Patent Application
20080278542
The disclosure relates to micro-fluid ejection heads and, in a particular exemplary embodiment, to micro-fluid ejection heads having a reduced number of input/output address lines for ejection heaters disposed on a non-semiconductor substrate.
Micro-fluid ejection devices such as ink jet printers continue to experience wide acceptance as economical replacements for laser printers. Micro-fluid ejection devices also are finding wide application in other fields such as in the medical, chemical, and mechanical fields. As the capabilities of micro-fluid ejection devices are increased to provide higher ejection rates, the ejection beads, which are the primary components of micro-fluid devices, continue to evolve and become more complex and more costly to manufacture.
Conventional micro-fluid ejection heads are designed and constructed with silicon micro-fluid ejection head chips that include both the ejection actuators for ejection of fluids and logic circuits to control the ejection actuators. However, the silicon wafers used to make silicon chips are currently only available in round format because the basic manufacturing process is based on a single seed crystal that is rotated in a high temp crucible to produce a circular bouts that is processed into thin circular wafers for the semiconductor industry.
The circular water stock is very efficient for relatively small micro-fluid ejection head chips relative to the diameter of the wafer. However, such circular wafer stock is inherently inefficient for use in making large rectangular silicon chips such as chips having a dimension of 2.5 centimeters or greater to provide a larger ejection swath dimension. In fact the expected yield of silicon chips having a dimension of greater than 2.5 centimeters from a circular wafer is typically less than about 100 chips from a six inch diameter wafer. Such a low chip yield per wafer makes the cost per chip prohibitively expensive.
In order to provide an ejection swath of greater than 2.5 centimeters, multiple semiconductor substrates may be attached to a fluid reservoir. However, alignment of individual multiple substrates is difficult and time consuming.
Another approach to providing a greater swath dimension is to provide separate substrates for the heaters and logic/driver devices. In that instance, the heater substrates may be made of relatively large, non-semiconductor materials while the logic/driver devices are provided on a semiconductor substrate that is electrically connected to the heater substrate. While this approach overcomes alignment problems associated with multiple substrates, it may require a significantly large number of input/output lines connecting the two substrates. For example, if a non-semiconductor substrate contains 10,000 heaters, 10,000 wiring connections may be required between the logic/driver substrate and the heater substrate in order to address each heater individually.
Accordingly, there is a need for improved structures and methods for making micro-fluid ejection heads, particularly ejection heads suitable for ejection devices having an ejection swath dimension of greater than about 2.5 centimeters.
In one of the disclosed exemplary embodiments, a micro-fluid ejection head is provided that has N actuators on a first substrate and logic capable of driving the N actuators on a second substrate. The first and second substrates are electrically interconnected with less than N electrical connections.
In other exemplary embodiments, each ejection actuator is associated with a diode that may be selected from vertically aligned and laterally aligned diodes. The diodes enable the use of row and column logic devices for activation of actuators with a reduced number of address line connections between the first and second substrates.
An advantage of the exemplary embodiments is that they may provide improved micro-fluid ejection heads of greater dimensions without adversely increasing a number of electrical connections required to activate the actuators. Another advantage of exemplary embodiments is that multiple process steps may be readily combined to provide structures having the reduced number of electrical connections.
Further advantages of the exemplary embodiments may become apparent by reference to the detailed description of the exemplary embodiments when considered in conjunction with the following drawings illustrating one or more non-limiting aspects of thereof, wherein like reference characters designate like or similar elements throughout the several drawings as follows:
As described in more detail below, embodiments of the disclosure relate to non-conventional substrates for providing micro-fluid ejection heads. Such non-conventional substrates, unlike conventional silicon substrates, may be provided in large format shapes to provide large arrays of fluid ejection actuators on a single substrate. Such large format shapes are particularly suited to providing page wide printers and other large format fluid ejection devices.
According to the disclosure, a substrate 10 (
In order to provide a surface finish suitable for depositing fluid ejection devices and thin film conductive layers on the device surface 14 of the substrate 10, the device surface 14 of the substrate 10 may be polished to a fine finish and, if desired, coated with a planarizing layer. Polishing alone may be sufficient to provide a surface roughness of less than about 7.5 nanometers, which is generally a sufficiently smooth surface. If not, a layer of glass (for example boro-phospho-silicate glass, BPSG) may be applied as by spinning or by chemical vapor depositing (CVD) onto the device surface 14 of the substrate 10. The techniques for applying the planarizing layer are well known in the semiconductor industry for coating silicon devices, but are not commonly used for coating non-conventional substrates such as substrate 10. There is a greater requirement for smoothness and planarity of the device surface 14 because of die deposition of fluid ejection devices 18 on the device surface 14 adjacent to a fluid supply slot 16 formed in the substrate.
After planarization of the device surface 14 of the substrate 10, a thermal conductive layer may be deposited in a fluid ejection actuator area of the substrate 10 adjacent to the slot 16 and the fluid ejection actuators 18 and conductors therefor, for example, a thin film resistor layer and an anode and a cathode conductor layer, may be deposited adjacent to the thermal conductive layer. The thin film resistor layer and conductor layer may be patterned and etched using well known semiconductor fabrication techniques to provide a plurality of the fluid ejection actuators 18 on the device surface 14 of the substrate. Suitable semiconductor fabrication techniques include, but are not limited to, micro-fluid jet ejection of conductive inks, sputtering, chemical vapor deposition, and the like.
In order to activate the ejection actuators 18 disposed on the non-semiconductor substrate 10, drivers and/or logic devices 20 are electrically connected to the actuators 18.
According to an exemplary embodiment of the disclosure, a diode may be provided in series with each actuator 18 so as to drastically reduce the number of wiring connections that are required between the actuators 18 and the driver and/or logic devices 20. For example, heater actuators 18 may be arranged in rows and columns with the heaters 18 in each row sharing a common low side ground switch (PET) and heater/diode pairs in each column sharing a common high side Vcc rail switch (PET). In this way the number of wiring connections required for individual heater 18 addressability may be reduced from N (e.g., 10,000) to 2N1/2 (e.g. 200).
The foregoing heater/diode pairs 18/30 in rows and columns are illustrated in the wiring schematic in
Of course if the diodes 30 are disposed on a remotely separate substrate from the substrate 10 containing the actuators 18, a wiring problem may still exist. For example, since each actuator 18 is uniquely connected in series with one diode 30 in the wiring scheme illustrated in
In order to solve the aforementioned wiring connection issues, the thin film diode 30 may be created on the substrate 10 in series with each actuator 18. However, diode fabrication requires a p-n semiconductor junction or a metal-semiconductor interface. Accordingly, a semiconductor material may need to be deposited on the non-semiconductor substrate 10 to provide suitable diode functionality.
In addition to the need for a semiconductor material, the diode should be capable of switching rapidly so as to avoid degrading a pulse signal experienced by the actuator 18 because fire pulses are on the order of 1 microsecond (μsec) duration. For a one μsec fire pulse, a diode's switching time should not exceed 100 nanoseconds (ns). A diode's switching speed is affected both by the charge carrier mobility in the semiconductor material and for a p-n junction diode by charge carrier recombination times in the depletion region when the voltage polarity is switched. Mobilities are much higher for single crystalline semiconductor materials than for polycrystalline or amorphous forms of the same materials. For example, single crystal silicon has an electron mobility of about 1400 cm2/V-s, amorphous silicon has a mobility of less than 1 cm2/V-s, and polycrystalline silicon mobilities may vary between 1 and 400+ cm/V-s depending on the degree of crystallinity, grain size, grain boundary charge trapping sites, etc.
In order to illustrate the importance of charge carrier mobility for a diode consider the following example in Table 1. A diode having a 0.2 μm diode channel length with a 1 V drop may require about 4 ns for an electron to traverse the channel length distance in single crystal silicon with mobility on the order of 1000 cm/V-s. The same channel length would take more than 4000 ns to traverse in amorphous silicon with a mobility of 1 cm/V-s and about. 40 ns to traverse in polysilicon with a mobility of 100 cm/V-s. Mobilities of less than about 100 cm/V-s are insufficient to guarantee switching speeds of less than 100 ns for this geometry.
As described above, as-deposited polysilicon has fairly small grain sizes and typically does not have the greater than 100 cm/V-s mobility required for switching speeds of less than 100 ns. Historically, silicon has been annealed, at a high temperature in excess of 1000° C. to achieve higher mobilities. Such harsh processing drastically constrains the range of substrate choices available and is expensive and/or time consuming.
However, laser recrystallization may he used to achieve higher mobilities in polysilicon without the harsh processing of high temperature annealing. Laser recrystallization of polysilicon provides what is referred to as “low temperature polysilicon” (LTPS) because heating is locally focused in a thin silicon layer of the polysilicon. Excimer laser (ELA) recrystallization applies focused energy pulses that melt silicon in a thin layer near a top surface of the polysilicon. Upon cooling, the molten silicon resolidifies to polysilicon with large grain sizes and fairly high mobilities (80+ cm/V-s). Layers or substrates underlying the molten silicon do not experience high temperatures. There are a number of different variations that fall under the category of LTPS. Among these are:
With respect to devices made by polysilicon recrystallization techniques, U.S. Pat. No. 6,541,316 to Toet et al. describes a method of integrating laser crystallized polysilicon P-N junction and Schottky diodes in an n×n array of MRAM cells in a memory device. Such Schottky diodes may perform a suitable addressing I/O reduction function similar to that required for an array of thin film heaters as described above with reference to
Schottky diodes typically have fast switching speeds. A Schottky interface is a metal-semiconductor interface. The difference between the work function of the metal and the Fermi energy in the semiconductor results in a potential barrier or impediment to current flow called the “Schottky Barrier”. Schottky diodes formed from n-type semiconductor-metal interfaces are fast switching relative to P-N junction diodes because such diodes are majority carrier devices (i.e., conduction is by electrons e- only and not holes). Schottky diodes typically do not exhibit the relatively slow hole/electron recombination that takes place in the depletion region in P-N junction diodes when voltage polarity is reversed.
As illustrated by the matrix of actuators 18 and diodes 30 in
P=I
2
R=I
2×(number of squares)×ρ
A thin film metallization layer having a thickness of less than about 2 μm and a width of less than about 100 μm may be incapable of carrying the kind of current that is expected from the multiple fire event described above without unacceptable power dissipation and associated substrate heating. However, larger ground bus widths may require a wider substrate for support of the ground bus 36. For example, the matrix addressing scheme of
One exemplary embodiment of the disclosure provides a thin film diode on a non-semiconductor substrate such as ceramic or glass having a length dimension of greater than 2.5 centimeters. The embodiment uses LTPS technology to provide a high mobility polysilicon layer on the substrate 10 for the thin film diodes 30. The process integrates the thin film processing steps required for actuator fabrication with those required for diode fabrication such that several steps are shared and the diodes may be created with a minimal number of incremental processing/mask/photo processing steps. Exemplary embodiments are provided for vertical and lateral diode construction and for Schottky as well as P-N junction diodes in
In a first exemplary process, an integrated actuator and vertically oriented Schottky or P-N junction diode process is described. The thin film processing steps for a vertically oriented Schottky diode are illustrated in
The purpose of the planarization/thermal barrier layer is two-fold. First, the presence of defects larger than about 75 angstroms in a heater surface can provide nucleation points for heterogeneous nucleation. Thermal micro-fluid ejection processes rely on homogeneous nucleation at the superheat limit of the fluid to provide the explosive energy required for jetting events to take place.
The second purpose of the planarization/thermal barrier layer is to provide a thermal barrier such that the energy from the electrical heating pulse is directed into the fluid drop and is not dissipated into the substrate. On a ceramic substrate, the thermal barrier layer is optimally about 1 to 2 μm thick. Thicker barrier layers inhibit thermal dissipation after the firing event and can limit maximum firing frequencies to those achievable on a glass substrate.
For a glass substrate, the steps described in
At this point in the process whether using a glass substrate or a glazed ceramic substrate it is typical in the display panel industry to deposit a diffusion barrier layer consisting of CVD SiO2 and Si3N4. The diffusion barrier layer is typically about 1000 Angstroms thick and prevents diffusion of alkaline species from the glass substrate or planarizing glaze layer into the semiconductor silicon layer which is deposited adjacent to the diffusion barrier layer. Accordingly, a diffusion barrier layer may be required for the same purpose in this process.
In a second step of the process, illustrated in
In step three, also illustrated in
In step four,
In step six,
In step seven, illustrated in
In step nine, a cavitation layer 54 such as sputtered Ta/TaN, may be deposited adjacent to the dielectric material 52 as shown in
In step ten, illustrated in
In an alternate embodiment, a vertical diode/actuator process may be applied to a p-n junction diode. In this embodiment, there is no longer a consideration of Schottky barrier potential in the selection of first metal layer 42 and second metal layer 58 (steps three and ten). However, the first metal layer 42 may still be selected for its refractory/diffusion barrier properties. In order to effect a P-N junction process the LTPS process (step six) above is modified by depositing amorphous silicon (CVD or Sputtering) with N or P type ion implant or deposition. The polysilicon has P+ doping (typically boron) for N type silicon or N+ (arsenic/phosphorus) implant or deposition for P type silicon to create a p-n junction within the silicon. A suitable LTPS process such as ELA, SLS or some other suitable LTPS laser recrystallization process is used to recrystallize the silicon and activate/drive the dopant into the silicon. In an alternative process, the doping profile may be reversed to reverse the polarity of the diode.
In a second exemplary process, an integrated actuator 18 and laterally oriented Schottky or P-N Junction diode is provided as illustrated in.
As described above, the purpose of the planarization/thermal barrier layer Is two-fold. First, the presence of detects larger than about 75 angstroms in a heater surface can provide nucleation points for heterogeneous nucleation. Micro-fluid ejection processes rely on homogeneous nucleation at the superheat limit of the fluid to provide the explosive energy required for jetting events to take place.
The second purpose of the planarization/thermal barrier layer is to provide a thermal barrier such that the energy from the electrical heating pulse Is directed into the ink drop and is not dissipated in the substrate. On ceramic substrates, the thermal barrier layer is optimally about 1 to 2 μm thick. Thicker thermal layers inhibit thermal dissipation after the firing event and can limit maximum firing frequencies to those achievable on a glass substrate.
For a glass substrate 10, the steps described in
At this point in the process whether using a glass substrate or a glazed ceramic substrate it is typical in the display panel industry to deposit a diffusion barrier layer consisting of CVD SiO2 and Si3N4. The diffusion barrier layer is typically about 1000 Angstroms thick and prevents diffusion of alkaline species from the glass substrate or planarizing glaze layer into the semiconductor silicon layer which will be deposited on top of if. Accordingly, a diffusion barrier layer may be required for the same purpose in this process.
In step two,
In step three,
In step four of the process, illustrated in
In step five, also illustrated in
In step six,
In step seven, illustrated in
In step eight, a cavitation layer 78 such as sputtered Ta/TaN, may be deposited adjacent to the dielectric material 76. The dielectric material 76 and cavitation layer 78 may then be photoimaged and developed using a photoresist material 80 as shown in
In step nine, as shown in
In another alternate embodiment, a lateral diode/actuator process may be applied to a p-n junction diode. In this embodiment, there is no longer a consideration of Schottky barrier potential in the selection of first metal layer 72 and second metal layer 82 (steps five and nine). However, the first metal layer 72 may still be selected for its refractory/diffusion barrier properties. In order to effect a P-N junction process the LTPS process (step two) above is modified by depositing amorphous silicon (CVD or Sputtering) with N or P type ion implant or deposition. The polysilicon has a P+ doping (typically boron) for N type silicon or N+ (arsenic/phosphorus) implant or deposition for P type silicon to create a p-n junction within the silicon. A suitable LTPS process such as ELA, SLS or some other suitable LTPS laser recrystallization process is used to recrystallize the silicon and activate/drive the dopant into the silicon. In an alternative process, the doping profile may be reversed to reverse the polarity of the diode.
For example, a variety of techniques may be used to form thick film metallization channels in ceramic or glass substrates prior to depositing metallization for the thin film address lines 88. The channels may be formed in the substrate so that ground bus lines 90 with adequate conductivity to accommodate simultaneous fire situations may be provided as described above. Additionally, techniques may be used to embed metal underneath and adjacent to the actuator 18 location for thermal dissipation reasons. Once the embedded thick film bus lines 90 in place the substrate 10 may be planarized as described above with reference to
For ceramic substrates obtained from fairly high purity fired wafers, such as wafers having an Al2O3 content ranging from 96 to 99.6 wt %, or as east tapes in a green state, several methods may be used for embedding the thick film bus lines 90 in the ceramic.
In a first process, microchannels for the thick film bus lines 90 are formed in a ceramic green tape. The base ceramic structure may be made from tape cast layers using available co-firable ceramic material sets, either Low or High temperature co-fired ceramics, based on Al2O3 or AlN ceramic materials.
Channels for the thick film bus lines 90 may be made in the ceramic substrate in a variety of ways. For a low temperature co-fired ceramic (LTCC) that is a composite of glass and ceramic Al2O3, a tape layer in the green state may be punched through or CNC milled/drilled before mating and laminating the tape to other green tape layers. Alternatively, hot embossing may be used to form channels in the LTCC material at pressures of about 1000 to about 3000 psi and temperature ranging from 50° C. to about 150° C. for 5-20 minutes. If necessary, the use of sacrificial volume materials (SVM) and other Insert materials can be used to fill in the channels to hold channel dimensions during firing of the green ceramic.
Once formed, the channels in the green state may be filled with conductor pastes containing particles of Ag, Au, Cu or other conductive metals, using a screen printing process. Metallization pastes, such as copper conductor paste, silver conductor paste, gold, conductor paste, and the like are selected to be compatible with the ceramic tape formulations. The conductor paste filled channels are co-fired with the green ceramic under appropriate conditions, e.g., nitrogen environment for copper, and temperatures ranging from about 600° to about 900° C.
The thick film bus lines 90 may also be formed in green ceramic tape using a hot embossing process. According to the process, fine metal wires having appropriate dimensions are placed on top of the green tape layer and then electric current is forced through the wires by a driving circuit designed to produce constant resistance in the load wires and thereby constant temperature of a level sufficient to allow the ceramic tape material to flow under the heat and applied pressure and the wires to become embedded into the ceramic tape. The wires are then disconnected from the drive circuit and trimmed flush with the edge of the ceramic at each end of the substrate. The wires may function as independent circuit conductors or bus lines 90.
Microchannels and conductor lines may also he formed in a fired ceramic layer. Channels may be formed in a fired ceramic substrate in a number of ways. One method of forming channels in a fired ceramic substrate involves the use of wet HF etching of the substrate through an appropriately patterned metal mask. Other standard approaches for cutting channels in the fired ceramic material include, but are not limited to, laser cutting, dicing, and water jet machining of the substrate. The bus lines 90 may then be screen printed or the ceramic channels filled by using appropriate metal pastes such as described above. For more closely spaced bus lines 90 that are 50 microns wide, photoimageable metal pastes may be used through UV exposure and developing with sodium carbonate or appropriate solvents and then firing at about 850° C. Yet another method for providing the thick film bus lines 90 is through the use of digital printing techniques followed by annealing at a temperature of less than about 400° C.
The thick film bus lines 90 may be formed in microchannels in a glass substrate by the following process. Open channels may he formed in the glass substrate by micromachining techniques involving the use of photosensitive masking materials and etching with HF solutions. After the open channels are formed, the glass substrate containing the open channels may be bonded anodically or by diffusion, to another glass substrate.
In an alternative process, illustrated in
The substrate 100 with the open channels 106 may be bonded to another unpatterned solid glass substrate 108 (or other substrate material) by soldering, diffusion bonding, gluing, and the like to close channels 106 from one side 110 thereof as shown in
In another alternative process, channels having the appropriate depth in a single photosensitive glass substrate may be formed by controlling the density of UV radiation energy and/or subsequent etch exposure times. It is possible to control energy density using variable gray scale mask patterns.
Once formed in the substrate 100, the channels 106 may be tilled with a screen printed metal conductor pastes 112 as shown in
If the thick film bus lines cannot be embedded in the substrate before thin film processing to provide the thin film address lines, the thick film bus lines may be applied over the top of the thin film address lines either on glass or ceramic substrates. In this case a dielectric material having vias/openings therein in appropriate locations to make a connection between thin and thick film metallization is applied to the substrate containing the thin film address lines. The challenge with an overprinted thick film is that most thick films require high temperature cure/sinter steps to burn off solvents/polymers. The thin films already patterned on the substrate may not withstand temperatures in excess of 400 to 450° C. Low temperature screen printable metal pastes are commercially available for the flat panel display industry and the photovoltaic cell industry.
At numerous places throughout this specification, reference has been made to a number of U.S. patents and/or patent publications. The relevant portions of all such cited documents are expressly incorporated in full into this disclosure as if fully set forth herein.
The foregoing embodiments are susceptible to considerable variation in their practice. Accordingly, the embodiments are not intended to be limited to the specific exemplifications set forth hereinabove. Rather, the foregoing embodiments are within the spirit and scope of the appended claims, including the equivalents thereof available as a matter of law.
The patentees do not intend to dedicate any disclosed embodiments to the public, and to the extent any disclosed modifications or alterations may not literally fall within the scope of the claims, they are considered to be part hereof under the doctrine of equivalents.
