DESCRIPTION
Ion beam doped Lithium Tan ta late or similar compound
SAW (surface acoustic wave) components are used in a wealth of different electronic applications and particularly in the realization of filters, resonators, delay lines and duplexers for IF and RF stages in telecommunication devices. SAW components can be manufactured small and at low cost, which makes them an ideal choice for in the design of consumer telecommunication apparatuses, like domestic TV sets and cellular phones. SAW components are however also employed in strain gauges, radar pulse processors and in many other devices.
Different piezoelectric materials can be employed in the construction of SAW devices. Lithium Tantalate, LiTaθ3, is gaining favor in this field thanks to its excellent characteristics of high piezoelectric coefficient, low dispersion and low frequency temperature dependency.
A shortcoming of known devices based on LiTaO3 is the high pyroelectric coefficient exhibited by this material. Pyroelectric effect causes an accumulation of electrical charges in the bulk of the device when the device is subject to thermal drifts. In this condition the surface of the substrate of the device collects, by induction or conduction, an electric charge. If the charge is not dissipated, the difference of potential generated by this thermally induced charge grows until an electric discharge is generated.
Thermally induced dielectric discharges induce spurious voltage spikes on the device's electrodes, which are especially harmful when the device is used in a digital circuit, like a digital telecommunication system. In this case the spikes generate false digital transitions and contribute directly to the error rate of the device.
Another shortcoming of LiTaO3 is that the coupling between pyroelectric and piezoelectric effects produces heavy mechanical stresses
when the crystal is subjected to thermal cycling, in particular during the manufacturing of the devices. It is believed that the high pyroelectric coefficient of LiTaO3 is the source of many wafer and device ruptures during manufacturing, and consequently that the yield and cost of the production are negatively affected by the pyroelectric effect.
This extreme fragility lowers manufacturing yields and negatively affects production costs. A LiTaθ3 that were less prone to breaking would be extremely advantageous, even independently from the fact that the pyroelectrically-generated spikes are suppressed. In particular it would simplify the assembly of circuits using the flip-chip process.
One aim of the present invention is to provide a LiTaO3 material which is free from the above shortcomings of the prior art.
Another aim of the present invention is to provide a LiTaO3 SAW device wherein the spurious spikes generated by the pyroelectric effect are reduced or absent.
Another aim of the present invention is to provide a method for treating LiTaO3 crystals in order to contrast the pernicious effects of pyroelectricity.
These aims are provided by the material, the device and the method of the claims of the corresponding categories, and in particular by materials, devices and methods in which the LiTaO3 is doped by ionic implantation in order to increase its conductivity.
The invention is based on the observation that, in normal condition that are found during device operation, the thermal drift responsible for the generation of pyroelectric charges is slow. It happens typically on a time scale of the order of some minutes, or of 102 seconds.
The dielectric relaxation time of a material is given by the relationτ = p/ε0ε. where p represents the bulk resistivity, and εr εo stand for
the relative dielectric constant and the dielectric constant of vacuum, as it is well known.
Dielectric relaxation time in undoped LiTaO3 is of the order of 104 seconds. Since this time is much longer than the timescale of the thermal drift, pyroelectrically generated charges can accumulate until they reach a harmful level. If, on the contrary, the resistivity of the material could be reduced, by doping, by a factor 1000 or more, the pyroelectrically generated charges would recombine in a few seconds, before that a discharge level is attained, thereby suppressing the harmful consequences of pyroelectricity.
The normal functioning of the device, on the contrary, would not be affected by the reduced resistivity. At the normal operating frequencies encountered in SAW devices the substrate behaves like a perfect insulator, because such frequencies are still much higher than the inverse of the dielectric relaxation time.
The useful properties of LiTaO3, in particular its piezoelectric properties, must however be preserved in the doping process. The doping process must neither cause excessive heating of the substrate nor induce an excessive number of lattice defects in form of dislocations or amorphous zones.
For testing the material, wafer slices cut from a LiTaO3 crystal have been implanted with ion beams of different chemical species and energy, and have therefore been used to produce a conventional SAW device, in this case a dual band pass balanced filter.
The devices so produced have been consequently tested with a network analyzer, in order to evaluate the electrical performance of the filter. Since the device under test was balanced, balun transformers have been used to match it with the unbalanced input/output of the network analyzer, according to standard practices. The frequency response obtained was compared with the response expected for a filter of the same model
realized on a standard substrate of undoped LiTaθ3. Concordance of the responses was taken of an indication that the piezoelectricity of LiTaθ3 was not affected by the doping process. An example of normal and defective frequency responses is shown on figure 1.
In a further testing step the amplitude and frequency of the spikes exceeding a gate of 100 mV has been monitored with an oscilloscope while subjecting the devices to thermal cycles. Reliably reproducible thermal cycles can be produced by many known means, for example by a ThermoStream® device produced by Temptronic® Corporation. The tests have been carried out at temperatures cycling between -10 °C and 80 °C, following a triangle function of 36 minutes' period, as it is visible on figure 2.
Doping LiTaO3 is preferably performed by means of an ion beam accelerator. In some tests more severe thermal cycles have been produced by a hot air gun delivering a stream of hot air at 120 °C, followed by application of a cold spray.
Different implantation schemes have been tried, as it is apparent from appendix 1. The irradiation tests performed can be grouped in the following categories:
- High energy implantation. Ion beam energy 4.5 5.2 MeV, chemical species employed: H, He, N, O;
- Medium energy implantation. Ion Beam energy 2 MeV, chemical species employed: H and He;
- Low energy implantation. Ion beam energy 0.06-0.15 MeV, chemical species employed: B and N.
The high-energy and medium-energy irradiations have been performed by means of a Van der Graaff accelerator. The low energy irradiations have been performed by means of an ion implanter capable of
producing ion beams whose energy is comprised between 10 keV and 200 keV.
The main difference between high and low energy irradiation is that high energy irradiation will produced a relatively deep doped layer, as it is well known that range and straggling increase with increasing ion energy. Low energy implantation, on the contrary, produces a relatively shallow doped layer. Medium energy irradiation results lay in between the two cases above.
Various possibilities of implantation can be evaluated by a computer program to calculate stopping and ranges in materials using the well known Monte Carlo technique.
Table 1:
Ion E (keV) Material Range / nm Straggling / nm
B 30-60 LiTaO3 100-200 -
B 45 LiTaO3 152 101
N 45-80 LiTaO3 100-200 -
N 60 LiTaO3 146 93
P 90-160 LiTaO3 100-200 -
P 120 LiTaO3 150 88
H 3300 LiTaO3 94000 6000
He 3300 LiTaO3 13700 900
H 3300 LiNbO3 111000 5000
H 3600 LiNb03 128000 6000
Table 1 gives the expected mean ranges and straggling of several ions in LiTaO3 and LiNbOaat various energies. In some cases of very shallow implantation the straggling parameter could not be reliably calculated. A selection of deposition profiles is shown below from Figure 1 to Figure 3.
One difficulty in implantation process is hat the ion implanted LiTaθ3 wafers became very fragile. Special precautions were elaborated and introduced in order to avoid the crack of the wafers, caused by internal stress. In particular, a 10 μg/cm2 Carbon film deposition on a front side or a Si wafer juxtaposed behind the LiTaθ3 wafer were used in order to increase
surface thermo- and electrical conductivity of the samples ( to decrease a possibility of local overheating or overcharging) . The irradiation current was limited, in order to avoid sample overheating or the ion beam has been periodically interrupted in order to allow cooling of the target. Other known target cooling devices could also be employed, for example a liquid cooling.
It was found that thick (0.5 mm) wafer slices break less easily than thin ones (0.35 mm).
After irradiation some of the samples have been examined by XRD (X-ray diffraction) and Ellipsometry measurements which showed, as it was expected, the formation of an internally stressed amorphous layer nears the surface of the wafer. It is known that known such defects can be eliminated or reduced by heat treatments. The effectiveness of said heat treatments start approximately at 200 °C, and increases quickly with the temperature. Temperatures above the Curie Temperature of LiTaO3, are however less preferable, because they may cause a degradation of piezoelectric properties.
The tests show that there is a critical implantation dose between 1015-1016 atoms/cm2, depending on the type of ions. This is the dose at which point defects start to form voids and bubbles.
By carefully choosing the implantation parameters, it has been possible to reduce considerably number and voltage amplitude of spikes, while keeping the parameters relevant to SAW devices (phase velocity, coupling coefficient and dielectric constant) in suitable range for the applications. The results are shown in the following examples
First example
Undoped wafers of LiTaO3 have been irradiated as shown in the following table.
Table 2: Irradiated Samples
C N 150 2.1 x1015
D B 150 2.1 x1016
E H 2500 5 1015
F O 2500 5 x1016
G N 5000 2 x1016
H N 2500 2 x1016
1 N 5000 5 x1015
J H 2000 1 x1015
After irradiation the samples have been used as substrates to produce a conventional SAW double band pass filter. The performances of the filter have been studied, in order to ascertain whether the material had retained its piezoelectric properties. The figures show the frequency responses of samples A, B, C, D, along with the response from a device realized on an undoped substrate. One can see that the material piezoelectricity of LiTaO3 is conserved in this case. The samples marked E, F, G, H, lhave been altered in their piezoelectric properties, but not to the point to be unsuitable for the application. The sample marked J has lost almost all its piezoelectricity. It can be concluded that low-energy implantation of electrically active ions has the least detrimental effect on piezoelectric properties of LiTaθ3-
The devices realized from the samples marked A, B, C, D have been tested for thermally induced spikes. 5 devices for each substrate have been tested, each devices has been tested three times, in order to filter
statistical fluctuations. The results are summarized in Table 3 below. The best results were obtained for the more heavily doped samples.
Table 3: Spike test results
Measure undoped A B C D LiTa03 spikes Level spikes Level spikes Level spikes Level spikes Level
Sample tests # [mV] # [mV] # [mV] # [mVJ # [mV]
1 1 10 150 10 110 4 180 3 70 1 50
2 6 85 7 220 5 200 8 190 8 200
3 8 150 7 150 6 200 1 150 6 120 avg 8.00 128.3 8.0 160 5.0 193.3 4.0 80 5.0 60
2 1 12 180 4 190 4 180 2 80 3 60
2 9 170 5 220 7 220 6 180 5 170
3 10 190 10 200 5 200 4 75 2 200 avg 10.3 180.0 6.3 203.3 5.3 200.0 4.0 111.7 3.3 143.3
3 1 8 200 4 150 10 155 5 60 2 55
2 7 220 7 150 6 200 6 175 8 200
3 4 200 12 200 6 200 7 100 3 150 avg 6.3 206.7 7.7 166.7 7.3 185.0 6.0 111.7 4.3 135.0
4 1 19 150 7 170 5 150 4 150 3 75
2 7 150 10 100 7 220 6 120 5 190
3 8 110 12 200 5 200 10 160 9 140 avg 11.3 136.7 9.7 156.7 5.7 190.0 6.7 143.3 5.7 135.0
5 1 18 200 3 125 3 70 5 40 2 50
2 8 200 6 200 6 125 5 125 5 200
3 7 80 8 220 9 200 8 140 2 210 avg 11.0 160.0 5.7 181.7 6.0 131.7 6.0 101.7 3.0 153.3
Average 9.4 162 7.5 173 5.9 180 5.3 121 4.3 138
Improvement Standard Slight Medium Good Good
Second example
The tests in the example above have been carried on without any heat treatment of the sample. A second batch of wafers has been irradiated with Boron and Nitrogen ions at energies between 100 and180 keV, current 80 and 150 nA at three doses: 1.0 x 1016, 2.0 x 1016and 3.0 x 1016 at cm2, while the target was kept at 450 °C for the whole duration of the irradiation. At this temperature the ion mobility is much higher, thus the implanted dose can diffuse in the material. At the same time the heat treatment allows to anneal the lattice defects caused by the ion implantation. The samples show a characteristic dark coloration, that is a clear indication of an alteration in the electronic band structure. The crystals so treated possess a very high conductivity, and by their use the spiking phenomenon is completely eliminated or vastly reduced.
Although the present invention has been here disclosed with reference to Lithium Tantalate, it may as well be applied to other compound of similar chemical and physical characteristics; In particular Lithium Niobate.