Nanoparticles in Homeopathic Dilutions? More Like, Wishful Thinking. Or Magic Pixie Dust.

Those who read my regular posts (Yes, that rare breed of people…) are amply aware that I am no fan of pseudoscience and quackery, as well as the relentless invasion of quackery into academia, leading invariably to scientifically implausible, nonsensical “research”, for which Dr. Harriet Hall had aptly coined the term “Tooth Fairy Science” several years ago over at Science Based Medicine.

You could measure how much money the Tooth Fairy leaves under the pillow, whether she leaves more cash for the first or last tooth, whether the payoff is greater if you leave the tooth in a plastic baggie versus wrapped in Kleenex. You can get all kinds of good data that is reproducible and statistically significant. Yes, you have learned something. But you haven’t learned what you think you’ve learned, because you haven’t bothered to establish whether the Tooth Fairy really exists.

Much of the “research” in the pseudoscience of homeopathy follows the same exalted tradition of Tooth Fairy science. The fact that homeopathy fails to show any effect in properly executed real-world clinical trials is not surprising if one considers that at the root of this quack nostrum are absurd and implausible dilutions at which it is physico-chemically impossible to have any molecule of the starting material left in the solution. However, that has not stopped some homeo-aficionado academics from attempting to provide a mechanistic basis for their claim of homeopathy’s biological actions.

These efforts have borne fruit in the generation of several hypotheses — all speculative, none sufficiently validated empirically — ranging from the scientifically preposterous “memory of water”, prodigious handwaving around quantam physics, as well as poor attempts to shoehorn Clathrate hydrates to fit in the homeopathic theory. The latest of such attempts comes from Prashant S. Chikramane and his colleagues at a prestigious Indian institution of higher studies, the Indian Institute of Technology, Bombay; in a 2010 paper published in the journal Homeopathy (of course!), Chikramane et al. claimed that the extremely diluted homeopathic solutions were not empty of the solute, but in fact contained the starting material — they had used homeopathic preparations of six metals in their tests, copper, tin, zinc, gold, silver, and platinum — in form of nanoparticles and aggregates, detectable by electron microscopic and atomic spectroscopic techniques. In addition, the author claimed that beyond the 6c dilution (dilution factor of 10¹²), further serial dilution of the homeopathic preparations did not result in any further reduction in the concentration of the starting material.

I remember this paper well, especially since it was soundly criticized by, among others, the veteran science-blogger and pseudoscience debunker par excellence, Orac, in an enlightening blog post, where he raised a quite plausible counter-hypothesis, with arguments, that the authors were actually seeing metal contaminants in the dilute solutions (something the authors appeared not to have controlled for), in order to explain the observations presented in the paper. Ordinarily, I have been trying to give these kind of nonsensical pseudoscience a wide berth. However, I was recently reacquainted with this group via their next paper, published in 2012 in the Americal Chemical Society journal Langmuir — basically an extension of the same study — and I wanted to find out what further work they had done.

Those of us who have taken even the most basic of analytical chemistry courses may find it difficult to wrap their mind around the following statement found in the introduction (italicized for emphasis, by me); it is nothing short of amazing to me that it came from a professor in the department of chemical engineering of IIT-B.

Our conclusions arise from our experiments indicating that in the successive dilution process of manufacturing, beyond a certain stage, the dilution is merely apparent and the concentration of the starting material in the diluted product reaches a non-zero asymptotic level no matter how much more the sample is diluted.

Just contemplate this: concentration remains same, no matter how much more the sample is diluted. For all units of measurement of concentration as mass of matter divided by the volume of the diluent, the ratio supposedly remains the same regardless of how much the denominator is changed.

Naturally, every core of my scientist being revolted against this idea that went against all the physics, chemistry and mathematics that I have ever learnt, but I steeled myself to look into how the good professor set out to prove this hypothesis. To be sure, I don’t want to reinvent the wheel here, since Orac has already done such a masterful job of pointing out the major problems in this hypothesis (do go read that post, now!); I would, however, like to first bring attention to the major questions which Orac raised, and follow it up with my own questions about the study. Orac’s questions were:

• Possibility of metalic contaminants in the preparations introduced during the manufacturing process via the environment and/or the solvents — and therefore, impossible to pin down.
• Possibility of impure solvents used in the preparations during the manufacturing process, whereas the experimental negative controls used in the comparison were reasonably pure.
• Possibility of metalic contaminants in the Nitric acid used for solubilization during the experimental steps, given that concentrated nitric acid may have “a fair number of heavy metal impurities. For instance, it can have 5 ppm iron and 10 ppm heavy metals.”

None of these questions were, of course, scientifically tested and answered in the paper. But in addition, once I got into the meat of the paper, especially the supplemental data, the methods and the explanations, I found that the situation even got murkier, especially the question of concentration of the substances in the homeopathic preparations tested.

Presence and Concentration of substances?

The technique that Chikramane et al. used to measure the amount of substances in their dilutions is a valid and widely-used spectrometric technique, Inductively Coupled Plasma Atomic Emission Spectrometry (ICP-AES), which provides accurate measurement of concentrations of substances in a sample. The authors took 2 liters of each dilution of each substance and concentrated it via solvent evaporation to a final volume of 4 ml, a 500-times concentration – which was used as sample for ICP-AES. The results obtained (presented in Table 2) were rather interesting, in that they cast a shadow on the authors’ claims (even though they gloss over it in the discussion).

1. It is difficult not to notice the WIDE variation in concentrations of claimed substances in the dilutions – at different runs, and from different manufacturers. Take, for example, gold (Aurum) and silver (Argentum) metal, which were not detected in most dilutions from both manufacturers. Also, where they were detected, there is no explanation for why a higher dilution (200C) turns out to have similar or more substance compared to a lower dilution (6C).
2. We see the same strange issue with copper, tin and zinc in some runs, too, but more importantly, how there is a substantial amount of copper and zinc in their negative control, 90% ethanol in miliQ water. Is this due to contamination of the instruments from previous runs? We don’t know, but the authors have asserted their system was contamination-proof.
3. If there was indeed some substance in the negative control solutions, why did it not show in the sophisticated technique of SAED (see below) that the authors used to identify substances in their dilutions?
4. By the same token, for the dilutions in which authors detected nothing, how did they manage to get SAED patterns for substances allegedly present in them?
5. Leaving alone the high dilutions, even the starting concentration of different substances in the 6C (i.e.10¹²) dilution ranged from ~80 to ~1350 picograms per ml, i.e.in the range of thousand millionth of a milligram per ml. Which means, that starting solution (of which they had 2L) would have contained ~0.2 to ~3 milligram of each substance. Considering this was 6C, the original stock would have contained 80 grams (of gold, too!) to ~1.4 kg of other substances per milliliter. As you can well see, these amounts are absurd, and one’d think a professor of chemistry would catch that.
6. From the numbers, it is also easy to see that the concentrations of the substances in the diluted samples don’t make sense, the most parsimonious explanation for which would be that the dilution process was inadequate or incomplete or horribly inefficient, or that the people who make up these dilutions have zero understanding of analytical chemistry (or for that matter, anything).

Nanoparticles?

Let me start with a bit of the technical backgrounds. First, the question of nanoparticles. Generation of nanoparticles (particles with at least one dimension ≤100nm) is an immensely high-energy process. A common method uses the ‘laser ablation’ system, which removes material instantaneously from a surface via irradiation with a high powered laser beam; the beam when focused on the surface of a target causes a rapid increase of the temperature of the material, vaporizing it. The vaporized material, alone or in combination with introduced new materials, condenses into particles and particle clusters with diameters in the nanometer ranges. Alternative methods use ultrasound energy (see below), and less preferably, controlled chemical reactions.

Accoustic cavitation (a term Chikramane actually uses) denotes the process of formation, growth and implosive collapse of bubbles in liquids irradiated with high-intensity ultrasound (with frequencies >20KHz); when the bubbles collapse, the gas contained therein is compressed locally, which generates heat and creates a localized, but extremely transient, hotspot (with effective temperatures of ~5000 K, and pressures of ~1000 atm) in which extraordinary physical and chemical conditions temporarily exist. Under such conditions, shockwaves generated can cause interparticle collisions and formation of nanoparticles.

A fundamental part of Chikramane’s claim rests on their observation that there was metal nanoparticles present in the 30C and 200C dilutions of the homeopathic preparations they purchased from two manufacturers. In order to explain the generation of these nanoparticles in situ, the authors considered it to be a consequence of the dilution process used by manufacturers: initially a triturition of raw materials in lactose, a process in which impurities soluble in lactose are leached away, followed by steps of a process peculiar to homeopaths, succussion — which involves pounding of the liquid container against an elastic stop prior to the next dilution. The authors have repeatedly claimed, throughout the paper, that this succussion process creates shearing forces which are responsible for generation of particles of various shapes and sizes; in addition, they have claimed that succussion generates ultrasound waves resulting in accoustic cavitation that generates nanoparticles.

Needless to say, both these hypotheses from Chikramane are suspect. Given the extremely high-energy processes required for the generation of nanoparticles, it is difficult to imagine how ‘pounding liquid in a glass container against a rubber stop 10 times’ would satisfy that requirement. Of course, there is no laboratory assay for evaluating the efficacy of succussion, and the authors haven’t adduced enough evidence in support of these musings. Additionally, the observations from the images as presented in Figure 1 — that (a) supposedly-identical substances bought from two different manufacturers show grossly different appearances, and (b) large aggregates whose sizes vary greatly — don’t inspire confidence either.

In addition, the first figure in supplemental data provides a range of sizes of zinc nanoparticles and aggregates that the authors claim are in the preparations. And yet, there is no convincing explanation of why there is so much of variation in the size range of zinc nanoparticles between two dilutions of the same substance, as well as between two manufacturers. To my mind, this points to noise, not signal, inherent in the system.

(Of note, a peculiarity: while serial dilution after succussion is done by adding more solvent to an amount (aliquot) of the mixture from the previous step, some homeopaths also follow the “Korsakovian” method of dilution, in which the liquid container is emptied and refilled with solvent, and the volume of fluid from the previous step adhering to the wall of the container is arbitrarily considered sufficient starting solute for the next dilution. Use of the Korsakovian method is often indicated by a ‘K’ label with the dilution, such as 200CK. I find this a fascinating illustration of the unscientific nature of the homeopathic preparations.)

Substance identification?

Secondly, the identification of the generated nanomaterials. This uses a combination of imaging and spectroscopic techniques available with a modern analytical Transmission Electron Microscope (TEM). Nanoparticle units exist in form of crystals in which the atoms of the material are arranged in various orientations and packing densities. The crystal lattice structure for each type of substance is, therefore, a unique combination of packed atoms and intervening gaps. This lattice structure lends itself nicely to techniques of electron diffraction, such as selected area electron diffraction (SAED).

In SAED, high energy electrons are made to pass through nanoparticle crystals; the electrons hitting atoms in the crystal lattice are diffracted, while other electrons pass through the gaps — which generates a very characteristic pattern – like a fingerprint – for an element or compound. The software computes intensities of the atomic scatter and distance between planes (d-spacing), generating a pattern profile. There are internationally available comprehensive databases containing physico-chemical and crystallographic information from crystalline substances; for example, the NIST Crystallographic Data Center has compiled a database with a quarter of a million well-characterized crystalline compounds. Software-search of the observed patterns against these databases can, therefore, help identify the substance in the nanoparticle crystal.

This is the technique Chikramane et al. used to identify the substances present in the extremely diluted 30C and 200C solutions, by comparison of their SAED patterns against a database. But if one looks at the presented data (including in the supplemental data), certain questions inevitably arise – especially related to the technique, questions that the authors have not addressed.

• In SAED, each of the spots in the pattern corresponds to a specific diffraction condition of the crystal lattice structure of the substance. However, if the crystal is tilted while under illumination, different diffraction conditions come into play, producing different spots. This is where the hkl (Miller indices) values, the relative intensities and d-spacing values become important. Although variations in these observed values are known to arise from inherent errors in instrumentation, the wide variations of d-spacing values for the same substance at different dilutions and from different manufacturers (as presented in Table 1 and supplementary data) make the identification of the substances less precise and suspect.
• From the SAED patterns presented in Figures 2 and 3, it is not clear why the same substance at different dilutions and from different manufacturers appears to be polycrystalline (discrete halo) and amorphous (diffuse halo). The authors’ attempted explanation of this observation via accoustic cavitation doesn’t make sense for reasons stated above.

Orac’s argument in favor of industrial/environmental contamination with metallic nanoparticles in the purchased preparations seems the most reasonable and parsimonious explanation. Nonetheless, in spite of all these unanswered queries and unconvincing observations outlined above, I wanted to understand how the metallic solute showed up under SAED in some of the dilutions at least. In their 2010 paper, Chikramane et al. had conjured up a handwaving hypothesis in the discussion, mentioning ‘nanobubbles’ with zero evidence. This was a major reason for my interest in this group’s 2012 Langmuir paper, which has been recently touted much by a prominent and indefatigable purveyor of pseudoscience on Twitter. There were some immediately striking features of this paper.

1. Choosing (thankfully) a proper method for generation of nanoparticles, the authors used a chemical synthesis method to generate gold nanoparticles.
2. The authors used various techniques to produce bubbles in the dilutions similar to what they observed during the succussion process.
3. It wasn’t immediately clear from the text whether the authors kept replacing the volume of fluid into the main bottle. If this was the bottle they were aliquoting the top layer and middle layer out from, this was not a serial dilution – since the withdrawal of solute at every step would have had a minimal impact. The aliquots seems to have been diluted 1:100 at every step, making at best a 2C dilution and change. If, however, they did process the new container after putting in the aliquot from the previous one, the dilutions would be fine.
4. For some reason, the authors used quite some paragraph space as well as techniques and time to establish that lactose acted as a stabilizer for the gold nanoparticles preventing their aggregation (which is not a new phenomenon; reducing disaccharide Trehalose is used for the same purpose with colloidal gold particles before desiccation), and that the dilution processing methods — succussion, blowing air, ultrasonication — generated bubbles in the liquid. Suffice it to say, the processes generated millimeter-sized bubbles (not ‘nanobubbles’ claimed in the 2010 paper).
5. As expected, with dilution, the concentration of gold nanoparticles gradually reduced. However, the top layer, containing the bubbles, managed to hold on to the gold nanoparticles for longer duration through dilutions. All that this effect, observed in succussion and air blowing (air blowing was even better than succussion), tells me is that in presence of air bubbles, the dilution process doesn’t work properly because the air bubbles retains solutes.
6. Figure 7 makes it apparent that there is something wrong with the experimental process, since at 1C, the measured concentration of gold nanoparticles in the top layer with lactose doesn’t reach the value shown in Figure 6 for the same conditions. The so-called asymptote formation beyond 6C, therefore, may simply be due to a failure to make a proper, homogeneous solution from the solutes, because 99% (according to the authors) is being retained in the froth.
7. Regardless of explanations, the measureable gold particle still goes down below estimable levels by the 15C (10³⁰) dilution — a far cry from the 30C and 200C dilutions that homeopaths are so fond of.

To my mind, the asymptote effect — retention of the solute nanoparticles — presents an interesting quandary to the principle of Classical Homeopathy. If the solute is retained through subsequent dilutions, doesn’t the whole concept of ‘potentization’ or ‘dynamization’ via dilution go up in smoke? Oh, the irony!

There are many fields in medicine testing out legitimate uses of nanoparticles, such as drug delivery, therapy including non-invasive surgeries and antimicrobials, diagnostics and so forth; homeopathy is most decidedly NOT one of them, despite the fatuous claims made by the homeopaths and homepathy aficionados to the contrary.