X-Message-Number: 21049
Date: Tue, 04 Feb 2003 00:27:14 +0100
From: Henri Kluytmans <>
Subject: Cost of MNT objects

Thomas Donaldson wrote :

>I was not, at least directly, claiming that biological systems
>would be more efficient. 

But you seemed to suggest it. :)

>Your message said that biological systems provided a proof 
>of principle for the existence of nanotechnological
>systems. I pointed out one way in which they were quite clearly
>different from nanotechnological systems, 

Yes, you are right about that.  

I said :

"There seems to be no theoretical barrier, even better, biological 
systems seem to be a proof of concept (like birds where for flying 

I was meaning to say here, that biological systems are a proof of 
the two fundamental things which the MNT concept and biology have 
in common, and which are :

"molecular building blocks" and "self-reproduction"

Indeed, biological systems cannot be used as proof that the MNT 
concept as a whole is possible. When I made this statement I 
was only having in mind these two central issues of MNT and 
biology. ( This was because Yvan was questioning the possibilty 
of "self assembly" of more than 20 components. )

The two main differences between the way biology works and 
the way MNT should work are : 

1) That biology uses molecular building blocks that are weakly 
bonded (e.g. Van der Waals interaction) to each other 
(e.g. proteins) while the MNT concept want to use very 
strong (covalent) bonds between its building blocks (mostly 

2) That biology uses transport of molecules mostly based on 
Brownian movement while the MNT concept will use mechanically 
controlled systems.

The first will result in MNT producing stronger objects than biology, 
the latter will result in MNT producing atomically precise objects.

Similarly, regarding the birds example, they are only a proof that 
flying objects heavier than air are physically possible, and not, of 
course, that a yet-engine or propeller driven flight is possible. 


>I raised the question of
>whether or not living systems might tell us that the kind of 
>nanotechnological devices you envisioned just wouldn't be
>so easily made that they could compete. I raised a question,
>but did not directly answer it.

And I tried to make clear that current research into MNT 
designs (or as Robert Freitas calls it "Exploratory Design") 
seems to indicate that they will compete, and by a very 
large margin indeed. 

>I WILL point out that in real life at least one factor which 
>you don't mention in your message becomes quite important: the
>energy cost of making a device in the first place. Sure, the
>device when made may use far less energy than a living cell,
>be packed more tightly, etc etc. But how much will it cost to

Hmm, this is a fairly straight forward calculation. I will try 
to do it here for two examples : the MNT computing device with a 
processing power equal to the human brain, and the artificial red 
blood cells (i.e. respirocytes).

The energy released when following materials are burned :

1 KG of lipids   = 38 megajoule 
1 KG of proteins = 17 megajoule 
1 KG of glucose  = 16 megajoule 
1 KG of diamond  = 33 megajoule 
1 KG of methane  = 50 megajoule 
1 KG of gasoline = 44 megajoule 

1) MNT computing device of 10^16 cps (calculations/second)

Merkle made following estimate in "Molecular Repair of the Brain" :

"Given that there are 100 grams of protein and 175 grams of 
lipid in the brain, this means there is almost 2,000 kilocalories 
of chemical energy stored in the structure of the brain, or 
about 8 x 10^6 joules"

Regarding the MNT computing device based on nano-rod-logic :
A 10^16 cps device, uses 1 watt , size 10^-11 m^3 (cube with 0.22 mm sides)
which would contain a chemical energy of about 1 joule when made of 
solid diamond. (Of course, the actual device will contain much less 
total chemical energy in its bonds because it's not a solid block of 
diamond. I guess even more than 2 or 3 times less chemical energy.)

A less conservative human brain processing power estimate would result 
in a 10^13 cps device, that uses 1 microwatt , and has a size of 
10^-14 m3 (cube with 0.022 mm sides), and it would contain less than 
0.001 joule of chemical energy !!!

(These estimates are based on chapter 12 in Nanosystems)

I did not include the cooling part of the device, but on the other hand, 
I took the chemical energy for a volume-equivalent mass of diamond, 
which should make more than good for the omission of the cooling part.
Also the energy cost for transportation of molecules, inefficiencies, 
computation and entropy reduction together are almost ten times less 
than the chemical energy.

2) Artificial red blood cells (or respirocytes)

From the article about respirocytes by Freitas :

- the operational lifespan of respirocyte is about 20 years.
- 1% per day of all red cells are destroyed naturally 
- baseline respirocytes can deliver 236 times more oxygen to the 
tissues per unit volume than natural red cells
- density of red cells = 1100 kg/m3
- a 1-micron respirocyte has a density of 679 kg/m3 with tanks empty 

Assuming a human lifespan of a 100 years :

Every 100 days, all red blood cells must be replaced. Thus for a 
100 years that's 365 times total replacement for red blood cells.
For the respirocytes only 5 times total replacement is required 
in 100 years.

For 1 milliliter of blood the chemical energy is = ~30 Kilojoule
(assuming a 50%/50% proteines/lipids proportion for red cells)

For 1 milliliter of respirocytes chemical energy is = ~22 Kilojoule

For a 100 years, the chemical energy for producing red bloodcells or
an equivalent mass of respirocytes compare as 2*10^4 :: 1  



The energy cost of making a device, and also the operational energy 
usage of device made using MNT should be several orders of magnitude 
better than for biological systems.

If these examples are typical, then it seems that biological systems 
are not very efficient in ANY WAY . 


I didn't even take into account that :

1) Of course, as I said before, MNT systems could be designed 
that would be based on building blocks with lower bond energies 
and Brownian movement transportsystems. Such systems should 
use less energy for building atomicely precise objects than 
biological systems do.

2) Furthermore it's to be questioned if the energy consumption 
for creation of objects is a valid general reference for 
efficiency or economic cost. Energy will be much cheaper in 
the future when solar power collectors can be manufactured 
much cheaper and better than now. One should also consider the 
fact that an enormous amount of energy is being wasted in our 
solar system. More than 99.999999% of the energy the sun is 
producing is radiated into space without being used at all.

3) Also, as Drexler notes in Nanosystems, when organic feedstock
molecules are used as raw material (in stead of CO2) then a
reasonably efficient molecular manufacturing process will be
net energy producer!!

>to make more nanosystems. The construction process you propose
>(not that I think you are alone) COULD turn out to be expensive
>enough that it just isn't practical when compared with how
>living things are made. 

Regarding the "economic price" of MNT produced objects, they 
should be VERY CHEAP indeed, compared to the price of currently 
manufactered products. 

Current products vary between 10-10000 $/Kg

Drexler estimated a price for MNT produced objects of : 0.1 $/Kg 
but here he assumed obtaining raw materials from agricultural 


The basic price of a product is determined by the cost of energy, 
raw materials and development of the design blueprints.

Because the cost of design can be neglected when large 
series of a product are made, the price will be determined 
primarily by the cost of energy and raw materials.

(And there will probably be a lot of designs that will 
be available as open-source, like there are for computer 
software now.)

The raw materials are mostly molecules containing C atoms (and 
some H and/or O). On Earth, at this moment, the easiest source 
to obtain these raw materials is the atmosphere (CO2 and H2O).
These resources are for free.

Energy can be obtained from solar power collectors, that in 
turn can be produced using MNT. Raw material and energy cost 
for creating the solar power collectors themselves can be 
neglected in the long term. The cost of surface area seems 
to be the principal remaining cost!

Merkle made a crude estimate in "Molecular Repair of the Brain" :

"The land cost $100 [for 1 year], so the cost per joule is 
0.9 nanocents and the cost per kilowatt hour is 3.3 millicents. 
Solar power, once we can make the solar cells cheaply enough, 
will be over several thousand times cheaper than electric power 
is today."

Making 1 KG of diamond out of CO2 would require about 33 MegaJoule,
so the energy cost would only be about 0.03 cents!

(Hmm, looks like the price Drexler estimated is very conservative.)

Furthermore, even cheaper would be, creating floating solar panels 
in international waters (or in space), this should eliminate 
the cost of surface area.

>(Not that if this were so, we could
>not use similar reactions to those of living things to make
>our nanodevices!).


But for the moment, it still looks like MNT systems based 
on covalent C-C bonds and mechanical transport systems 
producing atomicely precise objects seem feasible. 
And also that the products made using this technology 
will not be expensive at all.

>Basically I was pointing out one issue which some proposals
>for making nanodevices don't consider. 

Hmm, I seem to recall many articles/books where production and 
energy costs were considered (by Drexler and Merkle).

>In any case, if YOUR proposed systems are too costly compared 
>to the kinds of nanodevices based on chemistry and biochemistry, 
>then no one will buy them. 

From the examples available it looks like it will be the 
opposite : objects manufactured using MNT should be dirt cheap!

(Even compared to products based on biochemistry.)

>If they are put out to compete with
>living things, living things will win, not by being more
>efficient or superior, but simply by numbers alone, 

When considering self-reproducing systems, starting numbers are not 
relevant, in the long term reproduction efficiency is the only 
thing that counts.

>while your nanodevices struggle to reproduce themselves only
>a few times.

Ehh ... why will they struggle to reproduce only a few times ?


Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=21049