a = acceleration voltage, V C = capacitance, F Cos beta = cosine of impact angle Dia = diameter of projectile, micrometer d = projectile diameter, m E = energy of capacitor discharge, J KE = kinetic energy of projectile, J K1 = preliminary detector sensitivity parameter, (g)1/3 (km/s) K2, K3 = preliminary detector sensitivity parameter, (micrometer)(km/s) Mass = projectile mass, (g) m = projectile mass, (kg) p = momentum of projectile, (kg)(m/s) q = charge on projectile, C S = detector sensitivity parameter, (micrometer)(km/s)(micrometer)^-0.875 Thck = detector thickness, micrometer Vel = projectile velocity, km/s v = velocity of projectile, m/s Vb = detector bias voltage, V Vs = detector signal voltage, V beta = angle between velocity and detector normal, deg rho = projectile density, kg/m3 tau = time interval, s
The most basic and probably the most important measurement made concerning dust particles in space is a determination of the flux (the number of particles encountered per unit time by a spacecraft in a particular orbit). Ideally, particle flux would be measured as a function of the mass, density, velocity or other characteristics of the particles. Usually this is not possible and compromises must be made. The flux can sometimes be measured as a function of the type of damage or hazard associated with the particles counted. The pressurized cell, for example, has been used to measure fluxes of larger particles which represent a puncture1). However puncture is not the only meteoroid or debris hazard. Particles which cannot damage the spacecraft can damage critical surfaces. If such particles exist in large numbers, they will limit the useful life of optical surfaces, windows, thermal coatings or other spacecraft components that depend on fixed surface properties. The capacitor type detector has been developed to make time resolved flux measurements of small meteoroid and debris environments in space.
The capacitor type detector2) is a relatively simple device consisting of a parallel plate capacitor with a very thin dielectric. The top metal plate or electrode is made very thin and is the surface of the capacitor that is exposed to the impacting particles. The device is operated with an electrical potential (bias voltage) applied across the capacitor plates, hence a charge is normally stored in the capacitor. When a particle impacts the exposed thin plate with enough energy, it can cause the dielectric to breakdown and result in an internal discharge of the capacitor. The event can be measured by monitoring the charge required to recharge the capacitor. The mechanism responsible for the discharge is complex and will be discussed further in a later section. The sensitivity of the detector depends on a number of factors such as the dielectric thickness, the top electrode material and thickness, and the applied bias voltage, as well as the physical properties of the impacting particle.
Some calibration work has been done on detectors of this type and reported in the past3). In that work, limited tests were performed using the Goddard Space Flight Center accelerator (a two million volt accelerator) which is similar to the Langley Research Center accelerator described later and used for the present work. In that work the researchers investigated metal-dielectric-silicon detectors using silicon dioxide and silicon nitride as the dielectric with thicnesses ranging from 0.4 mirometers to 4.0 micrometers. They also investigated both Al and Au metal electrodes of different thicknesses ranging from 0.05 micrometers to 0.2 micrometers. The results of their test were very encouraging. However, the number of tests they were able to perform were limited and indicated the need for a more extensive study to determine the calibration needed for these type of detectors.
The purpose of the present work has been to repeat some of the characterizing experiments previously reported4) on the metal-oxide- silicon (MOS) capacitor detectors as well as to greatly expand the number of tests with the aim of developing a more accurate calibration of the detectors. Based on our results coupled with the previous work, it has been possible to design and carry out a very successful experiment (Interplanetary Dust Experiment) on the Long Duration Exposure Facility (LDEF)5).
DETECTOR FABRICATION
The technology used to fabricate the detectors is based on the silicon
technology widely used in the manufacture of microelectronic devices and
circuits. Figure 1 shows a cross-section and
Figure 2 a photograph of a sensor. The substrate
(bottom electrode) of the detectors used in these studies
consisted of silicon wafers, two inches in diameter, and polished on
one side (top side or sensitive surface) by a combination of mechanical
and electro-polishing. The silicon wafers were p type with resistivities
of less than 0.01 Ohm cm. The fabrication process consists of the following
steps: (1) substrate clean, (2) thermal oxidation of the silicon wafer to
form the capacitor dielectric, (3) removal of a portion of the oxide on the
unpolished (back) side of the wafer to gain access to the silicon to form
an electrical contact, and (4) evaporation of metal (normally aluminum )
on the top and bottom sides of the wafer to form the top capacitor electrode
and electrical contact to the silicon wafer.
Wafer cleaning consisted of a typical RCA clean followed by a rinse in
de-ionized water and dried using dry nitrogen. The thermal oxide was
grown in a standard oxidation furnace using a quartz tube. A dry-wet-dry
oxidation process was used in which the wafer was first exposed
to dry oxygen for five minutes followed by a wet cycle for the appropriate
time (the time being chosen to give the desired oxide thickness) which was
then followed by an additional five minute dry cycle. The wet process was
used to reduce the oxidation time over that which would have been required
if dry oxygen were used for the total oxidation cycle. Typically the
oxides were grown at 1100° C.
Following the oxidation, a hole was etched in the oxide on the back side of the silicon wafers by placing them on a flat surface with the unpolished side facing up. A crystallite of ammonium bifluoride was then placed in the center of the wafer. A drop of water was then placed on the crystal. The water added to the ammonium bifluoride crystal forms hydrogen fluoride ions which etch the oxide. The area removed depends on the size of the crystal and the amount of water added. The typical etch area was approximately one centimeter in diameter. Following the etch, which takes approximately five minutes, the wafer was thoroughly rinsed in de-ionized water. The wafers were then dipped in acetone and then in hot Transene(TM). This was followed by drying with dry nitrogen. The wafers were then loaded into a vacuum evaporator where the backside metal was evaporated on to the wafers. This step was followed by turning the wafers over and evaporating metal on the front side. A shadow mask was used to define the metal and an un-metallized edge was left on the wafers which was approximately 0.075 cm. Following metallization the wafers were annealed in dry nitrogen at 200° C to sinter the metal. A variety of dielectric thicknesses were investigated ranging from 0.4 to 1.7 micrometers. Aluminum with a nominal thickness of 0.1 micrometers was used for the top and bottom metal.
After completion of the fabrication process the detectors were tested electrically by connecting them to a curve tracer to measure the internal leakage of the capacitor. Typically the capacitors were found to be shorted at this stage. However, the shorts could be removed by connecting a resistor (5-100 Kilo-Ohm) in series with the capacitor and turning up the voltage on the curve tracer. As the voltage is increased, Joule heating, caused by high currents in the defect regions, burn (evaporate) away the thin top electrode metal in the vicinity of the defect and hence the defect region is removed from the electrical circuit of the capacitor. Also as the voltage is increased to higher values weak spots in the capacitor tend to breakdown again burning away the metal electrode at the point of breakdown. The maximum voltage that the capacitor can withstand depends on the dielectric strength of the silicon oxide which is approximately 10 MV/cm. When breakdown occurs it is usually accompanied by a flash of light which can easily be seen by the naked eye. The voltage used to clear a capacitor detector was chosen based on the thickness of the dielectric.
The major advantage of the electrostatic accelerator was that the projectiles were charged and therefore easy to measure and control. After being accelerated by the Van de Graaff accelerator, the projectiles entered the instrumented control section of the beam line. In this section of the simulator, a computer controlled the projectile and recorded the measurements made on the particle. The time of flight for each projectile was measured over a 1.0 meter flight path. If this time was within a pre-selected range, an electronic gate was opened to allow only that projectile to proceed to the target. For each projectile selected, out of the many accelerated, three measurements were recorded: (1) the charge 'q' on the projectile, (2) the time of flight 'tau' over a 1.0 meter flight path, and (3) the acceleration voltage 'a'. From these three measurements, projectile parameters such as kinetic energy 'q a', velocity 'v = 1/tau', mass '2 q a tau^2', and momentum '2 q a tau' were calculated. The projectile diameter 'd = (12 q a tau^2/(PI rho))^1/3' was also calculated assuming a spherical shape and uniform density (rho = 7.86 gr/cm^3) of the projectile.
There were some inconveniences associated with the electrostatic accelerator, however. The projectile mass and velocity could not be independently selected or controlled because the accelerator produced a sequence of projectiles which conformed to the relationship 'd v^2 = constant'. This was expected since the projectiles were charged to a uniform surface charge density. To alleviate this problem, the accelerator was operated at three different voltages (4, 1, and 0.5 Mvolts). Figure 3 is a plot of the projectile parameters for all the calibration tests included in this report. This figure illuminates some of the problems with this data set. For example, the clustering of projectiles along lines of constant energy was due to the resolution in the measurement of the accelerator voltage (10 k volts) and the projectile charge (1.5x10-15 coulombs). These measurements were the largest source of error in the determination of the projectile mass or diameter and in subsequent plots error bars are shown when this error exceeds 10% of the mass. Clustering of projectiles in velocity regions was the result of the accelerator control system rejecting projectiles near the edge of the velocity selection windows.
The projectiles were focused so that they impacted in an area approximately 1 to 3 mm in diameter. The position of the detector was controlled vertically and horizontally with a precision of approximately 1 mm. The angle of impact was varied from 0° to 75° from the normal to the target with an error of approximately 1 degree. For normal impacts, it was possible to have a well-defined target area of 5-7 mm in diameter. For thermal environmental impact tests, a target chamber with heating elements and a liquid nitrogen coil was used.
A schematic diagram of the instrumentation used to measure the detector signals along with a sketch of a typical signal is shown in Figure 4. The signal parameters of interest were the time of occurrence, the rise time, and the amplitude. The recover y time was fixed by the RC time constant of the detector circuit, where R is the bias resistance (1 M Ohm) and C is the detector capacitance. Recovery, itself, is more important than recovery time and a qualitative determination of the detector recovery was made by comparing the detector leakage current after a test to its value before the test. If the leakage current was unchanged, the detector was assumed to have recovered.
A dual-beam oscilloscope (Fig. 4) was the primary means of measuring the signal parameters and correlating the signal with the impacting particle. The oscilloscope was triggered by the accelerator when a particle was selected for entry into the target chamber. For the calibration test that are reported here, the instrumentation (similar to that used in the characterizing test of ref. 4) included an inflight projectile detector located at the entrance to the target chamber. This detector was similar to those used in the accelerator beam control section and its output was monitored with one beam of the oscilloscope to provide an independent measure of the projectile charge and velocity. This detector provided a reliable way of knowing, in the absence of a signal from the MOS detector, that a particle impact did occur. The second beam of the oscilloscope was used to look for low level signals (from the MOS detector) at the expected time of impact and the transient voltmeter measured the voltage level of discharge signals that exceeded the range of the oscilloscope. Also a galvanometer with a slow moving strip chart recorder was used to continuously monitor the current to the MOS detector. This galvanometer had a high current sensitivity but poor time response; however, it could detect discharge signals and would have indicated any signals that were not synchronized with the impacting particle.
The signals measured using the test arrangement described earlier are classified according to amplitude into two types - those greater than 1 volt (called discharge signals) and those less than 1 volt (called low-level signals). Although this classification may seem arbitrary, the difference between the two types of signals is substantial as can be seen in Figure 5 where signal amplitude is plotted as a function of applied voltage for four detector thicknesss from 0.5 micrometers to 1.8 micrometers (dielectric thicknesses of 0.4, 0.7, 1.0, and 1.7 micrometers). The discharge time (rise time of both types of signal) was approximately 0.5 microseconds for the 1.1 micrometers detectors and 1.5 microseconds for the 0.5 micrometers detectors. Figure 6 is a plot similar to that of Figure 5 except that a linear scale is used and all signals for a given bias voltage are averaged. As indicated in Figures 5 and 6, the discharge signals have been averaged for each test and show a strong dependence on the bias voltage. As expected, the greater the bias voltage, the fewer the number of low-level signals. Above some bias threshold there are few, if any, low-level signals indicating that the detector will essentially respond as an event counter, assuming that the impacting projectile has enough energy to cause any response at all from the detector. As can be seen in Figures 5 and 6, a bias voltage greater than approximately 35 volts on any of these detectors is enough to exceed the required bias threshold. One would expect the bias threshold voltage to scale with the electric field, however, this does not appear to be the case. Although there is a slight dependence on the thickness, it is far from what a direct field dependence would require. The dependence of the bias threshold voltage will be discussed later as it relates to the projectile parameters. Recovery of the detector after an impact was usually very good with the leakage current less than 10-9 ampere. Occasionally a low-level signal was followed by an increased or erratic leakage current (10-100 nA). In one test, a detector was completely discharged and did not recover when it was inadvertly subjected to a stream of particles (approximately 60/sec.) for a few seconds. The problem here is there is not enough current flowing in any given damaged region of the detector to burn away the top electrode material. This will be discussed further later. The magnitude of the discharge signal is almost independent of the detector dielectric thickness and is essentially set by the applied bias as can be seen in Figures 5 and 6.
The impacted areas were visually inspected following impact tests to determine the characteristics of the events. For the case of low-level signals, the impact damage is confined to an area just slightly larger than the crater caused by the impacting projectile. In sharp contrast, the impacts which caused a discharge signal resulted in a damaged area ranging from 10 micrometers to 100 micrometers depending on the test conditions - particularly the bias voltage. Figure 7 is a photomicrograph of a typical impact site which resulted in a discharge signal. As can be seen, the damaged area is nearly circular. Upon close examination one finds that the top electrode material is burned away throughout the total damaged area. However, the dielectric remained intact except for the central part of the area (~10 micrometers) where the projectile impacted. In this central region where the impact occurred, the dielectric is removed and the underlying silicon is exposed. The nature and extent of the damaged region is far greater than that caused by the projectile alone. More will be discussed on this subject later.
Tests were conducted to determine if a relationship exists between the type of signal and the projectile parameters. These data are plotted in Figures 8-11. In these figures each data point represents an impacting projectile according to its velocity and diameter. The type of symbol represents whether a discharge occurred, a low-level signal was recorded, or no signal was recorded. These tests repeat and extend similar tests reported in reference 4, but with much greater confidence in the 'no signal' events. Based on the earlier (ref. 4) preliminary calibration, all of these tests were expected to produce some response from the detector. As can be seen from the data plotted in Figures 8-11, there does not appear to be any correlation between the signal type and the projectile parameters that might indicate a threshold of projectile sensitivity.
What is very important is the fact that a minimum or threshold of bias voltage is required before a detector will respond to impacting particles in a consistent manner. This can clearly be seen by comparing the data in Figure 8 with that in Figure 9 where the detector bias was increased from 20 volts to 25 volts. As seen, the number of no signal events is drastically reduced with an increase of only 5 volts. Similarly, comparing the data of Figure 10 with Figure 11 (thicker dielectric and lower electric field strength) shows a decrease in the number of low level signals for an increase in bias voltage of only 5 volts.
There is a relationship between the discharge signal and the damaged area as can be seen in Figure 12. The detector signal energy, E, can be calculated and is the difference between the energy stored in the detector (capacitor) before impact and the energy remaining after impact:
where, C, is the capacitance of the detector, Vb, is the magnitude of the bias voltage, and Vs, is the signal voltage. This energy range is approximately 10 to 200 J. By comparison, the kinetic energy of the projectiles (for this data) was 0.2 to 6 micro J and their diameter range was 0.4 to 4 micrometers. This clearly shows that the energy stored in the capacitor plays an important role in the discharge mechanism and crater formation.
Another series of tests investigated the effects of operating at various temperatures that these detectors may be subjected to in the space environment. The results of these experiments are shown in Figure 13. As can be seen, the detectors operate over the temperature range -100° C to 95° C. However, there was a definite decrease in signal voltage for a given bias voltage at lower temperatures.
The effects of operating with a positive bias voltage on the bottom (Si) electrode were studied. As can be seen from the data plotted in Figures 5, 6 and 13, the signal voltage is significantly reduced if the polarity of the detector bias voltage is switched from a negative bias on the Si substrate to a positive bias. Also, the shape of the discharge region for the positive Si bias was not the usual round shape that was found for the negative bias. Although the discharge energy was less, the area affected was greater and irregular in shape. The area and shape of removed material also depended on whether the detectors were sintered at 200° C following evaporation of the top aluminum electrode. For the sintered samples, the areas were smaller and much more symmetrical and uniform.
Extensive tests (>4,000) were run to determine the sensitivity of the detectors to projectile mass, velocity, and impact angle for different detector biases and dielectric thicknesses. In these calibration tests, either a discharge or low level signal is regarded as a detector response and the minimum detectable voltage level for a no response was about 1 m volt. Four dielectric thicknesses were used for these tests: 0.4, 0.7, 1.0, and 1.7 micrometers; and the top electrode was 0.1 micrometers of aluminum for all detectors. Figures 14-17 are typical plots of mass versus velocity for fixed dielectric thickness, bias voltage, and impact angle. As can be seen from the plotted data, there are mass and velocity values that clearly cause discharges, values that do not result in discharges, and values at which discharges are sometimes seen and sometimes not. This overlap region occurs because the dielectric breakdown process is complex and statistical in nature. The discharge process depends on many other factors such as variations in the particle geometry (i.e., not perfect spheres as assumed), variations in the dielectric, and top electrode thicknesses. To use these detectors, it will be necessary to tolerate some statistical error and uncertainty in the sensitivity near the lower end of the detectivity curve.
Since no satisfactory theoretical model is available for the sensitivity, we have attempted to develop an empirical model through curve fitting of the experimental data. To do this, we first made logarithmic plots of mass versus velocity for each angle of impact, dielectric thickness, and bias voltage, as shown, for example, in Figures 14-17. From these plots it was found that the data could be reasonably fit to a power series of the form:
(mass)^0.33(velocity) = constant (K1), or
(diameter)(velocity) = constant (K2).
Once this relationship was established, we plotted the product (diameter and velocity) as a function of impact angle (cos beta), as measured from the normal to the target, for a given dielectric thickness and bias voltage. Figure 18 is an example of such a plot. From these data we established the following:
(diameter)(velocity)(cos beta)^1.5 = constant (K3).
Using this relationship, we computed (K3) as an impact calibration parameter and examined this parameter as a function of the applied bias voltage and sensor thickness. Since there is a large number of points to be plotted, a stacked histogram is used to display these relationships and, although the range of the impact parameter is from 0.5 to 15, only the data near the sensitivity thershold is plotted. Figures 19 - 22 are the primary calibration data (data at lower bias voltage can be compaired but not used for calibration). As can be seen in these figures, there is little dependence of the sensitivity on the applied bias voltage if the voltage is above the threshold as discussed earlier. The unexpected low level signals in Figure 20 occured at incident angles of 50° or greater and are a good example of the effects of bias voltage for impacts near the sensitivity threshold.
From these figures one can determine a value for the calibration parameter (K3) for each sensor thickness (dielectric plus the top metal) directly from the data at the higher bias voltage. For example, the data for the 0.8 micrometers sensor in Figure 20 indicate that for values of K3 less than 1.5 there will most likely be no response from the sensor and for values greater than 1.5 the sensor will most likely respond with a high level signal. For the 1.8 micrometers sensor (Figure 22), the statistics are improved by combining the data for both bias voltages.
By plotting these values of K3 as a function of the thickness for each detector, one can get the relationship of the sensitivity to thickness. Figure 23 shows this data with a best fit power curve which provides a general expression for the sensitivity of these sensors to iron particles as follows:
S = (Dia)(Vel)(Cos beta)^1.5(Thck)^-0.875 - 1.814
S > 0 ...Discharge Signal
S < 0 ...No Signal,
where, S is the sensitivity, 'Dia' is the projectile diameter in micrometers, 'Vel' is the projectile velocity is in km/sec, 'beta' is the angle of impact measured from the normal to the detector and 'Thck' is the sum of the detector dielectric and top metal thicknesses in micrometers. A caution is in order here and that is we have studied only iron particles in the above mentioned tests. The detector sensitivity is obviously a function of the density of the projectiles used and hence the sensitivity will change with projectile density. We would discourage simple scaling of the mass to arrive at a new sensitivity as this will likely be in error. Tests are needed using projectiles with different densities to determine the dependence of the sensitivity on projectile density.
Given the above equation for sensitivity and the projectile parameters used in these test (Figure 3), one can plot the calibration for various thickness of sensor or angle of impact. Figure 24 is an example of such a plot for the thickness used in these test and impacts at normal angle of incidence.
As shown by the impact tests reported here, the capacitor detectors respond to an impacting projectile in two different modes: (1) such that a low-level signal is obtained and the damaged area of the capacitor is limited to the impact crater region with little or no evaporation of the top electrode outside the impact crater and (2) such that a large signal (a significant portion of the applied capacitor voltage) is obtained and is accompanied by significant evaporation of the top electrode around the impact crater. Due to the occasional erratic nature of the low-level signals, we have concentrated on characterizing the detectors when operated in the discharge mode where the capacitor is partially discharged and a large signal is produced. For example, there has been no attempt to correlate the magnitude of low-level signals with the mass or velocity of impacting particles. The primary questions that have been considered are: Can the detectors be operated such that they reliably record impacting events and what is the threshold for detection as a function of particle mass, velocity, and angle of impact?
The breakdown mechanisms responsible for producing the signal are obviously very complex with a number of factors participating at various stages of the discharge event. First, the actual discharge mechanism is not monitored since it is an internal mechanism and only the recharge current or sensor voltage can be measured. The impact of a high velocity projectile on a capacitor-type detector generates very high pressures and temperatures (shock waves) in the capacitor as well as in the projectile. The high pressures resulting from the shock wave cause electro-mechanical breakdown of the dielectric. This can occur either from mechanical collapse of the dielectric, or the intrinsic strength of the dielectric can be exceeded. In either case there is a threshold pressure or shock that is needed to trigger the event. Temperature also causes changes in the mechanical as well as the electrical properties of the dielectric. Consequently the nature of the projectile as well as its mass and velocity affect the shock wave (temperature of the target) and hence the threshold for breakdown.
Although a pressure or shock induced discharge is the initiating mechanism for the discharge, it is not adequate to explain a sustained discharge. A second mechanism must be invoked if the discharge is to be sustained for any period of time. One way that has been suggested is to consider the impacted area of the capacitor to be a resistor in parallel with the capacitor - this assumes the impact area is small compared to the total capacitor area which is the case here. With this assumption, the discharge must be sustained for a time greater than approximately 3 times the value of the parallel resistor times the capacitance if complete discharge is to occur. This second mechanism contributing to the capacitor discharge is related to the large quantity of thermodynamic energy (high temperatures) available in the shocked region of the capacitor immediately following impact. This thermodynamic energy is sufficient to generate in the region a highly compressed plasma. The material in the impact region may behave like a liquid with a very high vapor pressure as opposed to a gas. In any case it will be highly ionized, so there are free electrons throughout the impact region, and hence the plasma region may act like a metallic conductor.
Experimentally the discharges are observed to be accompanied by a flash of light which confirms the presence of a plasma. We did not make a distinction whether or not light could be seen in cases where a low-level signal was observed. One way of locating the position of an impact is to observe where the flash of light occurs when operated in the discharge mode. Rarefaction waves from the impact can also add to the luminosity of the plasma. As mentioned earlier, when a new detector is first formed, it often has electrical shorts caused by imperfections in the dielectric. These shorts are removed by applying voltage to the capacitor which causes the shorted region to burn out. As these defects are removed they are often accompanied by a flash of light.
There is yet another mechanism which limits the amount of discharge the capacitor experiences. This mechanism is controlled by the metal type and thickness of the top electrode as well as the applied voltage. Since the metal is thin by design, it is not capable of sustaining high current densities for any significant time due to resistive heating. As the capacitor discharges through the short caused by the impact, the energy dissipated by the metal electrode simply heats the metal to the point that it evaporates away, leaving an open circuit to the shorted region. This is a valuable attribute of detectors of this type since it renders the detector ready to count again if it is being used as an event counter. The area of this burn-out is dependent on the applied voltage and the larger the discharge signal the larger the burn-out area. Figure 12 indicates a linear relationship between the amount of aluminum removed and the signal energy. From this data, one can caculate the ratio of signal energy to grams of aluminum removed (238 kJ/gr) and compare with the amount of thermal energy (13.6 kJ/gr) required to vaporize aluminum from 300 K. This comparison indicates that vaporizing aluminum accounts for less then 10% of the signal energy.
There appear, then, to be at least three mechanisms that play a role in the discharge of these detectors: (1) initiation of the discharge by the projectile, (2) limited continuation of the discharge by the plasma created by the impact resulting in low-level (millivolt) signals for low applied voltages, and (3) discharges sustained by the stored energy in the capacitor which last long enough to evaporate away a portion of the top metal electrode around the impact region, thereby opening the electrical circuit to the impacted area. The energy stored in the capacitor is clearly seen to play a significant role in the discharge mechanism, since it controls the area over which the top electrode metal is vaporized and removed. This is also seen, by comparing the magnitude of the signal energy to that of the projectile, where the maximum kinetic energy of the projectiles used in this study was approximately 6x10-6 J while the discharge energy was as much as 2x10-4 J.
From a theoretical view, it has not been possible to develop a model based on physical principles that can be used to describe the discharge mechanisms and the detector response to projectile parameters such as mass and velocity. On first glance one would think that it might be possible to relate the detector sensitivity to crater depth or diameter. If this were possible, then one could take advantage of existing work on cratering theory such as that given by Öpik 7) and Gault 8). For example, Gault empirically scaled his parameters with kinetic energy (mv^2) raised to a power between 0.370 and 1.133 depending on the parameter of interest such as penetration depth. Öpik scaled his parameters with a slightly lower value of impact velocity more closely resembling momentum (mv) rather than energy. As can be seen in our empirical formulation for sensitivity the mass-velocity relationship is mv^3 which is a stronger function of velocity than is kinetic energy. Based on the present work and that of previous researchers (Ref. 3), we have shown that it is possible to build detectors of the MOS capacitor type that can be used as event counters to measure hypervelocity particle impacts. When used as an event counter, bias voltages large enough to cause significant discharge of the capacitors are required. Each event removes a small portion of the detector. However, the area is so small (20-60 micrometers in diameter) compared to the area of the detectors (2 inch diameter) that unless thousands of impacts are counted they can be neglected. It is also possible to account for the area removed, since the area removed is approximately the same for all events, by keeping up with the number of discharges and removing that area from the active area.
Although it has not been possible to develop an analytical expression for the detector sensitivity based on a theoretical knowledge of the discharge mechanisms, it has been possible to fit an empirical expression to the data as shown in figures 18-23. We believe that detectors of the type described here are very reliable and offer a means of obtaining data on the flux of meteoroid and space debris in the low mass range where it has been very difficult to obtain real time data in the past. Detectors of this type were used on the Interplanetary Dust Experiment (ref. 5) that was flown on the Long Duration Exposure Facility. The experiment was returned to earth and the detectors analyzed following 5 1/2 years in space and they were found to still be in good working order.
Figure 2. Micrometeoroid Impact Detector
Figure 3. Projectile Velocity, Mass, and Kinetic Energy for Accelerator Voltages
Figure 4. Block diagram of instrumentation used for MOS detector calibration test.
Figure 5. Signal Amplitude versus Detector Bias Voltage for Various Detector Thicknesses
Figure 8. Projectile Parameters versus Type of Signal for 0.5
micrometers Detector at 20 Volts Bias
Figure 9. Projectile Parameters versus Type of Signal for 0.5
micrometers Detector at 25 Volts Bias
Figure 10. Projectile Parameters versus Type of Signal for 1.1
micrometers Detector at 25 Volts Bias
Figure 11. Projectile Parameters versus Type of Signal for 1.1
micrometers Detector at 30 Volts Bias
Figure 12. Diameter of Damage Area versus Discharge Signal Energy
Figure 13. Signal versus Bias Voltage at Various Temperatures
Figure 14. Typical Calibration Test Results for Detector with 0.4
micrometers Dielectric Thickness
Figure 15. Typical Calibration Test Results for Detector with 0.7
micrometers Dielectric Thickness
Figure 16. Typical Calibration Test Results for Detector with 1.0
micrometers Dielectric Thickness
Figure 17. Typical Calibration Test Results for Detector with 1.7
micrometers Dielectric Thickness
Figure 19. Calibration Data for 0.5 micrometers Sensor
Figure 20. Calibration Data for 0.8 micrometers Sensor
Figure 21. Calibration Data for 1.1 micrometers Sensor
Figure 22. Calibration Data for 2.2 micrometers Sensor
Figure 23. Metal-Oxide-Silicon Capacitor Detectors Sensitivity
to Iron Particles.
