Clottocytes: Artificial Mechanical Platelets

April 12, 2002 by Robert A. Freitas Jr.

Nanorobotic artificial mechanical platelets (“clottocytes”) may allow for complete hemostasis in as little as one second – 100 to 1000 times faster than the natural system and 10,000 times more effective in terms of bloodstream concentration. They could also work internally. Using acoustic pulses, a blood vessel break could be rapidly communicated to neighboring clottocytes, immediately triggering a progressive controlled mesh-release cascade.

Originally published June 2000 in Institute for Molecular Manufacturing Report Number 18: Nanomedicine, in conjunction with Foresight Update 41. Published on April 12, 2002.

People often ask for examples of the unique benefits that nanorobots can bring to medicine. That is, what sorts of simple things will robotic nanomedicine allow us to do, that an advanced biotechnology could not accomplish, even in principle? The respirocytes [1] – artificial mechanical red blood cells – are one answer to this perennial question. Respirocytes are micron-sized diamondoid storage tanks for transporting respiratory gases throughout the human body that can be reversibly pressurized up to 1000 atm in direct response to changing tissue requirements. Here, I’d like to describe another interesting example of a simple nanorobotic application that could provide a unique superbiological capability: "instant" hemostasis using clottocytes, or artificial mechanical platelets.

The structure and primary functions of the platelet are well known. In brief, platelets are roughly spheroidal nucleus-free blood cells measuring ~2 microns in diameter with an average bloodstream lifetime of ~10 days [2] and a mean blood concentration of ~250,000 cells/mm3 [3]. Platelets gather at a site of bleeding. There they are activated, becoming sticky and clumping together to form a plug that helps seal the blood vessel and stop the bleeding. At the same time, they release substances that help promote clotting. Natural blood coagulation is a complex process involving platelets, red and white cells, endothelial cells, an array of coagulation factors, fibrinolytic proteins and protease inhibitors whose contributions wax and wane over time. Interestingly, it has been found that platelets can slowly crawl across surfaces [4], and they have other well-studied ancillary abilities such as the phagocytosis of foreign particles [5] and the killing of microfilarial larval parasites [6].

A complete functional design of an artificial platelet is beyond the scope of this paper. Here, I want to focus on the purely mechanical aspects of the hemostatic function of platelets, and describe how this function might be served more effectively by a small in vivo population of medical nanorobotic devices.

After injury to a blood vessel, a natural hemostatic plug is formed which is composed predominantly of platelets. Platelet activation – primary hemostasis – normally proceeds in three phases [13]. The first phase is platelet adhesion, in which a cell monolayer carpet forms in response to the exposure of an appropriate surface to the blood. The relevant natural surface in vivo is thought to be the subendothelial matrix. This matrix lies just below the endothelial cells that coat the blood vessels and would become exposed if a vessel was injured. Artificial surfaces can also induce adhesion. The second phase is platelet aggregation into a plug, mediated by the interaction of fibrinogen with glycoprotein receptors on the platelet surface (the gpIIb/IIIa complex) in the presence of micromolar concentrations of calcium. The third phase is platelet secretion, in which platelet granules release their contents into the extracellular space. These contents include adenosine diphosphate (ADP), Ca++, and various proteins such as platelet factor 4 and thromboglobulin that contribute to the formation of a stable plug, along with other agents such as serotonin and epinephrine, which cause vasoconstriction.

Secondary hemostasis then ensues with the deposition of fibrin. Fibrin strands quickly form a fine meshwork of random fibrils, trapping more platelets and other blood cells to produce a solid clot. Total bleeding time, as experimentally measured from initial time of injury to cessation of blood flow, may range from 2-5 minutes [7-9] up to 9-10 minutes [10, 13] if even small doses of aspirin are present, with 2-8 minutes being typical in clinical practice; minor prolongations up to 15-20 minutes are not considered clinically risky [11-13], and medical dictionaries [14] give the normal coagulation time as 6-17 minutes (360-1020 sec). (Bleeding times begin to be prolonged in otherwise normal patients when their platelet count falls below ~50,000 cells/mm3 [12], ~20% of the normal concentration.) Over the next several hours, the fibrils slowly diffuse within the clot, much as spaghetti moves in boiling water, forming unstable side-to-side monomer associations and thereafter thick bundles, until finally they become cross linked with covalent disulfide bonds by factor XIIIa, making a dense clot. Note that bleeding time is a measure only of clotting due to platelet function and does not account well for the effect of fibrin (the actual coagulant cascade).

By contrast, the artificial mechanical platelet or clottocyte may allow complete hemostasis in as little as ~1 second, even in moderately large wounds. This response time is on the order of 100-1000 times faster than the natural system. Our baseline clottocyte is conceived as a serum oxyglucose-powered spherical nanorobot ~2 microns in diameter (~4 micron3 volume) containing a fiber mesh that is compactly folded onboard. Upon command from its control computer, the device promptly unfurls its mesh packet in the immediate vicinity of an injured blood vessel – following, say, a cut through the skin. Soluble thin films coating certain parts of the mesh dissolve upon contact with plasma water, revealing sticky sections (e.g., complementary to blood group antigens unique to red cell surfaces) in desired patterns. Blood cells are immediately trapped in the overlapping artificial nettings released by multiple neighboring activated clottocytes, and bleeding halts at once.

How much netting can each individual clottocyte carry? The required fiber volume of a mesh that covers an area Anet using fibers of working strength sigmafiber and thickness tfiber with a grid size of lmesh is Vmesh=2(Anet1/2+Anet/lmesh) tfiber2. Minimum fiber thickness is tfiber >~ (pbloodlmesh2/4 sigmafiber)(1/2), where the maximum blood pressure that may be resisted by the netting is pblood ~ 0.25 atm (190 mmHg). Taking lmesh ~1 micron and sigmafiber ~1010 N/m2 for diamondoid fibers, tfiber >~0.8 nm and Vmesh ~0.1 micron3 (taking up 3% of device volume) for a net of area Anet = 0.1 mm2. If instead of the strongest diamondoid fibers we use bioresorbable organic fibers for the netting — having, say, roughly the strength of cellulose or spider silk (sigmafiber ~109 N/m2 [3]) — then the fibers must be 2.5 nm thick and mesh volume becomes 1.3 micron3 (occupying 30% of nanodevice volume) to throw out a 0.1 mm2 net. Stokes law drag power [3] on the net during its 1 second unfurlment time is a fairly modest ~100 pW per nanorobot, assuming whole-blood viscosity at the normal ~45% hematocrit (Hct).

How many clottocytes are needed to stop bleeding in ~1 second? The required blood concentration nbot of nanorobots required to stop capillary flow at velocity vcap in a response time tstop, assuming noverlap fully overlapped nets, is nbot ~noverlap/(Anettstopvcap). Taking noverlap=2, Anet=0.1 mm2, tstop=1 sec, and vcap ~1 mm/sec [3] gives nbot=20 mm-3, or just ~110 million clottocytes in the entire 5.4-liter human body blood volume representing ~11 m2 of total deployable mesh surface. This total dose is ~0.4 mm3 of clottocytes, which produces a negligible serum nanocrit [3] of Nct ~ 0.00001%. During the 1 second hemostasis time, an incision wound measuring 1 cm long and 3 mm deep would lose only ~6 mm3 of blood, less than one-tenth of a single droplet. There are 2-3 red cells per deployed 1 micron2 mesh square, more than enough to ensure that the meshwork will be completely filled, allowing complete blockage of a breach.

Special control protocols are needed to guarantee that clottocytes don’t release their mesh packets in the wrong place inside the body, or at an inappropriate time. These protocols will demand that carefully specified constellations of sensor readings must be observed before device activation is permitted.

For example, the atmospheric concentrations of gases such as carbon dioxide and oxygen are different than in blood serum. As clottocyte-rich blood enters a breach in a blood vessel, nanorobot onboard sensors can rapidly detect the change in partial pressures, indicating that the nanodevice is being bled out of the body. At a nanorobot whole-blood concentration of 20 mm-3, mean device separation is 370 microns. If the first device to be bled from the body lies 75 microns from the air-serum interface, oxygen molecules from the air can diffuse through serum at human body temperature (310 K) from the interface to the nanodevice surface in ~1 second [3]. Detection of this change can be rapidly broadcast to neighboring clottocytes using acoustic pulses that are received in times on the order of microseconds, allowing rapid propagation of a device-enablement cascade. Similarly, air temperature is normally lower than body temperature. The thermal equilibration time [3] across a distance L in serum at 310 K is tEQ ~(6.7 x 106) L2, hence a device that lies 75 microns from the air-plasma interface can detect a change in temperature in tEQ ~ 40 millisec. Other relevant sensor readings may include blood pressure profiles, bioacoustic monitoring, bioelectrical field measurements, optical and ultraviolet radiation detection, and sudden shifts in pH or other ionic concentrations. At some cost in rapidity of response, clottocytes also could eavesdrop [3] on natural biological platelet control signals, using sensors with receptors for the natural prostaglandins produced by endothelial cells that normally induce or inhibit platelet activation.

The rapid mechanical action of clottocytes could interfere with the much slower natural platelet adhesion and aggregation processes, or disturb the normal equilibrium between the clotting and fibrinolytic systems. Thus it may be necessary for artificial platelets to release quantities of various chemical substances that will encourage the remainder of the coagulation cascade to proceed normally or at an accelerated pace, including timed localized vasodilation and vasoconstriction, control of endothelial cell modulation of natural platelet action, and finally clot retraction and fibrinolysis much later during tertiary hemostasis.

There is a small risk that a potentially-fatal catastrophic clotting cascade called disseminated intravascular coagulation (DIC) [18] could be triggered by excessive clottocyte activity. Coagulation is usually confined to a localized area by a combination of bloodflow, localized thrombin production, and circulating coagulation inhibitors such as antithrombin III (a potent thrombin inhibitor). But if the stimulus to coagulation is too great, excess thrombin is produced and enters the general circulation. This overwhelms the natural control mechanisms and leads to excess fibrin deposition, formation of large numbers of microthrombi (intravascular clotting), rapid depletion of platelets and fibrinogen (and other coagulation factors), secondary fibrinolysis, and often hemorrhage, the typical signs of acute DIC. One solution is to equip clottocytes with sensors to detect decreased serum levels of fibrinogen, plasminogen, alpha2-antiplasmin, antithrombin III, factor VII and protein C, and elevated levels of thrombin and various fibrin/fibrinogen-derived degradation products. If DIC conditions arise, nanorobots might respond by absorbing and metabolizing the excess thrombin, or by releasing thrombin inhibitors such as antithrombin III, hirudin, argatroban or lepirudin [19] or anticoagulants that reduce thrombin generation such as danaparoid [19] to interrupt the cascade. For example, a ~0.02% Nct concentration of nanorobots, suitably activated, could replace the entire depleted natural bloodstream content of antithrombin III from onboard stores.

Extended bleeding often serves to cleanse a wound of foreign matter and bacteria that might have entered the body along with the skin-penetrating object that caused the wound. By immediately staunching the flow, it might be argued that clottocytes would prevent this natural cleaning action from taking place. However, it is anticipated that clottocytes would only be one component of a complete hematological "upgrade" package, and that other species of circulating nanorobots would perform these scavenging and janitorial tasks.

Yet another possible complication is that the bare tissue walls of a wound will continue to exude fluid, and may begin to desiccate, if only the capillary termini are sealed but the rest of the tissue is left exposed to open air. Since clottocytes may remain attached to their discharged nets, and can communicate with each other via acoustic channels [3], it should be possible to precisely control the development of a larger artificial mesh-based clot via coordinated mesh extensions or retractions within the clot. Alternatively, clottocytes could allow blood fluids to flood small incised or avulsed wound volumes, allowing exposed tissue walls to be bathed in fluids but casting a watertight sealant net across the wound opening flush with the epidermal plane of the wound cavity.

What about internal bleeding? Clottocytes will require far more sophisticated operational protocols if they are intended to assist platelets participating in the sealing of internal blood vessel lesions, in order to avoid inadvertently blocking the lumen of the entire vessel. Similarly, prevention of bleeding at vascular anastomoses, hemarthroses, internal bruising, "blood blisters" and larger tissue hematomas, as well as forced local coagulation in tumors or in intracerebral aneurysms, may also require more advanced protocols, possibly including integration with pre-existing in vivo navigation systems [3]. For some of these applications, motile clottocytes may be required in place of the free-floating nanorobots described in this paper, along with a graduated recruitment response depending upon how many (intercommunicating) devices appear to be involved in the event.

Numerous significant design questions remain, including most importantly biocompatibility issues — are clottocytes truly inert? Will they interact with other blood cells or with endothelial cells? Will they activate complement pathways or elicit fibrin deposition? Diamond is indeed chemically inert [15, 16] and is generally considered noninflammatory relative to the complement system [17]. An enveloped clottocyte externally coated with autologous platelet membrane should be nearly as biocompatible as native platelets and could also assist in the recruitment of intrinsic coagulation mechanisms — especially important for severely thrombocytopenic (platelet-poor) patients. Another requirement is that the bioresorbable netting must be capable of being broken up into phagocytosable pieces, either by natural enzymatic pathways or by artificial fiberlytic enzymes (analogous to fibrinolytic plasmin) that may be released from the clottocyte at the appropriate time. The fiber material should also be nonimmunogenic, to avoid uncontrolled immune-mediated platelet activation [19]. Further analysis must await the completion of Volume II of Nanomedicine.

To summarize: an artificial mechanical platelet appears to permit the halting of bleeding 100-1000 times faster than natural hemostasis. While 1-300 platelets might be broken and still be insufficient to initiate a self-perpetuating clotting cascade, even a single clottocyte, upon reliably detecting a blood vessel break, can rapidly communicate this fact to its neighbors, immediately triggering a progressive controlled mesh-release cascade. Clottocytes may perform a clotting function that is equivalent in its essentials to that performed by biological platelets — but at only ~0.01% of the bloodstream concentration of those cells. Hence clottocytes appear to be ~10,000 times more effective as clotting agents than an equal volume of natural platelets.


The author thanks Stephen S. Flitman, M.D., C. Christopher Hook, M.D., and Ronald G. Landes, M.D., for helpful comments on an earlier version of this paper.


1. Robert A. Freitas Jr., "Exploratory Design in Medical Nanotechnology: A Mechanical Artificial Red Cell," Artificial Cells, Blood Substitutes, and Immobil. Biotech. 26(1998):411-430. See also:

2. J. Willis Hurst, Medicine for the Practicing Physician, Third Edition, Butterworth-Heinemann, Boston MA, 1992, p. 771.

3. Robert A. Freitas Jr., Nanomedicine, Volume I: Basic Capabilities, Landes Bioscience, Georgetown, TX, 1999; see at:

4. Etsuko Ito, Ken Suzuki, Masayuki Yamato, Masayuki Yokoyama, Yasuhisa Sakurai, Teruo Okano, "Active platelet movements on hydrophobic/hydrophilic microdomain-structured surfaces," J. Biomed. Mater. Res. 42(October 1998):148-155.

5. R.P. Awadhiya, J.L. Vegad, G.N. Kolte, "Demonstration of the phagocytic activity of chicken thrombocytes using colloidal carbon," Res. Vet. Sci. 29(July 1980):120-122; B.S. Dhodapkar, J.L. Vegad, G.N. Kolte, "Demonstration of the phagocytic activity of chicken basophils in the reversed Arthus reaction using colloidal carbon," Res. Vet. Sci. 33(November 1982):377-379.

6. A. Haque, W. Cuna, B. Bonnel, A. Capron, M. Joseph, "Platelet mediated killing of larvae from different filarial species in the presence of Dipetalonema viteae stimulated IgE antibodies," Parasite Immunol. 7(September 1985):517-526.

7. R. Kumar, J.E. Ansell, R.T. Canoso, D. Deykin, "Clinical trial of a new bleeding-time device," Am. J. Clin. Pathol. 70(October 1978):642-645.

8. HealthGate Medical Tests, "Bleeding Time," 27 July 1999, see at:

9. L. Ardekian, R. Gaspar, M. Peled, B. Brener, D. Laufer, "Does low-dose aspirin therapy complicate oral surgical procedures?" J. Am. Dent. Assoc. 131(March 2000):331-335.

10. L.R. Hertzendorf, L. Stehling, A.S. Kurec, F.R. Davey, "Comparison of bleeding times performed on the arm and the leg," Am. J. Clin. Pathol. 87(March 1987):393-396.

11. A. Barber, D. Green, T. Galluzzo, C.H. Ts’ao, "The bleeding time as a preoperative screening test," Am. J. Med. 78(May 1985):761-764.

12. Franklin A. Bontempo, "Use and Misuse of the Bleeding Time," April 1994, see at:

13. Stuart E. Lind, "Chapter 33. The Hemostatic System," in Robert I. Handin, Thomas P. Stossel, Samuel E. Lux, eds., Blood: Principles and Practice of Hematology, J.B. Lippincott Co., Philadelphia, PA, 1995, pp. 949-972.

14. Clayton L. Thomas, ed., Taber’s Cyclopedic Medical Dictionary, 17th Edition, F.A. Davis Company, Philadelphia PA, 1989.

15. R. Kornu, W.J. Maloney, M.A. Kelly, R.L. Smith, "Osteoblast adhesion to orthopaedic implant alloys: effects of cell adhesion molecules and diamond-like carbon coating," J. Orthop. Res. 14(November 1996):871-877.

16. R. Lappalainen, A. Anttila, H. Heinonen, "Diamond coated total hip replacements," Clin. Orthop. 352(July 1998):118-127.

17. M. Doherty, J.T. Whicher, P.A. Dieppe, "Activation of the alternative pathway of complement by monosodium urate monohydrate crystals and other inflammatory particles," Ann. Rheum. Dis. 42(June 1983):285-291.

18. R.L. Bick, B. Arun, E.P. Frenkel, "Disseminated intravascular coagulation. Clinical and pathophysiological mechanisms and manifestations," Haemostasis 29(1999):111-134; M. Levi, E. de Jonge, T. van der Poll, H. ten Cate, "Disseminated intravascular coagulation," Thromb. Haemost. 82(August 1999):695-705; 706-721 (related).

19. T.E. Warkentin, "Heparin-induced thrombocytopenia: a ten-year retrospective," Annu. Rev. Med. 50(1999):129-147; D.B. Brieger, K.H. Mak, K. Kottke-Marchant, E.J. Topol, "Heparin-induced thrombocytopenia," J. Am. Coll. Cardiol. 31(June 1998):1449-1459.

Copyright 2000 Robert A. Freitas Jr. All Rights Reserved. Used with permission.