US20140196274A1 - Thermoelectric evaluation and manufacturing methods - Google Patents

Abstract

A means for determining the electrical resistance and resistivity of thermoelectric material allows quality control at all steps in the construction of a bismuth telluride and antimony telluride thermoelectric generator. The method involves measuring negative thermoelectric voltage with no current flowing and then a measure of negative thermoelectric voltage while forcing known current through the material in the same direction as shorted to accurately determine thermoelectric resistance. A manual and automatic method of manufacturing thermoelectric rings using forcing current for in-process testing means.

Description

RELATED APPLICATIONS
• Several related applications disclose details for the manufacture of components of bismuth-telluride based thermoelectric elements that comprise thermoelectric generators and thermoelectric chillers. Application Ser. No. 11/517,882 is entitled “Thermoelectric device with make-before-break high frequency converter”. United States Patent Application 20030217766 is entitled “Torus semiconductor thermoelectric device” filed Nov. 27, 2003. Improved thermoelectric generators and thermoelectric chiller devices are disclosed in pending patent application Ser. No. 11/259,922 entitled “Solid state thermoelectric power converter” and Ser. No. 11/364,719 entitled “Bismuth-Tellurium and Antimony-Tellurium-Based Thermoelectric Chiller”.
TECHNICAL FIELD
• This invention relates to methods of assembly and novel testing procedures for the efficient manufacture of thermoelectric generators and thermoelectric chillers. With regard to the testing of components at various stages of manufacture unique methods allow reliable data for producing finished products with few defects. Various classical electrical evaluation methods are normally used for determining thermoelectric properties. Improved methodology involves thermoelectric material operated under heat flow conditions while measuring negative, resistive performance.
BACKGROUND ART
• The classical way to measure semiconductor and thermoelectrics alike has been to apply a voltage to “force current” through material while measuring voltage drop over a distance through a cross section to determine bulk resistance. The semiconductor industry was founded using V/I, 4-point probe methods to determine material resistance. From a determination of R, it is simple to mathematically determine bulk material resistivity based on resistance and physical dimensions. From Ohm's law, we know that I=V/R, R=V/I, R=resistance, V=voltage, and I=current R=ρ l/A, where ρ=resistivity, l=length, and A area. Using the same equation, and with careful attention to units;
• ρ=R A/l, so given the measured R and the dimensions of the material under the probes, the resistivity ρ (in Ohm-cm) can be easily determined. This methodology has been the gold standard for R and p determination for germanium, silicon, and some III-V compounds in semiconductor technology. However, this method has been found to be in error when thermoelectric materials are measured. Measurement values for thermoelectric material have been found to be off by a factor of 2 and more on the high side.
• There are problems associated with thermoelectric material measurement. Applying voltage to material under test seems to cause the material to develop a counter voltage that inhibits current flow in thermoelectric material thus changing the expected measurement. Some investigators have suggested pulse methods for determining resistivity measurements, thinking that measurements taken with very short voltage pulses might get around the material disturbance. Errors are found using direct current, pulse, and low and high frequency alternating current methods of measure.
• The following disclosure details a method that accurately predicts satisfactory the performance of components as well as the final properties of thermoelectric generators and chillers. A very simple measuring method that predicts exactly the performance of thermoelectric devices is based on operating the thermoelectric material under conditions that simulate actual solid state generator and chiller operation. With thermoelectric material tests under high current operation, the negative resistance characteristics of the material can be accurately determined. Negative resistance was found to have the same V/I slope regardless of the temperature difference, ΔT, and the voltage current ratio, V/I that was measured. Once ΔT is established in material when measuring negative resistance of thermoelectric material connected to large mass copper heat source and sink, changes in ΔT occur very slowly, allowing time to make accurate determination of current, and reduced negative voltage caused by the current flow. From measurements of −V₁, the voltage when the current, I=0 and −V₂, the voltage when current is allowed to flow, and the current, I_{(at −V2)}, the current at voltage V₂still at temperature ΔT, the negative resistance slope can be easily determined, including the zero-voltage-crossing for material with a particular ΔT. By forcing current to flow through thermoelectric material with ΔT induced, accurate in-process material measurements can be made. In-process material measurements are absolutely essential to successful, low cost manufacture of solid state generators and chillers, because completed solid state generators and chillers are composed of many elements and ohmic connections. Any element found faulty in assembly can be exchanged for one that performs to specification.
• Accurate measurements of thermoelectric material can be made by using the forcing current method to determine negative resistance for heated junctions, with or without metal contacts. Thermoelectric device performance can be measured accurately in ingot, wafer, coupon, and ring form, before and after contact bonding. This method can be used during all stages of assembly and final test, even used for non-destructive evaluation of returned defective product. Because thermoelectric products consist of an assembly of large numbers of elements, numbering into the hundreds and thousands, one faulty or misplaced part can render a sophisticated, high performance solid state generator or chiller useless. This measuring invention has the ability to accurately test elements of the product during assembly, as the product moves along the assembly line, and this makes all the difference in having all products shippable at final test, otherwise realize a less than 10% final test yield.
• This invention relates to a measuring system for determining negative resistance in thermoelectric material under thermal operation. Negative voltage measurements made on thermoelectric material while large current is caused to flow along with heat, is different than negative voltage measured with the same heat flow but without current flow. When the current flow through the thermoelectric material is increased until the negative voltage produced by a certain ΔT is reduced to zero volts, the current through the material is at a maximum for the ΔT of the device under test. Zero crossing current for the material under test can be increased by increasing ΔT, but the −V/I slope representing negative resistance of the material will stay the same as long as current does not exceed 1,000 Ampere per square centimeter for metal coated contacts. If current through the material is further increased, the voltage becomes zero and then a portion of the current flows in the positive resistance region. With make-before-break shorted ring thermoelectric generator and chiller devices, the thermoelectric driven ring operates in the negative resistance region and the up-converter and primary portion of the circuit operates in the positive resistance region, forcing current and current drag being equal, or −V=+V. This means −V measured across the thermoelectric ring will be driving current and +V measured across the make-before break switch part of the secondary is dragging with the same polarity as a resistor with current flowing in the same direction.
• Thermoelectric devices have been used for many years for specific applications where the simplicity of design warrants their use despite a low energy conversion efficiency.
• Resistance and resistivity measurements of thermoelectric material are difficult to make because the act of making measurements tends to disturb the material in such a way as to render measured results questionable. When current passes through bulk thermoelectric material, the current also drags heat with the current and this disturbs any voltage readings.
• This invention uses heat flow defined by the measuring system, ΔT determined by the physical dimensions of the device under test and the heat input and heat removed from the device by heat sink. The current passing through the device under test is limited to less than the thermoelectric device could produce on its own if it were short circuited by a super conducting wire. This way, the device operates with heat flow and current, the negative drive voltage at current being less negative than open circuit. Using only this measured data, all electrical parameters can be determined. The measuring technique is simple and accurate, allowing in-process use throughout the assembly, final test and even provides a non-destructive means to analyze customer returns.
• Previous to this invention, methods of measure for thermoelectric material included the “Harman Method”, as in T. C. Harman, J. H. Chan, and M. J. Logan, J. Appl. Phys. 30, 1351 (1957). These methods were previously considered the gold standard for thermoelectric measurements and investigation.
• A survey of measuring methods was made by J. D. Hinderman, “Thermoelectric materials evaluation program”, 3-M, Saint Paul. 04/1979, NASA/STI Keywords: Evaluation, Technology Assessment, thermoelectric material, thermoelectricity.
• Material test methods were also outlined by H. Iwasaki, M. Koyano, Y. Yamamura, and H. Hori, School of Material Science, JAIST, Tatsunokuchi 923-1292, Ishikawa, Japan. Center for Nano Materials and Technology, JAIST, Tatsunokuchi 923-1292, Ishikawa, Japan.
• It is a purpose of this invention to provide an accurate in-process evaluation method for determining the heat to electrical energy performance for thermoelectric material beginning with starting materials, carrying through to assembled and bonded devices. It is a further purpose of this invention to provide electrical evaluation for non-bonded and bonded elements that go into the fabrication of a thermoelectric device. It is a further purpose of this invention to provide electrical evaluation for non-bonded and bonded thermoelectric devices at the end points of manufacture.
• It is a further purpose of this invention to provide details for manual and automated methods for assembling thermoelectric generators and chillers.
• It is a further purpose of this invention to disclose unique manufacturing and assembly aspects for efficient manufacture of thermoelectric generators and chillers.
BRIEF DESCRIPTION OF DRAWINGS
• FIG. 1 illustrates the first stages of a manual assembly line method of making thermoelectric rings that provide the current to be converted to standard AC voltages.
• FIG. 2 illustrates subsequent assembly stages for completion of a bonded ring of thermoelectric coupons and an up-converter.
• FIG. 3 illustrates the classical method to determine material resistance in wafer form, as used in the semiconductor industry.
• FIG. 4 shows how a silicon ingot can be measured using a constant current source to determine material resistivity.
• FIG. 5 shows a fixture for testing a novel forcing current bench test method that can be used to accurately determine thermoelectric wafer and coupon performance.
• FIG. 6 shows the electronic circuit ofFIG. 5 where both p-type and n-type wafers can be evaluated simultaneously.
• FIG. 7 shows how a coupon is measured with a go-no go test fixture,
• FIG. 8 shows how a raw and coated wafer can be evaluated with a test system.
• FIG. 9 shows an evaluation method similar toFIG. 8 that can be used to evaluate green un-bonded and bonded thermoelectric rings.
• FIG. 10 shows how a completed thermoelectric generator or chiller ring can be evaluated for output capacity, resistance and voltage without a teardown.
• FIG. 11 shows how the slope of the V/I line represents a negative resistance of −1.04E-5 Ohms to mean that current can flow in this region without resistance until the negative driving voltage −V becomes a positive voltage on the right side of zero volts on the chart.
• FIG. 12 shows what happens to thermally induced junction(s) voltage when a fixed current is forced in the direction that current would normally flow if the junction(s) were short-circuited with a piece of super-conducting wire where the negative voltage decreases becoming zero at a current of 385 Amperes.
• FIG. 13 shows what happens when current is increased beyond the zero-voltage crossing point (385.5 amps), negative voltage becomes positive voltage to the right of zero-volts and system resistance also becomes positive.
• FIG. 14 shows what happens in the negative resistance region when the ΔT across the thermoelectric material is changed, V/I slope stays the same while negative voltage and current at zero-crossing change.
• FIG. 15 shows a V/I plot for a ring with 60 n-type wafers and 60 p-type wafers arranged as a ring as the ΔTs for thermoelectric material junctions are changed.
• FIGS. 16a, b, andcshows how a thermoelectric ring can be evaluated open-circuit voltage as in16a, in16bwith shorted current and in16cas a thermoelectric ring with up-converter attached.
• FIG. 17 illustrates an exploded view of a conductive wedge, p-type crystalline wafer, a hot fin, an n-type crystalline wafer, a cold fin, that comprise a coupon.
• FIG. 18 illustrates the final positions of the elements of the coupon with all elements situated on an indexing substrate carrier.
• FIG. 19 illustrates an assembled coupon situated on an indexing substrate, parts held in alignment with a clamp.
• FIG. 20 illustrates how a robot picks up a coupon and inserts it into a test fixture for on-line testing.
• FIG. 21 illustrates an on-line test fixture and dynamic test method for determining un-bonded coupon performance both before and after thermal bonding.
• FIG. 22 illustrates an automated solder paste dispenser that facilitates the automatic coating of coupon elements and coupons by robotic intervention.
• FIG. 23 illustrates a vacuum operated parts pickup device attached to robot jaw used for parts coating and parts placement.
• FIG. 24 illustrates a multi-color-marking device to aid in marking parts determined faulty by on-line testing by failure mode.
• FIG. 25 illustrates a coupon-making machine capable of assembling, on-line testing, marking, bin disposition, and brazing passed coupons.
• FIG. 26 illustrates the top view of the coupon-making machine with part handlers, delivery chutes, robot and conveyor bonding furnace.
• FIG. 27 illustrates a means of removing coupons from bonding furnace oriented for further processing by a second robot.
• FIG. 28 illustrates a side view of coupons coming off furnace belt to orient into an orienting chute.
• FIG. 29 illustrates a top view of a second robotic work station that assembles thermoelectric rings.
• FIG. 30 illustrates a top view of a second robotic work station that assembles and tests thermoelectric rings and places them on a belt furnace for bonding.
• FIG. 31 shows a ring entering and exiting a conveyer furnace that presents bonded product for further generator and chiller assembly.
• FIG. 32 illustrates a system for creating a casting mold for net-shaped thermoelectric wafers.
• FIG. 33 illustrates a system for pouring a tree of thermoelectric wafers causing each net-shaped crystal to grow single crystal as molten metal slowly cools to solidify and then reduced to near room temperature.
• FIG. 34 illustrates a tree of net shaped thermoelectric wafers removed from mold as single crystal caused by a seed crystal mounted in the mold runner.
• FIG. 35ashows a rectangular arrangement of linear coupons for a thermoelectric device.
• FIG. 35billustrates a triangular arrangement of linear coupons for a thermoelectric device.
• FIG. 35cshows an octagonal arrangement of linear coupons for a thermoelectric device.
DISCLOSURE OF THE INVENTION
• To illustrate this invention the figures listed above are drawn to show components of a few implementations of the invention. It should be understood that these figures do not in any way limit this invention as described in the claims.
• FIG. 1 shows the top view1 of a first phase of an operator assisted assembly line method for making use of wafers, fins and wedge to form ring components of a thermoelectric generator.Operator2 picks up an offsetfin3 from bin4 and inserts it in a two-sided paste dispenser5. The offset structure is shown in detail inFIG. 5.Operator2 then places p-type wafer6 from bin7 on the offset side solder region offin3 forming a partially completed p-type hot fin and places it on movingbelt8 forming a wafer hot fin assembly9.Operator10 takes afin3 frombin11 and applies paste from one-sided dispenser12 then places an n-type wafer13 frombin14 on the solder applied side offin3.Operator10 then places thisassembly16 onbelt8 with the fin-offset side upwards.Operator15 first picks up the n-typewafer fin assembly16 and places it in hand-heldalignment template17.Operator15 then picks up hot fin-wafer assembly9 and turns it over placing it over the n-typewafer fin assembly16 intemplate17 forming a straight hot-fin-cold-fin partially completed coupon assembly.Operator15 then picks upwedge18 frombin19 and coats it with a one-sided paste dispenser12 and places it solder face down over the p-type wafer of the fin-wafer assembly9 with the wedge taper end facing opposite to the direction of the belt.Operator15 takes aclamp20 frombin21 and secures the completed but un-bonded coupon formed intemplate17 to form a competed clampedunbonded coupon22.Operator15 places the completely assembledun-bonded coupon22 onbonding furnace belt23 offurnace24. Periodically complete and un-bonded coupons are tested intesting unit25 to be described in detail below inFIGS. 5,6 and7.
• FIG. 2 shows atop view26 of an operator assisted assembly line method for converting bondedcoupon assemblies27 into a ring component for a thermoelectric generator.Bonded coupons27 are shown emerging frombelt furnace24 falling into holdingchamber28. Coupons are cooled by fan29.Operator30 removesclamps20 from bondedcoupon27 and places clamps20 inbin21 for reuse.Operator31 places cooled bondedcoupons27 intotemplate32 whereoperator31 applies solder paste fromsupply33 to the wedge side ofcoupon27.Template32 is constructed to receivecoupons27 in only one orientation with the wedge side upwards.Operator30 serially places wedge-pastedcoupons34 intoassembly fixture35.Ring assembly fixture35 is comprised ofinternal ring support36 that supportsfins34 in the vertical position.Fixture35 has an irregular bottom support that only allows eachcoupon34 to be inserted into the fixture in one direction with regard to the wedge side. An up-converter37 is supported in theassembly fixture35 at the midpoint of thecoupons34. When the prescribed number ofcoupons34 have been installed inring assembly fixture35 aninsulated compression ring38 is placed and tightened around the formed ring39 at the wafer level. Each assembled ring39 is tested atstation40 prior to thermal bonding of the ring. The details of testing atstation40 are described below in detail. After completion and successful testing of ring39 usingstation40 the ring assembly39 is removed fromassembly fixture35. The assembled ring39 is placed on movingbelt41 for thermal bonding inoven42 where solder-pasted bondedcoupons27 become thermally bonded as a continuous ring when heated to a temperature of 260 degrees C. for 5 minutes.
• The above detailed presentation of a preferred manual assembly process is only one of a wide variety of methods that can be used to assemble a completed thermoelectric ring. Other procedures than accomplish the same final assembled product can be developed by those skilled in the art of assembly line processes.
• Periodic testing of components in the assembly method described above prevents components from compromising the performance of the final product. The testing portion of this invention, comprises a dynamic method for accurate measurement and evaluation for thermoelectric material that can be used throughout the thermoelectric device manufacturing process. The following discussion will show with voltage to current ratio graphical plotting, how a novel forcing current test method for determining negative resistance in thermoelectric material works by describing the significance and meaning of each plot. The discussion of the data demonstrates a simple, quick, and valid way to actively measure thermoelectric properties for several in-process configurations.
• FIG. 3 illustrates the classical method of determining material resistance in wafer form, as used by the semiconductor industry. Resistance is calculated using the voltage to current ratio, V/I, data along with physical dimensions of the material under the probe, i.e., thedistance47 between voltage probes. Determination includes wafer thickness and an estimation of current spreading in the wafer, along the test sample's width.Test apparatus43 usesvoltmeter44 andammeter45 to measurewafer46 using a constant current source of approximately 0.1 ampere while measuring the voltage. From wafer thickness, measured voltage fromvoltmeter44, known current fromammeter45, and thespacing47 between voltage probes the resistance in the vertical direction of the wafer can be calculated. Resistivity can be determined mathematically using Ohms law and the definition of conductivity, ρ, in Ohm-cm. This analysis can only be made on wafers without metal contacts. Wafers with metal contacts must be tested by other means that require either the removal of metal contacts or taking into consideration the conductivity of the metal contacts. The values obtained for semiconductor material differ widely from those obtained using the same procedure with thermoelectric material. A novel method for determination of the electrical characteristics of thermoelectric material is described beginning withFIG. 5.
• FIG. 4 shows the classical means48 for measuring electrical characteristics of asilicon ingot49 usingvoltmeter44 andammeter45 with a current source. Knowing the distance betweenprobes50 across the voltmeter probes, the resistance of sliced wafers from theingot49 can be determined in advance of slicing.
• FIG. 5 diagrams atest system51 that more accurately predicts thermoelectric behavior in wafers with and without metal contacts, coupons bonded and non-bonded, and thermoelectric rings bonded and non-bonded more accurately than semiconductor methods. The forced current technique with heat flow produces valid, reproducible measurements throughout thermoelectric assembly process. Measuringsystem51 is used inFIGS. 1 and 2 as in-process test instruments25 and40. InFIG. 5 resistance is being measured for acoupon27 consisting ofhot fin52 shown with offset pins,cold fin53 shown with offset pins, taperedwedge54, n-type wafer55 and p-type wafer56. Components oftest system51 includeheater57,clip58,fan59, and a constantcurrent power supply60 consisting of anammeter61,power source62 and a means to vary current63, andmomentary switch64.Probe65 and probe66 apply current frompower supply60 tocold fin53 andwedge54.Probe67 and68 connectvoltmeter69 acrosscold fin53 andhot fin52 respectively to measure n-type wafer55. To analyze the p-type wafer56,voltmeter70 connects acrosshot fin52 andwedge54 withprobes68 and71 respectively.Heater57 heatshot fin52.Fan59 blows air oncold fin53 to remove heat to ambient.
• Measuringinstrument51 is used to accurately determine thermoelectric wafer and coupon performance for various stages of assembly. The internal resistance of the thermoelectric wafer material, wafer contacts through heat conducting nickel coated copper fins can be obtained. Theheater57 is used to heat the hot fin of a thermoelectric coupon causing heat flow through p-type and n-type wafers. Heat flows through thermoelectric wafer material conducting through wafers tocold Cut53 andwedge54 heat drawn with the assist offan59. Voltage across the n-type wafer of the coupon is measured withprobe67 on the cold fin and probe68 onhot fin52. Voltage across the p-type wafer of the coupon is measured withprobe68 on the hot fin and probe71 onwedge54. As the heat begins to flow fromhot fin52 through the n-type thermoelectric material of the coupon tocold fin53, this also heats the p-type wafer56 with heat traveling to thewedge54. The voltage determines the temperature differential just as a thermometer or thermocouple. The negative voltage with no forcing current flowing is measured usingmeter69 is recorded to measure n-type negative voltage values at a heat flow thoughcoupon27. Negative voltages are obtained first with no current flowing throughmomentary switch64 ofpower supply60 in the of position.Meter69 is used to measure n-type negative voltage.Meter70 is used to measure p-type negative voltage.Momentary switch64 inpower supply60 is closed and negative voltage measurements are again noted along with the current frompower supply60. Reduced negative voltage measured withvoltage meter69 and70 are obtained by proper connection of the power supply leads from forcingcurrent supply60. This convention allows voltage values to be less negative when current is applied frompower supply60. Themomentary switch64 activated and new negative voltage recorded for eachvoltmeter69 and70 at known current, the internal resistance of n-type, p-type and coupon can be mathematically determined from the voltage and current measurements. The resistance data can be used to pass or fail a wafer, coupon or even a thermoelectric ring before thermally bonding. Testing thermoelectric elements in process saves material and increases final test yields of thermoelectric rings composed of 300 and more perfect elements for proper operation.
• FIG. 6 shows atest system72, similar totest system51 except only onevolt meter74 is used withprobes73 and67. In this case voltage drop toward zero when forcing current enabled represents both n-type and p-type wafers in series, therefore the resistance of the complete coupon in the bonded or un-bonded state can be determined.
• A typical coupon has ahot fin52, acold fin53 and awedge54. Thewedge54 can be used as a temporary heat sink to allow quick evaluation of thermoelectric material attached to wedge54 while it is cold and before it soaks too much heat to operate as an effective heat sink. This way, the material attached to the wedge can be evaluated using the forcing current method if measurements are performed quickly, the wedge serving as a temporary heat sink for short duration until it becomes too heated. If tests are performed quickly, the resistance is calculated from the voltage-current ratios, VA. These ratios define the slopes for both types of materials, n- and p-type. Resistance standards are set for a coupon to pass quality assurance testing.
• FIG. 7 showsproduction fixture75 inbox76 providingcurrent probes77 andvoltage measuring contacts78 to collect data from n-type and n-type material incoupon27. This method lends itself to automation under computer control with umbilical79.Fixture75 can also incorporate a temporary heat sink adjacent to the coupon'swedge54 to remove heat equally across both wafer types.FIG. 7 illustrates how the test procedure ofFIGS. 5 and 6 can be packaged inbox76 to conveniently evaluate non-bonded and bonded coupons as a part of the assembly line process. Under computer control negative voltage is measured after the coupon achieves a pre-set temperature betweenwedge54 andcold fin53, the −V1 recorded, current forced and −V2 measured at known current I. A computer computation is automatically made resulting in the illumination of either green go-lamp80 to indicate pass or red no-go lamp81 to indicate the coupon resistance value exceeds operating limit for proper thermoelectric ring operation.
• FIG. 8 shows anapparatus82 for using the forcingcurrent power supply60 ofFIG. 6 connecting tohot fin83 throughprobe84 andprobe85 tocold fin86 cooled byfan87. Raw thermoelectric n-type and p-type wafers as well as awafer88 with metal coating onwafer88 can be evaluated this way. Individual non-bondedthermoelectric wafer88 is aligned so as to protrude fromhot fin83 heated byheater89. Protrudingwafer88 is cooled on the other side bycold fin86. Voltage probes90 and91 fromvoltage meter92 make contact with the bare orcoated wafer88 or metal coating onwafer88. The same measuring technique used inFIG. 6 allows determination of individual wafer resistance parameters with and without metal coating.
• FIG. 9 illustrates a measuringsystem93 for determining electrical resistance in an assembled thermoelectric ring before bonding and after bonding.System93 withbox102 shows a constantcurrent source95 connected bycontacts96 and97 to power output leads98 and99 ofring94.Voltage meter100 also connects tooutput terminals98 and99 on thering94 side of current forcingconnections96 and97 so as to be independent of the contact resistance of the current source terminals. The hot fins of the ring are heated and negative voltage ofring94 is measured at temperature withvoltage meter100 and the value recorded as −V1.Momentary switch101 is closed and the negative voltage, −V2, ofring94 is recorded as measured withmeter100 at current I produced by constantcurrent power supply95. From −V1, −V2, and the current at −V2, the internal resistance ofring94 is determined. The same forcing current test method can be used as a nondestructive evaluation method for generator products returned from the field as shown inFIG. 10. A thermal gradient is first applied to the open circuitthermoelectric device94. A thermoelectric negative voltage measurement is then made with the open circuit ring onterminals98 and99, with zero current flowing. The accumulated voltage is indicative of ΔT between n-type and p-type wafers connecting the hot and cold fins for thering94 as a first measurement. This measurement is actually a negative voltage, −V₁, produced by ΔT that is caused by heating the hot fins and cooling the cold fins withmomentary switch101 in the open condition. Themomentary switch101 is closed to force approximately 50 amps frombox102, through the highcapacity load resistor95 properly adjusted to serve as a constant current supply. By proper application ofvoltage meter100 leads to ring94 to output leads98 and99, −V1, −V2 voltage can be measured. Current flows momentarily through thethermoelectric ring94 under test in the same direction current would normally flow if the thermoelectric device were shorted on itself or operating the primary windings of an up-converter through MOSfet switch banks. When forcing current flows through the thermoelectric device, the open circuit negative voltage −V1 is reduced and is made more positive, as −V2 with application of known current. The second negative voltage −V2 is recorded, along with the magnitude of the actual forced current I from powersupply load resistor95 flowing through thering device94. From the −V1, −V2 and I the resistance of the ring can be calculated.
• FIG. 10shows apparatus103 that represents athermoelectric generator ring94 with complete electronic drive and switch circuitry attached. Heat flow through hot fins to cold fins causes current104 to flow in thering94 whenMOSfet switch banks105 and106 are switched “on” and conducting current throughring94. Ring current is maintained continuous by thermoelectric effect in only onedirection104 by normally open make-before-break switches105 and106 located acrosspower output terminals107 and108. Current direction around the high frequencymagnetic core109 is switched oppositely by electronic circuit elements driven by inverted drive from the pulsewidth modulator chip110. Forcing current tests can be made to theheated ring94 of a thermoelectric generator by measuring across the thermoelectricring output terminals107 and108 with MOSfet switches105 and106 in the inactive normally open mode. Forcing current frompower supply95 as inFIG. 9 allows voltage probe measurements to be made atterminals111 and112 withvoltage meter100. Voltage −V1 is measured withvoltmeter100 onring94 with hot fins heated and cold fins cooled and with zero current. Voltage −V2 measured withvolt meter100 with forced current frommomentary switch101 switchedcurrent source95 allows the determination of internal resistance ofring94 from calculation including current104 I at voltage −V2. This measurement of the thermoelectric ring can be made with and without electronic drive circuitry connected with electronic drive inactive. InitiallyMOSfet switch banks105 and106 are normally open when pulsewidth modulator chip110 is open and inactive. Voltage probes111 and112 fromvoltmeter100 are connected acrosspositions107 and108 but closer to the ring. Theprobes111 and112 are connected with polarity to indicate a minus voltage V1 when the ring is heated and the MOSfet switches are open. The hot fins of the ring are heated and voltage across the ring is measured with thevoltmeter100 connected to result in negative voltage values. The voltage value is noted as −V₁with no current in the ring.Momentary switch101 is closed to force a known amount of current, I, about 50 amps thoughring94. This reduces negative voltage −V1 to a higher value, that is, less negative voltage as recorded fromvoltmeter100 and designated −V₂. From the values −V₁and −V₂and measured current I, the resistance of the ring is calculated. The resistance of the ring of 60 coupons should be less than 0.001 Ohms, both before and after solder paste bonding.
• The relevance of the test method was demonstrated using the material from a thermoelectric material library developed over a five-year period of time when wafer contact research was conducted. Samples from this library were re-measured over and over using the negative voltage, forced current method, comparing these results to classic 4-point and voltage current ratio methodology. The negative voltage, forced current method is by far the most accurate and can be used all along the production process for quality assurance. The examples below provide detail data to illustrate the methods and provide quality values for n-type wafers, p-type wafers, coupons, thermoelectric rings and thermoelectric generator.
• FIG. 11 shows the slope of the voltage to current line, V/I line, representing a negative resistance of −1.04E-5 Ohms when a hot fin of a coupon is heated. A negative voltage of −4E-3 was measured as −V1 at113 with no current flowing. With a forcing current of 71 amperes, a −V2 voltage of −3.0E-3 at114 is recorded.Power supply60 inFIG. 8 causes current flow in the same direction current would flow if coupon were shorted. A negative voltage in this region means that current can flow in a thermoelectric ring without resistance, up until the time the thermoelectric driving voltage becomes positive voltage, crossing zero from the left side of the graph to operate in the positive side (right) of zero on the X-Y chart. The slope of the line represents resistance (V/I) and because voltage is negative, −V/I represents negative resistance operation. When a current of 71 amperes is forced through the coupon inFIG. 11, the negative voltage is reduced to −3.0E-3 shown assloping line115. This data: −4.0E-3 volts, 0 amps and −3.1E-3 volts, 71 amps, amounts to a resistance of −1.04E-5 Ohms as calculated. The resistance is negative because voltages are negative and the current is in the positive direction that the up-converter allows current to be driven in the ring.
• FIG. 12 shows that the same resistance values are derived when the forcing current is increase to the point that −V2 becomes zero at a current of 385 amps at116. Before current is applied to the device under test with a thermal gradient present using either a heater block or the flow of hot air across hot fins, the coupon produces a voltage of −4E-3, and a zero current at117. A forcing current of 385.5 moves V1 voltage from −4E-3, zero current to a −V2 at zero volts,116. The forcing current of 385.5 is the same current at zero crossing that would occur if the heat-induced ring were shorted with a micro-ohm short. This performance is measured for a single thermoelectric wafer of a 2-wafer coupon plotted as tested inFIGS. 5 and 6 with a ΔT of approximately 80 C. Metal-coated wafers can also be accurately evaluated for electrical performance using the test setup shown inFIG. 8 by measuring voltage on protruding wafer sides rather than on the thermal or current driving connections.FIG. 12 also shows negative voltage (−V₁, −V₂) data plotted against currents zero and I at V=0, on X-Y coordinates with thermoelectric junction(s) maintained at an arbitrary ΔT byheater87 inFIG. 8.
• FIG. 13 shows the thermoelectric junction ofFIG. 12 when forcing current is further increased beyond that of zero-crossing118. The voltage becomes positive as the current increases beyond that of zero crossing as at points120. Current is forced beyond that of zero crossing for thermoelectric material at a set ΔT. The forcing current is in the direction current would normally flow if the junctions were short-circuited with a piece of super-conducting wire. Rather than pushing the current higher each time to achieve a zero crossing, V=0, a zero crossing can be easily calculated using the measured data −V₁, and −V₂, and then measuring I at −V₂. The linear equation Y=mX+b can be used to determine V/I which is −R, which is also the slope of the line and the resistance of the device under test. One value at119 shows −V with negative current. This condition occurs when forcing current in the ring is reversed. The linearity of the slope of voltage to current is seen for 5 measured points, 2 of which at 120 show positive voltage values. The resistivity (ρ) can be accurately obtained by ρ=R(A/L), which is resistance, times the area of the wafer, divided by the wafer thickness, all in centimeters, because the units for ρ are defined as being Ohm-cm in the SI system.
• FIG. 14 shows −V/I plots for various ΔTs for the same thermoelectric junction for the wafer measured inFIG. 11. The −R is the same for the material regardless of ΔT because all V/I slopes are parallel. For each line a different temperature was used for which a zero voltage current applies. At each different temperature differential the forcing current was applied until the voltage became zero. The calculation for resistance is the same because eachslope121,122, and123 are the same. This shows that any differential temperature between hot fins and cold fins can be used to analyze wafer, coupon, and ring resistance.
• FIG. 15 plots values measured for a thermoelectric ring of 60 coupons. The hot fins are heated and the cold fins cooled to produce −V1 shown at124, which amounts to −3 volts when the ring is open circuit and a self-induced current of 4,808 amperes due to ring shorting at125. The ring was then shorted partially by placing an up-converter with a resistance of +6.0E-4 Ohms, not shown, across the ring terminals. This addition reduces −V1 to −1.5 volts shown at126. The up-converter's resistance is mostly in the switch bank that is composed of parallel-connected MOSfets switches. Two make-before-break MOSfet switch banks allow 2,500 amperes to flow through the ring as shown at127. The same current produced in the ring circulates through ring shorted by up-converter.
• The R slopes124,125 inFIG. 15 are at a particular junction temperature differential for the ring. This represents the internal resistance for the ring. The X-Y line shifts to the right when the load of the up-converter is connected. If junction temperature differential is increased, the loaded −V1 becomes more negative, increasing the current in the up-converter. Wafer contacts are a limiting factor for the amount of current a ring may circulate reliably. Nickel contacts on tellurium based thermoelectric wafers have been found to operate reliably at 750 amps per square centimeter. Shorting of a thermoelectric ring on itself can lead to high current catastrophic ring failure. To operate properly at high power level, thermoelectric generator material must have proper metal contacts to have high current capacity. A load resistor must always be placed in series with a ring when shorting to prevent over-current destruction. An up-converter used as an output device for a thermoelectric ring can modulate ring current at a safe and continuous controlled current in the ring portion of the generator.
• FIG. 16athrough16cillustrates three different test modes for a heat driventhermoelectric ring94.FIG. 16ashows heated and cooledthermoelectric ring94 in open circuit mode with current being zero.Thermoelectric ring94 is made up of 60 bondedcoupons27 that measure −3 volts open circuit withvoltmeter128.
• FIG. 16billustrates another test mode for heat inducedring94 described in16a.FIG. 16bshowsring94 operating in the ring-shorted mode with heavy-duty ammeter129 conducting and measuring current130 inring94. Voltage acrossring94 is measured withvoltmeter128 and found in this test to be zero. Current130 circulating inring94 in shorted mode is 4,808 amperes as measured withammeter129 and plotted inFIG. 15 as125.
• FIG. 16cillustrates a test mode for heat inducedring94 described in16aand16bconnected to secondary winding131 of up-converter132 with two ultra-low impedance make-before-breakMOSfet switch banks133. Resistance of each of the make-before-breakMOSfet switch banks133 is on the order of 6.0E-4 Ohm or 0.0006 Ohm. When a current of 2,400 Amperes flows throughswitch bank133 driven by voltage from partially shortedring94,voltage meter134 measures +1.5 volts acrossswitch banks133. Heat inducedthermoelectric ring94 uses roughly half of the open-circuit voltage to force current through up-converter primary winding131 andswitch bank133. This can be verified usingvoltmeter134 andcurrent meter135. High frequency, 50-kHz to 200-kHz, overlapped switching is required for efficient, switching power supply energy transformation, ring to secondary output windings. The on-resistance of the MOSfet switch banks can be varied with the number of MOSfet switches used in parallel. This is the same technique used to limit ring current through each switch bank. A typical resistive load for a 60 coupon thermoelectric ring connected to up-converter is 6.4E-4 Ohms for each switch bank which limits ring current to approximately 2,400 Amperes while the generator is operating. Ring-generated current flows due to the difference in temperature from one side of n-type and p-type wafers, which creates voltage. Energy output from the ring-driven up-converter can be adjusted by changing heat flow throughring94, which varies temperature differential across each thermoelectric wafer. With heat flow adjustments voltage can be regulated, current adjusted by the number of switches used, controlling the energy output ofring94.
• A simple way to calculate potential output of a closed-loop type shorted thermoelectric generator is to use the following equation:
• P_o=V²/R; where P_ois the potential power in the ring over time, V is the open circuit voltage of the ring squared, and R is the resistance or AC impedance of the up-converter and MOSfet switch-bank.
• Typical values for a 5-kW generator are: Po=(3)²/1E-3=9,000 Watts
• This energy is achieved with a 9-inch diameterthermoelectric ring94 made up of 60coupons27, with an operating ΔT of approximately 200 C. Transformer losses in the switching up-converter system is from heat dissipated in the MOSfet switches, magnetic core losses, and primary winding losses in the up-converter circuit. Roughly 5-kW of useful energy output is realized from 9-kW circulating in the ring as potential energy. To increase this output, ΔT across wafers can be increased to 250 C however no test data exists at this time to support reliable operation beyond a 250 C ΔT.
• Another way to increase generator output performance is to manufacture a larger diameter ring made up of a larger number of thesame coupons24. A 24-inch diameter ring has been modeled with an output capacity of 333-kW, 400-hp, that uses the same wafers and heat transfer elements as the 9-inch ring. A 13″ diameter ring, estimated to produce 150-kW or 200-hp, would satisfy electric automotive requirements.
• The forcing current evaluation method uses three simple to obtain measurements. The are; −V₁, −V₂at I, and I (at −V2) to determine R. The exact ΔT of the measurements is unimportant because temperature changes occur slowly due to large copper mass of coupon fins. To determine R for a combination of coupons, the test circuit shown inFIG. 5 is used. Heat is applied to one side of the junction(s) while the other side(s) is cooled or remains cooled. An automotive starter relay with a foot-switch is used to force current through the junction(s) in the same direction current will flow when ring elements are driving up-converter. Test current flows through load resistor, the device under test, and back to battery ground. Should −V become more negative across the device under test with current application, the polarity of the forcing current is backward, but the results will still predict the correct R. The apparatus ofFIGS. 5 through 10 is reliable for several months of use, after which test fixture can either be overhauled with new parts or discarded and completely replaced. From the open circuit V₁at ΔT, I₀, and then V₂at the same ΔT with current flowing through the junction(s) the R and the ρ of the material can be calculated using the equation Y=m′X+b. The −R for a completed circuit or ring can also be calculated. From these results, everything is know to predict the performance of a thermoelectric generator operation from the beginning as a material ingot, raw wafers, wafers with metal contacts, assembled into coupons, and as a ring at final assembly and for product rings returning from the field as defective. If acoupon34 is found during assembly that functions below acceptance level, it can be scrapped, replaced with one that operates in the normal performance range. The test apparatus described inFIG. 5 through 10 was used to determine R, p, and zero crossing current based on a library of material samples cataloged over five years.
• In a preferred embodiment the quality of wafers are tested before being assemble in a coupon arrangement. The power output performance Po of a completed thermoelectric coupon, or generator is a function of the summation of the voltages squared, caused by the temperature differential of individual wafers of the coupon or coupons of the ring, divided by the sum of the resistances of individual wafers, and wafer connections in the coupon that make up the ring; Po=V²/R. Therefore, thermoelectric wafer(s) and wafer connections within useful coupons must have resistances in the sub-micro-Ohm range to perform well as a ring. For control of voltage in the build cycle, incoming thermoelectric wafers are measured using a temperature-controlled soldering iron as a hot-point probe. The probe is placed on a wafer's topside surface as a negative connection to a digital voltmeter. The plus connection of the digital voltmeter is connected to a room temperature heat sink that maintains the bottom surface of the wafer at room temperature. A hot-point-probe measurement of 0.055 volts is typical for either p-type or n-type wafer when the soldering iron is controlled at 250 C, and the room temperature at 25 C. The polarity of the voltage measured in this way is positive for p-type and negative for n-type wafers. The in-process measurement system of this patent is all about measuring and selecting of coupons consisting of cold fin, n-type wafer, hot fin, p-type wafer, and conductive wedge to form a coupon. Coupons that are assembled as a ring in a thermoelectric generator then operate with predictable electrical power output performance. This evaluation method can determine go, no-go performance for individual wafers, individual coupons and assembled coupons of a ring. Being able to remove high resistance coupon components from the build cycle allows the passing of only acceptable performance elements for building generator rings. This is a very important procedure because should a single coupon in the building process have an internal resistance much higher than all others, the power output of the completed ring is degraded.
• Manual assembly methods for coupon and ring manufacture are illustrated inFIGS. 1 and 2. These methods can be replaced with automatic methods that test parts as they are manufactured and remove failed parts from line,FIGS. 17 through 31 de a robotic means of building and testing rings.
• FIG. 17 explodedview136 illustrates acold fin137, and an n-type crystalline wafer138, ahot fin139, a p-type crystalline wafer140 along with aconductive wedge141 that comprises parts of a coupon of this invention along withsubstrate support piece142. Analignment substrate142 is shown beneath the exploded view of the elements of the coupon and the relative position each element will occupy when assembled as a complete coupon. N-type wafer138 is soldered tocold fin137 and on the other side tohot fin139. P-type wafer140 is soldered tohot fin139 on one side and to aconductive wedge141 on the other side. In a preferred embodiment solder paste bonding is 95-5 tin-silver alloy solder having 4% additional pure silver powder added, applied in the region of contact between thesemiconductor wafer138,140 and the hot fin, cold fin and wedge. It should be understood that reversing relative positions ofwafer138 andwafer140 creates an electronically equivalent device with opposite electrical polarity.
• FIG. 18 illustrates acold fin137, ahot fin139, a p-type crystalline wafer140 and an n-type crystalline wafer138 andconductive wedge141 that comprise acoupon136 of this invention shown along withalignment substrate142.
• FIG. 19 illustrates a final arrangement of the elements of thecoupons143 mounted onalignment substrate142.FIG. 19 uses clamp144 to maintain the element positions included in 143. In a preferred embodiment sixty of these coupons complete a 5-kW; 9-inch diameter thermoelectric ring. This number can be varied depending on the operating voltage desired. A 333-kW generator uses 142 coupons to compose a 24-inch diameter ring. The Seebeck voltage also effects how much voltage is produced for a given temperature differential between the hot and cold fins. It should be understood that the cold fins need not be directed at 180 degrees to the hot fins. Furthermore it is possible to fashion the shape of either the hot fin or the cold fin or both to preclude the need for the conductive wedge component. In a preferred embodiment tin-silver solder paste containing an additional 4% silver is applied to each side of each coupon except where the coupon is adjacent to an insulator component of the ring.
• FIG. 20 illustrates the robotic method of the coupon making machine placing an assembled and green solder pastedcoupon143 and substrate into an on-line test fixture145. Green solder paste means solder paste applied to parts but not yet thermally melted or in the case of one-part epoxy cured and un-cured. Placing the coupon into the tester activates a computer-aided tester such as LABview manufactured by National Instruments of Austin, Tex., not shown. Current and voltage probes of145 connect to test the coupon under real operating conditions. The computer-controlled tester determines whether the coupon under test passes or fails. InFIG. 20, thecold fin137, the n-type wafer138, thehot fin139, the p-type wafer140, and theconductive wedge141 can be seen in their assembled position onsubstrate142, held byclamp144 being positioned to insert intotest fixture145 byrobot jaws146.
• FIG. 21 illustrates theelectrical components147 oftest fixture145 inFIG. 20.Heating element148 is used to heat one end of thecoupon143, whileblower149 is used to cool the other end ofcoupon143. The test program involves heating one end of the coupon while cooling the other withblower149.Meters150 and151 are negative voltage meters that first detect −V₁at zero-current voltage across the n-type wafer and the p-type wafers at a temperature that is caused byheater148 andblower149. While the temperature differential across wafer changes very slowly with time caused byheater148, computer controlled test measurements occur very fast, much faster than temperature changes can occur. This allows accurate measurements of −V₁and −V₂made at150 and151 first at zero current and then with forced current, throughcoupon143. The direction of forced current, measured at152 is in the same direction that is caused by normal thermal heat flow through the ring. The faster the −V₁and −V₂measurements can be made with and without forced current, the less effect a changing temperature differential causes and affect the measured results. The current intest fixture147 is sourced bybattery153. Current through coupon is regulated byvariable load resistor154. Reliable resistance measurements can be made on coupons clamped with green solder paste, as well as with coupons after heat bonding. The test method of147 was described earlier in this patent. The test method allows on-line testing of components through all stages of manufacture, providing reliable pass or fail data that can be acted upon during processing. This prevents incorrectly assembled coupons with inferior electrical performance from entering the building cycle.
• FIG. 22 illustrates a pressure operatedsolder paste dispenser155 used with a robot to coat one side of awafer138, fin or wedge during robotic coupon assembly. Air pressure to thedispenser155 is pulsed when the robot arm nears thedispenser155, causing a measured amount ofsolder paste156 to appear onpaste applicator157. Lid flange andcap158,dip tube159 andpaste flask160 are installed in the tabletop, mounted over dowels to secure theapplicator device155 to a location where the robot can be programmed to accurately home.
• FIG. 23 describes acombination tool161 that consists of avacuum pickup chuck162, mounted torobot jaws146.Vacuum supply line163 draws and holds articles topickup chuck162 for precise holding while the robot arm positions the part in the proper place as programmed. Thevacuum supply164 is controlled by the robot's program to operate with arm movement, turning on-off vacuum synchronized with arm movement for programmed parts pickup, placement and processing.
• FIG. 24 illustrates a markingfixture165, composed of markingpens166 that allowsrobot jaws146 to markcoupon143 withmark167 using various colored pens held in holdingfixture168. Depending on computer controlled electrical test results, the robot, is programmed to place the coupon in contact with one of the colored paint pens that indicated the results of the on-line testing. The robot is then programmed to place thecoupon143 in a fail bin, or to place it on the conveyor furnace belt for thermal bonding.
• FIG. 25 is top view of acoupon making machine169 composed of a one-armed robot170 and various parts placement and test aids.Robot170 is mounted on aspecial tabletop171 with various fixtures positioned and attached firmly to the tabletop. Markingfixture165 is located nearfail bin172, which is located near computer controlledelectronic tester145. Central to the robot's radius of movement is adock fixture173 where the parts making up acoupon143 are assembled and aligned. Located within the robot's radius of movement and aligned to the working radius areparts delivery chutes174 that deliver the various elements making up acoupon143. Elements making up a coupon are individually picked up byvacuum chuck162 attached torobot jaws146.Paste dispenser155 is located ontabletop171 along the robot's operating radius to allow solder paste application as required. Coupon assembly involves the vacuum pickup of thealignment substrate142 withrobot170 and placement ofsubstrate142 intodock fixture173. Aftersubstrate142 is placed in the dock, robot arm175 then moves to vacuum pickup acold fin137 placing this on top of thesubstrate142 already located in thedock173. The robot arm175 then lifts and swings to vacuum pick up an n-type wafer138, moving the wafer over and lightly touching thesolder paste dispenser155. With wafer held in place overpaste dispenser155, a short pulse of air pressure delivers paste to one side of the n-type wafer138. The robot-arm175 then lifts and swings the n-type wafer over the dock position and places thewafer138 with solder past ontocold fin137. Then robot arm175 lifts and movesvacuum chuck162 to a position over the chute that delivershot fin139. The robot arm175 then picks up the hot fin usingvacuum chuck162 and locates the hot fin over thepaste dispenser155 wherefin139 is one-side coated with solderpaste using dispenser155. Robot arm175 then liftshot fin139 with green solder past applied and swings to dock173 position loweringhot fin139 to place in contact with n-type138 wafer andsubstrate142. Releasing vacuum, robot arm175 lifts and swings over p-type wafer140. P-type wafer140 is picked up by robot arm175 using vacuum and delivered over,paste dispensing station155, lowered, paste coated and delivered intodock173, placed ontohot fin139. Robotic arm175 then moves to vacuumpickup wedge141, moving141 tosolder paste dispenser155 position, where141 is green solder coated, moved and the one-sidecoated wedge141 is placed onto the p-type wafer140 positioned at thedock173. Robot arm175 then lifts and moves to dispensingchute containing clamps144, grasps and squeezes clamp144 from thedelivery chute174, then moves the clamp near the dock position. Robot arm175 squeezes clamp144 and moves to cover the assembled parts ofcoupon143 withclamp144 and then releases squeeze pressure applied to clamp144. Without moving robot arm175 away fromdock173,jaws146 again graspclamp144 so as to hold thecoupon143 without releasing clamping pressure oncoupon143. The robot arm then delivers the assembled coupon over computer controlledtest fixture145 and arm175 rotates to insert the clampedcoupon143 assembly intotest fixture145 to begin an electrical test under computer control, not shown. Depending on the results of electrical test ofcoupon143, the programmable robot is instructed to move to markingstation165 wherecoupon143 is marked with one of three colors depending on test results. A red mark indicates fail and the robot moves to place failed coupon infail bin172. A blue mark indicates pass and the coupon is moved by robot to thefurnace conveyor belt176 placed for thermal bonding on in-going conveyor belt.
• FIG. 26 is an expandedtop view177 ofFIG. 25, showing the relationship ofvibration parts feeders178 and in-goingconveyor belt179 andthermal bonding furnace180.Vibration parts feeders176, also called “Centron bowls”, are used to separate, orient, and deliver individual parts used in machine automation. Individual bowls can hold and process enough parts to operate for 24-hours, to place parts into individual partsdelivery feed chutes174 that are used in the make-up ofthermoelectric coupons143 bycoupon assembly machine169. Testedgreen coupons143 are placed on the in-goingconveyor belt179 ofthermal bonding furnace180 with oxygen-free atmosphere. Heat bondedcoupons181exit furnace180 on out-goingconveyor belt182. A controller supervises the thermal profile and conveyor belt speed ofatmosphere furnace180.
• FIG. 27 is atop view183 of the backend ofatmospheric furnace180 showing bondedcoupons184 exiting on out-goingconveyor belt182 sliding one at a time downslide185 and catchchute186 where bonded coupons can be picked up byrobot187.
• FIG. 28 is aside view188 of the backend of out-goingfurnace conveyor belt182 ofatmospheric furnace180showing slide185 and catchchute186 with bonded, clampedcoupon184 in catch placement position. This arrangement makes it possible forrobot187 jaws to capture andlift coupon184 for further processing.
• FIG. 29 shows a top view of thermoelectricring making machine189.Bonded coupons184 exit out-goingconveyor belt182 offurnace180. Coupons slide downchute185, and become positioned forrobotic pickup187.Chute186 will incorporate a parts buffer not shown to storeextra coupons184 during times whenrobot187 is busy with ring making activities.Assembly robot187grips coupon184 individually to place in clamp-and-substrate-off dock190.Robot187 removesclamp144 fromcoupon184 and places clamp inbin191,substrate142 inbin192.Robot187 then grips bondedcoupon184, rotates to insert into computer controlledtest station193. Depending on computer controlled test results at193,robot187marks coupon184 atmark station194 with red or blue paint mark.Robot187 then places red-marked failed coupons infail bin195 and either picks up another coupon for clamp-and-substrate removal fromchute186, or moves a bluemarked coupon184 to solderpaste dispenser station196 for solder paste application. Once solder paste is applied to blue mark passed, testedcoupon184,robot187 rotates blue-marked coupon into vertical position and placescoupon184 within up-converter-strap assembly197 for 9-inch rings and198 for 24-inch rings located atring assembly stations199 and200. Up-converter-strap assemblies,201,202 are positioned byrobot187 atring assembly stations199,200 to begin ring assembly. Up-converter-strap assemblies201,202 are acquired frommaterial supply bin203 where up-converter-straps are stored until needed to begin the build cycle for a new ring. Up-converter-strap assemblies are manufactured with legs to stand in proper position to receive green-pastedcoupons184 that are used to form a ring.Testers204 and205 are computer controlled that perform electrical tests on completed rings afterstraps201,202 are tightened by strap-tighteningmotors206,207.Swing arms208,209 move the completed, tested rings onto conveyor belt furnace, not shown, for thermal bonding or pushed further onto a fail ring bin ramp depending on test results of computer controlledtesters204,205.Robot187 usesswing arms208,209 to effect completed ring positioning.
• FIG. 30 is an extended top view ofFIG. 29 ring making machine showing connection to ring-bonding atmospheric furnace in211.Robot187 is shown usingswing arm209, ofset208,209 to position completedring94 onto in-goingconveyor belt212 of oxygen-freeatmospheric furnace211. Finished, tested,thermoelectric rings94 move through oxygen-free conveyor furnace211.
• FIG. 31 is an extension of top viewFIG. 30 showing out-goingconveyor belt213 discharging finished bondedthermoelectric ring214 ontoslide chute215 to roll slowly downroller ramp conveyor216 for further processing.
• In a preferred embodiment finished bonded thermoelectric rings would be cued to undergo further automatic processing. For instance, a secondary coil with printed circuit controller board interconnecting to up-converter, connected to computer controlled tester will be inserted between up-converter straps onring94 inFIG. 9. The switch-bank will be solder pasted and inserted between the up-converter strap201,202 and a functional green-test will be made of the complete thermoelectric system with switch-bank installed. Upon pass of this test, the heater-equipped jaws forcing up-converter strap201,202 into switch-bank, not shown, will be activated to make connections complete. A final functional electrical test will follow. This completes the ring making process and passed thermoelectric systems will move by conveyor to final assembly and test.
• In a preferred embodiment the finished and passed thermoelectric systems will be placed by robot into die-drawn stainless case, on top of a bead of silicon, high temperature RTV dispensed by the robot. The connecting controller board will be positioned and snapped into place by the same robot. Top cover is computer placed, held in place by a bead of RTV, A motor-driven double fan with supports and hot-to-cold section baffle, is glued and installed by computer to complete the assembly. The assembly is loaded into a box for shipment, instruction and warranty papers included. Box is then sealed by the computer and computer placed on pallet for shipment.
• FIG. 32 illustrates asystem217 for creating a castingmold218 for net-shaped thermoelectric wafers. Wafer manufacture of thermoelectric devices is a time consuming and labor intensive part of thermoelectric manufacturing process. Shaping damage-free wafers is problematic and shrinkage due to breakage is a major cost adder in wafer production. The system of217 consists of abox218 filled with pressed hollowceramic beads219 with a trace quantity of low-dropping-point high temperature grease to hold shape. Slottedimpression mold pattern220 withslots221 allowpattern222 to be pressed into the hollowceramic beads219 forming wafer-shapedcavities223 inbeads219 along runner indentation in the otherwise flat surface of pressed beads. The wafer cavities are connected by trunk runner, which later receives molten thermoelectric metal.Re-useable polycarbonate parts222 are shown insystem217 some shown pressed throughslots221 in pattern and one222 carefully removed frommold pattern220 bywire hook224 leaving acavity223 underpattern220 in pressedbeads219. After allparts222 are inserted throughslots221 and carefully removed to revealmold cavity223, thepattern220 is removed to yield a ready to pour net shape mold for casting thermoelectric material as net-shaped wafers.Wafer cavities223 are connected in pressed beadimpression connecting pattern220 as a ready to pour casting mold.
• FIG. 33 illustrates asystem225 for pouring a tree of thermoelectric wafers causing each net-shaped crystal to grow single crystal as molten metal slowly cools to solidify and then continuing to grow single as temperature reduces to near room temperature.Mold system225 illustrates the casting method for thermoelectric material resulting in a low cost, net-shaped, high yield method for producing single crystal thermoelectric wafers. To create single crystal net shaped wafers as poured, asingle crystal seed226 is inserted into the pressed bead mold with half of the seed protruding intocavity227 that will become therunner230 when poured. Formulatedthermoelectric material228 is melted incrucible229 and poured into the pressed bead imprint beginning at theseed crystal226 end of themold218. Moltenthermoelectric material228 poured from the seed end of the mold withcrucible229 fillsrunner cavity227.Thermoelectric material228 inrunner230 micro-melts intoseed crystal226 while flowing intocavities223 formed by the removal of the replicatepolycarbonate wafers222. Asmolten runner230 cools to a solid beginning at theseed crystal226 end of runner imprint,runner material230 solidification proceeding away from theseed226, into filledwafer cavities223 to become single crystal net-shaped wafers with the same orientation asseed crystal226. As the material inwafer cavities223 slowly cool, they too become single crystal, with minimum number of grain boundaries and traps for high flux electronic carriers. Experience shows that a 5-minute cool down undisturbed inmold fixture218 is required to complete the single crystal wafer growth process.
• FIG. 34 illustrates asystem231 for tree growth of net shaped thermoelectric wafers for mold growingsingle crystals232 from aseed crystal226 in themold runner230.System231 shows a single crystal tree of wafers removed frommold system225 with connecting net shapedwafers232 of solidifiedthermoelectric material230 connecting runner allows material to grow single from seed crystal toindividual wafers232 throughrunner230.Wafer233 is shown separated fromrunner tree230 by scribing along aline234 with carbide tipped pencil on wafer material then carefully snappingwafer233 fromrunner tree230. Another method of partingwafer233 fromtree230 is by partly scoring alongparting line234 with a small rotary grinding tool and snappingwafer233 fromrunner tree230. When care is taken to prevent the inadvertent mixing of p- and n-type materials when selecting seed materials for a type pour, the runner and seed alter wafers are removed from the tree runner can be rough-crushed and used for subsequent pours to reduce material waste and cost of crystal wafer manufacture.
• FIG. 35aillustrates a rectangular arrangement of linear arrangedcoupons235 without individual wedges between coupons using instead three 90degree wedges236 located at ends of coupon sets235 and one corner of the rectangle arrangement of coupons terminating with up-converter237 closing the circuit for current flow in the thermoelectric device.
• FIG. 35billustrates a triangular arrangement of linear arrangedcoupons235 withspecial wedge238 in two places and an up-converter237 closing the circuit for current flow in the thermoelectric device.
• FIG. 35cillustrates an octagonal arrangement of linear arrangedcoupons235 withspecial wedge239 in seven places and an up-converter237 closing the circuit for current flow in a thermoelectric device. Thermoelectric rings or loops can be configured to work with any number of geometric shapes, even stacked as a′ coiled helix with ends terminated with an up-converter237 closing the circuit for current flow to output useful electrical energy or pump heat as a chiller.
• Solder paste bonding can be replaced with silver epoxy formulated with high temperature one-part polyamide epoxy. Epoxy Technology, 14 Fortune Drive, Biterica, Mass. 01821, single component E3084 was evaluated. When high temperature silver epoxy is used for bonding of thermoelectric, no metal coating of nickel is necessary on the wafers contacting surfaces of coupons to achieve the same electrical performance as nickel and tin-silver solder bonding. Thermoelectric voltage evaluation results, expressed as α, measured as micro-volts/C for Sn/Ag solder paste and then silver epoxy coupons, is charted below:

• Hot-point probe	Clamped green	After bond	Green ring	Ring bonded
  Sn/Ag paste 270	240	170	140	120
  				poor
  Silver epoxy 275	280	270	275	280
  				excellent

  
  The forcing current test methodology works well for measuring contact resistance with both tin-silver solder paste, cured and un-cured and one-component high temperature silver epoxy from Epoxy Technology. The above chart shows how thermoelectric voltage holds up during process for tin-silver solder paste and silver epoxy.
• Also, just like silicon electronic devices, thermoelectric semiconductor wafers require edge passivation for long-lifetime operation, high reliability. An edge coating of high temperature curing varnish with dye coloring is used, red for p-type and blue for n-type. A suitable dye for the edge-coating varnish is Rit dye used sparingly from Phoenix Brands, Indianapolis, Ind. Wafers are edge-only coated and cured at 150 C prior to assembly into coupons. Edge coated wafers with polyester have 200 microvolt per degree C. a and higher operating voltages after varnish cure. Delphon CC-1105, one-part polyester pre-catalyzed, from John C. Dolph Company, Monmouth Junction, N.J. is used for edge junction coating. Unlike nickel coated wafers that tend to degrade to as low as 20 micro-volts per degree centigrade during coating, coupon bonding and ring formation using one-part silver epoxy for wafer bonding has significantly higher voltage and higher operating temperatures than the nickel-plated, tin-silver, solder bonding process. This high voltage operating performance continues through the coupon and ring forming process, to final operating tests and on into long term generation and chiller operation in the field as demonstrated in the above chart.
• Thus having described the method of fabricating the test instrument, a means of measuring thermoelectric material, and a variety of examples as to how said measuring instrument may be used in a broad range of thermoelectric components, and having described a manual means for manufacturing generator rings as well as an automatic means for manufacturing generator rings, we claim:

Claims (12)

1-43. (canceled)

44. An automated method for assembling and testing a ring of coupons adapted for inclusion in a thermoelectric generator, the method comprising steps of:

A. robotically assembling an un-bonded coupon that includes at least:

i. a pair of wafers;

ii. a hot fin that is juxtaposed on one side thereof with one of the wafers; and

iii. a cold fin that is juxtaposed on one side thereof with the wafer that is juxtaposed with the hot fin;

B. using a computer, testing electrical resistance of said un-bonded coupon by:

i. robotically attaching a volt meter to said un-bonded coupon via a probe;

ii. robotically connecting an amp meter in series with said un-bonded coupon;

iii. without any external electrical current applied to the un-bonded coupon, using the volt meter to measure a first voltage produced by said un-bonded coupon;

iv. while applying a known forced electrical current externally through said un-bonded coupon in a direction current would flow if the un-bonded coupon were short circuited, using the amp meter to measure the forced current, and using the volt meter to measure a second voltage of said un-bonded coupon; and

v. calculating electrical resistance for said un-bond coupon using:

a. the first voltage measured in step B.iii. above while no external current flows through said un-bonded coupon;

b. the second voltage measured in step B.iv. while the forced current flows through said un-bonded coupon; and

c. the known, externally applied forced current flowing through the un-bonded coupon measured in step B.iv. using the amp meter;

C. if the electrical resistance calculated for said un-bonded coupon is less than a pre-established coupon resistance threshold, transferring the un-bonded coupon to a furnace for bonding;

D. repeating steps A. through C. until a number of un-bonded coupons required for assembling said ring have been bonded;

E. robotically assembling the bonded coupons into said ring of coupons;

F. using a computer, testing electrical resistance of said un-bonded ring of bonded coupons by:

i. robotically attaching a volt meter output terminals of said un-bonded ring of bonded coupons;

ii. robotically connecting an amp meter in series with output terminals of said un-bonded ring of bonded coupons;

iii. without any external electrical current applied to the un-bonded ring of bonded coupons, using the volt meter to measure a first voltage produced by said un-bonded ring of bonded coupons;

iv. while applying a known forced electrical current externally through said un-bonded ring of bonded coupons in a direction current would flow if the un-bonded ring of bonded coupons were short circuited, using the amp meter to measure the forced current, and using the volt meter to measure a second voltage of said un-bonded ram of bonded coupons; and

v. calculating electrical resistance for said un-bonded ring of bonded coupons using:

a. the first voltage measured in step F.iii. above while no external current flows through said un-bonded ring of bonded coupons;

b. the second voltage measured in step F.iv. while the forced current flows through said un-bonded ring of bonded coupons; and

c. the known, externally applied forced current flowing through the un-bonded ring of bonded coupons measured in step F.iv. using the amp meter; and

G. if the electrical resistance of said un-bonded ring of bonded coupons is less than a pre-established ring resistance threshold, transferrin the un-bonded ring of bonded coupons to a furnace for bonding.

45. The automated method for assembling and testing ofclaim 44 comprising an additional step of:

H. estimating electrical power output producible by said un-bonded coupon using:

the first voltage measured for said un-bonded coupon in step B.iii.; and

the electrical resistance of said un-bonded coupon calculated in step B.v.

46. The automated method for assembling and testing ofclaim 44 comprising additional steps of while measuring the first voltage in step B.iii. and while measuring the second voltage in step B.iv.:

H. heating hot fin of the un-bonded coupon; and

I. cooling cold fin of the un-bonded coupon.

47. The automated method for assembling and testing ofclaim 46 comprising an additional step of:

J. estimating electrical power output producible by said un-bonded coupon using:

i. the first voltage measured for said un-bonded coupon in step B.iii.; and

ii. the electrical resistance of said un-bonded coupon calculated in step B.V.

48. The automated method for assembling and testing ofclaim 44 comprising an additional step of:

H. estimating electrical power output producible by said un-bonded ring of coupons using:

i. the first voltage measured for said un-bonded ring of coupons in step F.iii.; and

ii. the electrical resistance of said un-bonded ring of coupons calculated in step F.v.

49. The automated method for assembling and testing ofclaim 44 comprising additional steps of while measuring the first voltage in step F.iii. and while measuring the second voltage in step F.iv.:

H. heating hot fins included in all of the coupons of the un-bonded ring; and

I. cooling cold fins included in all of the coupons of the un-bonded ring.

50. The automated method for assembling and testing ofclaim 49 comprising an additional step of:

J. estimating electrical power output producible by said un-bonded ring of coupons using:

i. the first voltage measured for said un-bonded ring of coupons in step F.iii.; and

ii. the electrical resistance of said un-bonded ring of coupons calculated in step F.v.

51. The automated method for assembling and testing a ring of coupons adapted for inclusion in a thermoelectric generator ofclaim 44 comprising an additional step of:

H. using a computer, testing electrical resistance of said bonded ring of coupons by:

i. robotically attaching a volt meter output terminals of said un-bonded ring of bonded coupons;

ii. robotically connecting an amp meter in series with output terminals of said un-bonded ring of bonded coupons;

iii. without any external electrical current applied to the un-bonded ring of bonded coupons, using the volt meter to measure a first voltage produced by said un-bonded ring of bonded coupons;

iv. while applying a known forced electrical current externally through said un-bonded ring of bonded coupons in a direction current would flow if the un-bonded ring of bonded coupons were short circuited, using the amp meter to measure the forced current, and using the volt meter to measure a second voltage of said un-bonded ring of bonded coupons; and

v. calculating electrical resistance for said un-bonded ring of bonded coupons using;

a. the first voltage measured in step H.iii. above while no external current flows through said un-bonded ring of bonded coupons;

b. the second voltage measured in step H.iv. while the forced current flows through said un-bonded ring of bonded coupons; and

c. the known, externally applied forced current flowing through the un-bonded ring of bonded coupons measured in step H.iv. using the amp meter.

52. The automated method for assembling and testing ofclaim 51 comprising an additional step of:

I. estimating electrical power output producible by said bonded ring of coupons using:

i. the first voltage: measured for said bonded ring of coupons in step H.iii.; and

ii. the electrical resistance of said bonded ring of coupons calculated in step H.v.

53. The automated method for assembling and testing ofclaim 51 comprising additional steps of while measuring the first voltage in step H.iii. and while measuring the second voltage in step H.iv.:

I. heating hot fins included in all of the coupons of the bonded ring; and

J. cooling cold fins included in all of the coupons of the bonded ring.

54. The automated method for assembling and testing ofclaim 53 comprising an additional step of:

K. estimating electrical power output producible by said bonded ring of coupons using:

i. the first voltage measured for said bonded ring of coupons in step H.iii.; and

ii. the electrical resistance of said bonded ring of coupons calculated in step H.v.

Images