The invention relates to a method for optical detection of leaks in gas-tight housings, particularly of micro-electronic systems (MEMS), in which the deformation of the object surface (membrane) resulting from pressure impact on the objects disposed in a pressure chamber covered with a glass plate, and the subsequent reverse status change of this deformation are optically measured.
Such a method is described in the article “Optical Leak Testing of hermetic opto-electronic devices,” John W. Newman, 1995, although this is a method for testing opto-electronic devices, whereby the housings of these devices are covered with metal lids that are soldered, welded, or glued onto the housing opening. In order to prevent oxygen or other contaminants from getting on the contacts and precisely ground optical surfaces situated in the housing, these lids have to be applied to the housing in particularly sealed manner.
The object to be tested is laid into a pressure container that can be filled with helium. The helium now exerts a pressure onto the object, whereby the lid bends slightly into the interior of the housing. If there is now a leak, pressure equalization takes place over time, and the bending is caused to reverse.
The leak rate can be calculated from the time progression of the return to the starting position.
In the state of the art, the bending of the lid is measured using a digital holographic camera, whereby of course a great optical effort must be performed for generation of the hologram. The optical system, which consists of several beam splitters, lenses, and mirrors, requires precise adjustment and is therefore susceptible to shocks and contaminants.
The invention is therefore based on the task of conducting a method of the type stated initially with significantly less technical effort, and nevertheless achieving excellent measurement accuracy.
The invention accomplishes this task in accordance with the characterizing part of claim 1, in that the optical detection takes place by means of a profilometer that works without contact, using a chromatic confocal sensor, whereby the glass plate that serves as a cover for the pressure chamber is part of the optical system of the sensor.
The optical system of such a surface measurement device consists essentially of a polychromatic point light source (in other words no laser is required), a lens, and a dispersive plate disposed between lens and object.
To take the picture, all that is required is a semi-permeable mirror, which transmits the picture taken to a CCD camera, for example.
This optical structure, which is already quite simple, in and of itself, is simplified even further, according to the invention, in that the dispersive plate is now part of the pressure chamber, namely the lid of this chamber, which is configured as a glass plate.
The chamber is mounted on the table of the device, which table can be moved in the x-y direction. In this manner, it is now possible to scan the objects (MEMS) to be examined, which are present in multiple numbers on a wafer, which is disposed in the pressure chamber as a whole, point by point, i.e. object by object, whereby the deformation of the membranes or other surfaces of the objects to be tested, brought about by the gas pressure in the chamber, is measured. Such objects can be, for example, pressure sensors or acceleration sensors on a micro-scale.
The advantage of this optical-mechanical structure with the glass lid of the pressure chamber as part of the optical system of the confocal sensor lies in the fact that a very short work distance and a very great z (height) resolution exist. The glass lid, which should have a thickness between 5 and 10 mm, preferably 7 mm for use in the case of wafers, does not result in an increase in the work distance, since it is part of the optics. As part of the sensor, it also does not disturb the sensor itself, either.
It was possible to show that the mechanical attachment of the glass plate on the pressure chamber is sufficient. The bending of the glass plate, which is dependent on the pressure, and the inclination of the lid, which is dependent on the pressure, are eliminated by means of electronic image processing.
The method is conducted as follows:
The complete wafer is placed into the pressure chamber and the test is conducted fully automatically. All the hermetically sealed components on the wafer are individually scanned with a scanning time tS (between 10 and 20 sec), resulting in a scanning time tSW=tS*N, whereby N is the number of objects on the wafer. After the first reference scanning procedure, at a pressure of p0, the helium pressure is switched to pW. Afterwards, the scanning procedure is repeated over the complete wafer, specifically at least twice. Thus the complete testing time per wafer is twf=3*N*ts, and the helium impact time is tb=tSW=N*tS.
As an example, a typical wafer has 2400 objects, whereby a scanning time tS of 12 sec per object is provided. Since measurements are taken three times, in the present example, this results in a cycle time of 36 sec per object. This means that a helium impact time of 8 hours per wafer is required, and thus a high resolution for the leak rate at a total measurement time of 24 hours per wafer.
The leak rate is calculated from the time-dependent relaxation of the membrane deformation.
Since the method of operation of such a confocal sensor is known, this technology will not be discussed in greater detail.
As is evident from the right drawing of
As is furthermore evident from
The pressure regulator 11 is connected with a computer 12, in terms of data. The surface measurement device 1 is also connected with this computer 12, in terms of data, just like the table that can be moved in the x-y direction, which is furthermore controlled by the PC.
Number | Date | Country | Kind |
---|---|---|---|
10 2004 018 895.5 | Apr 2004 | DE | national |
Filing Document | Filing Date | Country | Kind | 371c Date |
---|---|---|---|---|
PCT/DE05/00795 | 4/15/2005 | WO | 12/4/2006 |