Abstract
A diaphragm and suspension for an electroacoustic transducer are formed by depositing a layer of compliant material on a first surface of a solid substrate and removing material from a second surface of the solid substrate. The removal leaves a block of substrate material suspended within an inner perimeter of an outer support ring of the substrate material by the compliant material, the block providing the diaphragm.
Claims
1. A diaphragm and suspension assembly for an electroacoustic transducer, the assembly comprising: a piston having a surface to serve as the diaphragm; a support ring surrounding the piston and separated from the piston by a gap; and a layer of compliant material adhered to the support ring and to the surface of the piston, forming a suspension that suspends the piston in the gap; the layer of compliant material having a Young's modulus and thickness resulting in a mechanical stiffness in the range of 5-100 N/m.
2. The diaphragm and suspension assembly of claim 1, wherein the compliant material has an elastic strain limit of at least 50 percent.
3. The diaphragm and suspension assembly of claim 1, wherein the compliant material has an elastic strain limit of at least 150 percent.
4. The diaphragm and suspension assembly of claim 1, wherein the compliant material is a cured compliant material.
5. The diaphragm and suspension assembly of claim 1, wherein the compliant material comprises liquid silicone rubber.
6. The diaphragm and suspension assembly of claim 1, wherein the compliant material has a thickness in a range of 30-80 m.
7. The diaphragm and suspension assembly of claim 1, further comprising attaching a bobbin to the piston, the bobbin located adjacent to an inner perimeter of the support ring.
8. The diaphragm and suspension assembly of claim 7, wherein the bobbin is attached to the piston by adhesive, the adhesive being contained by a side wall of the piston such that it does not contact the suspension.
9. The diaphragm and suspension assembly of claim 1, wherein an outer diameter of the support ring is around 4 mm.
10. The diaphragm and suspension assembly of claim 1, wherein an outer diameter of the support ring is less than 4 mm.
11. The diaphragm and suspension assembly of claim 1, wherein the gap is around 300 m.
12. The diaphragm and suspension assembly of claim 1, wherein an underside of the piston includes voids.
13. The diaphragm and suspension assembly of claim 1, wherein an underside of the piston includes a pattern of at least one of rings, ribs, and voids.
14. The diaphragm and suspension assembly of claim 1, further comprising a ferromagnetic housing coupled to the support ring.
15. An electro-acoustic transducer comprising; the diaphragm and suspension assembly of claim 14; a voice-coil mechanically coupled to the piston; and a magnet coupled to the ferromagnetic housing to form a magnetic circuit such that the voice-coil is suspended within a magnetic field created by the magnetic circuit.
16. A method of forming an electroacoustic transducer having a diaphragm and suspension, the method comprising: depositing a layer of compliant material on a first surface of a substrate; and removing material from a second surface of the substrate, the removal leaving a block of substrate material suspended within an inner perimeter of an outer support ring of the substrate material by the compliant material, the block providing the diaphragm.
17. The method of claim 16, wherein the compliant material has a Young's modulus and thickness resulting in a mechanical stiffness in the range of 5-100 N/m.
18. The method of claim 16, wherein the compliant material comprises liquid silicone rubber (LSR).
19. The method of claim 16, wherein the compliant material has a thickness in a range of 30-80 m.
20. The method of claim 16, wherein an outer diameter of the support ring is around 4 mm or less.
Description
BRIEF DESCRIPTION OF THE DRAWINGS
[0044] FIG. 1 shows a cross-sectional view of a complete electro-acoustical transducer.
[0045] FIGS. 2A, 2B, and 2C show a top perspective, bottom perspective, and cross-sectional view of the diaphragm and suspension of the transducer.
[0046] FIGS. 3A and 3B show an assembly process for the transducer.
[0047] FIG. 4 shows a partial sectional view with dimensions of an example of the transducer.
[0048] FIG. 5A through 5K and 6A through 6M show MEMS fabrication processes for the piston and suspension of the transducer.
DESCRIPTION
[0049] As shown in FIG. 1, an electro-acoustic transducer 100 built using the technique disclosed below includes a diaphragm 102 suspended from a support ring 104 by a suspension 106. Unlike conventional loudspeaker suspensions, the suspension 106 consists of a layer of compliant material extending over the entire surface of the diaphragm, as shown more clearly in FIG. 2A. The diaphragm itself also differs from typical loudspeaker diaphragms, in that its radiating surface is a flat plane, hence we refer to it as a piston. The remaining parts of the transducer match those of a conventional electro-dynamic loudspeaker: a voice coil 108 wound around a bobbin 110, surrounding a coin 112 and magnet 114. The coin 112 and magnet 114 are connected to the support ring by a back plate 116 and housing 118, which, like the coin, are formed of ferromagnetic material, such as steel. Electrical current flowing through the voice coil within the field produced by the magnet 114 and shaped by the ferromagnetic parts produces a force on the voice coil in the axial direction. This is transferred to the piston 102 by the bobbin 110, resulting in motion of the piston, and the production of sound. The same effects can be used in reverse to produce current from sound, i.e., using the transducer as a microphone or other type of pressure sensor. In other examples, the voice coil is stationary and the magnet moves. Such a small transducer is described, aside from the fabrication of the piston and suspension as disclosed below, in U.S. patent application Ser. No. 15/182,069, Miniature Device Having an Acoustic Diaphragm, filed Jun. 14, 2016, the entire contents of which are incorporated here by reference.
[0050] One potential material for the compliant suspension is liquid silicone rubber (LSR), a product based on polydimethylsiloxane (PDMS). To properly suspend the piston, while allowing it to move as needed at acoustic frequencies, the material of the suspension should have an elastic strain limit of at least 50 percent and a Young's modulus and thickness resulting in mechanical stiffness of the suspension in the range of 5-100 N/m. Various elastomers will meet this requirement. LSR is one example. In addition, even larger elastic strain limits, as high as 100 or 150 percent may be desired to accommodate large forces applied to the transducer when an ear-sealing earbud of which it is a component is inserted into or removed from an ear canal. Conversely, for applications where less displacement is needed, an elastic strain limit as low as 10 percent may be sufficient.
[0051] The piston and suspension are shown in more detail in FIGS. 2A-2C. FIGS. 2A and 2B show top and bottom views of the piston and suspension surrounded by the silicon substrate 200 from which they are formed. In FIG. 2A, the layer of material 202 (wavy lines) from which the suspension 106 is formed can be seen to extend over the entire top surface 204 of the piston 102, and over the support ring 206 that forms the top edge of the housing 104 in FIG. 1. The material 202 is cut out above the gap between the support ring 206 and the surrounding substrate in FIGS. 2A and 2C but intact in FIG. 2B, to assist in visualizing the construction. The bottom view 2B and side sectional view 2C show that the underside of the piston may consist of a pattern of rings 208 and ribs 210, with voids 212 between them etched in the silicon. This provides stiffness to the silicon piston while decreasing its weight relative to a solid disk. In other examples, a flat plate of silicon is sufficiently stiff, and the ribs and rings are not needed for stiffness, though similar structures, or just the outermost ring 208, may be needed due to the fabrication process, as discussed below. The sectional view also shows a layer 216 of SiO.sub.2, which will be explained below.
[0052] FIGS. 3A and 3B show one example of how the piston and suspension can be connected to the rest of the transducer. In FIG. 3A, the housing and bobbin, with the magnet, coin, back plate, and voice coil already assembled to them, are dipped into a shallow pool of adhesive 300 in order to apply a uniform bead of adhesive to one end of the housing. Preferably, the bead is sized to fill the gap between the outer support ring and the inner surface of the housing without excessive squeeze-out of adhesive. In other examples, the magnet, coin, and back plate are not attached until later. Then, in FIG. 3B, the bobbin is set on the piston 102, and the housing 118 is set on the outer ring 206. The adhesive is cured, and the transducer is ready for further processing, such as attaching or dressing lead-outs from the voice coil. In some example, the lead-outs extending from the voice coil are dressed before the bobbin is attached to the piston. In some examples, the bobbin and housing are attached to the piston and ring, respectively, before the ring is cut away from the rest of the substrate. This can make it easier to fix the location of the piston and ring when making the attachment. Further, a large number of bobbins and housings can be attached to a full wafer of pistons and rings all at once, using an appropriate fixture.
[0053] FIG. 4 shows a detail of the cross-section of the transducer, with dimensions of one example implementation. Other implementations may have quite different dimensions. In this example, the suspension is formed from a layer 202 of liquid silicone rubber (LSR) 10-500 m thick depending on desired suspension stiffness, formed by spin-coating the LSR on the silicon substrate. In some examples, the LSR layer is 30-80 m thick, and in one particular example, it is about 50 m thick. The piston top is between 10 and 100 m thick, and in some cases around 50 m thick, and is separated from the LSR by a 0.25-2 m thick layer of SiO.sub.2 thermal oxide and/or 5-50 nm of Cr or other suitable material, as discussed below with regard to the fabrication process. The outer ring 208 of the piston 102 is 50 m thick, and it is separated from the support ring 206 by a small gap 214 of around 300 m. The support ring provides an adhesion area for the LSR at the top surface of the substrate, and includes a thinner wall, around 75 m thick, extending down the inner face of the gap, providing a lip where the wall of the main housing may be attached. These dimensions allow the completed transducer to have an outer diameter only 4 mm acrosssubstantially smaller than typical electrodynamic (voice coil moving a diaphragm) transducers (only one outer edge is shown in FIG. 4). Smaller sizes may be achieved, though with less space available inside the bobbin for the magnet and coin. With a magnet as small as 1.5 mm, a total transducer diameter of 3 mm may be achieved. Larger sizes may also be built using this method, though the piston may need to be thicker or have more reinforcing ribs as the aspect ratio (diameter to height) increases.
[0054] As shown in this example, the bobbin has an outer diameter matched to the inner diameter of the outer ring of the piston, so that the bobbin is contained inside the outer ring. This design contains any extra adhesive to the inside of the piston and outside of the housing ring, i.e., away from the gap between the piston and the housing, unlike in the example of FIG. 3B. Similarly, attaching the housing 118 to the outer periphery of the support ring keeps the adhesive for that joint out of the gap.
[0055] FIGS. 5A-5K show a cross-section of a silicon wafer as it goes through an example MEMS fabrication process to form the piston and suspension. Other MEMS processes, with different technologies used for patterning, masking, and etching may be used, with accordingly different process steps. The etch depths mentioned below are based on a 300 m thick Si wafer and may be adjusted to achieve the desired characteristics of the Si piston, e.g., mechanical stiffness, moving mass, etc. The process steps are as follows: [0056] 1. Layers (504, 506) of thermal oxide (SiO.sub.2) are grown on the top and bottom surfaces of a 300 m thick Silicon wafer 502. (FIG. 5A) [0057] 2. A 5-50 nm thick layer 508 of Chromium is deposited on the top by physical vapor deposition (PVD). The Cr will serve as an etch-stop for later steps; other appropriate materials may be used. (FIG. 5B) [0058] 3. A 50 m thick layer 510 of LSR is spin-coated on top of the Cr and cured. Thinner or thicker layers of LSR may be used, based on the properties of the LSR and the desired amount of excursion and stiffness in the speaker. (FIG. 5C) [0059] 4. Photoresist 512, 514 is spin-coated onto both sides. (FIG. 5D) [0060] 5. The bottom side is masked (516) and exposed to an appropriate light source to activate the photoresist 512. (FIG. 5E) [0061] 6. The photoresist layer is developed and used to mask reactive ion etching (RIE) or HF etching of the bottom SiO.sub.2 layer 506. (FIG. 5F) [0062] 7. The developed photoresist 512 on at least the lower surface is stripped and a new coating 518 is spin-coated. (FIG. 5G) [0063] 8. Another mask 522 is used to expose the photoresist 518 on the bottom side. (FIG. 5H) [0064] 9. The photoresist 518 is developed and used to mask deep reactive ion etching (DRIE) through 50 m of the bottom of the Si wafer to create channels 524, 525 (note that these are circular channels in the wafer, viewed twice each in the cross-section). (FIG. 5I) [0065] 10. The bottom layer of photoresist 518 is stripped, and DRIE is used again to etch through the remaining 250 m of the silicon wafer (FIG. 5J). Where the first DRIE etch was performed, the second etch goes completely through the wafer, extending the channels 524, 525 to the SiO.sub.2 layer 504; the area that was protected by the second mask during the 50 m etch remains 50 m thick, as only 250 m is removed, forming the plate 526 of the piston and the top surface of the support ring. The areas protected by the first mask remain protected by the SiO.sub.2 506 left behind after the RIE etch in step 6, and form the rings of the piston and housing and any other full thickness features, such as the stiffening ribs and rings mentioned above (not shown). In some examples, full-thickness features are also used to manage the DRIE process. [0066] 11. The remaining SiO.sub.2 506 at the bottom layer and at the top of the now-open channels 524, 525 between the piston and the housing is removed using RIE or HF, with the Cr layer 508 serving as an etch-stop to prevent the RIE or HF from etching the underside of the LSR layer 510 after etching the top SiO.sub.2 layer 504 via the channels 524, 525. (FIG. 5K). The remaining photoresist layer 514 covering the LSR 510 is stripped.
[0067] The process shown above etches a channel 525 through the wafer around the outer support ring, allowing the piston/support ring/suspension unit to be cut out of the substrate. Many such units can be formed simultaneously in a single substrate, held in place by the LSR layer, and cut out as needed by either mechanical means, RIE, or laser-cutting. The inner wall of the bulk Si remaining outside the outermost channel 525 may serve as an alignment guide to the cutting process. As noted above, housings and bobbins may be attached to the support rings and pistons in bulk before they are cut out of the substrate, and the housings may also serve as alignment guides for the cutting operation. Curing the LSR layer helps control the pretension in the surround, to make the stiffness of the surround more linear. Without pretension, bending stiffness dominates near the neutral axial position of the piston (with no magnetic forces applied to the voice coil). At some piston excursion, the tensile stresses in the surround begin to dominate and cause the stiffness to increase. The pretension due to curing makes the overall stiffness greater but much more linear. In some examples, curing the LSR at 150 C. roughly doubles the near-neutral position stiffness.
[0068] Another process flow is shown in FIG. 6A through 6M. This process begins with a Silicon-on-insulator (SOI) wafer 600 and delays the application of the LSR layer to late in the process, which may be more compatible with some MEMS fabrication workflows. The process steps are as follows: [0069] 1. The process begins with a SOI wafer having a first layer 602 of Si, oxide layers 604 and 608 on either side of the first Si layer, and a very thin (2-10 m) second Si layer 606 bonded on top. (FIG. 6A) [0070] 2. A single layer 610 of photoresist is applied to the bottom of the wafer. (FIG. 6B) [0071] 3. The bottom side is masked (612) and exposed to an appropriate light source to activate the photoresist 610. (FIG. 6C) [0072] 4. The photoresist layer is developed and used to mask reactive ion etching (RIE) or HF etching of the bottom SiO.sub.2 layer 608. (FIG. 6D-E) [0073] 5. The developed photoresist 610 is stripped and a new coating 614 is spin-coated. (FIG. 6F) [0074] 6. Another mask 616 is used to expose the photoresist 614 on the bottom side. (FIG. 6G) [0075] 7. The photoresist 614 is developed to create a new mask that covers the remaining SiO.sub.2 608 and part of the main silicon layer 602. (FIG. 6H) [0076] 8. Deep reactive ion etching (DRIE) through 50 m of the bottom of the Si layer 602, masked by the photoresist 614, creates channels 618, 620 (note again that these are circular channels in the wafer, viewed twice each in the cross-section). (FIG. 6I) 9. The bottom layer of photoresist 614 is stripped, and DRIE is used again to etch through the remaining 250 m of the silicon wafer (FIG. 6J). As before, where the first DRIE etch was performed, the second etch goes completely through the wafer, extending the channels 618, 620 to the top SiO.sub.2 layer 604; the area that was protected by the second mask during the 50 m etch remains 50 m thick, as only 250 m is removed, forming the plate 622 of the piston and the top surface of the support ring. The areas protected by the first mask remain protected by the SiO.sub.2 608 left behind after the RIE etch in step 4, and form the rings of the piston and support ring and any other full thickness features, such as the stiffening ribs and rings mentioned above (not shown). In some examples, full-thickness features are also used to manage the DRIE process. [0077] 10. The remaining SiO.sub.2 608 at the bottom layer and at the top of the now-open channels 618, 620 between the piston and the housing is removed using RIE or HF. (FIG. 6K) [0078] 11. A 50 m thick layer 622 of LSR is now spin-coated on top of the top Si layer 606 and cured. Thinner or thicker layers of LSR may be used, based on the properties of the LSR and the desired amount of excursion and stiffness in the speaker. (FIG. 6L) [0079] 12. To release the piston 622, the Si of the thin top layer 606 is etched using an isotropic XeF.sub.2 etch. This etch is effectively masked by the much thicker (even where nearly etched through) bottom Si layer 602while 5 m of the piston layer may be lost, 45 m remain, combined with the 5 m of the top layer that are protected between the bottom layer and the LSR. Vertical Si areas will not be etched as they are still protected by a passivation layer deposited during the DRIE step. Other isotropic or anisotropic etching techniques (e.g., RIE using chlorine or fluorine chemistries, KOH, TMAH) may be used instead of XeF2 for this release step.
[0080] As compared to the first example, because the LSR is added late in the process, the top layer of photoresist is not needed.
[0081] A number of implementations have been described. Nevertheless, it will be understood that additional modifications may be made without departing from the scope of the inventive concepts described herein, and, accordingly, other embodiments are within the scope of the following claims.
