Patent Application
20100123082
This invention relates generally to electron microscopy.
Image capture on an electron microscope has traditionally been accomplished by exposing the electron image directly onto film and then processing and digitizing the film or by impinging the electron image onto a scintillator which converts the electron image into a light image transferring that image using light optics either in the form of a fused fiber optic plate onto an image sensor and then detecting and reading out that light image by the image sensor, or by first reflecting that image off a metal mirror that acts as a low backscatter beam dump and transferring that image using light optics in the form of a transfer lens onto an image sensor and then detecting and reading out that light image by the image sensor.
Both techniques are non-ideal. Film suffers from incomplete coverage of the detection area by the granular detection medium and from a relatively high admixture of noise from grain development in the case of low exposure dose. A fiber-optically coupled scintillator suffers from grain noise and from imperfect resolution of the optical coupling. In addition, the two types of optical coupling have addition drawbacks. With fiber-optical coupling which also acts as a substrate for the scintillator, electrons which have traversed the scintillator scatter in the substrate and return to the scintillator to produce additional confounding light usually at some distance from entry. Lens-coupling with a low-Z fold mirror avoids electron backscatter but suffers from the poor optical collection efficiency of even the high-numerical aperture lens suitable for image transfer over a large area through a folded optic with high resolution.
The non-ideality of traditional electron image capture methods (granularity, heaviness and poor collection efficiency) can all be addressed through the use of a low-Z semiconductor as the primary detection medium. Semiconductors such as Si have low Z compared with scintillators such as Gadolinium oxy-sulfide or YAG, are uniform through their volume in terms of efficiency at generating electron-hole pairs, and are significantly more efficient than indirect coupling methods at collecting the secondary electrons than optical coupling means which have numerical aperture and absorption issues along the coupling chain. There are several ways known to segment the volume of the detector so that lateral resolution is conferred, all relying on implants, oxides and gate structures driven to given voltages to shape the electric fields in the detection volume such that generated electrons are swept into pockets isolated one from another without dead spaces between the swept volumes. Illustration of these ways are provided by the examples of charge-coupled devices (“CCDs”), charge injection devices (“CIDs”), and CMOS active pixel sensors (“APSs”).
Direct electron detection by semiconductor has difficulties as well, however. First, there is the issue of damage to the sensor. Ionization caused by the incident high-energy electron which occurs in the insulating layers of the device instead of in the charge-collecting parts, causes fields which alter functioning of the device, either by creating charge traps which reduce charge transfer efficiency or increase lag or by inducing increased dark current generation. Knock-on damage caused by electrons above approximately 150 kV causes dark current spikes. This damage mechanism is considerably less important in an electron detector than ionization, at least for electron energies below 500 kV. Both forms of damage have the strongest effect on CCDs due to the heavy reliance in a CCD on charge transfer. Active pixel sensors are considerably more robust than CCDs and for this reason have been used more and more extensively for high-energy particle detectors. Recently, Active Pixel Sensor (“APS”) detectors have been used in electron microscopy as well. However, data shows a limited range of usefulness due to damage mechanisms in the pixel. U.S. Pat. No. 7,262,411 shows significant rise in dark current with only 500,000 electrons of total illumination, requiring special measures such as beam-current-interlocked shuttering to limit exposure and prevent accidental exposure and annealing to release insulator-trapped charge to maintain usefulness of the detector over time. All references cited herein are incorporated by reference. The need for many tens of thousands of electrons per exposure in some applications and the rising demand for thousands of images in many applications drives a need for greater robustness than can currently be offer by APS technology.
Second, there is the problem of backscatter. Charge collection can be limited in vertical extent by the use of a thin epitaxial layer. This layer is typically between 5 and 20 um in thickness. For electron energies above 100 kV, the scattering of an incident electron will spend a significant portion of its time after having traversed the collection volume of the epitaxial layer, with many electrons returning to generate electrons a second or third time, usually at a significant distance from the original point of entry. This creates the exact same source of noise for electron imaging at kVs above 100 kV that occurs with a thin scintillator deposited on a fiber-optic substrate. However, since low-Z materials also allow the electron to travel far before stopping, the lateral spatial extent of scatter can be significantly higher than for a fiber-optic substrate.
Further problems include pre-scatter and oversensitivity. Pre-scatter: At lower kVs, the extensive structures over the APS pixel often as thick as 8 um, cause SNR degradation and resolution loss before the electrons even enter the charge-collection region. Oversensitivity: One of the advantages of direct electron detection by a semiconductor can also be a disadvantage. The number of electrons generated by an incoming primary can be so high as to saturate the charge storage capacity of a pixel after only a few primary electrons have been detected.
A method of acquiring an image formed by electrons in a sensor having a substrate side and a front side is disclosed, comprising exposing the semiconductor image sensor to electrons that approach the sensor from the substrate side. In a further embodiment, the electrons are decelerated prior entering the sensor. In a further embodiment, the deceleration is achieved with lens electrodes. In a further embodiment, the deceleration is achieved by placing a material, such as a scintillator between the source of the electrons and the sensor. Examples of suitable sensors include CCDs, CIDs, active pixel sensors and sensors not having a buffer amplifier for each charge collection pixel. In a further embodiment, the sensor is periodically annealed to remove charge build-up. In a further embodiment, the method includes shuttering the sensor to reduce damage during said exposure.
In an embodiment, electron back-illumination of any semiconductor image sensor is disclosed. Thinning or full depletion would allow back-illumination by electrons with still-complete charge collection provided the electron energy were low enough. This condition would apply to special applications requiring electron imaging at low kV and also to the sensor in a decelerated camera. Deceleration of electrons prior to entering an image sensor is described in Mooney U.S. Pat. No. 6,414,309 (Mooney), and in U.S. Pat. No. 5,998,790 (Downing). All references cited herein are incorporated by reference. Electron back-illumination completely eliminates electron exposure of oxide structures on any of the above-mention devices (CCD, APS, CID) or the passive-pixel sensor mentioned in an associated disclosure by the applicant, filed concurrently with this application and claiming priority to provisional application Ser. No. 61/115,821, thereby reducing damage mechanisms to effectively zero since knock-on damage does not occur below ˜150 kV. With back-illumination, the other non-damage-related problems with direct semiconductor detection listed above could also be addressed. Prescatter is eliminated because electrons are no longer traversing the overlying electronic structures to reach the electron collection volume. Because lower kV electrons generate tracks with smaller spatial extent and deposit less energy in the form of electron-hole pairs, the above-described problems of backscatter and oversensitivity are addressed by virtue of the lower kV requirement for the applicability of this design.
This utility application claims the benefit under 35 U.S.C. §119(e) of Provisional Application Ser. No. 61/115,837 filed on Nov. 18, 2009 and entitled Method for Electron Back-illumination of a Semiconductor Image Sensor, the entire disclosure of which is incorporated by reference herein.
| Number | Date | Country | |
|---|---|---|---|
| 61115837 | Nov 2008 | US |
