CA1262847A - Silicon implant devices - Google Patents

Abstract

ABSTRACT
A silicon intravascular device for implantation into animals including humans comprises a small semiconductor device of size less than 500 µm adapted to pass along blood vessels in large numbers, the devices carrying signal processing means for collect-ively providing an output in response to an input signal. The de-vice is preferably less than 7 µm or 3 µm overall. The input signal can be acoustic, electromagnetic, temperature, or pH value or chemical.

Description

2~ 7 SILICON IMPL~TT DEVICES
This invention relates to silicon devices for implanting into animals including humans.

Various devices have been used for implants. ~or example heart pacemakers are used to maintain an adequate heart beat rate in humans. These can be self contained and are implanted under a surgical operation in the chest cavity where they remain until replaced after months or years.

Probes carrying devices have been temporarily inserted into animals to monitor temperature etc. These are short term uses and intrusive.

Small spheres have been injected into the blood stream and their progress monitored. For example 10 to 25,um diameter spheres have been labelled with a radio nuclide and their movement used in cnecking blood flow. Such spheres are passive and can only move with blood flow and in some cases slowly release chemicals by dissolving.
Lyposomes, almost invariably submicron in size and incorporating drugs have been employed therapeutically notably in the treatment of leishmaniasis. ~gain, these devices are wholly passive.

,~ , , ~, '' ; ,:
, ' :
' .

~i2~

These problems are overcome in the present invention by using very - small discrete active devices which are injected into the blood circulation to collectively perform a required function such as drug release or temperature monitoring. As an alternative to injection direct into blood vessels the devices may be inhaled and absorbed into the lungs for circulation within the vascular system,or injected into joints, the cerebral ventricles, and the urinary and genital tracts.

According to this invention a medical implant comprises a small silicon device, less than 500 /um, capable of ~assing along blood vessels or inhalation into lungs, and carrying signal processing means for providin~ an output in reponse to an input signal.
For circulation in the blood system the device are preferably less than 7 Jum e.g. < 3 ym. For limited circulation within the large blood vessels the device may be 250 ,um or more depending upon where the devices are injected.
The input signal may be acoustic, electromagnetic, temperature, nuclear radiation, p}l, or chemical.

The output signal may be acoustic, electromagnetic, explosive, or chemical.

Energy to operate the device may be from a battery on the device or external such as acoustic or electromagnetic in co-operation with piezoelectric material or aerial on the chip.
3o , ': ., , ' ~ ' ; ' ' ` : :

'7 The invention will now be described, by way of example only, Nith reference to the accompanying drawings of which:-Figure l is a sectional view of a device for carrying a chemical to be released on receipt of an acoustic signal;
Figure 2 is a circuit diagram for the device of Figure 1;
Figures 3.l to 3.4 are sectional view3 showing the processing steps in the production of the device of Figure l;
Figure 4 is a sectional view of an alternative device for carrying a chemical to be released on receipt of an acoustic signal;
Figure 5 is an alternative form of the device of Figure l;
Figure 6 is a circuit diagram for the device of Figure 5;
Figures 7.1 to 7.4 are sectional views showing the processing steps in the production of the device of Figure 5;
Figure 8 i9 an alternative to Figure 5;
Figure 9 is a ~ectional view of a device carrying a chemical 7 with a self contained battery and signal processing circuit;
Figure lO is a circuit diagram for the device of Figure 9;
Figures 11.1 to 11.8 are sectional views showing processing steps to the production of Figure 9;
Figure 12.1 to 12.3 are sectional views showing altern~tive processing steps;
Figure 13 is a graph of current against voltage for an FET;
Figure 14 is an alternative form of the device of Figure 9;
Figure 15 is a circuit diagram of Figure 14;
Figure 16 is a circuit diagram for a temperature-current sensor;
Figure 17 is a circuit diagram for a temperature-voltage sensor; and Figure 18 is a circuit diagram for a device carrying a chemical to be released on receipt of ionising radiation.
Figures 19, 20 are circuit diagrams for devices larger than those of Figure 1~ for lodging in selected organs.
~5 .

.
~..

" ' : ; . ,: , . .

~L2~ 7 ~ 4 ~
The device 1 of Figure 1 comprises a closed chamber carrying a chemical 2. This chamber is formed by walls 3 of SiO , a bottom p-t Si plate 4 and a top n-type Si plate 5. A layer 6 of Ti makes electrical contact with the n top plate 5 and encloses a layer 7 of piezo electric material e.g. ZnO. The Ti layer 6 is surrounded by a passivating layer 8 of SiO whilst the n top plate 5 i8 covered by a p+ Si layer 9. A cermet resistor 14 connects the p~ ~i plate and the titanium plate of the piezo e:Lectric material 7.

As seen in Figure 2 the chemical is held between conducting plates 4 and 5. An a/c voltage is generated in the piezo electric 7 when it is illuminated by an acoustic beam. The output impedance of the piezo electric is capacitive so in order to provide a source 10 which can pass a net current a resistance 21 corresponding to the conducting path 14 is placed in parallel with the piezoelectric material 7. The junction between Ti layer 6 and n top plate 5 forms a rectifying diode 11 ~hich passes a d.c.
electric current through the chemical 2. This electrolyses the chemical 2 causing gas to be generated which ruptures the cell top plate 5, 9 allowing the chemical to escape.

In view of the small size of the devices the illuminating ultrasound beam must be of very high intensi-ty in order to generate an adequate voltage. In order to achieve this without heating the tissue too much or producing streaming or cavitation effects the ultrasound consists of very short (e.g. 5 ~s) pulses of very high int~nsity sound r~peated at reeular intervals (e.g. 10 ms).

~ _ 4 _ , ~ .
''~
, :
:~ '' ` ', . ' . .: :
-.,~
. , :
. . : , :

. .

~2~j~8'~7 For treatment in the body of a human about 10 of the Figure 1 devices are mixed into about 10cc of a saline solution and injected into a suitable blood vessel. The blood flow causes the devices to be carried along with the blood corpuscles. When they reach the desired place in the body they are illuminated by a beam of acoustic energy eg at 2.10 Hz. This causes release of the chemical at a highly localised position in the body. The blood flow may be normal, and go right rounci the body or blood containing the particles may be artificially circulated round specific organs or regions of the body. The latter approach requires surgery but has the advantage that the flow can then avoid the lung, liver and spleen where significant trapping occurs.

Alternatively the devices can be diluted into a carrier gas, such as a fluorinated hydrocarbon of chain length of typically 11 or l2 (obtainable from I.C.I. Ltd., England), and inhaled. Provided the devices are within the range < tOum they will remain in the lung for absorption into the blood vessels. Larger devices e.g. up to 300 um can also be inhaled into the lungs.
Enhanced positioning of the devices can be achieved when treating tumours. In this case the devices are coated with an antibody tailored to attach itself to the tumour site only. Antibodies are prepared from cultured samples of the tumour. Examples are P.L.A.P. (placental alkaline phosphatase), H.M.F.G. (human milk fat globulin), C.E.A. (Carcino Embryonic Antibody), H.C.G. (human chorionic gonadotrophin).

:.
i :
. . . . :, .

When coated with an antibody the devices 1 readily attach - themselves to the tumour site as they flow along the adjacent bloodvessels and capillaries. After a sufficient number have attached, a time dependent value, they are illuminated by the acoustic beam to release the chemical. ~y this means very high localised drug dosages can be achieved, much higher t;han can be tolerated by the body as a T~hole.

Coating the devices with anti-bodies may also be used for treating against bacteria. In this case the chemical 2 carried by the device may be gentomycin. Similarly antibody coated devices attach themselves to bacteria and deliver very high local drug concentrations.

The device of Figure 1 is formed by a series of steps illustrated in Figures 3.1 to 3.4.

1. A layer 3, 1.5 um thicX of silicon oxide is grown on an n-type Si substrate 15 of doping density < 10 cm . The oxide layer

3 i3 grown e.g. by flowing steam over a heated substrate.
~J'"~ 2. A layer 16 of photo resist e.g. Shipley AZ1470 obtainable from Shipley Chemicals Ltd. of Herald Way, Coventry CV3 2~Q, is spun onto the SiO layer 3 and dried. This resist 16 is exposed through a mask using ultraviolet light. Unexposed resist is dissolved in a developer obtainable from Shipley Chemicals ~td.
Alternatively an electron beam resist such as P.M.M.A. can be used.
After exposure with an e beam it can be developed using a 1:1 mixture of isobutyl methyl ketone and isopropyl alcohol.

;~f r~ie ~k ,, 1~

~ ' `''' ~ ' :
':
.' ' ': ~

~ i2 ~ ~'7 3. Using the remaining resist as a mask the oxide 3 is removed using a plasma etch. This leaves a series of holes 17 in the oxide 3, Figure 3.1, typically 1.5 um square.

4. The remaining re~ist 16 is removed by immersion in fuming nitric acid or an oxygen plasma and rinsing with deionised water and dried.

5. A layer 4 of p+ Si is formed in the Si substrate 15 at the bottom of each hole 17 for example by diffusion of boron. The layer 4 is ty ically 0.3~m thick with a doping concentration of 2.5 . lO cm

6. A discontinuous layer 12 of Pt may be deposited e~g. by electron beam evaporation or sputtering on the exposed p+ layer 4 in the holes 17. This Pt layer reduces the voltage subsequently needed to operate the device.

7. The top surface of SiO is covered with a thin layer of glue 18 for example by evaporation or printing. Suitable glues are indium evaporated on the oxide, an epoxy resin, or rubber adhesive printed onto the oxide.

8. An upper wafer 19 is pre~ared of n-type Si typically having a carrier concentration of < 10 cm : '~

: -- -: . , :- : : -.

~ ~"

, '7

9. This upper wafer l9 has formed thereon a p+ layer 0.2ym - thick 9 followed by an n-layer 5 of 0.2 ~m. q'hese layers 9, 5 may be formed by vapour phase epitaxial growth using dopants o~ arsenic for the n-layer and boron for the p-t layer.20 Ty3pical doping levels are 5 . 10 cm for the n layer and 2 . lO cm for the p+ layer.

10. A discontinous Pt 13 film may be deposited on the n layer.

ll. A chemical 2 is deposited in each device hole 17 by pouring a liquid over the whole substrate and spinning or blotting off excess chemical or squeezin& out the e~cess when the upper wafer is fixed to the lower wafer. Possible chemicals to attack a tumour are toxins such as nitrogen mustard, the toxin secreted by corynebacterium diptheriae or rycin. A surfactant may be added to help the chemical into the holes.

12. q'he upper wafer 19 is placed over the bottom substrate, Fi&ure 3.2, with the n layer 5 in contact with the glued surface of the oxide. Additional to or instead of glue 18 on the oxide 3 the n layer may be coated with a glue or a hardener for the glue.
Pressure is applied to seal the contactin& surfaces. When the glue 18 is an epoxy it must be electrically conducting, e.g. by containing metallic powder, or be removed in the area of the chemical so that the n layer 5 makes electrical contact with the chemica. 2.

. :
~,,, :, ~

: ,.

, ....

~2~

g The n layer 5, oxide 3, and p+ layer 4 form closed chambers - containing the chemical 2. One apparatus, not shown, for pressing the upper wafer 19 onto the substrate 15 comprises a hydraulic press. Jaws on the press are slightly curved to exert maximum pressure at the centre of the wafer and to squeeze out sideways excess chemical. The upper wafer 19 is placed over the substrate 15, after holes 17 are filled with chemicals, both placed between deformable sheets e.g. 50 ~um thick polythene, and inserted between the press jaws. As the jaws are brought together excess chemical is squeezed out and the layer 5 bonded firmly to the walls 3.

13. The lower n-type substrate 15 is removed by a selective etch such as EDA or alcoholic XDH (eq), Figure 3.2. This leaves islands of p+ Si 4 on the SiO 3. This is described in ~.F.
Raley et al, J.Electrochem.Soc.: Solid State Science, Technolog~, Jan 1984, 131(1) pp. 161-171; K Petersen, Proc.I.E.E.E., ~lay 1982 70(5) pp420-457; and "Thin Film Processes" edited by J. ~.
Vossen, U. ~ern, Academic Press 1978 pp 443-444.
14. Using the p~ Si islands 4 as a mask the exposed oxide is re~oved with a plasma etch. This etching is stopped when the n Si material 5 is reached, Figure 3.3, leaving the chamber walls 3 of oxide about 0.5 ~um thick.

, ..

: ' : ' ~ ' :. , '' 2~

- lo -15. Any glue 18 on the exposed n type layer 5 is removed by - etching, e.g. oxygen plasma for an epoxy glue, or dilute hydrochloric acid wet etch for indium glue. For an electrically conducting glue it is advisable to etch it back under the oxida 3 to a small amount. This prevents short circuits to the chemical from any metal layer deposited subsequently.

16. Piezo electrical material 7, e.g. ZnO , is evaporated or sputtered onto the exposed p+ Si and oxide walls, Figure 3.3.
The ZnO i9 deposited at an angle so as not to cover the n layer 5. Since the current which electro]yses the chemical 2 mus-t pass through the zinc oxide 7, either the zinc oxide 7 must leak or a parallel conducting path across it must be provided e.g. by evaporating or sputtering a thin cermet film 14. This resistance f the film 14 must be accurately controlled since if it is too low the piezo electric will be short circuited and if it is too high it will impede the flow of electrolysing current excessively.
A typical resistance value is approximately 5/w.C. where w is angular frequency of illuminating ultra sound and C is capacitance of piezo electric layer.

17. An 0.1 ,um layer 6 of Ti is evaporated over the ZnO and exposed oxide walls 3 and onto the n layer 5 at its junction with the oxide 3. This is achieved by evaporating at an angle.
18. A passivating layer 8 of SiO is evaporated or sputtered over the Ti 6 and part of the n layer 5 whilst still leaving exposed parts of n material, Figure 3.3.
: :
19. The exposed Si, n and p+, 5, 9, is etched through to the n-type Si 19 using the passivating oxide 8 as a mask and KOH or EDA
or a plasma as the etchant, Figure 3.4.

, ,, : : .:

-: .. :~' ` ''~, ~: , ~ .'.
,. ;

~2~

20~ A wax or polymer e.g. APIEZON W~40 wax, coating 20 iS flowed over the separate de~ices to provide support, Figure 3 ~ 4 ~

21~ The top n-type Si 19 is remo~red by a selective etch e.g. EDA
or alcoholic KOH which doe~ not etch the p+ material 9 ~ Each device 1 is now separate and held together only by the wa~ 20.

22~ ~hen required the wax or polymer binder 20 is di3solved away to provide unattached de~ices.
23~ If required the separate devices can be coated with an antibody, at least on the passivating oxide.

Details of techniques for coating solid surface~ with antibodies are contained for example in the following together with their a~sociated reference~:-H. H. Weetall Meth Enzymol 44 P 134 R. A. Messing Metch Enzymol 44 P 148 P. J. Halling & P. Dunnill, ~iotechnology and Bioengineering Vol ~YI P 393~416 (1979) In addition to direct coating of the surfaces, the surfaces can be coated with a lipid layer to which antibodies can be attached, as described in TJ D. Heath, R. T. ~raley, D. Papahadjopoulos, Science 210 PP~ 539~541 (1980) A. Huang, Y. S. Tsao, S. J. Kennel, L. Huang, 3iochem.
Biophys. Act 716, pp 140~150 (1982) J. Barbet, P. Macky, L. D. Leserman, J. Supramolecular Structure and Cellular Biochemistry 16 pp 243-258 (1981)~

~e~

.. `

:

~6~ 7 Figure 4 is an alternative form of Figure 1 using a p-n junction rather than a Schottky barrier for the diode 11. As before a chemical 25 is held in a chamber 26. The chamber side walls are formed of SiO 34; the top by a p~ Si layer 27 coated with an In glue layer 28; and the bottom by a p+ Si layer 29. A p-n junction 30 is formed between the p+ bottom 29 and an n region 31 of Si. ZnO 32 surrounds the n region and is itself enclosed by a Ti layer 33 which connects with the In glue 28. A cermet layer or discontinuous metal film layer 35 is deposited to provide a resistive link between the n-silicon and the titanium layer 33.
Processing steps are similar to that for the device of Figure 1.

~Ihen illuminated with ultrasound~ an a.c. voltage is generated in the piezo electric which drives a current up to the indium 28, through the chemical and the diode formed by 25, 30 and the resistor 35. The effect of the diode is to ensure that d.c. flow flow can only occur this way round. Hydrogen is evolved at the p+ cathode 29 and at the anode 27 the indium glue 28 is dissolved.
The combination of these two effects lead to rupture of the cell.
The device of Figures 1 and 4 leak the electrolysing current through the zinc oxide 7 or resistor 14. A leakage path may instead be provided through a Schottky diode as explained with reference to Figures 6, 7.

,~

.

-. ~

'7 As seen in Figure 5 a device 79 for releasing chemical 81 in response to an acoustic signal comprises a closed chamber 80 formed of' a bottom plate 82, ~ilicon oxide walls 8~, and a top plate 84. The bottom plate 82 is of n type silicon with an n~
region 85 in it3 upper surface. A thin layer 86 of Pt covers the bottom of the chamber. The top plate 84 is p type silicon fixed to the walls with indium glue 87. A layer 88 of piezo electric ZnO covers the bottom plate 82 and part of the side wall 8~.
This ZnO 87 is partly enclosed by a layer of titanium 89 which makes electrical contact with the top plate 84 and bottom plate 88 but not to the indium 87 and is itself enclosed by a passivating layer 90 of silicon dioxide.

Figure 6 shows the circuit of the device of ~igure 5. A
generating source 91 of electricity i3 produced across the ZnO
layer 88 when illuminated by an acoustic beam. Diode Dl is formed by Schottky contact between Ti layer 89 and top plate 84, and diode D2 is formed between Ti layer 89 and the bottom plate 82.
~hen point A becomes negative with respect to point B current flows through diode D2. When point A becomes positive, with respect to B~ then current flows through the chemical 81 and diode D1. The chemical electrolyses as before and is released from the chamber. Since silicon will form an anodic oxide on the passage of current in many solutions the bottom plate, which forms an anode, is covered with the Pt layer with the heavily doped n+
layer providing a contact.

.
.. . .
.

, :

- l4 -The device of Fjgure 5 may be formed by the following steps as shown in ~igures 7.l to 7.4:-l. Produce a silicon wafer 92 the bulk of which is lightly doped, with a very heavily doped (of the order of 10 cm ) p-type layer 93 (typically 1 micron thick) covered with a lightly doped (about 10 cm ) n-type layer 82 typically 0.5 microns thick. These can be produced by implanting a large does of boron into p-type silicon 91, annealing, and growing an n-type epitaxial layer on the p+ layer 92.

2. Deposit a layer typically 2 microns thick of silicon dioxide 83 e.g. by chemical vapour deposition, evaporation, or sputtering.

3. Deposit a 400 nm layer of polycrystalline silicon 94.

4. Photolith and etch polycrystalline silicon 94 with a plasma.

5. Grow 400A of oxide 95 on the polycrystalline silicon 94.
This will help to protect layer 94 in step 7 and densifies the oxide 83.

o. Photolith to produce a resist mask 96 outlinin~ the devices.

7. Plas~a etch the oxide 83 down to the silicon layer 82 using the resist mask 96. The resulting structure is shown in Figure 7.1 with spaces separating devices.

8. ~ Remove resist 96.

. ~ : :: :

9. Etch n type silicon 82 using the oxide 83, 95 as a mask~ going down to the p~ layer 93. This can be done using an alkaline wet etch such as 33~ potassium hydroxide in water or EDP which will stop automatically when it reached the p~ layer. Alternatively a plasma can be used, but it in this case the resist 96 must be kept until this is done.

10. Plasma etch silicon dioxide 83 down to silicon layer 82.
using the polycrystalline silicor. 94 as a mask, see Figure 7.2.
This fo m s side walls to the chamber 80.

11. Remove the polycrystalline silicon 94 with an alkaline etch, a plasma, or by ion beam milling.

12. Implant arsenic and anneal to produce a shallow heavily doped n-type region 85 in the surface of the wafer 82.

13. Evaporate a layer 86 200A thick of platinum directly downwards. This will form the lower electrolysis electrode.

14. Use angled ion beam milling to remove platinum from the top of the oxide 83 while leaving it in the bottoms of the chambers 80.

15. The top surface of the silicon dioxide is covered with a thin layer of glue 87 for example by evapororation or printing.
Suitable glues are indium evaporated onto the oxide, an epoxy resin or a r~bber adhesive prlnted onto the oxide.

:

~ ' ;.; ' ~ - ' ' :
, -- ., - :.. . ~, ~.

~2~ 7

- 16 -16. An upper waver is prepared from a p type substrate 97 carrying a 1~um thick p+ layer 98 covered by a 0.5 ~m thick p-type layer 84. Preparation is similar to step 1.

17. A chemical 81 is deposited in each device chamber 80 and sealed in by sticking the upper wafer down on to the lower one by the techniques described earlier.

18. The lightly doped substrate 92 of the original wafer is removed up to the p+ etch stop by means of an alkaline etch e.g.
EDP or a mixture of potassium hydroxide, ethanol and water.

19. ~he p+ la~er 92 is removed preferably by plasma etching. An etch consisting of 1 part of hydrofluoric acid, 3 parts of nitric acid and 8 parts of acetic acid, which will remove heavily doped but not lightly doped silicon, may also be used.

20. If indium solder is used as a glue 87 it is etched back slightly using a dilute acid.

21. Piezo electric material 88 e.g. zinc oxide i3 deposited by evaporating onto the exposed n-type silicon 82. The deposition is conducted at an angle so that some of the n-type silicon 82 is not covered.

22. A metal 89 i9 deposited which makes Schottky diodes onto both n-type and p-type silicon e.g. titanium, tungsten, nickel, or chromium. This may be done by angled evaporation inclined to tha opposite side to that used for step 21. This metal must make ~0 contact to both the p-type silicon 84 on top of the device and to the n-type silicon 82 at the bottom of the device, and cover the piezoelectric layer 880 :

, ~
~1 - :. ~ . ~ : .

: : ::. ,: :-, ~ 7

23. A passivating layer 90 of silicon nitride, alumina or silicon dioxide is deposited over the metal 89 and the zinc oxide 88. This may be achieved by angled electron beam evaporation.

24. The p-type layer 2C between devices is etched through e.g. by means of a plasma or alkaline etch similar to those described earlier.

25. The devices are supported by a wax or polymer e.g. Aplezon ~40 wax which covers the wafer and sticks it to a rigid support e.g. another wafer whose surface is passivated with silicon nitride.

26. The substrate 97 is removed by a selective etch which stops at the p+ layer 98. This may be achieved using EDP or mixtures of water, potassium hydroxide and ethanol.

27. The p+ layer 98 is removed, either by plasma etching, or by using a mixture of 1:~:8 hydrofluoric acid: nitric acid: acetic acid. This separates the devices which continue to be held by the wax.

28. The WAX or polymeric support is dissolved, releasing the devices 79 which can be filtered out, and washed.

29. Antibody coatings can be applied i~ desired, and the devices can be suspended m aline ior injeotion.

, -: : . .,~ : : .

A variant on the process is to omi-t steps 25 and 26, and remove the p+ layer 98 with the l :3:8 hydrofluoric acid:nitric acid:acetic acid etch by etching from below and filtering ou-t the devices. In this case it is desirable that the passivating layer 90 deposited 5 in step 23 should be of silicon nitride since it is attached much more slowly than silicon dioxide or heavily doped silicon by the etch. Further variants involve replacing the p+ etch stop layers 98 by silicon dioxide layers. Buried silicon dioxide layers can be produced by high dose high energy ion implantation by recrystallising polycrystallire silicon on top of silicon dioxide (see S. M. Sze, VLSI Technology. McGraw Hill 1983 p 83 ff), or by anodising a p-type layer of silicon under a surface layer of n type material to form a buried layer of porous silicon which can be readily oxidised to give a buried oxide layer (K. Imai, H. Unns, I.E.E.E. Trans. Electron Devices, ED 31 (3) ~larch 1984 p 297 ff, U.S. Patent 3,919,060 H ~ Pogge et al 1974). A general review of methods of growing producing silicon on silicon dioxide is given by L. Jastrzebski, J. Crystal Growth 70 (1984) p 253-170.

Figure 8 is an alternative to Figure 6. As before device 100 comprises a chamber 101 formed by a p-type silicon bottom plate 102, silicon dioxide side walls 103, and an n type sllicon top plate 104. A ring of indium glue 105 holds the top plate 104 in place. To avoid a Schottky barrier at the glue joint a thin n+
25 silicon layer 106 is formed on the lower surface of the top plate 104. A piezo electric layer 107 of ZnC is formed on the bottom plate 102 and part of the side walls 103. A layer of Ti 108 covers part of the ZnO layer 107 and makes electrical contact with the top plate 104. Contact by the layer 108 with the glue 105 is avoided by reces~ing the glue layer 105. A passivating layer 109 of silicon dioxide covers the ZnO and Ti layers 107, 108.

., ..,..,. ~ .
~ .: ' .' ~ ' :

The device of Figure 8 differs from that of Figure 5 in that the top plate 104 acts as an anode and the bottom plate 102 as a cathode. Diodes are formed, as before, between the Ti layer 108 and bottom plate 102 and Ti layer 108 and top plate 104.

'.~J'hen a current flows gas is generated at the ca-thode 102. At the top plate anode 104 the silicon passivates through anodic oxide formation and the current then flows through the indium resulting in anodic dissolution of the indium glue 105. Thi3 provide3 both a build up of gas pre3sure and a dissolution of the indium glue.
These two effects together lead to rupture of the device and release of chemical 110 in the chamber 101.

The proces3 for making the device of Figure 8 is similar to that for making Figure 5 except that the lower substrate is p type and the epitaxial layer on the upper wafer is n type with a thin n+
layer on it. In step 20 the exposed area of this n' layer is removed e.g. with potassium hydroxide etch, so the metal 8 makes contact to n silicon and not to the n+ layer, while the contact to the indium is made via an n+ layer so an ohmic contact is formed there.

- 1 9 _ ~L~b`~8~ ~

~ 20 -A self contained and self powered device 41 is shown in Figure 9 with its circuit diagram in Figure 10. It comprises a chamber 42 containing a chemical 43 such as a nitrogen mustard. The chamber walls 44 are formed of SiO , the base 45 of p-type Si and the top 46 of p+ Si. A layer 47 of indium holds the top to the walls.
The bulk of the base 45 is electrically isolated from the cheMical 45 by an oxide layer 40.

~elow the chamber 42 is a processing circuit formed by four FET
10 devices T1, T2, T4, T5 and a CHEi~FET T3. The transistors T4 and T5 are not visible in the cross section shown in Figure 5. In the bottom plate in Figure 9 are four diffused regions 48, 49, 50, 51, with a metal strip 52 connecting the region 51 with the top plate 46. A layer of SiO forms gate insulators 53, 54, 55. Gate 15 electrodes 56, 57 are formed by deposited Al, gate 57 is also connected to the gate of T5 and gate 56 is connected to the output of an inverter formed by T4 and T5. A layer 58 of Si M
covers the gate elec-trodes 56, 57 and oxides 53, 54, 55, with a hole 59 to the n+ diffusion 49. This hole 59 is covered with a layer of refractory metal 60 forming both a conductor and diffusion barrier. A layer of Ag 61 covers the metal 60 and gate electrodes 56, 57 to form a switch terminal 62. Since silver has a large work function and is also connected to the most negative point in the circuit, the silicon beneath it will tend not to invert. Thus silver on the surface 79 can be used to supplement or avoid the need for channel stop implants to isolate the transistors.

~"

" .. "
- , ~ ~., , :
" , . ' ": '' ~ ~''' ., -: , ~l~62~ 7 Above the chamber 42 is a battery 68 formed by a bottom electrode 63 ? electrol~te 64, top electrode 65 and Ag switch terminal 66 in serial layer order. An insulating alumina ring 67 is arranged on top of the p+ top plate 46 and extends below the indium glue 47 onto the side walls 44.

Figure 10 shows the circuit diagram for the device of ~igure 9.
The batter~ 68 is formed by the negative electrode 65, electrolyte 64, and positive electrode 63. The Ag layer 66 forms one terminal 69 of a switch 70, with the lower Ag layer 61 forming the other switch terminal 62. Immersion of th0 device 41 in a suitable solution, e.g. blood, completes the circuit between the t-,ro terminals 62, 69. '~hen ~ept under dry conditions the two terminals 62, 69 are unconnected and so the battery 68 does not run down.
Transistor T3 is formed by the n+ regions 49, 48 acting as source and drain respectively; gate insulation is provided by the oxide 53 and the gate electrode provided by the Al region 56.

Transistor T2 is formed by the n+ region 50 forming a source; the n+ region 51 acts as a drain; oxide 55 forms the gate insulation;
and Al region 57 forms a gate electrode and is connected to the source.

The CHEI~FET T1 has a source formed by n~ region 49 and drain formed by n-~ re~ion 50; oxide 54 forms the gate insu].ation. As seen in Figure 5 there is no gate electrode. In a CHEMFET a gate voltage appears at the nitride surface 5~ due to the pH Yalue of the solution in which the device is immersed. Thus the source drain current is a measure of the pH at the nitride/solution interface.

Details of CHEMFETs are given in:-Ion Selective Elecrodes in Analytical Chemistry Vol 2, Editor H Freise, Plenum Press ;~ 19~0 article by J Janta and R J Huber, P 107-173 Chemically Sensitive Field Effect Transistors.

The device transistors T1 and T2 and T~ operate in what is termed the sub-threshold mode. Figure 13 shows the voltage-current curve for a field effect transistor (FET). Mormally such a device is operated with a gate voltage above a threshold value shown as VT. Below V the current consumption is very small but large changes occur with voltage. Sub-threshold mode operation is described for exanple in:-M B Barron - "Low Level Currents in Insulated Gate Field Effect Transistors", Solid State Electronics, Vol. 15 (1572) p.29~
R. M~ S. Sanson ~ J. D. Meingl - "Ion Implanted Complementary MOS Transistors in Low Voltage Circuits", I.E.E.E.
Journal of Solid-state Circuits SC-7 No. 2 (1972) p.146 a-,i, . . ,, ~ : -, ~., :: , ~2~ 7 -- 2~ -R. J. VanOverstraeten et al - "The Influenee of Surfaee Potential Fluetuation on the Operation of the MOS
Transistor in Weak Inversion", I.E.E.E. Transaetion3 of Electron Deviees, ED-20 No. 12 (1973) p.1t54 R. J. VanOverstraeten et al - "Inadequacy of the Classical Theory of the MOS Transistor Operating in Weak Inversion", I.E.E.E. Transactions on Eleetron Deviees, ED-20 Mo. 12 (1973) p.1150 R. R. Troutman - "Subthreshold Design Consideration for Insulated Gate Field-Effect Transistors", I.E.E.E.
Journal of Solid-State Circuits, SC-9 ~lo. 2 (1974) p.55 R. R. Troutman - "Subthreshold Slope for Insulated Gate Field-Effect Transistors", I.E.E.E. Transaction on Eleetron Devices (1978) p.1049 ~. W. J. Barker - "Small Signal Subthreshold Model for I.G.F.E.T.S.", Electronie Letters, Vol. 12 No. 10 (1976) p.260 E. Vittoz & J. Fellrath - "CMOS Analog Integrated Cireuits based on l~reak Inversion Operation", I.E.E.E. Journal of Solid-state Circuits, SC-12 No. 3 (1977) p.224 P. Antognetti et al - "CAD Model for Threshold and Subthreshold Conduction in MOS~ETS", I.E.E.E. Journal of Solid-state Cireuits, SC-17 No. 3 (1982) p.454.

~ : : : :

, -- 23 -- ~ ~
, :: : : : :

... ,: : ~:

. - ~ :- .. .

, .. ~-. . .
- , ~ : , , The gate voltage of a nitride pH detecting CHEMFET changes by about 55mV per pH unit. A change in 1 pH unit will therefore change the resistance of an FET, operating in the sub-threshold mode, by a factor of more than 3.5. This changes the voltage on the T5 gate from about 65~ to 35p of the battery voltage. For a 3 volt battery this is a change of 0.9 volts which is sufficient to turn T5 on or off. Changes in pH of aboul: 0.5 units will switch T5.
T5 and T4 form an inverting switch so when T5 is turned off, the gate voltage to T3 rises and T3 turns on.
In use about 10 devices of Figure 9 are mixed into about 10cc of saline solution and injected into a suitable blood vessel. Normal blood flow circulates these devices within the vascular system.
The blood pH varies within the body, around a -tu~our i-t may drop by around 0.4 units. For a large intraperitoneal injection of glucose this differences increases to one pH unit as described by:-M. Eden, B. Haines, H. Kahler, J. Nat. Cancer Inst., 16(2) p.541ff (1955~
H. Kahler, ~. V. B. Robertson, J. Nat. Cancer Inst., 3,pp.495-501 (1943) P. Gullin et al, J~ Nat. Cancer Inst., 34(6) p-857 ff (1965) S. A. Shah, R K. Jars, P. L. Finney, A. L. Yee, 35th Annual Conference on Engineering in Medicine and Biology, Marriott Hotel, Philadelphia, PA, 22-24 September 1982, p.1~8.

- 24 ~

:.

! . ' .

~ .

~L~62~`7 When in the blood the battery is switched on since blood conducts electricity. The battery voltage i5 divided bet~een the CH~FET
T1 and FET T2 ac-tin~ as a resistor because its gate and source are connected together, both operate in sub-threshold mode and so 5 consume very little power. If the blood pH falls, the drain voltage of T1 changes and comparatively large changes are produced in the gate voltage of T5 thus turning it OFF. This causes the gate voltage of T3 to rise turning it on.

In this condition the battery 68 is connected across the cheMical chamber 42, i.e. between p+ plate 46 ~nd n+ region 48, so the chemical 43 is electrolysed. The resulting gas pressure ruptures the chamber 42 and releases the chemical 4~ into the blood at the position of low pH.
Processing steps to produce the device of ~igure 9 are shown in Figures 11.1 to 11.8.

1. A p-type layer 45 0.5 um thick is formed on a p+ silicon 20 subgtrate 75. The p+ substrate may extend for the ,rhole thickness of the wafer or may itself by a thin layer on a lightly doped wafer. Typically the p-type 45 layer has resistivity greater than 0.06 ohm cm and the p+ 75 is less than 0.01 ohm cm.

2. Clean the p-layer 45 and deposit SiO 1.5 um thick 44.

3. Use photo lithography and a plasma etch to produce chamber walls 44, Figure 11.1. Typically the walls 44 are 0.5 /um thick with an internal diameter of 1.5 pm.
~0 O
4. Grow a 1000 A thermal oxide layer 40.

.._ .
` - ,-,: . -:
. . :

'~
''' ~

: . .
. : ~ ~,,, ,:

'7 5. Ion beam mill or reactive beam etch ~"ith the beam incident at an angle to make 2 holes 38, 39 in the oxide layer 40.

6. Form n+ regions 48, 51 in the base 45 by implanting phosphorolls through the holes 38, 39. Implanting 5 x 10 cm of phosphorus at ~0 keV allo~ls the 1000A oxide 40 to act as a mask, Figure 11.1.

7. Evaporate a refractory metal 52, at an angle to provide a strip 52 connecting the n+ region 51 with the top of the chamber 42, Figure 11.2. Platinum is preferred. This will also act as an additional region of anode for electrolyising the chemical.
This is desirable since silicon tends to form an anodic oxide in many electrolytes.

8. Coat the whole of upper side of the substrate ~!' th a refractory non-contaminating inorganic oxide 76. One possible support is magnesium oxide. An alternative is to deposit 1800A of chemical vapour deposited silicon nitride followed by 250 ,um of polycrystalline silicon. Processes for the deposition of thick polycrystalline layers have been developed for bipolar SOI
applications (see L. Jastrzebski, J. Crystal Growth 70 ( 1984 ) p.
253-270). This oxide 76 acts as a support for subsequent processing so a thick layer is required.
9. Etch away the p~ material 75 of the original ~qubstrate. A
suitable etchant is 1 part HF (aq.): 3 parts HNO (aq.): 8 parts CH COOH. Thia removes p~ b~t not p-tvpe Si 45.

:

~ .

,, , , .~" .., : . . :
- . : :

: ~:
-: :
- ~

:~ .

'7 ~ 27 -lO. Form four n-~ regions 48a, 49, 50, 51a by implanting phosphorus or ar~enic through a resist mask and annealing. Two of the n+ regions 48a, 51 a connect through the p-layer 45 to the n+
regions 48, 51, inside the chamber 42. Threshold adjustment and 5 channel stop implants can also be done at this stage as required using resist masking layers.

11. Grow a SiO layer 0.15 ,um thick 53, 54, 55.

lO 12. Remove SiO layer between separat;e deviceR.

13. ~emove the p-type Si 45 between devices to separate them, Figure 11 . 3. This is achieved using the oxide 53, 54, 55 as a mask and plasma etching or a chemical etch such as hydrazine water 15 or ethylene diamine pyrocatechol water.

14. Open up holes in oxide layer to form connections with the n+
regions 49, 50 Figure 7.3 for three transistors T1, T2, T3. At the same time the silicon dioxide 40 between the devices is 20 etched through.

15. Form electrodes 56, 57 by depositing and etching a conducting layer of Al, a refractory metal, or polysilicon, Figure 11 .3.
These electrodes form the gates of T2, T3, T4, T5 and connect the 25 source of T2 to the gates of T2 and T5 and the source of T4 to the gates of T4 and T3.

, .~

:: . : , ::: :
, .
': '' . ~:

~2~j28~7 16. Depo~it an Si N (nitride) layer 58 by a chemical vapour - deposition (C.V.D.) or (if Al metalisation is u3ed) a plasma assisted C.V.D. process. The nitride 58 covers the whole base of the device and extends up to overlap the chamber walls 44, 5 Figure 11 .4.

17. Open up a hole 59 in the nitricle 58 to the n region 49 using a resist mask and plasma etch. In this process nitride connecting the devices is removed, Figure 11 .4.
18. Deposit a conducting refractory metal, e.g. molybdenum or tantalum, as a diffusion barrier 60. This prevents diffusion of ~odium into the oxide 53, 54, when the device is in use.

15 19. Deposit Ag 61.

20. Etch away Ag 61 and refractory metal 60 to leave them shaped as in Figure 11.5 with exposed nitride 58 at the gate of the CHE~IFET T1. A suitable etchant for silver is nitric acid or 20 potassium cyanide. Alternatively ion beam milling with a resist mask can be used.

21. Apply hydrochloric acid to form a silver chloride layer on the silver electrode 61.
22. Coat bottom surfaces with a polymer or wax e.g. APIEZON ~140 wax 77 applied molten and allowed to cool. Preferably thi~
coating is thin (e.g. a few ~um) and serves to stick the chips to a rigid support e.g. a silicon wafer or glass disc. This provides a

30 support for later processing steps.

:

.

:: ~' ~ `" . , -: " '` ` ~ :
. . ~
.:
. : ' - 29 ~
23. Remove inorganic support'76 from the top of the devicesO A
suitable etchant for MgO is hydrochloric acid. Polycry3talline silicon can be removed by potassium hydroxide solution or by mixture of hydrofluoric acid, nitric acid and acetic acid. A
silicon nitride layer can be removed by a plasma or by hot phospshoric acid. The structure is shoim in Figures 11.5, 11.6.

24. Deposit a thin e.g. 0.1 ,um layer of indium 47 onto the top of the chamber 44 by evaporation at a shallow angle Figure 11.7.
Thiq acts as a glue for the chamber top.

25. Fill chamber 42 with the desired chemical 43. ~emove surplus by spinning or wiping with an absorber or on fixing chamber top as described earlier in the context of the piezo electric powered device.

26. Place a p-type Si wafer 78 with a 0.4 ~1m p+ layer 46 and a coating 0.05 um of In 47 onto the chamber walls 44. Pressure of about 3 x 10 Nm and/or ultrasound e.g. at 20-60 kHz fixes 20 the wafer 77 to the chamber walls Figure 11.7. Typically the p+
layer 46 has a carrier concentration/doping level 2 x 10 cm Figure 11.7.

27. Remove p-type Si 78 with a selective etch e.g. alcoholic KOH.
This does not remove the p+ material 46.

28. Evaporate or sputter SiO onto the p~ layer 46.

29. Using photolithography expose the p+ layer 46 over the gaps between the devices.

. , ~,...
''.- .. , ~, ;

.
::
, ' . .
, . . ..

~j2~7 30. Etch p+ layer 46 to separate each devicev Figure 11.8.

31. Remove SiO from p~ layer 46, e.g. by plasma etching, thereby avoiding damaging the oxide on the sides of the devices.

32. Evaporate or sputter Si M or alumina 67 over the p~
layer and extend do1mwards onto the chamber walls 44 thus covering the indium 47 and refractory metal 52 with an insulator.

33. Form a hole in the nitride 67 to expose the p~ layer 46 Figure 11.8 by photolithography. The nitride can be plasma etched.

34. Form a battery bottom electrode 63 by evaporation or ~puttering. A suitable material is V 0 , or CoO or V O /B 0 with a typical thickne~s of 0.3 ~m. 2

35. Form an electrolyte layer 64 typically 0O3 ~m thick by evaporation or sputtering. Suitable materials are (LiP0 ) (LiI) glass;
3 0.67 0.~3 2 0-37 2 5 0.18 (LiI)o 4~ glass, Alternatively 25 polyethylene oxide doped with lithium chlorate (PE0) LiClO
~ 4 could be applied by dip coating or spinning.

36. Form the negative electrode of the battery 65 by evaporation 30 or sputtering. Suitable materials are Li or LiIn or LiAl typically 0.~ ~m thick. This ia followed by a refractory metal e.g. W, Mo, Ta which can be electron beam evaporated, and which prevent moisture from diffusing into the lithium.

.

:: :
, .

, ~ : . `~ ' , ,: :

' ~ ~`' ' ': '`'' ' : ' .
`:' :: , :
, .

:ll2~ '7

37. Deposit Ag 66 by evaporation or sputtering.

38. Expose Ag to HCl to form AgCl electrode 66.

5 39. Dissolve black wax 77 or polymer support. The devices are now detached.

40. Wash and dry the devices and store in dry atmosphere. The devices are then ready for mixing into a saline solution when needed.

Figures 12.1 to 12.3 show an alternative fabrication process, which avoids the need for supporting the devices at intermediate high temperature stages and is suitable for devices containing circuits.
l 5 This has the following steps:-1. Provide an Si substrate 114 with a layer of` about 0.5 ~m of silicon 116 on silicon dioxide l l 5 . Pattern and etch the silicon l l 5 to produce islands of silicon 116 on silicon dioxide l l 5 and 20 fabricate circuits 117, 118 in them. References to silicon on silicon dioxide technology are given above.

2. Passivate circuits 117, 118 and protect with a layer 119 of 2000 A of evaporated silicon. Pattern this layer 119 where contact 25 holes are going to be needed.

_ ~;1 _ ., -: :
.
:

- .

~6~
3. Deposit 2 ~m of silicon dioxide 121 by chemical vapour deposition or evaporation or sputtering.

4~ Photolith and plasma etch the oxide 121 back to the silicon 116 thus forming chamber walls.

5. Deposit lOOOA of silicon nitride 122 by chemical vapour deposition or sputtering and plasma etch to remove it from non-vertical surfaces.
6. Use angled evaporation to produce contact 128 to the circuit from the top of the silicon dioxide.

7. Produce a top wafer consisting of a 0.3 ~m p-~ (approximately 15 2 . lO cm ) layer 123 on a lightly doped 3ubstrate 124.

8. Deposit indium adhesive 125 on the top of the silicon dioxide 121 and/or on the top wafer p+ layer 123.

20 9. Coat with chemical payload 127 and push substrate 114, 124 together as described earlier to seal in chemical 126, Figure 12.2.

10. Remove the lightly doped substrate 124 with a selective etch which does not attack the p+ layer 123 e.g. potassium hydroxide ; ~ 25 solution.

: ~
: ~ :

~ 32 -,: . . , . , -, - . :

~ : : , : ::
; ,, ~ :

::

~j2~ 7 ll. Evaporate or sputter a silicon dioxide layer over the whole p+ layer.

12. Photolith and etch to remove the silicon dioxide over the gaps between the devices.

1~. Etch through the p' layer 123 u~ing the oxide from step 12 as a mas~ to separate the devices.

14. Fabricate and encapsulate a battery 127, Figure 12. 3 ~s in steps 29-36 in the previous process.

15. Etch in hydrofluoric acid to dissolve silicon dioxide layer 116 and separate devices. I~ash and dry devices and store in a dry atmosphere.

~

:"

~: . , , .: , .

i28~ 7 ~ 34 -An alternative implementation of a similar device in PMOS is illustrated in Figures 14 and 15 and given the same reference numerals as Figures 9, 10. As the solution becomes more acidic, V of a CHEMFET T1 becomes more negative and turns off. This causes a negative voltage to be applied to the gate of T3 which conducts causing a current to flow through the chemical which electrolyses leading to rupture of the cavity and release of the chemical. The mode of operation is the same as tha-t of the l~OS
device of Figure 9 except that no inverter is required so two fewer transistors (T3, T4) are needed. The construction process is identical except that an n-type wafer and p-type implants are needed to make the device and n+ channel stops 70 are required to isolate the transistors. ~o oxide is needed to separate the bulk of the silicon from the solution but qome platinum 71 needs to be deposited on the positive electrode and patterned to provide an electrolysing contact since a silicon contact might form an anodic oxide. The platinum can be deposited by angled evaporation and ion beam milling used to remove material from the upper parts and tops of the container walls.
Devices similar to those of Figure 9 can be made to sense temperatures and discharge their chemical when a given temperature is reached. For example tumours are at a higher temperature than surrounding tissue. Thus drugs can be released at a tumour site.
To measure temperature diodes may be used instead of a CHEMFET.
The reverse bias leakage of diodes exhibits very strong temperature dependence. For example in silicon the leakage approximately doubles for each 8 ~ rise in temperature.

~ 34 -... :

::
. ~
' '; ' `

,.

~;2~

Alternatively temperature can be measured by means of the te~perature dependence of the subthreshold conductance employing circuits such as those illustrated in Figures 16 and 17. In the circuit of Figure 16, the transistors T1-T4 set the gate voltage of T5 to N kT loge / S1 . S4 ¦
n ~

where S = width/length of the channel of a given transistor.

Nn = Ideality factor describing the subthre~hold slope of an n channel i~lOSFET equal to q 1 ~

T = Temperature k = Boltzmanns constant q = electronic charge.

The current through T5 And thus the current through T3 therefQre rises exponentially with temperature. Since the drain and &ate of T3 are linked, the drain voltage of T3 is therefore linearly 25 dependent on tem~erature. The transistor T6 forms a current mirror from T3 so IT increases exponentially with temperature. In the circuit of Figure 12, the gate voltage of T1 rises linearly uith temperature. Transistors T3, T5, T7, T9 and Tl1 are current mirrors. The pairs of transistors T2, T3; T49 T5, etc form source followers, and the source voltage of each one is equaI to its gate voltage plus a constant which varies linearly with temperature so V is roughly half way between the rails V
out DD
and V and varies llnearly with temperature.

. :,: '..................... : ~ :

: ::
. .

As before the device may be partially coated with ar antibody or with an enzyme. An enzyme may react with a specific substrate to change the local pH ~Ihich may be detected by a CHEl~lFET, thus increasing the range of materials which the same silicon structure can respond to.

Alternatively, the device may be encapsulated in a cell, e.g. a white cell. This may be achieved by allowing white cells to engulf the device in vitro and to inject the resultant white cells and devices. Since the body sees the white cells as frie~dly the devices are not trapped.

Figure 18 is a circuit diagram for a device similar to Figure 9 but which releases a chemical 4~ in the presence of ionising radiation.
The device includes a battery 68, diode D, capacitor C and FET T1, T2 as before. The transistor Tl only passes a very small current.
When an ionising event e.g. the passage of an alpha or beta particle occurs in the diode D, the current f'lows in the diode.
This raises the potential of point P, charges up the capacitor C
20 and turns T2 on. Current flows through T2 and into the chemical which it releases by electrolyte rupture in the usual manner. The device is used in conjunction with a radio labelled antibody.
First the targat tissue e.g. the tumour is labelled using a radio labelled antibody. The isotope should emit alpha or beta particles. The devices are then introduced into the bloodstream.
This allows the targets to be attacked simultaneously with both radiation and chemical agents.

. . , , .:
'. ~ ~'' ";
.. .
, , ~ ' .

~2~ 8~'7 ~ 37 -A further example of this invention uses devices tha-t are comparatively long e.g. up to 500 ym. These cannot circulate through capillaries but can pas~ along larger blood vessels. Thus by choosing device size and point of injection into the body devices can be carried deep into selected organs where whey will lodge. Providing the shape is arranged not to block blood vessels the devices may relatively safely remain in the vessels and perform their required task. Suitable device shapes are L-shape and cross-shape.
These larger devices can cary relatively complicated processing circuits for sensing a required parameter, e.g. temperature or pH, and provide an output signal for exernal detection or a slow release of drugs on command.
An example of a circuit for use in these larger devices is shown in Figure 19. The device is formed as an integrated device 135 in the same manner as in Figure 1 or 14. A sensor 136 of temperature or pH provides a variable voltage signal 137 to a voltage to frequency converter 138 e.g. a voltage controlled oscillator.
Output 139 from this converter modulates a signal flowing throueh an FET 140 from a transducer 141. This transducer 141 may be a layer of piezo electric material between two electrically conducting plates. The plate may be of dipole dimensions. When irradiated by a sound source the transducer 141 provides a voltage signal at the source frequenoy. A stabilised power supply 142 takes an input from the transducer 141 to power the converter 138 and sensor 136. Modulation of the signal passing through the transducer 141 and F.E.T. 140, representing sensor output, is detected externally e.g. by circuits similar to those used for detecting doppler shifted radar returns. Conditions within a patient's organ under study can therefore be monitored continuously.

: :

, ... . ...

: ~ ' ' ': ' :

12~

The circuit of Figure 20 is similar to that of Figure 18. It has a sensor 136, voltage to frequency converter 138 and sensor 136.
Instead of power being taken from an ex~ernal source the device contains its own battery 143. As in Figure 17 the transducer and F.E.T. 140 provide a modulated signal when irradiated by a sound source.

Instead of or in addition to transmitting information the device of Eigures 19, 20 may release drugs 144. Under the control of the sensor 136 output or under the control of the externally applied sound signal. Such drug release may be achieved by gas pressure generated by electrolysis and may be over a prolonged period continuously or intermittently.

More than one drug may be carried independently in the same device and released together or in sequence.

The sensor 136 may be sensitive to ionising radiation e.g. X-rays or alpha particles. Thus when an organ is irradiated by alpha particles or X-rays a drug ~ould be released to reinforce the radiation treatment. Alternatively the sensors 136 may be sensitive to radio frequency signals to release drugs on command.

.
:

; ~ . . : ~ ` .
, .
:

'7

- 39 -~or devices such as those in ~igure 1 designed to circulate within an animal blood circulation system it is necessary to know:-(i) how lon~ devices stay in the blood:
(ii) how they are removed:
(iii) whether any major adverse effects are produced by the devices.

To provide these results solid microdiscs silicon were neutron activated, mixed into a saline solution, and injected into a pig'8 blood vessel. Blood samples were taken at regular intervals and the radioactivity of the blood measured. This showed how long the microdiscs stayed in the blood circulatory system. After blood sample measurements indicated nearly all the microdiscs had been removed from the blood the pig was killed. The radioactivity of various tissues was measured to determine where the microdiscs had finally lodged. Sections of tissue were prepared for histological examination to see if there was any grouping together of microdiscs and to determine the microvascular site of trapping.

Three experiments were performed, the design of each experiment being largely dictated by the results of the previous one.
~iscussion prior to the first experiment had concluded that microdiscs having a long clearance half life in blood would offer greater clirical usefulness and that the first experiment should assess the clearance rate and trapping sites of very small microdiscs, smaller than red blood cells, which could reasonably be expected to remain in circulation for long periods (hours). The microdiscs were however large enough to be made using present day technology, and for microelectronic drugs of their size to be manufactured using technology expected to be available in fifteen years time to carry a useful electronic circuit and drug payload.
All experiments were performed with polycrystalline silicon microdi-qcs labelled with arsenic 76 (a gamma e~itter)O The discs did not contain circuits or drug payloads.

`::

: ~'' ' ' ;
:
.
. . .
~ .

- : '

- 40 -Experiment l was performed with three micron square microdiscs which were 700nm thick. These microdiscs were cleared from the blood extremel~ quickly following intravenous administration; the vast majority were trapped in the lungs on either the first or second circulation of the microdiscs. The extremely rapid clearance was ascribed at the time to the square shape and sharp corners of the microdi~cs.

Experiment 2 used 3 micron diameter circular microdiscs with a lower specific activity of arsenic 76. Clearance was nearly as rapid as with the 3 micron square microdiscs, resulting in barely detectable levels of radioactivity in the blood at the time of the first blood sample which was taken after 2 minutes. These clearance rates were extremely surprising. Pulmonary capillaries are reported to be seven to nine microns in diameter. The high clearance rates suggest that mechanical trapping may not be the sole mechanism responsible for the pulmonary accumulation of microdiscs.

Experiment 3 used 1.5 micron circular microdiscs in an attempt to reduce the rate of mechanical entraplment of the microdiscs.
Clearance rates were even higher than for the 3 micron circular discs. Experiments were then suspended to review progress and to identify further avenues of experiment.

, .
" .:
... ..
:, :

~L2~

- 41 -Procedure Pigs of R ~ar~e ~Ihite strain were sedated with an air/halothane mi~ture and aneasthetised with sodium pentobarbitone. Pigs were tracheotomised and allowed to breathe spontaneously. The femoral artery and vein of one leg were exposed and cannulae passed into the abdominal aorta and inferior vena cava. The pig was placed in a dorsal recumbent position and maintained by bolus intravenous administration of sodium pentabarbitone as required. Table l presents details of the specific activities of the microdiscs.
The microdiscs were suspended in 2 ml of 0.9~ saline ~y ultrasonification. Eighty per cent of the solution was e~tracted for experiments 1 and 2 (93~ for experiment 3) and made up to 5 ml with 0.9~ saline. Small aliquots (either lO microlitres or 25 microlitres) were withdrawn as standards to estimate the total activity ultimately injected into the animal and to provide a check on the half life of the radio isotope. Arsenic 76 should be the predominate isotope with a half life of 26.3 hours.

The microdiscs were administered intravenously over l minute with multiple rinsing of the catheter and stock solution syringe.
Blood samples were withdrawn intra arterially at two minute intervals for Experiments l and 2 and initially at o~e minute intervals in Experiment 3. One millilitre of each sample of withdrawn blood was centrifuged and radio activity measured in a ~ilj Model 2001 Gamma Counter. At the end of the blood sampling period the animal was killed by exsangrination whilst still under the influence of anaesthetic and then subjected to a post mortem examination. Samples of tissues of 0.5-l gramme were taken, weighed and the radioactivity measured. A total of 15 samples were taken for Experiment 1, 22 for Experiment 2 and 60 for E~periment 3. Further samples of lung were taken for hi~tological examination to assess the site of trapping and whether agglomeration of the microspheres had occurred. All measured radio activities of the standards, blood and tissue samples were corrected for decay to a common time.

.
:
::
.: .
': :, ~L2~
-- 42 ~-Results The entrapment of the microdiscs was so efficient for the 3.0 micron square and circular discs that the measured activity within the blood was barely above background even though 720,C00 cps and 300,000 cps respectively had been administered intravenously (Table 1). Table 2 shows measured activities expressed as cps/ml whole blood for these discs. In the light of these observations the 1.5 micron microdiscs were prepared with a substantially greater arsenic 76 activity and were administered into a pig of lower body mass (Table 1). Significant entrapment of the 1.5 micron particles in the first minute still occurred (Table 2) but measured activities in the blood still showed a three fold increase above background, even after 60 minutes.

Table 4 shows isotope activity in the blood expressed as a fraction of projected blood activity assuming instantaneous and uniform mixing of the microdiscs within the projected blood volume.
The projected activities were 249 cps/ml, and 2,464 cps/ml for Experiments 1, 2 and 3 respectively (Table 1). An extremely large proportion of the injected microdiscs were cleared prior to the first blood sample at 2 minutes. Approximately 98~o of the 3 micron squares, and 95,~; of the 3 micron circles were trapped by this time. The fraction of the microdiscs remaining in the blood which were trapped in the mext 28 minutes was quite low. The high specific activity of the 1.5 micron circular microdiscs administered in Expermiment 3 permitted a more accurate estimate of the clearance rate in this case. Table 3 shows that after one minute, only 0.95,0 of the administered microdiscs remained in circulation. Further entrapment occurred over the ensuing minutes and after 59 minutes only 0.24~o of the administered microdiscs were still circulating. The relative blood activity as a fraction of time 19 shown in graphical format ln Figure 7.

:

;: :

:

8'~'~

In all three experiments the majority of the microdiscs were trapped within the lung following intravenous administration. The isotope activity expressed as cps/g wet weight of tissue are presented in Table 4. Table 5 shows the same data expressed as tissue activity divided by the level of activity which would have occurred if the injected activity were spread uniformly through the body with a constant level of activity per unit mass of tissue.
In Experiments 1 and 2 (3 micron microdiscs) notable quantities of microdiscs crossedthe pulmonary vasculature and could be detected in, as expected, those organs receiving a significant blood flow per gramme of tissue, namely the liver, kidney and spleen.
Activity could also be detected in samples of pancreas, heart, bowel, brain and skeletal muscle - the activity in these organs was very low being only approximately double the background level.
The 1.5 micron microdiscs injected in Experiment 3 showed the same general pattern of distribution. The vast majority of the microdiscs were again trapped in the lung with the "relative activities" of the samples extending over a large range of 13.0 to 199.9; the mean +/- standard deviation "relative activity" of the 2~ lung samples being 113.2+/-53.8. A small proportion of the 1.5 micron microdiscs has traversed and pulmonary circulation and had been trapped predominately in the liver and kidney ("relative activities" of 1.10 ~/- 0.21 and 0.091 +/- 0.017 respectively).
Unlike the 3 micron microdiscs, the 1.5 micron microdiscs could not be detected in the samples of bowel, skeletal muscle and brain (in spite of an approximately ten fold increase in the administered activity per unit body weight compared with Experiment 1 and an approximately 25 ~old increase compared with Experiment 2). These vascular beds appear to allow the passage of the 1.5 micron discs.
There was no evidence of clumping of microdiscs - all of the 25 discs found were single discs and were not in direct close association with additional microdiscs.

.

, ,~' ',' ; :
,: : .
:;,:
, ,~ , ' "' ' Table 1.
Sumrnary of E~perimental Details and Microdisc Characteristics Quantity i~licrodiscs used: 0.7 ~urn thick 3 ,um squares ~ ~m squares 1.5 ~rn circles Fraction of activity 0.8 0.8 0.93 extracted and injected Number of microdiscs 2.88.10 2.&8.10 1.14.10 Activity injected (uCi) 18.4 8 110 Animal weight (kg) 45 48 21 Projected tissue activity 15.5 6.16 194 (cps/g) E~timated blood volume (l)* 2.88 3.02 1.65 Projected b].ood activity 249 98 2462 (cps/ml) * Estimated from formula presented in (7) Blood Volume (l) = 0.179 (body weight (kg) **0.73 Assuming instantaneous uniform mixing and no trapping of the microdisc -:
- ;' ,:

`' ' ,', - : :, - 45 ~
Table 2 Activity of whole blood in cps/ml above background as a function of time after administration of microdiscs. Background /ras 5.5 cps for Experiment 1 and 2 cps for Experiments 2 and 3.
Time after injection (mins) 3 ,um squares 3,um discs 1.5 pm discs 1 23.4 2 5.5 5.3 16.6 ~ 13.27 4 2.4 ~.7 11.98 10.6 6 1.6 3.5 7 9.9 8 1.9 2.9 9 9.4 1.5 2.9 11 9.0 12 1.6 2.5 13 8.3 14 1.5 2.8 7.6 16 1.8 2.6 17 7.6 18 1.8 2.2 : l9 7.
1.8 2.5 22 1.6 2.3 24 1.6 2.1 6.9 26 1.4 2.2 ~ 28 l.9 2.4 29 6.8 ~0 2.1 2.1 34 6.7 39 6.4 6.1 49 6.0 54 6.1 ', ': ' , ' :
:' ,", ': ~ ,. , ` :

- 46 ~
Table 3 "Relative activity" of whole blood as a function of time after injection. "Relative activity i~ expressed as the level of radioactivity above background divider by the projected blood activity if the injected microspheres were uniformly and instantaneously distributed in the total blood volume (Table l) Time after injection (mins) 3 ym squares 3 ~um discs 1.5 /um discs 0.0095 2 0.022 0.0541 0.0067 3 0.0054 4 0.0096 0.0378 0.0048 O.C043 6 0.0064 0.0347 7 0.0040 ~ 0.0076 0.0296 9 0.0038 0.0060 0.0296 11 0.0037 12 0.0064 0.0255 13 0.0034 14 0.0060 0.0186 0.0031 16 0.0072 0.0255 17 0.0031 18 0.0072 0~0224 19 0.0030 0.0072 0.0255 22 0.0064 0.0235 24 0.0064 0.0214 0.0028 26 0.0056 0.0224 28 0.0076 0.0245 29 0.0028 0.0084 0.0214 34 0.0027 39 0.0026 ~4 0~0025 49 0.0024 . - ,.
.
~, :
.
.

~ . ...

Table 4 Activity of samples of various organa e~pressed as cps/gm sample weight about background Organ 3 ~m squareq 3 ~m discs 1.5 /um discs Lung:Right Diagrammatic Lobe 1546 620.8 16104+/-8935 Right Middle Lobe 1438 366.9 25600+/-4640 Right Apical Lobe 898 480.5 27478+/-4181 Left Diagramma-tic Lobe 1893+/-141 61.9 21756+/-4100 Left 21iddle Lobe - 349.3 21000 Left Apical Lobe 1632+/-233 77.8 9300 Accessory Lobe 582.5 31941 ~/-6aO7 Liver: Right Lateral Lobe 6.4 3.5 238.5+/-17.5 Medial Lobe - 3.3 227.5+/-15.5 Left Lateral Lobe 5.5 3.3 184+/-57 Kidney:Right - 18.1 l8+/-5 Left - 17.1 17+/-3 Adrenal - 3. 6,4~ 3 Spleen 7.7 3.3 300+/_71 Pancreas - 2. 6 Heart:Left Ventricle - 1.7 6.8+/-0.7 Right Ventricle 2.3 1.9 lO+/-4 Small Bowel - 2.6 0.0 Colon 1.2 0.0 Skeletal Muscle - 0.8 0.0 Brain:Cerebral Hemispheres - 0.6 0.0 35 - means that no measurement has been made.

.
.. . .
` ~ : ` ` ',; '' ,,, :. ` . ' ,~ . :
-. .
- ~

- ~ .. . . :
, , . .; . .-~ j2~3 Table 5 "Relative activity" of various organs, expres~ed as a ratio o~' tissue activity ( Cp9 above background/g sample weight) to the total injected activity divided by the body weight.
5 Organ 3 jum squares 3 ~um discs 1.5 /um disc~
Lung:Right Diagrammatic Lobe 90.8+/ 8.6 100.7 83.2+/-80 Right Iliddle Lobe 90.1+/-6.53 59.5 132~/-41.2 Righ-t Apical Lobe 56.3+/-11 78.0 142+/-37 Left Diagrammatic Lobe 118.65 10.0 112.3~/-37 Left ~iddle Lobe - 56.7 109+/-43 Left Apical Lobe 102+/-4.6 12.6 48.13+/-36 Accessory Lobe 94.5 165+/-31 Liver: Right Lateral Lobe 0.401 0.568 1.23+/-0.128 Iledial Lobe 0. 535 1.18 Left Lateral Lobe 0. 345 0.535 0.62,1.03,1.19 ~idney:Right - 2.94 0.094+/-0.025 Left - 2.78 0.088t/-0.144 Adrenal - O. 698,0.584 Spleen 0. 48 0.535 1.55+ /-0.366 25 Pancreas - O. 422 Heart:~eft Ventricle - 0.276 0.035+/-0.003 Right Ventricle 0.144 0.308 0.031+/-0.02 Small Bowel - O. 422 0.0 Colon 0.195 0.0 Skeletal Mu~cle - 0.130 0.0 Brain:Cerebral Hemispheres 0.100 0.0 - means that no measurement has been made.

~, "~ . , . ' ,. ~ '',: ' ' ~ :
:- ~

.~

~2~:i2~

Table 6 Relation between activity and blood f'low Tissue Relative blood Relative activity** from experiment flow in tissue* 3 /um squares 3 lum discs 1.5 ~um discs Lung 60 83~/-10.5 52.7~/-16 104+/-34 Skeletal muscle 0.15 0.13 0.0 Kidney 12 2.84+/-0.08 0.09 Heart 6.3 0.126 0.291+/-0.016 0.042 10 Brain 4.8 0.097 O.C

* Relative blood flow = (blood flow per unit mass of tissue 2.blood flow from one side of heart/mass of animal ** Relative activity = activity per unit mass of tissue (activity per unit mass if activity were uniformly distributed) 9 ~ :

' ~ ~

: -.. :, :: ~. : . , , , ~.:, : ~ -The above experimental work shows:-1. The techniques for filtering, drying and resuspending theparticles work and agglomeration is not a problem;

2. There were no acute medical problems despite the fact that the doses were larger than those which would be used as a drug. Thids has also been reported in the literature as follows:-Chemical evaluation of acute cardiopulmonary toxicity of microspheres.
D. R. Allen, J. M. ~errens, F. ~l. Cheney, W B. ~elp.
1978. J. Nucl. Med. Vol. 19 No. 11 p. 1204-1208 Pulmonary perfusion imaging: Acute toxicity and safety factors as a function of particle size.
M. A. DaYi~, R. A. Taube.
1978. J. Nucl. Med. VolO 19 No. 11, p. 1209-1213.

Pathological changes in the lungs of mice following injection of human albumin microspheres.
J. Szymendera, O. Mioduszewska, I. Licinska, A. Czarnomska, B. Lucka.
1977. J. Nucl. Med. Vol. 18 No. 5 p. 478-482.

Blood flow measurements with radiolabelled particles.
M. Heyman, B. D. Payne, J. I. E. Hoffman, A. M. Rudolph 1977. Prog. Cardiovascular Diseases Vol. XX No. 1 p. 55-79.

30 3. The data, particularly for the 1. 5 um devices indicate that the removal by the circulation outside the lung, liver and spleen is small, so perfusing individual limbs or organs with bIood containing IIEDICs is viable.

35 4. To get general unrestricted circulation, coatings which will prevent attack by the recticulendothelial system are needed.

' ~ : ` ;`' :'~`' :`
. .
:~ ~
::: : - ` .
`

Claims (18)

Claims:-

1. An intravascular device comprising a small semiconductor device of size less than 500 µm adapted to pass along blood vessels in large numbers, the devices carrying signal processing means for collectively providing an output in response to an input signal.

2. The device of claim 1 wherein the device is less than 7 µm overall.

3. The device of claim 1 wherein the devices is less than 3 µm overall.

4. The device of claim 1 wherein the input signal is acoustic.

5. The device of claim 1 wherein the input signal is electromagnetic.

6. The device of claim 1 wherein the input signal is temperature.

7. The device of claim 1 wherein the input signal is pH
value.

8. The device of claim 1 wherein the input signal is chemical.

9. The device of claim 1 wherein the output is acoustic.

10. The device of claim 1 wherein the output is electromagnetic.

11. The device of claim 1 wherein the device encapsulates a chemical compound or composition which is released on receipt of an input signal to the device.

12. The device of claim 1 wherein the device includes a battery.

13. The device of claim 2 wherein the device includes piezo electric material and a rectifying structure for generating an electric signal.

14. The device of claim 4 wherein the device includes a diode for measuring temperature.

15. The device of claim 5 wherein the device includes a CHEMFET for measuring pH values.

16. The device of claim 1 wherein the device is coated with an antibody.

17. The device of claim 11 wherein the chemical compound or composition is a pharmaceutical compound or composition.

18. A pharmaceutical preparation comprising a plurality of the devices of claim 17 in a pharmaceutically acceptable carrier or diluent for injection into a blood vessel.