JPH0220039A - Evaluating method for single event resistance of semiconductor element - Google Patents

Abstract

PURPOSE:To secure a region having weak single event resistance and to calculate the sectional area of a single event phenomenon due to the sum of the areas of the regions by forming the same connecting structure as that of a real element, elongating or contracting a flying stroke by ion energy change, isolating and evaluating charge collecting step. CONSTITUTION:When energy of alpha-ray is varied and radiated by regulating the vacuum degree of a vacuum chamber in which a 241Am beam source and a sample are secured, for example, to an N<+>-P bond formed on a P-type substrate and a charge collecting step is discussed, its incident energy is proportional to the charge collection amount in a region A (less than 1Mev), and a funneling step in the P-well is generated. A diffusing step in the P-well occurs in a region B, ranged at least 3mum. In a region C, the charge collection amount is reduced by half due to the ion shunt phenomenon of the N<+> type substrate. Further, one responsive region P<+>-N junction is discussed, compared with the magnitude of the charge collection amount, a region in which single event resistance is deteriorated is secured, and the sectional area of the single event phenomenon can be calculated by the sum of the areas of the regions.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a method for evaluating single event resistance in a semiconductor device, and particularly to soft error and latch-up resistance evaluation of a silicon semiconductor device.

[Conventional technology]

Semiconductor integrated circuits mounted on artificial satellites have become larger in capacity and more highly integrated as satellites become more sophisticated and performant, and there is therefore a trend towards higher density and miniaturization.

However, miniaturization of semiconductor devices reduces the amount of information charge per bit, making them susceptible to damage due to soft errors caused by incidence of charged heavy particles and single events such as latch-up. Therefore, the single event phenomenon has come to be recognized as a serious problem in space electronic equipment.

Next, regarding the mechanism of single event phenomenon, N”
An example will be described in which a charged particle is incident on a -P junction. FIG. 6 shows the time course of the charge collection process. Figure 6 (a
) The electric field in the depletion layer is distorted by the incidence of charged particles, and the depletion layer is substantially elongated (FIG. 6(b)). Then, the electric charges in the elongated electric field are collected by the electric field (funneling process), and the strain in the depletion layer is eliminated. This process is known to be a fast phenomenon that occurs within several hundred picoseconds. Next, as shown in FIG. 6(c), charges (N, electrons, and holes) remaining in the silicon substrate diffuse and reach the depletion layer region, thereby causing charge collection. This process is called a diffusion process and depends on the range of the charged particles and lasts several nanoseconds to several tens of nanoseconds. Charges are collected in the junction structure through the two processes described above, namely the funneling process and the diffusion process. If the amount of collected charge is sufficient to cause memory inversion or latch-up, a single event phenomenon will occur.

In actual semiconductor devices, the area of the junction is small, and the amount collected by the diffusion process is small. Therefore, substantial charge collection occurs primarily during the funneling process.

Establishment of a method for evaluating the single event resistance of semiconductor devices as described in the present invention is important both for determining whether or not to actually mount a semiconductor device on a satellite, and for considering methods for enhancing resistance.

Conventionally, single-event resistance evaluation has been performed by irradiating an actual semiconductor device with charged particles emitted by an accelerator or the like and observing the number of memory inversions. Although this method is an essential procedure for the final evaluation of each individual satellite-mounted device, it does not provide information about the mechanisms that occur inside the device, that is, the charge collection process due to funneling or diffusion.

Therefore, in order to study the charge collection process inside the device, the single event phenomenon mechanism is being investigated by forming a junction structure that simulates an actual device and observing the amount of collected charge.

[Problem to be solved by the invention]

The conventional evaluation method using the junction structure described above has the disadvantage that when the junction area is made smaller in order to reduce the charge collection process due to the diffusion process, the number of charged particles incident on the junction is significantly reduced, which requires a large amount of time for observation. There is. Therefore, a junction with a relatively large area of 100 μm×100 μm or more is used, but in this case, two components, the funneling process and the diffusion process, contribute.

In order to separate these two processes, attempts have been made to examine the temporal changes in the amount of charge collected and to separate the fast component from the funneling process and the slow component from the diffusion process.

However, the funneling phenomenon is a fast phenomenon (on the order of several hundred picoseconds), and the reliability of the measured values of the time and amount of charge is low. Furthermore, the measuring equipment becomes expensive because it tracks high-speed phenomena.

[Means to solve the problem]

The single-event resistance evaluation method of the present invention separates the charge collection process by varying the energy and range of the incident particle and evaluating the total amount of charge collected on the junction using a highly accurate pulse height analysis measurement system. It is something to do.

That is, in a region where the energy of charged particles is small and their range is shorter than the funneling length, charge collection occurs only by the funneling process, and the incident energy and the amount of charge collection are proportional. As the energy increases further and the range becomes longer than the funneling length, some of the charge collection occurs by diffusion. However, collection through the diffusion process is less efficient than the funneling process. Therefore, the amount of charge collected is not proportional to the incident energy, and the slope gradually decreases.

If the energy is further increased and the particles reach a depth where no charge collection occurs, the amount of collected charge will hardly increase even if the ion energy is increased.

By expanding and contracting the range by changing the ion energy as described above, it is possible to separate three regions, namely, a region where a funneling process occurs, a region where a diffusion process occurs, and a region where no charge collection occurs.

Furthermore, by forming a junction structure that is the same as that of the actual device and separately evaluating the charge collection process using the above method, it is possible to fix the region with weak single event resistance and derive the cross-sectional area of the single event phenomenon from the sum of the areas of that region. It becomes possible to do so.

〔Example〕 Next, the present invention will be explained with reference to the drawings.

FIG. 1 shows the results of one embodiment of the present invention.

An N''-P junction (area 500 μm x 500 μm) fabricated on a P-type substrate (impurity concentration 5 x 1015 cm-') was irradiated with alpha rays and the amount of charge collected was observed.The alpha rays used here were Americium 241 (241A
The initial energy of the charged particles emitted from m) is approximately 5.
5 M e V, range is approximately 33 μm. The ion energy of the α rays can be arbitrarily changed, for example, by adjusting the degree of vacuum in the vacuum chamber in which the 241 Am ray source and the sample are fixed.

Figure 2 shows how the incident alpha rays (energy 5.5 MeV) lose energy and the stopping power value (energy lost per unit length).

Furthermore, the range of α-rays with lower energy can be determined by moving the origin of the horizontal axis to a predetermined α-ray energy value. It can be seen that the stopping power value increases as the energy of the α-ray decreases.

As shown in FIG. 1, charge collection is performed in three regions I.

It can be divided into ■ and ■. Region I is a region where only the funneling process occurs, and the incident energy and the amount of charge collection are proportional, and the slope thereof is 0.9. Therefore, most of the electron-hole pairs generated in the funneling length are collected at the junction electrode. The maximum incident energy value in this region is 1.8-2.
Funneling length is 7.0~8.0μ from 0 M e V
It is expected to be about m. Region ■ is a region where both the funneling process and the diffusion process occur, and the incident energy value is ~3.
It is thought that the region where the diffusion component occurs extends to about 17 μm from 75 MeV. Further, region m is a region that hardly affects the charge collection process and corresponds to a range of 17 to 33 μm.

As described above, by using this method, it becomes possible to easily separate the charge collection mechanism based on the N''-P junction.

Next, the evaluation method of the present invention will be applied to a structure simulating a 0MO8SRAM. FIG. 3 shows the positions of the cell portion and sensitive region of a CMO8SRAM, and FIG. 4 shows an example of the cross-sectional structure of a MO8) transistor.

FIG. 5 shows the results of examining the charge collection process due to α rays incident on the position shown in FIG. 4 (the drain portion of the N-channel transistor).

The area is divided into three parts. That is, area A (~I
MeV), the incident energy and the amount of charge collection are proportional.

Therefore, it can be seen that a funneling process occurs in the P-well portion (funneling length ~3 μm). Region B is considered to be a region where the diffusion process in the P-well occurs, and its range is thought to be 3 to 5 μm. In region C, the amount of charge collection decreases rapidly, and the amount of charge collection is halved. This is due to the ion shunt phenomenon (electrons generated along the trajectory) with the N+ substrate.
This is because a phenomenon in which a hole pair is divided into two N+ regions has occurred.

As described above, the charge collection mechanism can be divided by changing the energy of the α ray and observing the amount of charge collection.

Furthermore, by considering the other sensitive region P''-N junction (P+/N well/N epi/N+ substrate structure), it is possible to compare the amount of charge collection and fix the region with poor single event resistance. .

Furthermore, the cross-sectional area of the single event phenomenon can be derived from the area sum of the area.

Further, although each of the above embodiments uses α rays, the same method can be implemented using other charged particles. At this time, the ion energy can be changed by adjusting the vacuum level of the reaction chamber, introducing other gases, closing the reaction chamber with a thin metal film, etc.

〔Effect of the invention〕

As explained above, the present invention expands and contracts the range of ions by varying the ion energy of charged particles, and by observing charge collection by a junction structure that simulates a real device, the charge collection mechanism is isolated and evaluated. There is an effect that it can be done.

As a result, it becomes possible to fix a region that is vulnerable to a single event, and furthermore, it becomes possible to predict the cross-sectional area by calculating the sum of the areas.

[Brief explanation of the drawing]

FIG. 1 is a graph showing the relationship between the incident energy of α-rays and the amount of charge collected according to an embodiment of the present invention. Area I: Area where the funneling process occurs, Area I: Area where the diffusion process occurs, Area ■...
...A region with a small amount of charge collection. FIG. 2 is a graph showing the attenuation of the ion energy of α rays and the change in stopping power. Figure 3 is an equivalent circuit diagram of 0MO3SRAM, Figure 4 is 0M
O8) is a diagram schematically showing the incidence of charged particles on a transistor. FIG. 5 is a graph showing the relationship between incident energy and charge collection amount when α rays are incident on a junction structure simulating a 0MO3) transistor structure. Region A: region where the funneling process occurs in the P-well region, region B: region where the diffusion process occurs in the P-well region, region C: region of the substrate N+ region. A region where the ion shunt (splitting of electron-hole pairs) phenomenon occurs. FIGS. 6(a) to 6(c) are schematic diagrams showing the charge collection process due to the incidence of charged particles. Agent Patent Attorney Uchihara Uchihara Otoku I's Nutopisi 2' 8゛wa (MeVβC) Deaf N (A aH)-, Bill's responsibility ding'n ω) Ac < the incident energy N'- (,'4e :V) Figure 3 Figure 4 Incident energy without MeV (MeV) 5th problem

Claims (2)

[Claims] (1) In single event resistance evaluation of semiconductor devices, a junction structure simulating a semiconductor device is irradiated with charged particles while changing the energy and range of the charged particles within the semiconductor, and the amount of charge collected is measured. A method for evaluating single event resistance in a semiconductor device, characterized in that: (2) A method for evaluating single event resistance in a semiconductor device, characterized in that the mechanism of the charge collection process is separately evaluated by measuring the amount of charge according to claim 1, and the single event resistance and single event generation cross section of the semiconductor device are derived. .