX. Film type and thicknesses have been selected to optimize QE in defined wavelength regions. These wavelength regions are 170–200 nm (16.5 nm Al2O3), 190–240 nm (23 nm Al2O3), and 230–300 nm (23 nm HfO2/2 nm Al2O3) and were chosen based on the index of available UV transparent materials.
Performance of current typical UV detectors in major space missions.
Article metrics loading...
Full text loading...