Method for producing a batch of acoustic wave filters

Abstract

A method for the batch production of acoustic wave filters comprises: synthesizing N theoretical filters, each filter defined by a set of j theoretical resonator(s) having a triplet C.sub.0ij,eq, .sub.rij,eq and .sub.aij,eq, these parameters grouped into subsets; determining a reference resonator structure for each subset, naturally having a resonant frequency .sub.r,ref, where .sub.aij,eq<.sub.r,ref<.sub.rij,eq; determining, for each theoretical resonator, an elementary building block comprising an intermediate resonator R.sub.ij, a parallel reactance Xp.sub.ij and/or a series reactance Xs.sub.ij, the intermediate resonator R.sub.ij having a triplet C.sub.0ij, .sub.r,ref and .sub.a,ref, the parameters C.sub.0ij, Xpij and/or Xs.sub.ij defined so the elementary building block has a triplet: C.sub.0ij,eq, .sub.rij,eq and .sub.aij,eq; determining the geometrical dimensions of the actual resonators R.sub.ij of the filters so they have a capacitance C.sub.0ij; producing each actual resonator; associating series and/or parallel reactances with actual resonators in order to form the elementary building blocks.

Claims

1. A method for a batch production of a set of N filters F.sub.i, where 1iN each acoustic wave filter comprising M.sub.i actual acoustic wave resonators R.sub.ij, where 1jM.sub.i, and comprising center frequencies f.sub.i and bandwidths l.sub.i, which are different for at least two of them, comprising the following steps: synthesizing N theoretical filters, each filter being defined on the basis of a set of M.sub.i theoretical resonator(s), where 1iN, such that said theoretical filters have said center frequencies f.sub.i and said bandwidths l.sub.i, the theoretical resonators of each filter respectively having a triplet of parameters: a theoretical static capacitance C.sub.0ij,eq where 1jM.sub.j, a theoretical resonant frequency r.sub.ij,eq and a theoretical antiresonant frequency .sub.aij,eq, these parameters being grouped into one or more subsets; determining a reference resonator structure for each subset such that said reference resonator naturally has a resonant frequency .sub.r,ref that is lower than the lowest of the theoretical resonant frequencies .sub.rij,eq of said subset and an antiresonant frequency .sub.a,ref that is higher than the highest of the theoretical antiresonant frequencies .sub.aij,eq of said subset; determining, subsequent to the preceding step for each theoretical resonator of each subset, an elementary building block comprising an intermediate resonator R.sub.ij, at least one of a parallel reactance Xp.sub.ij associated with the actual acoustic wave resonators R.sub.ij to form the elementary building blocks and a series reactance Xs.sub.ij associated with the actual acoustic wave resonators R.sub.ij to form the elementary building blocks, the intermediate resonator R.sub.ij having a static capacitance C.sub.0ij, the resonant frequency .sub.r,ref and the antiresonant frequency .sub.a,ref, parameters of at least one of the static capacitance C.sub.0ij, the parallel reactance Xp.sub.ij and the series reactance Xs.sub.ij being defined such that the elementary building block has the static capacitance that is equal to the theoretical static capacitance C.sub.0ij,eq, the resonant frequency that is equal to the theoretical resonant frequency .sub.rij,eq and the antiresonant frequency that is equal to the theoretical antiresonant frequency .sub.aij,eq; determining geometrical dimensions of the actual acoustic wave resonators of the filters base on geometrical dimensions of said reference resonator structure such that they respectively have the static capacitance C.sub.0ij of the intermediate resonator R.sub.ij; producing each of said actual acoustic wave resonators.

2. The method for the batch production of the set of N filters F.sub.i as claimed in claim 1, wherein the elementary building block comprises a reactance connected in parallel which is connected, on the one hand, to an input/output terminal and, on the other hand, to an intermediate node between the intermediate resonator and a reactance connected in series.

3. The method for the batch production of the set of N filters F.sub.i as claimed in claim 1, wherein the elementary building block comprises a reactance connected in parallel with the series reactance and with the intermediate resonator.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) The invention will be better understood and other advantages will become apparent on reading the following description, which is given by way of non-limiting example, and by virtue of the appended figures in which:

(2) FIG. 1 illustrates the principle of co-integrating two filters by using a mass overload layer in order to lower the center frequency of one thereof, according to the prior art;

(3) FIGS. 2a to 2d relate to an exemplary embodiment of two filters of a WCDMA duplexer and illustrate stacks of the reception and transmission filters, revealing the addition of an extra layer to the resonators of the transmission filter (FIG. 2a); to an optical photograph of the duplexer thus produced on a single chip (FIG. 2b); to the respective electrical responses of the resonators (FIG. 2c) and the obtained filters (FIG. 2d);

(4) FIG. 3 illustrates the electrical measurement of the co-integrated WCDMA reception and transmission filters according to the article by A. Reinhardt, J. B. David, C. Fuchs, M. Clement, J. Olivares, E. Iborra, N. Rimmer, S. Burgess;

(5) FIGS. 4a and 4b illustrate the effect of a parallel capacitor and of a series capacitor on the electrical response of a resonator according to the prior art;

(6) FIG. 5 illustrates the principle of a filter containing resonators to which variable capacitors are added in series and in parallel according to the prior art;

(7) FIG. 6 illustrates examples of filters produced on the basis of the topology illustrated in FIG. 5, obtained by varying the capacitance values of the capacitors of the filter circuit according to the prior art;

(8) FIG. 7 illustrates the mismatch of the filters when they are centered beyond 0% or 11% of the resonant frequency of the series resonators according to the prior art;

(9) FIG. 8 illustrates a first example of an elementary building block used in the present invention in order to form the batch-produced filters;

(10) FIG. 9 illustrates the equivalent circuit diagram of a resonator referred to as a modified Butterworth-Van Dyke resonator;

(11) FIG. 10 illustrates the definition of the resonant and antiresonant frequencies of a piezoelectric resonator in relation to the equivalent circuit diagram of the modified Butterworth-Van Dyke resonator;

(12) FIG. 11 illustrates a second example of an elementary building block used in the present invention in order to form the batch-produced filters;

(13) FIG. 12 illustrates the architecture of a ladder filter used for each filter in an exemplary embodiment of the present invention;

(14) FIG. 13 illustrates the frequency response of the architecture of FIG. 12, configured for the bands TX 28, 17, 13 and 5 (scale on the left, in dB) and the frequency response of the doublet of resonators based on LNO, designed to produce these filters (scale on the right, in ohms);

(15) FIG. 14 illustrates the frequency response of the architecture of FIG. 12, configured for the bands RX 28, 17, 13 and 5 (scale on the left, in dB) and the frequency response of the doublet of resonators based on LNO, designed to produce all of these filters (scale on the right, in ohms);

(16) FIG. 15 illustrates a schematic view in cross section of a portion of a filter including a resonator and three co-integrated capacitors according to the method of the invention;

(17) FIG. 16 illustrates a schematic view from above of a portion of the transmission filter of the band 28 and produced by co-integrating the capacitors and the resonators.

DETAILED DESCRIPTION

(18) In general, as explained above, the applicant proposes the production of elementary building blocks to form the filters that it is desired to synthesize, namely the resonator/series reactance/parallel reactance triplets, such that they have the impedances or admittances (the inverse of impedance) required for the satisfactory operation of the filter into which they will be inserted.

(19) First Example of an Elementary Building Block that may be used in the Present Invention:

(20) The elementary building block of this example is shown in FIG. 8: it is composed of a resonator R, of a reactance (a capacitor in the example shown) Xp connected in parallel with the resonator, and another reactance (a capacitor in the example shown) Xs connected in series with the resonator. It should be noted that the reactances connected in series or in parallel with the resonator may be inductors. It is also possible to envisage combining capacitors and inductors while keeping the assembly joined to the resonator reactive overall.

(21) In general, the impedance Zr of an electromechanical resonator is characterized, regardless of its nature (SAW, BAW, Lamb, etc.) and in a first approach, by the impedance of its equivalent model, referred to as a Butterworth-Van Dyke model and shown in FIG. 9, and given, ignoring losses, by the relationship:

(22) $Z_r=\frac{1}{Y_r}$
where

(23) $Y_r=j\left(\frac{L_m{C}_m{C}_0{}^2-C_0-C_m}{L_m{C}_m{}^2-1}\right)$

(24) The curve of FIG. 10 corresponds to the plot of the impedance as defined by the formula above.

(25) The components of the motional branch (L.sub.m, C.sub.m and R.sub.m) are linked to the acoustic resonance and hence to the characteristics of the technological stack of the resonator, while the electrostatic branch (C.sub.0) is linked to the geometrical dimensions of the resonator.

(26) For this reason, the values of the elements of the motional branch are calculated so as to account for the resonant .sub.r and antiresonant .sub.a frequencies according to the following equations:

(27) ${}_r^2=\frac{1}{L_m{C}_m}{}_a^2=\frac{C_0+C_m}{L_m{C}_m{C}_0}={}_r^2\left(1+r\right)$
where

(28) $r=\frac{C_m}{C_0}$

(29) Furthermore, the characteristic impedance Z.sub.C of the resonator outside of noteworthy frequencies may be defined by omitting the piezoelectric effect, Z.sub.C then corresponding to the impedance of a capacitor of capacitance C.sub.0.

(30) In a simplified manner, it is therefore possible to express the electrical response of a resonator on the basis of the characteristic impedance (or admittance) of the resonator.

(31) $Y_r=j{C}_0\left(1+\frac{r}{1-{\left(\frac{}{{}_r}\right)}^2}\right)=Y_C\left(1+\frac{r}{1-{\left(\frac{}{{}_r}\right)}^2}\right)$

(32) Where Y.sub.c=jC.sub.0 is the admittance of the electrostatic branch only, and hence the quantity fixing the characteristic impedance of the resonator.

(33) When components are connected in series and in parallel with this resonator, a new impedance Z.sub.eq is obtained, characterized by new resonant and antiresonant frequencies and a new characteristic impedance.

(34) Letting Y.sub.eq denote the equivalent admittance of the assembly composed of a resonator of characteristic admittance Y.sub.C, of frequency .sub.r and two admittance elements Yp and Ys that are connected in series and in parallel, respectively, the following is obtained:

(35) $Y_{\mathrm{eq}}=\frac{Y_S\left(Y_P+Y_C\left(1+\frac{r}{1-{\left(\frac{}{{}_r}\right)}^2}\right)\right)}{Y_S+Y_P+Y_C\left(1+\frac{r}{1-{\left(\frac{}{{}_r}\right)}^2}\right)}$

(36) The characteristic admittance of this assembly is obtained when the high-frequency response is observed (as for a simple resonator):

(37) $Y_{C,\mathrm{eq}}=\frac{Y_S\left(Y_P+Y_C\right)}{Y_S+Y_P+Y_C}$

(38) The effective resonant frequency is characterized by a cancellation of the denominator of Y.sub.eq and therefore corresponds to:

(39) ${}_{r,\mathrm{eq}}={}_r\sqrt{1+\frac{r.Math.Y_C}{Y_S+Y_P+Y_C}}$

(40) Similarily, the effective antiresonant frequency is characterized by a cancellation of the numerator of Yeq and therefore corresponds to:

(41) ${}_{a,\mathrm{eq}}={}_r\sqrt{1+\frac{r.Math.Y_C}{Y_P+Y_C}}$

(42) These latter three equations show that, on the basis of an actual resonator having given natural resonant and antiresonant frequencies, it is possible to generate, by means of:

(43) adequate dimensioning of the static capacitance (obtained by defining the active area of the resonator for a BAW resonator, or the number of interdigitated combs for a SAW resonator) and by means of;

(44) adding reactive elements in series and in parallel (adding non-reactive elements leads to complex resonant and antiresonant frequencies, i.e. to a lossy resonator),

(45) any equivalent resonator response having chosen equivalent resonant and antiresonant frequencies and impedance.

(46) Stated otherwise, there is always a unique set of parameters (Yc, Ys and Yp) making it possible to set up an equivalent resonator having a chosen triplet (Y.sub.c,eq, .sub.r,eq and .sub.a,eq).

(47) Hence, by varying the geometrical dimensions of the resonator and the values of the series and parallel reactances, it is possible to freely position a pole and a zero, which are associated with a chosen characteristic impedance.

(48) Associating a Resonator with Series and Parallel Capacitors:

(49) The most natural case is that in which a resonator is associated with one capacitor in series and with another in parallel. Equations (1), (2) and (3) are simplified in this case:

(50) 0${}_{r,\mathrm{eq}}={}_r\sqrt{1+r.Math.\frac{C_0}{C_0+C_p+C_s}}{}_{a,\mathrm{eq}}={}_r\sqrt{1+r.Math.\frac{C_0}{C_0+C_p}}$

(51) Furthermore, this basic building block has a characteristic impedance Z.sub.C,eq that is different from that of the original resonator, characterized by an equivalent static capacitance that is equal to:

(52) $C_{0,\mathrm{eq}}=\frac{\left(C_0+C_p\right).Math.C_s}{C_0+C_p+C_s}$

(53) The operating range of these parameters is limited to positive values of C0, Cs and Cp. Moreover, the resonant and antiresonant frequency values that can be attained are limited to within the range between the resonant and antiresonant frequencies of the original resonator.

(54) Similarly, it is possible to simplify the equations of other configurations associating a resonator with reactive elements, in order to bring other potential solutions to light.

(55) Associating a Resonator with Series and Parallel Inductors:

(56) In a manner similar to the above, the equivalent resonant and antiresonant frequencies are obtained by means of equations (2) and (3), where Ys and Yp are the admittances of the series and parallel inductors. By means of a few calculations, the following is obtained:

(57) ${}_{r,\mathrm{eq}}=\sqrt{\frac{1}{2}\left(\frac{1}{L_{//}{C}_0}+{}_a^2\right)\left(1\sqrt{1-{\left(\frac{4{}_r^2}{1+{}_a^2{L}_{//}{C}_0}\right)}^2}\right)},{}_{,\mathrm{eq}}=\sqrt{\frac{1}{2}\left(\frac{1}{L_p{C}_0}+1+r\right)\left(1\sqrt{1-\frac{4{}_r^2}{{\left(\frac{1}{L_p{C}_0}+1+r\right)}^2}}\right),}$$C_{0,\mathrm{eq}}=\frac{1-L_P{C}_0{}^2}{\frac{L_p+L_s}{L_p{C}_0}\left(1-\frac{{}^2}{L_{//}{C}_0}\right)},\mathrm{where}$$L_{//}=\frac{L_p.Math.L_s}{L_p+L_s}.$

(58) It is found that it is still possible to write the equations allowing (C.sub.0,eq, .sub.r,eq and .sub.a,eq) to be determined as a function of (C.sub.0, Ls, L.sub.p).

(59) However, in this case, two resonant and antiresonant frequency solutions are possible, caused by a duplication of the resonance. Moreover, the equivalent static capacitance now depends on the frequency, and has a pole at the resonant frequency of the circuit formed by the static capacitance of the resonator and the joined inductors. For this reason, this type of circuit is not favored.

(60) Associating a Resonator with a Series Inductor and a Parallel Capacitor:

(61) In a manner similar to the above, the following is obtained in this case:

(62) ${}_{r,\mathrm{eq}}=\sqrt{\frac{1\sqrt{1-\frac{4L_s\left(C_0+C_p\right){}_r^2}{{\left(1+L_s{C}_0{}_a^2\right)}^2}}}{2L_s\left(C_0+C_p\right)}},{}_{a,\mathrm{eq}}={}_r.Math.\sqrt{1+r.Math.\frac{C_0}{C_0+C_p}},C_{0,\mathrm{eq}}=\frac{C_0+C_p}{1-L_s.Math.\left(C_0+C_p\right){}^2}.$

(63) The antiresonant frequency therefore depends only on the parallel capacitor, while the resonant frequency is sensitive to both added components. Furthermore, the resonant frequency is duplicate, and the equivalent static capacitance depends on the frequency. As in the preceding case, this association is not favored.

(64) Associating a Resonator with a Series Capacitor and a Parallel Inductor:

(65) In a manner similar to the above, the resonant and antiresonant frequencies are obtained by canceling out the numerator and the denominator, respectively, of equation (1). The following is obtained in this case:

(66) ${}_{r,\mathrm{eq}}\sqrt{\frac{1+L_p\left(C_0{}_a^2+C_s{}_r^2\right)}{2L_p\left(C_0+C_s\right)}\left(1\sqrt{1-\frac{4L_p\left(C_0+C_2\right){}_r^2}{{\left(1+L_p\left(C_0{}_a^2+C_s{}_r^2\right)\right)}^2}}\right),}$${}_{,\mathrm{eq}}=\sqrt{\left({}_a^2+\frac{1}{L_p{C}_0}\right)\left(1\sqrt{1-4\frac{L_p{C}_0{}_r^2}{{\left(1+L_p{C}_0{}_a^2\right)}^2}}\right)}{C}_{0,\mathrm{eq}}=C_s\frac{{}^2{L}_p{C}_0-1}{{}^2{L}_p\left(C_0+C_s\right)-1}$

(67) It is found that this configuration also produces duplicate resonances and antiresonances. Furthermore and again, the equivalent static capacitance depends on the frequency and has a frequency range for which it becomes negative.

(68) Second Example of an Elementary Building Block that may be used in the Present Invention:

(69) It should be noted that the configuration of FIG. 8 is not the only configuration liable to exhibit the type of behavior described above. It is also possible to envisage another example of an elementary building block. For example, if it is considered that the reactive element connected in parallel with the resonator is instead this time connected in parallel with the series association of the resonator and the reactive element connected in series, the diagram of FIG. 11 may be obtained.

(70) Throughout the rest of the description and without losing generality, the developments are dealt with below according to the case that is the most useful in practice, i.e. that in which two capacitors are added to the resonator.

(71) In this case, the equivalent impedance of the assembly is written as:

(72) $Y_{\mathrm{eq}}=\frac{1}{\frac{1}{Y_c\left(1+\frac{r}{1-{\left(\frac{}{{}_r}\right)}^2}\right)}+\frac{1}{Y_s}}+Y_p$

(73) And in this case it is shown that:

(74) ${}_{r,\mathrm{eq}}={}_r\sqrt{1+\frac{1}{}.Math.\frac{C_0}{C_0+C_s}}$${}_{a,\mathrm{eq}}={}_r\sqrt{1+\frac{1}{}.Math.\frac{C_0}{C_0+\frac{C_s{C}_p}{C_s+C_p}}}$$C_{0,\mathrm{eq}}=\frac{C_s+C_p}{C_s+C_0}.Math.\left(C_0+\frac{C_s{C}_p}{C_s+C_p}\right)$

(75) These expressions are sufficiently close to those obtained in the development of solution (1) to be able to draw the same theoretical conclusions.

(76) Thus, in light of the preceding developments, it is possible, according to the present invention, to define the following main steps of the method for the batch production of a set of N filters F.sub.i, where 1iN, each acoustic wave filter comprising j actual acoustic wave resonators R.sub.ij, where 1jM.sub.i, and comprising center frequencies f.sub.i and bandwidths I.sub.i, which are different for at least two of them.

(77) Advantageously, the filters may all be produced on one and the same chip and associated with parallel and series reactances, according to the design steps summarized below:

(78) Step 1: in this first step, N theoretical filters are synthesized, each filter being defined on the basis of a set of j theoretical resonator(s), where 1jM.sub.i, such that said theoretical filters have said center frequencies f.sub.i and said bandwidths I.sub.i, the theoretical resonators of each filter respectively having a triplet of parameters: a theoretical static capacitance C.sub.0ij,eq, a theoretical resonant frequency .sub.rij,eq and a theoretical antiresonant frequency .sub.aij,eq, these parameters being grouped into one or more subsets.

(79) Step 2: a stack allowing all of these resonators to be formed is produced. By means of techniques known to those skilled in the art, the designer produces a reference resonator structure for each subset allowing a resonator to be produced that naturally has a resonant frequency that is lower than the lowest of the desired theoretical resonant frequencies .sub.rij,eq and an antiresonant frequency that is higher than the highest of the desired theoretical antiresonant frequencies .sub.aij,eq.

(80) Step 3: An elementary building block is determined for each theoretical resonator of each subset, which building block comprises an intermediate resonator R.sub.ij, a parallel reactance Xp.sub.ij and/or a series reactance Xs.sub.ij, the intermediate resonator R.sub.ij having a static capacitance C.sub.0ij, a resonant frequency .sub.r,ref and an antiresonant frequency .sub.a,ref; the parameters C.sub.0ij, Xp.sub.ij and/or Xs.sub.ij being defined such that the elementary building block has a static capacitance C.sub.0ij,eq, a resonant frequency .sub.rij,eq and an antiresonant frequency .sub.aij,eq.

(81) Step 4: The geometrical dimensions of the actual resonators R.sub.ij of the filters are defined on the basis of the geometrical dimensions of the reference resonator structure such that they respectively have the capacitance C.sub.0ij of the intermediate resonator R.sub.ij.

(82) First Example of Batch Production Using Bulk Wave Filters:

(83) In this example, the applicant has produced, on one and the same chip, the eight transmission and reception filters of four duplexers in accordance with the LTE protocol in bands 28, 17, 13 and 5. The center frequencies and bandwidths of each of the transmission and reception filters for these bands are summarized in Table T1. In order to simplify the design phase, a single filter architecture is used for these eight frequency bands.

(84) TABLE-US-00001 TABLE T1 Center frequency Band Transmission Reception Bandwidth 28 725.5 MHz 780.5 MHz 45 MHz 17 710 MHz 740 MHz 12 MHz 13 782 MHz 751 MHz 10 MHz 5 836.5 MHz 881.5 MHz 25 MHz

(85) This architecture, illustrated in FIG. 12, consists of a ladder structure with three stages, each stage of which comprises a pair of elementary building blocks. A building block contains a resonator, characterized in particular by its static capacitance and its natural resonant and antiresonant frequencies. This building block also contains a series capacitor and a parallel capacitor. For each filter, it is possible to modify these three parameters in order to synthesize, one by one, the responses of each filter.

(86) The various production process steps of the present invention are followed as described below:

(87) Step 1:

(88) It is necessary to identify the triplets (C.sub.0eq, .sub.r,eq and .sub.a,eq) specific to each building block of the architecture, in order to meet the required specifications. For example, FIG. 13 shows the response of the four transmission filters thus formed.

(89) Table 2 summarizes the set of values obtained in the first optimization phase

(90) TABLE-US-00002 TABLE T2 Elementary building TX filters blocks C.sub.0eq (pF) .sub.r,eq (MHz) .sub.a,eq (MHz) Band 28 S 4.6 716 769 S2 1.7 720 759 P 7.8 673 711 P2 9 673 716 Band 17 S 3 713 738 S2 1.4 712 734 P 14.4 673 715 P2 16.3 673 710 Band 13 S 3.9 784 805 S2 2.1 783 805 P 10.4 750 783 P2 10.5 750 784 Band 5 S 2.7 827 871 S2 1.2 827 874 P 8.3 790 826 P2 8.3 790 825

(91) An identical operation is also carried out for the four reception filters, which allows a new set of values to be identified for each building block, summarized in Table T3 below:

(92) TABLE-US-00003 TABLE T3 Elementary building TX filters blocks C.sub.0eq (pF) .sub.r,eq (MHz) .sub.a,eq (MHz) Band 28 S 2.1 779 836 S2 1.1 778 842 P 6.1 744 775 P2 8.8 722 777 Band 17 S 3.4 744 774 S2 1.6 743 774 P 6.6 714 742 P2 8.3 700 742 Band 13 S 2.7 749 776 S2 1.3 749 775 P 8.8 726 748 P2 9.3 726 748 Band 5 S 2.1 878 919 S2 0.7 871 919 P 9.1 838 877 P2 12.3 823 875

(93) Step 2:

(94) On the basis of these elementary building block specifications, the applicant synthesized a stack based on bulk wave resonators. These resonators must have a natural resonant frequency of less than 673 MHz (the lowest frequency of all of the elementary building blocks), and an antiresonant frequency of more than 919 MHz (the highest frequency of all of the elementary building blocks), i.e. an electromechanical coupling coefficient of more than 60%. In order to achieve this, the applicant considered these elementary building blocks as belonging to two subsets (series building blocks and parallel building blocks), and used a piezoelectric layer made of lithium niobate (X-orientation). The stack used is as follows: Lower electrodes made of molybdenum: 190 nm thick Lithium niobate (X-orientation): 1757 nm thick Upper electrodes made of molybdenum: 150 nm thick Passivation made of silicon nitride: 300 nm.

(95) The frequencies of the resonators connected in series are differentiated from those of the resonators connected in parallel by using a localized mass overload, such as commonly used on bulk wave filters, for example, by means of an additional layer. This differentiation is obtained here by partially etching 100 nm, for example, from the passivation layer at the series resonators of the various filters. The electrical response of the pair of resonators thus produced is shown in FIG. 13, scale on the right. A frequency coverage that is compatible with the production of two subsets of elementary building blocks is thus ensured.

(96) Steps 3 and 4:

(97) Lastly, the equations presented above are used to calculate the areas of each resonator for each filter as well as the capacitance values of the series and parallel capacitors joined to each thereof. These values are grouped together in Tables T4 and T5 below:

(98) TABLE-US-00004 TABLE T4 Elementary Area TX filters building block resonator (m.sup.2) C.sub.0 (pF) C.sub.s (pF) C.sub.p (pF) Band 28 S 100 100 2.3 10.2 6 S2 60 60 0.8 5 1.8 P 70 70 1.1 6.7 P2 80 80 1.5 7.5 Band 17 S 70 70 1.1 5.7 5.2 S2 50 50 0.6 2.8 2.4 P 100 100 2.3 12.1 P2 100 100 2.3 14 Band 13 S 200 200 9.2 4.9 10.4 S2 150 150 5.2 2.6 5.9 P 250 250 14.3 15.2 18.1 P2 250 250 14.3 15.5 18 Band 5 S 180 180 7.4 3.7 2.1 S2 125 125 3.6 1.7 1.1 P 300 300 20.7 11.1 11.9 P2 300 300 20.7 11.2 12.1

(99) TABLE-US-00005 TABLE T5 Elementary Area TX filters building block resonator (m.sup.2) C.sub.0 (pF) C.sub.s (pF) C.sub.p (pF) Band 28 S 115 115 3 3.8 1.8 S2 85 85 1.7 1.9 1.1 P 185 185 7.8 9 11.5 P2 170 170 6.6 19.5 9.3 Band 17 S 115 115 3 5.3 6 S2 80 80 1.5 2.5 2.9 P 135 135 4.2 11.4 11.4 P2 125 125 3.6 22.2 9.7 Band 13 S 115 115 3 3.9 5.9 S2 80 80 1.5 1.9 2.8 P 190 190 8.3 12.8 20 P2 195 195 8.7 13.5 20.9 Band 5 S 200 200 9.2 2.8 0 S2 125 125 3.6 0.9 0 P 400 400 36.7 11.6 4.9 P2 400 400 36.7 17.3 5.3

(100) It should be noted that the resonators of the parallel branches of the transmission filters covering bands 28 and 17 do not have a series capacitor, since in this case direct use is made of the natural resonant frequency of the technological stack, which was positioned at 673 MHz, i.e. below these two bands.

(101) In a similar manner, the reception filter in band 5 does not have a capacitor in parallel to the series resonators. The upper edge of this band is effectively positioned on the natural antiresonant frequency of the technological stack, namely 919 MHz.

(102) It has therefore been shown that the same technological stack allows the transmission and reception filters of four complete duplexers, namely eight different filters, to be produced conjointly.

(103) The response of these filters is shown in FIGS. 13 and 14.

(104) 2) Co-Integration of the Reactances with the Resonators:

(105) The most natural way in which to produce these filters consists of first producing the various resonators of these various filters on one chip and then transferring this chip to a second chip dedicated to the production of passive components, in particular the capacitors required for the filter circuit. Proceeding in this way makes it possible to benefit from technologies optimized for the production of capacitors and hence not to have to strike a compromise between capacitor, resonator and filter performance.

(106) It has however been shown in the field of AlN bulk acoustic wave filters that considerable area gains could be expected by co-integrating capacitors with the resonators of a filter. The article by A. Volatier, G. Fattinger, F. Dumont, P. Stoyanov, R. Aigner, Technology enhancements for high performance BAW duplexer, Proceedings of the 2013 Joint UFFC, EFTF and PFM Symposium, p. 761, presents for example the production of capacitors by using the dielectric layers of the Bragg mirrors of SMR-type resonators.

(107) Furthermore, it may be noted that equations (2) and (3) may easily be rewritten by normalizing the admittances Ys and Yp with respect to C.sub.0. Stated otherwise, the resonant and antiresonant frequencies are set by the ratio between the capacitance values of the capacitors connected in series and in parallel to the resonator, and the static capacitance of this resonator. As the latter is set by the thickness of the piezoelectric layer in the case of bulk wave devices, it is sensitive to any variation in thickness occurring during the fabrication process. This will therefore lead to frequency variations. In order to overcome this issue, the applicant has already proposed and described, in the patent application FR 2 99 60 61, the production of these capacitors on the basis of the same piezoelectric layer, simply by locally modifying the stack.

(108) In order to achieve this, the resonators are produced on the basis of a lower electrode/piezoelectric layer/upper electrode/passivation layer stack produced above a free cavity. In proximity to these resonators, the capacitors allowing the resonant and antiresonant frequencies to be positioned are produced on the basis of a bonding layer/lower electrode/piezoelectric layer/upper electrode/over-metallization stack produced directly on the substrate so as to attenuate the acoustic resonances that may be generated.

(109) Likewise, the sacrificed layer and over-metallization thicknesses are selected so as to push these attenuated acoustic resonances into frequency zones where they can be ignored, as indicated in the patent application FR 2 99 60 61.

(110) FIG. 15 shows a schematic view in cross section of a portion of a filter including a resonator and three co-integrated capacitors. This view corresponds to the cross section denoted by AA in FIG. 16 which represents a view of a portion of the transmission filter of band 28 comprising the blocks S, S2 and P2 of the diagram of FIG. 13. The resonators of each of the blocks are referenced, Rs, Rs2 and Rp2, respectively, the series capacitors are referenced Css and Css2 (the block P2 does not have a series capacitor), and the parallel capacitors are referenced Cps, Cps2 and Cpp2.

(111) Second Example of Batch Production Using Surface Acoustic Wave Filters:

(112) The present invention is also advantageous for the co-integration of surface acoustic wave filters. It may seem trivial to co-integrate SAW filters on one and the same chip, those skilled in the art tending to modify the period of the interdigitated combs so as to obtain resonators having different resonant and antiresonant frequencies for each of the co-integrated filters.

(113) Nevertheless, in this case, if the electrodes of these various resonators are produced on the basis of the same metal layer, it will result in metal thickness/wavelength ratios that differ from one filter to another, thereby affecting the propagation conditions of the acoustic waves under the arrays of electrodes, and risk making them less than optimal for some of the filters.

(114) Conversely, according to the optimization by the applicant, all of the resonators of the various filters to be co-integrated may be produced on the basis of resonators all having the same period of interdigitated combs, and therefore all operating with the same metal thickness/wavelength ratio. Thus, all of the resonators use the same propagation conditions, thereby obviating the need for specific optimization from one filter to another.

(115) In order to illustrate the application of the present invention to the case of surface acoustic wave filters, the use of surface waves of Love type propagating on the surface of a LiNbO.sub.3 substrate of cut (YXI)/15 is considered, as employed in the article by T. Komatsu, K. Y. Hashimoto, T. Omori, M. Yamaguchi, Tunable radio-frequency filters using acoustic wave resonators and variable capacitors, Japanese Journal of Applied Physics 49, 2010, and reprising the various steps of the method of the invention.

(116) Insofar as the same set of filters as above is designed to obtain the same frequency bands, it is possible to start from the filter design produced for the first example, and hence from the resonator specifications provided in Table T2 and Table T3.

(117) A surface acoustic wave resonator is dimensioned by setting the following parameters:

(118) Electrodes: copper, 360 nm thick;

(119) Interdigitated combs: 2.39 m period, metallization ratio of 0.5.

(120) The use of copper electrodes is necessary in order to have very dense electrodes that will slow the surface waves by a considerable amount, so as to better guide them along the surface of the piezoelectric substrate and thus prevent them radiating into the substrate. The combination of metal thickness, its metallization ratio and the period of the interdigitated combs allows the electromechanical coupling coefficient to be maximized at 32%. This makes it possible to have a resonant frequency localized at 673 MHz and an antiresonant frequency at 773 MHz.

(121) These conditions are similar to those of the preceding example, despite the electromechanical coupling coefficient being smaller. According to Table T2 and Table T3, it is possible to cover only transmission bands 28 and 17. For these two bands, the resonators are dimensioned so as to have the static capacitances required to synthesize the resonators for transmission bands 28 and 17 listed in Table T2 and Table T3. This is achieved by adjusting the length of overlap between teeth, as well as the number of electrodes, according to techniques known to those skilled in the art.

(122) This example makes it apparent that a smaller electromechanical coupling coefficient quickly limits the number of bands that can be covered.

(123) Third Example of Batch Production Using Plate Wave Filters:

(124) A third embodiment consists of producing the preceding filters by using plate waves. These waves propagate through a thin film of piezoelectric material. Remaining with the case of lithium niobate, a particularly advantageous crystal orientation for the purpose of maximizing the electromechanical coupling coefficients of these plate waves is the cut referred to as (YXI)/30. This orientation allows the excitation of waves referred to as SH0 (shear horizontal zero-order) waves, which have an electromechanical coupling coefficient reaching 55% when the thickness of the plate is close to one tenth of the wavelength as described in the article by M. Kadota, S. Tanaka, Y. Kuratani, T. Kimura, Ultra wide band ladder filter using SH0 plate wave in thin LiNbO.sub.3 plate and its application, Proceedings of the 2014 IEEE International Ultrasonics Symposium, pp. 2031-2034.

(125) As above, the implementational procedure differs from the first example only in the second step: the definition of the resonator. As mentioned in the article cited above, the resonator consists of a film of lithium niobate of cut (YXI)/30 added to silicon and suspended above a cavity machined into the substrate. The waves are excited by interdigitated combs made of aluminum, which are positioned on the upper surface of the film and have a width of 1.4 m and a period of 2.75 m, the thickness of the film itself being 550 nm and the thickness of the electrodes being 80 nm. A resonator produced according to these geometrical parameters has a resonant frequency of close to 664 MHz and an antiresonant frequency of close to 824 MHz, and hence an electromechanical coupling coefficient of the order of 48%.

(126) These conditions are more favorable than in the preceding example, since they allow, according to Table T2 and Table T3, transmission bands 13, 17 and 28 and reception bands 13 and 17 to be covered.

(127) For these five bands, the resonators are dimensioned so as to have the static capacitances required to synthesize the resonators for these five bands listed in Table T2 and Table T3. As above, this is achieved by adjusting the length of overlap between teeth, as well as the number of electrodes, according to techniques known to those skilled in the art.

(128) Particular Advantages Afforded by the Invention

(129) The present invention makes it possible to produce a bank of acoustic filters on one and the same chip.

(130) While, in the production processes of the prior art, each separate filter requires its own technological stack and hence a different production batch, the invention allows one and the same production batch to be used for a substantial variety of filters. In this way the production process is optimized.

(131) As a result, this approach makes it possible to minimize the number of different components to be assembled in order to produce the radiofrequency stage of a communication system and, at the same time, to minimize the area occupied by all of these components.

(132) If the data of Table T4 are examined in greater detail, it is possible to estimate the size of a chip containing the eight filters presented by way of example, by assuming that the capacitors joined to the resonators are positioned on a separate chip, positioned opposite (interposer or active circuit). By reprising the resonator areas given above and considering the technological margins and the dimensions of the various associated interconnects, the total area occupied by the eight filters may be calculated as approximately 2.5 mm.sup.2. This is to be compared with the dimensions of an individual duplexer (a pair of filters), at around 1.5 mm.sup.2 in the best case, for these frequency bands. It can therefore be seen that the co-integration of the various filters of the four duplexers in question on one and the same chip allows a substantial area to be gained.