Seven decades of precious metals refining with response and responsibility

Partners in profitability

partners in profitability

Partners in profitability

Regular recovery of platinum and rhodium losses in nitric acid plants has become a significant factor in overall plant economics. Alan E. Heywood of Sabin Metal Corp. presents a general overview of the causes of PGM losses, the recovery of these valuable precious metals, and the sampling and refining procedures used to maximize their recovery


The production of nitric acid via the oxidation of ammonia over a rhodium- platinum alloy catalyst, generally known as the Ostwald Process, is based on the following equation:

partners in profitability

This highly efficient exothermic reaction results in significant platinum losses. Subsequent reactions are:

partners in profitability

The extraordinary increase in platinum and rhodium values that has occurred over the past five years (Figs 1 and 2) dictates that regular recovery from nitric acid plants is now a significant factor in overall plant economics, and thus, profitability. In fact, global annual Platinum Group Metal (PGM) losses from the Ostwald process at today’s prices amount to approximately $500 million.

Fig 1: Platinum prices 1992-2007

partners in profitability

Fig 2: Rhodium prices 1992-2007

partners in profitability

Platinum loss mechanism

It is generally accepted that the loss of platinum from ammonia oxidation catalysts occurs primarily in the form of volatile platinum oxide, typically PtO2, as opposed to the elemental metal. Indeed, it’s widely believed that the formation of platinum oxide is a prerequisite for this extremely efficient reaction to proceed.

However, some loss does occur as the metal, by physical attrition, the "cauliflower" growths which necessarily form from the initially smooth wire (Figs 3 & 4) becoming extremely fragile, particularly towards the end of an extended run, when the core wire may become extremely thin (Fig 5).

That the loss of platinum, and rhodium, is primarily as the oxide is supported by the fact that, at the catalyst operating temperature of about 900°C, the vapour pressures of the oxides are higher than that of the metal by many orders of magnitude1,2 (see Fig. 6), and the lower vapour pressure of rhodium oxide would explain the preferential loss of platinum from catalyst gauzes e.g. a used 5% Rh-Pt catalyst gauze typically assays 6-7% Rh.

Further credence is given to this hypothesis by the fact that the somewhat similar Andrussow process, used for the synthesis of hydrogen cyanide

partners in profitability

operates in an essentially neutral atmosphere at a temperature approximately 100°C higher than the Ostwald process, but nevertheless exhibits extremely low PGM losses.

The unit PGM losses from an Ostwald catalyst are generally in the range 0.1-0.4 gram/tonne (0.0029-0.0116 Troy oz/ton) 100% HN03, this typically equating to an overall loss of 30-60% of the originally installed weight, but the overall losses from an Andrussow catalyst are, normally, only 2- 3% of the originally installed weight.

The PGM losses for the Ostwald process are far higher than might be expected simply by operating platinum at 900°C in air, pure ammonia, oxygen or nitric oxide, and it has been suggested that the true surface temperature of the catalyst may in fact approach, or even exceed, that of the melting point of the alloy; i.e. 1800°C or more.3

This factor, allied with the high vapour pressures of the oxides versus the elemental metals, would readily explain the very significant PGM losses that occur during the ammonia oxidation stage of the Ostwald process, and therefore strongly suggests that the losses are primarily in the form of the oxide vapours rather than the metals.

Figs 3, 4, 5: Restructuring of catalyst gauze wires

partners in profitability
Fig 3: Typical used Rh-Pt catalyst gauze.
partners in profitability
Fig 4: Close-up of Fig 3 showing "cauliflower" growths

partners in profitability 9
Fig 5: "Cauliflowers" detached from an excessively thinned catalyst wire. The wire diameter is approximately 25 microns (0.001").

Fig 61: Vapour pressure of Pt and Rh and their oxides over 10% Rh-Pt in air

partners in profitability

What happens to the lost PGMs?

Reaction (2) has a negative temperature coefficient and is therefore favoured by low temperatures.

To reduce the temperature of the products from reaction (1) as quickly as possible and also to recover the large amount of heat generated by the reaction, a series of heat exchangers is placed in the process gas stream e.g. as illustrated in Fig. 74. These heat exchangers may be waste heat boilers, reheaters, steam superheaters, etc., depending upon the plant technology.

PGMs will deposit on these heat exchangers, the amount and PGM ratios depending on the location (Fig. 8)5, the exchanger design, the temperature and, considering Fig 6, the local pressure in the plant.

Fig 74: Ammonia oxidation – sample locations of PGM deposition

partners in profitability

Fig 85: Deposition of PGMs

partners in profitability

PGMs in the hottest areas, will also, to a degree, diffuse into the base metal of the heat exchanger tubes.

Over the years PGM recovery has been affected from various sources, although generally not in concert.

Examples, in roughly historical order of application, are:

  • Physical retrieval of dusts and sludges from various parts of the plant e.g. absorption columns and acid storage tanks;
  • Destructive recovery of redundant heat exchangers;
  • Filters which trap PGM particles in the process gas stream;
  • Getter ("catchment") gauzes;
  • Abrasive/hydroblasting cleaning of surfaces exposed to PGMs;
  • In-situ, acid cleaning techniques, using inhibited acids to remove oxide/PGM layers from heat exchangers without destruction;
  • Liquid acid filters which trap PGM particles, generally in the weak acid stream.

The various recovery methodologies in order of regular PGM returns are detailed below.

Getter ("catchment") gauzes, installed immediately under the catalyst, represent an extremely efficient, predictable – and very importantly with regard to cash flow – regular method of recovering PGMs, the getter typically being changed with the catalyst. The basis of this recovery system is that volatile platinum oxide is readily dissociated on the palladium surface, the free platinum forming a homogeneous series of continuous solid solutions with palladium.6 At the operating temperature of the catalyst the diffusion of platinum into palladium is extremely rapid, resulting in a homogeneous Pd-Pt alloy (Figs 10 a-e) which can be readily refined.

Some rhodium is also recovered, but may largely be in particulate form (Figs 10 c and f) and lightly adherent to the palladium/platinum surface. The recovery rate of rhodium can, therefore, to an extent be somewhat variable, particularly if the plant experiences numerous involuntary shutdowns.

Fig 9: Platinum and palladium prices 1992-2007

partners in profitability

Palladium has a high vapour pressure7 and, in use, getter gauzes lose some palladium. Consequently their economics can be sensitive to platinum/palladium price ratios but, at this time, the case for their use is irrefutable (Fig. 9), net savings well in excess of over $1 million/annum being readily possible with large high pressure plants.

Additionally, the lost palladium is largely recoverable from process gas filters and/ or heat exchangers (Fig. 11).

Getter gauzes being originally of European origin8 where, historically, most nitric acid plants have operated at relatively low pressures and gas velocities, tended to exhibit excessive pressure drop when used in the high pressure/high velocity plants which are prevalent in North America. These problems have recently been overcome with systems specifically designed to operate in the high pressure/high velocity regime – e.g. the "SRS" series developed by the Sabin Metal Corporation – and it is now possible to run getter gauze systems in high velocity plants for 100+ days with less than 0.055 bar (approx 0.8 psi) pressure drop being recorded at the end of the acid making campaign.

Figs 10 a-f: PGM mapping of a used Pd getter gauze

partners in profitability partners in profitability

10a: Used Pd getter gauze

10b: SEM image of surface from Fig 10a (Light coloured particle, centre, is Rh)

partners in profitability 16 partners in profitability 17

10c: PGM mapping of Fig. 10b showing the homogeneous distribution of Pd and Pt (Note Rh particle, centre)

10d: Pt mapping of Fig 10b

partners in profitability 18 partners in profitability 19

10e: Pd mapping of Fig 10b

10f: Rh mapping of Fig 10b (Same Rh particle as seen in Figs 10b and 10c)

Process gas filters can be a very efficient method of recovering PGMs, particularly in high pressure plants running for extended periods and where the Pt particles may be relatively large (Fig 5). Such filters are designed to collect particulate, not "molecular" PGMs (see above re: getter gauzes) and the two technologies are therefore, to a significant extent, complementary, it being readily possible to "tune" the recovery rate of each system and optimise their respective recovery characteristics as a "total system."

It is extremely important that the filter design is such that it will withstand the rigours of the hot process gas and effectively balance pressure drop and recovery efficiency, as such filters will also collect other particulates that pass through the plant (e.g., metallic oxides, airborne dust, etc.), not just the larger PGM particles.

Fig 11: Mapping of palladium on a tube

partners in profitability partners in profitability

11a: Surface deposit from a secondary ("warm") tube.

11b: X-Ray mapping of 11a: Fe+Pt+Pd.

partners in profitability partners in profitability

11c: X-Ray mapping of 11a: Pt

11d: X-Ray mapping of 11a: Pd

Note the preponderance of Pd indicating the potential for recovery of any Pd lost from getter systems

Some of the most effective process gas filter systems use ceramic fibre, rather than glass fibre, as the filter medium and their servicing is best left to the industry experts.

As with destructive recovery (see later), used filter elements should be securely packed after their removal from the plant to avoid the loss of loosely adherent PGM particles.

Abrasive/hydroblasting techniques are, essentially, line of sight cleaning processes and therefore cannot always get into the remote recesses of the system but they can nevertheless generate significant residues which are frequently very rich in PGMs (e.g. Figs 12 (a-e), and which are readily recoverable.

In-situ acid cleaning techniques rely on a very thorough understanding of the "controlled dissolution" that the process is essentially based upon (Figs 13 & 14). Possible problems include the removal of corrosive products that are sealing incipient leaks, and the evolution of hydrogen during the "leaching" process. This work needs to be carried out by experts.

A significant and complementary benefit of in-situ cleaning is the restoration of the tube surfaces to their original, unoxidized condition thereby significantly improving heat transfer (Figs 14 a and b).

Liquid acid filters which trap the very fine PGM particles that can remain in the acid stream after the getter, heat exchangers and PGM filters, are a fairly recent development and, to date, the results from their use seem to vary widely, as does their reliability. Much more work needs to be done in this very promising area of PGM recovery.

Retrieving dust and sludges from plant equipment and absorption columns is not always rewarding; known PGM assays from storage tanks vary from less than 500 ppm to 50%, depending upon factors such as the residence time of the acid in the tank and the outlet geometry. That said, no downstream residues recovered from a nitric acid plant should ever be disposed of without having a "check" assay carried out by a reputable refiner. Many a fortune has been lost because the residue in a storage tank was simply disposed of in a landfill.

It is, however, extremely important to recognize that "check" assays are not an indication of the actual PGM assay, but they can nevertheless act as a very useful "go/no-go" guide.

The extent to which the PGMs diffuse into the base metal parts of the heat train is highly dependent upon factors such as the operating temperature and oxide scale barriers and destructive recovery, as its name suggests, involves the dissolution of redundant parts, to the point where all surface and diffused PGMs have been removed, the remaining base metal then being available for sale as scrap.

One vital point that is frequently missed when removing and storing redundant parts is that they should be securely wrapped immediately after their removal from the plant as the dusts are often loose and frangible (Fig 12). Indeed, analysis of the soil beneath one heat exchanger that had been left to "weather in the boneyard" revealed significant quantities of PGMs. Heavy duty "shrink wrap," such as that used to protect recreational boats in winter, is ideal for wrapping redundant equipment destined for destructive recovery.

Whichever recovery technique is used the next key step is the refining of the residues resulting from the recovery process.


A key consideration in selecting a PGM refining organization is that its PGM outturn should be accepted without question by the bullion markets because "stranger" metal will always need to be sampled and assayed – and at the owner’s cost. Irrefutable evidence of the provenance and purity of the refining outturns is therefore of paramount importance.

Furthermore a clear understanding of, and compliance with, the refiner’s regulatory permits and waste disposal programme is necessary to avoid any possible environmental regulations – including the recently enacted Anti-Money Laundering Rule of the USA Patriot Act.

Fig 12: SEM/EDAX images of PGM containing dust lightly brushed from a tube surface

partners in profitability
12a: SEI image
partners in profitability
12b: X-ray mapping of surface dust from Fig 12a
partners in profitability
12c: Pt X-ray map from Fig 12a
partners in profitability
12d: Pd X-ray map from Fig 12a
partners in profitability
12e: Rh X-ray map from Fig. 12a

The refining of PGMs is an extremely complex, closely monitored series of chemical processes, notably where the lesser metals (e.g. Rh) are concerned. That said, of no less importance is the process of sampling the material to be refined and this is one of the most critical aspects of assuring highest possible PGM returns.

Much work has gone into the practice and statistics of sampling, notably by those such as Jacobsen9, whose work is both entertaining as well as educational, and some of the essentials of refining of even low grade materials, such as sludges and dust as might be recovered from nitric acid plants, have been covered by Beirne.9

With regard to materials sampling, solution sampling is extremely accurate as long as cognisance is taken of the fact that some "insolubles" (rhodium oxide, in particular) may remain and that these will need to be separately sampled and refined.

Fig 13: In situ acid cleaning of tubes

partners in profitability
13a: In situ acid cleaning of a primary ("hot") tube showing detaching Pt-rich surface layer
partners in profitability
13b: Analyses from position 2 in Fig. 13a a=Pt rich, b=Cr rich, c,d,e and f=Fe/Cr rich

Fig 14: Tube surface after in situ cleaning

partners in profitability
14a: Tube surface after removal of the deposit shown in Fig. 13 after non-destructive in-situ cleaning
partners in profitability
14b: Elemental mapping of Fig. 14a showing the relative absence of PGM residues on the tube surface after non-destructive, in situ cleaning. Magenta=Fe

It also makes sense from the perspective of both parties to have independent referees present and that "referee" (sometimes also known as "umpire") samples are reserved in case of dispute. Any reputable refiner will welcome this because, whilst the customer’s financial interests are clearly paramount, equally no refiner wants to be bankrupted by erroneous assays, so the use of witnessing referees serves the interests of both parties.

In short: given current PGM prices it is worthwhile for nitric acid producers to investigate all methods of PGM recovery, both individually and in concert, establishing the optima for its plant(s).

Then, if the maximum benefit from PGM recovery operations is to be realized, the producer and the refiner need to work very closely together, agreeing to the most suitable recovery, refining, sampling, and assaying techniques.


This article is not meant to be exhaustive rather, given the currently prevailing high price of all PGMs, to encourage their recovery throughout all the post-catalyst locations in nitric acid plants.

It is clear that there is still much work to do in the identification of PGM deposition mechanisms, and the resulting PGM ratios, but some immediately obvious rules are:

  • In the short term: Ally the recovery rates of getter gauzes with those of process gas filters for optimum, predictable, and regular recoveries.
  • In the medium term: Never ignore any residues found in the plant. However, be prepared to be surprised, either way.
  • In the longer term: Review opportunities for both destructive and non-destructive recovery.
  • And finally: Work very closely with your refining organization. It is as keen as you are to recover the maximum amount of PGMs from your plant. After all, it’s their business to save you money and enhance your profitability.


  1. Selman G.L. and Rushforth R.W., Platinum Metals Review, 1971, 15, (1), 84.
  2. Knapton A.G., Platinum Metals Review, 1979, 23, (1), 4.
  3. Garton G., Turkevitch J., J. Chim. Phys., 1954, (51), 516.
  4. Fierro J.L.G., Palacios J.M. and Tomas F., Platinum Metals review, 1990, 34, (2), 63.
  5. Ibid. p 64.
  6. Hansen M. "The Constitution of Binary Alloys," Second Edition, McGraw- Hill,1958, 1121.
  7. Chaston J.C., Platinum Metals Review, 1965, 4, (4),125 -129.
  8. Holzmann H., Chem-lng-Tech, 1968, 40, 1229
  9. Jacobsen R.T, "Simply Sampling," Hydrocarbon Engineering, (Sep 2003).
  10. Beirne K. M., "Choice of Refinement," Hydrocarbon Engineering (Jul 2003).


The contribution of Mr. B. L. Wibberley and Dr. Xian Wei Liu of the University of Hertfordshire, Hatfield, Herts, UK, through the preparation of the SEM images and X-ray mapping, is gratefully acknowledged.

Contact the author at