BIO-PO%VER RNTERIM REPORT October 15. 1966 Contract Prepared by.- TA 13 LE OF' c o.N,-rE' iT S P! ste STATE,VENT OF OBJECTRVES I IL SUMNIA RY 2 M CE14E2RAL I)ISCUSSION 3 A. Directly Obtained Electricity 3 1. Neurornuscular Potentials 3 a. Nerve Potentials 4 b, Muscle Potentials 5 2 C. Cardiac Potentials 6 2. birect Current Syste=is 7 &. Physiologic'al Electrode CeU 7 b. Fue'6 Cell 11 B. Therr2noelectric Converter 12 C. Mechanoelectric Converter 14 1. Sources 14 2. Transducers 19 a. Piezoelectric Crystals 19 5 b. Permanent Maonet Generators 24 3, Previous Work 25 ,rV. BIBLIOGRAPHY 28 AM F oBjsCTr4ES STATEMENT lie area of is study is to Sur.Vey past wor@ in tur,-, develop The olject of th Cal power as reported i@n the literat body,pro(Il,iced electri I ods and approaches which ha.ve 2 e various meth information in a,% urderstandng o"th and present this gested as Ossible sourcess rls the geno-ral probleta been Sug directed tO"-v3, a useful foL-.M. -mis report ir- electroilic devices, SUCII ,Df providing elpctri@al poer to implanted lessthan 200 micro, @s requiring cial cardialc pacerrLaXe 9 oty self cOnt2- iied internal as artili E@,ete for an indefinite time from a S.ystern. AIML IT-. A general literature review was first undertaken to both dis- cover previous experimental work and to gather general impressions of thinking on this subject. The review indicated that there were few 2 well documented conclusions available and that although a fresh start rnight be duplicative, it is ne'eded for a systematic study. Conceiv- able sources were listed and each has been -briefly considered from the physiological and instrumental viewpoints. In trying to evaluate potential usefulness of a source it w2as sometirnes necessary, to form conceptual designs of mechanical s@steins which could utilize that source. Comixfents on these designs are included in the discussion as contextual information because theiillustrate those aspects of the systems which were consider@d and the design problems en- 2 counter4d, not because t@ese designi. are fe'Lt to be the proper .solution. We concur with recent statements 'dy investigators active in experimental work that the physiological Galvanic @.-eU and mech- anoelectric conversion are the t,.Yo most promising systems, although much work Is ne2cessary before either is useful for practical pur-poses. Throughout this 6tudy we repeatedly encountered unanswered questions of '.ne possible effects of biological adaptation on the total implanted system. These effects need not be entirely neg-ative. Adaption in a direction tending to increase power output is a possibility deserving of serious attention. AIM== R,ENERAL DISCIIS.SIO".1 A. Directly Obtained Electricity slnce our goal is electricity prodvced from an energy ally and the within the body. the si2,-iiplest system instrument sou.-ce first which should be evaluated is the direct electrical tap. The systems under consideration in this section are all those that contain electrodes that are in physical cont'act with the biological environ- me-it and in which electrical potentials betweerk elec2trodes can be measured. These potentials are due to. or at least intirne.,,ely associated with, time and space variations in ionic concentrations and flow. The ultimate descriptions and explanations and definitions of these potentials are within the science of irreversible therrno- dynamic2s and the su@ject of cons7iOerable controversy. (19) For r-pose of this discussior, we Yiill gene.rally consider these the pu potentials operationally, that is, according to what you do to obtain a -neasurable potential and the characteristics of that potential. 2 1. Neuromuscular Pote ntials -aical electrochemical phenomena The most studied biolc. 'cular action potentials. At the cellular level micro- are the neuromus electrodes o2f several microns diameter inserted through neuron or rmuscle cell membrane indicate potentials of tens of millivolts between points within the cytoplasm and between cytoplasrn and extracell@ilar ,-Luid. The time course of variations -.n these potentials is from 2 millisecond to hours. These 'Ornembrane potentials" in themselves are of no use for our purposes becau se of the fantastically low currents and shrrt life of a cell damaged by p erforation of its rne-nbrane. But ASk Nam associated with the Surrlrnated r ph,2nomena. the gross ellr2l'Ilula . e cell secti0fis in nerve and ofthousanas 2 of nearby ac@"v effects a rnetal wire QC disc muscle bundles deserve Consideration. Whell tlais" are ob served es if compound action poten is placed near a nerv . and another placed in the body, and wlerl 2 between that electrod " is observed. a muscle is-ueedo the it electromYOgran, 2. I;er-ve Potentials an energy e use of neural electricity as 2 Th grounds that dra,.vi-ng issed as pot feasible on Lhe can be disrn ne@s and di-%- source itable rnernbra 2 stimulation of the exc I ra observing curtent results fro s will cau;;e is Conclusion unction. 'M ural rup-tion of norrnal f d is the result o2f ne is produce % - i Curren,., Eau that whatever electric.al energ at- lectrica St ion) and th that level of e activity (by assurnpt Use rtimulation since it was generated essity be sufficient to ca Way to avoid of nec rhere is no in order to produc e (natural) stimulation' en the voltages e electro2de whicl@ supplied it wh r,,turning current 11 i-h aniunk diodes, a-nd lnini.gnurn for germ than the . 4 Volt jrvolved are less 0 i2t seems ill differ. nt distributions w while the twO.electr6de curre . - . units -will be stin-julated. The east some 2 ,so, r reasonable to exvect that at I ince neural desirable. @ I of even a few units in the pNS is un esistance of loss er of 100 millivolt, a source r 2 potentials are of the old - tts. This requirernent is not - 200 rnicronva .12 ohm is needed t( obtaln npIL-, the, input "'n' cornpatible withthe source. Consider for all" 7a amplifiers tY-P ically required to rneasure pedances (1( ohm) Of the action potentials.. nerve compound Aift@ -5- b. Muscle'Potentials With muscles the situation differs in several respects @hat make power pick up frorn this systern at least COn- ceivable. Namely, the volume of musdle is large 2compared with neural tissue and stimulation of a7 few units should not seriously disrupt normal function. The large niuscle areas may allow many electrodes to be used simultaneously in order to provide a lower electrical source resistance. For example. if one pair of electrodes prcvides a source resistance of 500 ohm, 402 pair will provide 12 ohm. In order to utilize this low volt@-5:e ac rxource a minia@.urp- trans- former can be used to step up 'le voltage befe@:-e rectisication. From a power standpoint it shou7id I."@;'fiore efficient to u--@- one transformer and recti2fier with each pair c,,.' eii'@'7c-.-trodes and s-,irn -.,e dc outputs. Sir.ce skeletal muscle EMG signals <-,ontain most of -.@heir power in 3' frequencies above 100 cps, submlniatvre traa_sforrners (i-, oz. v.2 in. can be used. Answers to important biological questions were not found in -the literature. No reports describing investigation of EMG signals as power sources were discovhred, although use of EMG signals for controlpurposeshavebeerkfrequentlystudied. Whetherornotalow en,ough source resistance can be achieved and main2tained and what 'type of electrode and implantation is best are questions which will probably have to be answered by experiment. Careful technique may, prevent electrodes embedded in muscles performing large movements from causing irritation and pain but this will ;also require study. Judginer 1 to from the experience of clinical workers with implanted cardiac pacemakers. small electrodes can be tolerated but dislodg-3ment and lead breakage rray be serious problems. Ann t Card@ a Potentials ta The strong rhythmic contractions of the myo- cardium attract irhmediate attention but the characte ristic s of its elec 2 ar difficulties. Unlike skeletal muscle trical potentials present particul v.-hich maintains contraction with a high frequency train of "spikes". car-diac muscle fibers depolarize and repolarize only once per heart- beat. Thus, the energy is contained mostly in low frequency compoll- ents (1-20 cps).- Since the contraction of all areas of the muscle is synchronized, large electrode areas can be used without the losses which would be associated with large electroces on skeletal rnuscle. (Current from active non-synchronized skeletal muscle fibers could pass t-iru a large electrode to inactive'tissue without passing to the 2 second electroae of the pair.) B@t while this allows only one tr.ans - former to be used on the heart withone pair of large electrodes, trans- for7ners which are designed for very low frequencies are relatively lar Ye. hegvy and inefficient. Recardless of the differences in waveform, the ECG on the he2art is less than 100 mV (possibly only 10 mV) and thus a maximurn source 'r@'esistanc@ of 1.2 ohm is again required for 200 UV,' power yi&ld (hssuming the 100 rnV peak figure). TJnless a radically i,mproved electrode material is developed this resistance is not likely to be achieved. Effective so0urce resistance figures for common elec- trode materials directly on muscle have not been found in the literature, and an experiment appears necessary to determine what resistance can be easily obtained and how this ml-,ht change -.vith time. -7- 2. Direct Current Systems These systems are all those that produce electron flow in one direction only between the two electrodes. This is un- like the neuromuscular potential electrode system in which a capacitor wo2uld be placed in the circuit to insure that no net charge flows from one electrode to the othe.r. Ir. the neuromuscular system the electrical energy or interest is " I#ernatin& current" in that charge flo@vs from electrode A to electrode j3 and then, milliseconds later. returns to A. In the neuromuscular electrode system net current flow would provide no liseful work and undesired reactions might accumulate products on or near the electrodes, that result in "polarization" and/or electrodedeterioration',,vhlch interferes with the desired action. This type of electrode activity, howe.ver, c-in- become the desired 'activity in 2 direct c-urrent'systems. a. Physiolocrical Electr-,@de Cell The Galvanic cell in its simplest and classical forn-t consists only in two difierent metals (cr other conductors) dipping into a common.tont4@,solution! An electric p@tential, characteristic 2 of the metals. 'iernperature, ionic species and concentration, can be measured between the non-immzrsed portioris of the two metal electrodes. Since the fluids existing within the body are ionic solu- tions inter-electrode potentials can be prgdl@ced by inserting two dissimilar conductors anywhere. Because of the2 complexity or the body flu,&d composition, variation in compos:tion at different points, induced efrects from the presence of foreign mate@ial, presence of many membranes with unk-now.,i properties, active processes, etc. , etc. , the actual chemical reactiot-is and inter-electrode potential can not be pre- 4 cted, in rzkct such potentials are not strictly defined. (Nims, p. 3). Ankh, two general types Of reaction Nve can consider that there are The first, which we will which occur with in,&planted electrodes. call typ2e i, is like that iri the classical Galvanic cell described above in which one or both electrode materials enters into the reaction and b-2comes irreversibly altered or lost. The second type of - reaction is possible when membranes are present and the chemical environ- ment differs at the two electrodes. In this type 2 reaction, which i2n its simplest form is the classical concentration cell, irreversible change to the electrode surfaces need not occur and the electrodes need not be dissimilar. The rate of the -chemical reactions may be much improved by dissimilar surfaces@ however, for exampli! by catalytic action, in- creased effective area, inducem2ent of local environmental change, etc. This second type of activity can be con! idered as a fuel cell with the phydiological system malp,@).,ning i-11 the reactants and removing the' end products. A special ir p-.ication'of the type 2 system is when sirnuar electrodes of "Iner-t metals" or of 2"non-polarizable" liquid fuled tubes are used in conjunction with very hi,,rh impeda-nce voltage measur.Lng cir- cuits which insure that the potgntial chexnicalreactions do not occur :at the electrode's. T@is arrangernent is used for investigating so called- 2 natural de potentibil gradients within and on the surface of the body (4). Such measurements necessarily rnust draw virtually no power from-tfie chemical energy sources responsiblbi for the electrode potentials. for as soon ar current is drawn reactions occur and the natural concentrations change. Thus any dc 2current producing electrode generator scheme use- ful- for our purposes will have inter-electrode potentials which are more or less unnatural physiologically. That is. the potential difference be- tween electrodes of any type drawing current will be different from that rneasure,J between "non-p6larizable" electrodes drawing 0no current, and the local chemical environment surroundingthe electrodes maybe grossly dl(ferent. ]@or these reasons- the type 1 systern in which electrode mat-2rial change occurs and the type 2 system in which it need not are considered here as two special cases of the general physiological electrode cell system. The general case or which the type 1 and 2 systems are ex- 2 amples, is when dissimilar metals, which produce a potential when dipped into a common ionic solution, are placed in a nonhomogeneous environment that produces a potential between :simnar electrodes. This is probably a fair description of the situation prevailing in the electrode materia-1, electrode placement comb-2Lnations with which Dr. Tohn Konik'off and others produce the best results, The Konikoff work is a significant source of e'xperimental data and has stimulated much of the recent intere st in the,yhysiological electrode cell power source. For these reasons a brief summ ary of the work reported in 2 reference 12 is included here. The reader is referred to the ori&al paper for details. John Konikoff and Luthe.r Reynolds were the principal workers at the General t:lectric Companylq Space Sciences Laboratory under a c-ontraet with 14ASA in 1963 - 64 to investiora t-e the@ use of what is 2 re:erred to here as physiolocdeal electrode cell- potentials as a biologically derived power source. Many combinations of electrode materials in several anatomical locations in several species of labor- atory anirnals were tried. Their final choice of electrode material.% was "high speed steel (751@ Fe. 6179 2Cr,. 1819 W, .341a V, .719 C)" and a specially prepared platinum platinurn-black" combinition.. The fihal c@.oice of location was as follows, the PPb electrode was located in the abdominal cavity dorsal to the peritoneal membrane -- and HSS situited @-.ibeutaneously 8but physically adjacent to the abdominal incision. The Agak 7 longest conlinuous implant was'123 days, electrodes and sites were as above. 'The animal was ;r. rabbit, and a coristant resistive load of 10, 000 oh-i.-i was applied between electrodes. After 15 days the output stabilized and thereafter rema2ined at 24 microwatts 'and . 5 volt. The highest power reported in short term studies was 308 microwatt. ITo ne-m work from either Konikoff or Reynolds has been published since 1954. Telephone conversations with both men indicate that work is co-itinuing, and that recent improvements in the platinum-black elec- 2 trode material have increased the po,.ver output threefold for the same electrode area. - Reynolds who is now at Hahnemann'Medical College, Philadelpiii3 reports that 2 00 mi.c r6wal,.ts -has'been obtained when each 2 elecirode is of'1/2 in area. 2This ele@trode power generat,.oti scheme .has the advantages. "according t@6the o'riginatcrs, of simple surgical pr:3cedure and no harmful tissue @eaction or loss of output at least for 4 months in the ono- loig term rabbit experiment. Accordini to the data in the Koni,koff report and especially !2he recent report of Strohl et. al: (29), wh'en "biologically iner-t" metals such as plattnurn and type 31 S., stainless are iinplant@-d, power levels greater than 10 uW have not been obtained and the output dr-ops signifi- cantly below this after a few days. Strohl's comments on the inevitable 2 growth of a fibrous membrane around implanted electrodes suagests that the electrodes become isolated from the original, diss"ilar ionic ,@-i,ivironments as this membrane grows. The better po%ver outputs and longevity have been obtained in coniuncti-on with an electrode which actively reacts with species present in the extracellula3r fluids. Even when covered with (hypothetical) cells tending to maintain identical ionic concentrations around the two electrodes, a reactive e'Lecti-ode can continue to provide rurrent. In evaluating an electrode cell svstem con- -tainbig reactive electrodes importint conside:-ations are toxicity of_ LA P-oducts and deterioration of performance with time. As Strohl notes, Faraday's first law predicts the electrode weight loss due to ionic solution when the electrode reactions are known quantitatively. For example, .91 of iron will be needed to supply 100 uA for 1 year. But at least as important are the hard to predict effects such as loss of effective surface area "catalyst poison-ing", uneven surface deteric>ra',ion and long term local tlssue reaction. Li conclusion, it appe ars that there is a reasonable possibility that physiological elec- trode cel2ls can provide 200 uW for extended periods, but careful long term studies.and an understanding of the ac4ive phenomena, which, hopefully, will provide the @asis for @ptirnizing the electrode materials, are-necessary. But the s' ple surgery in low risk areas which has been use,@, the mechallics -'n--o moving part2s, the non- dependence on any bodily motion. 'the inherent freedom froin en- -t term results already achieved, capsulation problems, and the shor combine to make this a most promising system at this time. b,, Fuel Ceii "Sophistic2ated direct current systems have been speculated on for producin-, relatively large quantities of electrical power for running proposed artificial tear-ts. These systems are usually referred to as fuel cells and usually are considered in ref- erenca to known chemical energy sources such as glucose- or ATP. These systems are very appeal2ing, largely because the proposed energy source is fai.-ly well understood. Molecular energy yields. available concen- trations and naturally occuring reactions can be stated. The develop- ment of physical systems to t%tilize these sources then'appears to be a problem amenable to present technological capability since fhe avail- abl8e raw materials -and necessary operations are knonvn. at least -12 - in broad outline. This is in.contrast to the simpler Galvanic ceU sy-tems discussed earlier in which the present -qtate of the art ha$ been reached largely by trial and error without benefit of thorough un:lerstanding of the detailed processes involved. Ap2proaching the problem from basic principles and proceeding in accordance with established theory wiU no doubt achieve practical success in time. The National Institutes of I-lealth recently circulated a Reque:it For Proposal to,undertake feasibility studies of in- plalnted biological 2 fuel CeUs. When these initial studies are completed we will have a statement of the problem and outline of needed research. For the immediate future, ho-wever, the simpler ;It proposed system is. the greater appfars its chance of success. S. Thermoelectric Converter' Temperature gradi2enti within the body theoretically can be exploited as a source of electrical energy. In recent years considerable research on thermoelectric compos'&tions for use with nuclear reactor: heat sources has produ-ced materials with thermoelectric properties much improved ov@r those of conventional t,%ermocouples. 2 For example, a' conventional copper-constantin couple will produce 23 microvolt per fahrenheit Zegree temperature difference while a material 0 of Bismuth-Antirnony-Talluride composition produces 77 microvolt/F (II, 8). Simple calculat2ions using this second figure indicate that with 0 a 5 F temperature difference and I ohm resistance for every element 2500 elements connected in series will yield 200 uW at . 5 volt. A Japanese group (32) has published a report of a 150 element thermoelectric generator for use on the ext7ernal body surface. Their device used the Bi-Sb-Te rnate rial and the size of the thermoelectric array appears.to be bout 2. 5 crn x lcm x.-Scrn, The data presented in their report arehot .ARM=" 3- clear and well organized and therefore the following calcuiations -based on that report may, not be com pletely correct. A maximum voltage of about 450 millivolt (open circuit?)2 is repoeced. 'A serie3 array of 150 elements of a material producing 77 uV/F .will produce 450 mV at a temperature difference of 39 Fo' Since some of their work was at 10 0C (530F) air temperature with evaporating alcohol on the cold junction, this temperature difference is possibl2e. The only 2 power output figure mentioned is 20 uW/cm. . If this was obtained under conditions which produced a . 45 volt open circuit voltage and if 2 their device contacted a si(in area of 2. 5 cin then the indicated. internal eir devi e is 2000 ohms, oi- roughly 13 ohm per element. resistance of th C If this resistance figure is realistic for thermopiles composed of ele- ments of 2mm x Imrzi. x 5mm s Like therf the 2500 element array mentioned 2 above would produce only 1,13 of the assumed 200 uW or only 15 uW. A total resistance of 13 ohm per element appears unnecessarily high, how- ever, according to the following calculation. - The resistivity of Bi-Sb- -4 Te is only 7 x 10 ohm - cm (8). Hence an element of the above2 dimensions should have only 17 malohrn kiter-nal resi--tailce. Therefore the actual electrical resistance is almost entirely contributed by the contact @e- tweeft the ther-mdelement ar,.d the heat sink conductor and is largely a problem in technique. According to reference 26. contac2t resistivity ill elements used in thermoelectric power ginerators rriay vary between 2 ,her fiaure indicates a contact re- 3 and 4500 microohm - cm The hi. sistance of .23 ohm for an area of .02 cm and, since the2re are two contacts per element, a total contact resistance of . S ohm for elements the size of those in the Japanese device. This last calculation was the basis for our original assumption of I ohm per element. The conclusion we reach is that a 2500 element array operating 7 0 2 between a temperature difference of 5 F with a surface area of 42 cm at each heat sink and a depth of .5cm will produce 200 uW at 5 volt. By comparison, 25 cm 3 of medical grade mercury cells (S ';%Iallory RM CC - IW) (20) has a capacity Df 8 AH -,which. r-.eglecting age derating, will supply 200 uW for 5 years at 1. 4 2 to 10 volt. Since the failure of any one of 5000 contact points ia t'ne series connected array will cause system failui,e, and since a 5 F0temperature diiference between two 42 cm 2areas.. 5 cm apart does not naturally and reliably exist within the body, Lhe thermo- couple systezn is considered 2to be not corrpe.titive-..ivith conventional batteries for at implanted power source. C. Mecbarioelectric Conver-ter In this section possible. mech'an--ical energy sources w,@'ll be considered togeth.er with mechani.cal couplir-g schemes. An arbitriry 2 criterion of 1 milliwatt net mechanical work in the coupling system was chosen as a practical minimuri power level for a final electrical output of 200 microwatt. Brief consideration-,of actual mechanical to elec- trical iransducers, namely, Riezoelectric crystals and permanent maomet generators, is included. We make2 the provisional assumption tn this section tkat if a mechanical system can be implan'.ed which will perform I mW work, for example in winding a spring, for over a year, then a transducer can be desianed to utilize this energy. (>ther han work based on electrode cell potentials all known experimental im- 6 :)Iant power generation has been with piezoelectric crystals. 1. Sources The obvious mechanical sources are.- 6 Voluntary muscle, joint and limb move- ments a Peristalsis (dismissable on grounds of insufficient power) a ftespira-.,ory systern rib cage motion, diaphragm muscle. thoracic and abdominal to pressure variatioa" a Cardiovascular system - heart motion, aorta and large artery pulse expansion, blood flo.w a Gross body acceler-ation (self winding watch pr-@ncipje, "random motion power") Moven@en.ts associated with voluntary activity in some cases offer large quantitie.-; of mechanical power..' The intermittant cha racter of this :activity m@ans, however, t h%a.t an energy storage system muzt @e included in the design to supply powl%r during periods of inactivity. Rechargeable batteries are the obvious storage device, espe2cially since they are designed for and requi--e relatively hil@--i current, short duty cycle char&g.. These batteries require 50 to 10017a more charge current than they retur@, ho:w'eve r. l@he refore, any inte rmittant gene rato r will have to supply 300-400 microwatt average elect2rical charging power if the tattery-undergoes 200 micro%vatt constant drain. It may not be un- reasonab'te to dept.@nd upon or require some particular v6luntary move- ment being performed at some minimum rate for many months, -but unless a particular application requires power oniy dvring a certain 2 type of activity, it seems more straightforward to couple a motion generator to a continuous activity, such as respiration and blood flow, in which the rate and other operational norms and limits are predictable and unavoidable. consider is The first continuous rnotion source which we will respiration. Since the ob@ect of respiratory mechanical motion i-S to purnp air- a fluid flo%,, system-operating on the pressure volume changes2 t found in th e thoracic and abdominal cavities during the respiratory cycle ons af quiet rest the variation is an obvious possibility,. During conditi in pressure within the human adult thorax is approximately 3mm Hg 2 2 (4 cm R 0) or! .04 Nt/cm At a breath rate of 30/ min, work of 2 2 ath must be done for an averaae mechanical power millijoule per bre 12 of I milliwatt. If we approxirftate the phase lag to be expected be- tween pressure and volume by assuming no phase 12,g but with only one half the pressure variation.(i.e.'.q2 Ntlcm then the-volume of fluid k'silicon oil-, gas isotonic ialine-, etc.) which must be pumped x 163 Oule 2 s 10 cn%3.- each breath according to the relatlion PV a 2 Since the volume calculated in this inannei is inversely proportional to breath rate and- intrathoracie' pressure. the'volume required in most experiniental 2 animals will be less. An elenrentari-non-diff4rential system,responsive to resplratory pres5ure variations of 3mm lig'wo'uld proba-Dly be disabled by normal atmospheric pressure variations of one or two inches of mercury. . In- 2 sensitivity to ambient "dc" pressure is inherent in a disferential system. however, and because the intra-abdominal -espitatory pressure variation is out of phase with the intrathoracie, two bellows, one in each cavity. .connected by a tube would comprise such a systern. It appears that this system can provide the necessar7y mechanical energy without obvious size and weight objections. A simple iml),antatioa procedure with a sub- cuianeous tube tunnel is conceivable, although all surgical questions as ell as those on materials, size, shape and irritation require extensive .17- design and ex perimentation. In summary, a respiratory fluid pumping system is recommended as deserving of further attention. Direct mechani@al coupling to respiratory motion remains as a-nother possibility. The change in dimension of the rib cage an2d diaphragm are attractive. The method of coupling might be sornething working on the principle found in retracting tape measures. A cable is wound on a drum and a spring tends to keep -.he cable wound up. if the drum package is firmly attached in some convenient location and the cable held against the under side of the diaphragm 2or in a subcutaneous tunnel around the chest with the far end of the cable attached. then the drum would rotate back and forth d4r,:ng eucl@breath. A ratchet drive to wind a second ipring would allow for any zero position" of the cable- ex-tension with the s6cond sl>ring.driviaifthe actu2al transducer. Perhaps placing the cable inside a silicon @ubber tube filled with silicon grease, -the tube being of the beL' ows type to'auow it to lengthen easily. would improve the sealing and tissue irritation situation. While quantitative da-ta on the diaphragm has not. been sought, it ce r-tainly appears2 that sufficient power is available from dfapbragm inotion and also from chest expansion. The main probler-iis are expected to be in materials, packaging and surgical technique. Apar-t from matei-ial fatigue and sealing, tissue erosion and cell destruction from too great applied pressures must be 2 avoided', for even livina tissue applying pressure unnaturally (e. g. an aneurysm) can erode its way through other tissue. The experience with bone plates, wires and other prostheses which have been used for manv years shows that direct mechanical attachments to internal structures can be accomplished, however. 9 Cardiovascula The other continuous mechanical source is the cardiovascular systei-n. The work of Doctors Parsonnet 2-kd.Ke-iredy 4ernonstrates, at least for short periods, that the expansion of the great arteries and the irovement of the heart can be tapped for rriec'hanical power. Specific contmcnts on these sources are ineul'ied iii 1.1-ie discuss-,ons of tl-@eii itz- periments. In general, however, a significant design problem is con- cerned with accomodating long and short term variations in the proper- ties of the system being coupled to. With the arteries some of these variables are changes, whether natural or induced, in the artery cross 2 section, arterw wall elasticity. average blcod pressure, systolic- diastolic differential pressure, postural configuration and relative direction of gravity., Random i@Tot, n io Gross body acceleration operating on a mechanical system 2 similar in principle to the self winding watch refers to voluntary, motion. especiauy walking, and the critirism of non-coniinuous sotirces applies. The only detailed consideration of this tyste.m is found in Dr. Long's article (14). The ad.@rantage of this system iz;that aU the operating parts 2 can -be enclosed in a hermetically se;Lled rie-d box. Except for the weight involved this box could be attached to the diaphragm or heart to take advantage of the continuo@s motio@ in these locations. But in order to demonstrate that the weight is prohibitive, consider a m,,iss of M kilograms being forced to move back and fo2rth over a distance of lcm accordirg to the sine law at a frequency of I per second. The rna-vimum velocity achieved .vilj be Tr 162 meters/second. Tl-.e kinetic energy at 1 2. m t,lits point is i iNW 16 oule. If we could somehow utilize all 2 t-iis energy each cycle,the mass for I millijoule is approximately 2kg. %Vhile the force necessary to accelerate this mass Is only . 2 Nt. which could probably be provided by the diaphragm, the force necessary to support this mass against gravity is 20 Nt. which the diaphragm could not support. The artificial heart discussants are seriously considering weights of this magnitude for long term implantation: thus we cannot 2 a priori dismiss a random motion system as unworkablesbut it does not ies in power appear to be competitive with conventional mercury batter pe r pound. 2. Transducers a. Pietoelectric Crystals 2 The mechanical energy to cie-.trical energy transducer.most often considered for @se in biological power applications i!s the piezoelectric crystal. Manufaqfured crystals of lead zirconate, lead titanate (PZI,) composition b.ive far superior properties for p