Ensure Fair Market Value Through Accurate Sampling

catalyst recovery - part 2

Catalyst Recovery – Part 2
Robert T. Jacobsen Sabin Metal Corp.
Sampling is the key to determining the precious-metals content of spent catalysts, and ultimately the value of the metals.


THE FIRST ARTICLE (1) IN THIS SERIES described some of the issues involved in precious-metals recovery, including sampling and assaying, processing turnaround time, metals leasing and financing, and environmental concerns. While each of these functions is important, sampling procedures for spent catalyst materials are perhaps the most critical elements with regard to maximizing the recovery of precious metals from spent process catalysts.

This article provides additional information on specific sampling techniques and their influence on the value received for spent precious-metal-bearing catalysts. As a catalyst user, it is in your best interest to understand how sampling processes help refiners determine the precious-metals content of spent catalysts and, ultimately, the amount (and value) of these metals that is returned to you.

Fundamentally, sampling spent catalysts containing precious metals is similar to sampling any homogenous mass. The problem is, even new catalysts on substrates (carriers) such as soluble and insoluble alumina, alumi-no-silicate, zeolite or carbon supports are not homogenous masses. After years of exposure in the harsh environment of a catalytic reaction process, spent catalysts are far from homogenous. They accumulate many different contaminants of various densities, including sulfur, carbon, solvents, oils and water, during their useful lives (typically 5–6 yr).

A portion of the spent catalyst must first be reduced to a size amenable to laboratory analysis. This involves working with large quantities of spent catalysts (as much as many tens of tons) and ultimately extracting quantities as little as a few grams, while maintaining a precious-metal concentration representative of the original lot. During this process, contaminants that might interfere with that determination must also be eliminated.

The goal, therefore, is to obtain a sample that is as precisely representative of the overall material lot as possible, and as homogenous as possible, in order to make as accurate a determination as possible of the actual value of recoverable precious metals within the lot. Clearly, this is easier said than done, since there are many processes, evaluations, equipment and systems involved in the sampling process, and there are also vastly different sampling methods used depending on circumstances. Another critical factor is the refiner’s experience and expertise, since some sampling procedures — and their ultimate outcome — are affected solely by judgment. No matter what, all of these steps must be accomplished to arrive at an accurate, final value determination.


The three sampling techniques for spent catalysts are dry sampling, melt sampling, and solution sampling. Each of these uses specific methods and equipment and offers specific advantages. The most appropriate sampling method for a particular spent catalyst depends on the type of material being processed as well as its estimated precious-metals content.

Because of their composition and chemistry, precious-metal-bearing catalysts are usually sampled using dry sampling processes. Dry sampling is employed whenever materials cannot be dissolved in solution or are inappropriate to melt either because of their structure or because of the cost associated with melting vs. the possible return.

Because it is difficult to achieve homogeneity, dry sampling is more complex and potentially less precise than melt or solution sampling. In fact, this method requires more judgment and skill than the others.

An ideal dry sampling system would be capable of drawing representative samples from free-flowing catalyst at a rate of 2,000–3,000 lb/h according to the principles and practices of Pitard (2), Gy (3) and Merks (4).

Because sampling is considered the most important procedure in the precious-metals recovery and refining process, it must be viewed from the perspective of the refiner as well as the catalyst user (the refiner’s customer). Clearly, the customer’s goal is to receive the maximum possible value for the remaining precious metals in its spent catalyst materials. The refiner, on the other hand, must not only consistently meet that goal for its customer, but it must also provide the customer with detailed documentation as to how that value was determined. The refiner and the customer each have responsibilities that must be met to ensure a mutually rewarding relationship.

As previously mentioned, dry sampling procedures involve converting many tons of spent catalyst materials to as little as a few tens of grams with the same precious-metal concentration as was in the original lot of material. This is essentially a statistical process, and the laws of random sampling must be observed closely.


Any sampling method requires a free-flowing material. If the catalyst is agglomerated into large balls because of oil or residual coke, or if it is covered with standing liquid, etc., it simply cannot be sampled. Pretreatment is then necessary.


While a grain thief (a device consisting mainly of a long tube used to take samples at various depths) can produce a fairly representative sample if proper precautions and multiple samples are taken (4), it becomes impractical when the lot size to be sampled is greater than a few hundred pounds. Large lots require some means of drawing the sample from a flowing stream that ultimately contains all of the material. The design characteristics of these continuous samplers (auto-samplers) vary considerably, but all must have certain characteristics:

  1. The device that extracts the sample periodically must cut across the entire flowing stream of particles, taking an equi-vol-umetric sample from all portions of that stream. (This is often violated in commercial automatic samplers.) Many factors in a flowing stream cause the material to segregate, for example by particle size, shape, density, surface charge, etc. Thus, if one portion of the stream is favored in the selection process, it will inevitably bias the sample and, therefore, its apparent precious-metals content).
  2. The extraction device must be large enough that the size of the particles being collected is small by comparison. If this is not the case, the sample will be biased against the larger particles in favor of the smaller pieces. Many refiners consider that the ratio of the smallest part of the collector to the largest particle size should be a minimum of a factor of ten.
  3. Many collections, or cuts, of the stream (increments) should be taken at regular intervals during the passage of the entire lot through the device. The number should be well into the hundreds at a minimum to avoid the vagaries that the statistics of small numbers might introduce to the sample.
  4. All of the material contained in the subject lot of catalyst must end up either in the collected sample or in the uncollected material bin (the reject). This may appear to be obvious, but it is not always realistic.

    For example, material may be ejected from the system by higher-speed parts of the device or might fall between collection bins. In either of those cases, a biased sample would result because some difference in the particles would make it more likely that a certain type of particle is missed and therefore biased against, and there is no way of knowing whether those particles were higher or lower in precious-metals content.

    Dust collection makes it very difficult to keep all of the material in the sampler as either sample or reject. Most samplers employ integral dust collection for reasons of health and safety, as well as material conservation. This material should be collected in an easily observed manner and sampled separately, with credit given to the customer for its value. The precious-metals concentration (grade) of this fine material is usually substantially different (higher or lower) than that of the bulk lot.

  5. The sampling system should be easily cleanable to avoid cross-contamination.
  6. Finally, the sampling system should be easily observable by its operators and by the customer or his or her representative. Customer representation at precious-metals refiners is a generally accepted standard in the industry. It is a false economy to minimize representation to save a few dollars. Sampling is a complex process, and small mistakes or misunderstandings can be very costly.


Sampling is not as simple as merely loading an arbitrarily large lot of catalyst of arbitrarily sized particles into the device and reduce its size to a sample bag suitable for laboratory analysis.

Consider first the lot size itself. Assume that we can achieve a sample that represents the lot to a precision and accuracy of 1%. Also assume that the total lot size is 100 tons and that the grade is approximately 0.3% platinum. (These assumptions are typical.) The total platinum value in the lot — at current Pt prices of about $900 per troy ounce — is over $7.8 million. One percent of that value is $78,000. Most organizations would not tolerate such a variance.

Most refiners and customers will insist that the original lot be broken up into sub-lots of more manageable value, perhaps a few hundred thousand dollars each. For example, the 100-ton $7.8-million lot could be divided into 26 $300,000 sub-lots that would then be sampled independently. Assuming the error is unbiased, the 26 independent determinations, each at ± 1%, would yield a much lower financial risk than if the material were treated as a single lot at the same relative error. This assumes that the 26 determinations are unbiased — that is, the chance of error on the high side is equal to the chance of error on the low side of the true content.


Once it has been established that the sampling system is unbiased and that the sub-lot size is appropriate, the next task concerns producing a laboratory sample from a lot of 10,000 lb.

To illustrate this point, consider the following extreme analogy: Assume a 40,000-lb truckload of a mixture of bowling balls, some of which are made of glass and some of black ebonite plastic. We want to determine the percentage of glass in the mixture. We have a large auto-sampling system that meets the specifications above and can produce a random sample — to a point. Obviously, we cannot continue to split the lot down to a laboratory sample-bottle size, since a bowling ball will not fit in a typical sample bottle. The solution is to crush the bowling balls into small particles and then analyze the powdered mixture. But how small a sample should the bowling balls be split into before crushing them? Crushing is expensive and time-consuming, and therefore optimization of that grinding point is essential to efficient sampling.

The point to which the balls should be crushed is greater than one might suppose. For instance, assume that 10% of the original truckload of balls were made of glass and we wish to determine the percentage to the 2% level. Assume as well that the bowling balls each weigh 16 lb. Now let us take a 10% unbiased sample with our hypothetical bowling ball auto-sampler. We now have 4,000 lb in the sample and 36,000 lb in the reject bin. Since 4,000 lb of bowling balls is still a lot of bowling balls, we will take a 10% sample of the 4,000-lb sample to obtain a 400-lb bulk sample. If we now crush the 400 lb to a powder and properly split out a laboratory sample, can we possibly hope to have achieved our desired 2% accuracy?

Remember that in this hypothetical example we know the actual percentage (grade) of the glass balls to be 10%. Therefore, 10% of the balls in the 400-lb sample should be glass. But 10% of 400 lb is 40 lb, and each ball weighs 16 lb. Therefore, the number of glass balls in the sample should be 2.5, which is impossible, since we have not yet ground the balls. Even after discounting random sampling aspects that the entire 400-lb sample had only 25 balls in it, the best we could do would be either 2 of 25 (8%) or 3 of 25 (12%). In either case, we have a 20% relative error.


Ignoring this rule is one of the most common errors associated with dry sampling in general and catalyst sampling in particular. Pitard (2) and Gy (3) have derived formulae for estimating the minimum sample size from the physical parameters of the particulate system that are beyond the scope of this paper. A graphical representation of the theory is shown in (Figure 1). The abscissa is maximum particle size and the ordinate is sample size. For this particular hypothetical catalyst, points to the left of and above the “safe line” have a probable relative standard deviation ≤1%, while those to the right and below are ≥1% relative.

catalyst recovery - part 2

Figure 1. Particle size determines sample size.

It has happened that a 5,000-kg lot of catalyst with obvious color inhomogeneities and a maximum particle size of 1 cm was split all the way down to 1 kg before it was ground to –100 mesh and further split to laboratory size. Laboratory analysis agreement between customer and refiner was assured, but agreement between the sample and the lot in question was not. The refiner ground the bulk catalyst lot that was left down to the size required to achieve a sampling precision of 1–2%. Since the particle size was not appropriate prior to that, there is no assurance that the final material that was ground was, in fact, a representative sample to the degree of precision required.






A typical sampling procedure for a lot of spent catalyst materials sent to a precious-metals refiner begins when incoming catalyst materials are inspected, weighed, assigned tracking numbers and stored prior to sampling. The assignment of tracking numbers is critical; each lot — from the time it enters the refiner’s plant — is segregated from all other materials at the facility to eliminate all possibility of mixing with other lots.

When removed from a reactor, spent catalysts are often contaminated with organic materials, which must be removed to ensure accurate evaluation of the remaining precious metals. The catalysts must first be processed to remove those contaminants and achieve free-flowing properties that will ensure accurate sampling, provide a reproducible settlement weight, and also permit safe and efficient operation of the process that will ultimately recover the remaining precious metals. Contaminant removal will be discussed in more detail in the third article in this series.

After any necessary pretreatment, the lot or sub-lot is screened to separate any oversize, or “tramp,” material and any fines from the bulk of the catalyst. The tramp material is inspected by the refiner and the customer’s representative to confirm that it may be discarded. The fines fraction is sampled separately, since it obviously is very different from the bulk of the catalyst in particle size and possibly in precious metal grade.

The main portion of the catalyst lot, which typically consists of extrudates of the order of 1 cm in length, is then divided by an automatic sampler into one or two bulk samples of a size suitable to achieve the desired precision. This is often in the region of 100 lb. This may require two or more passes through the sampling system, depending on its design.

If two bulk samples were cut from the lot, one of them is used for drawing an approximately 1-lb sample that is known as a loss-on-ignition (LOI) sample. Alternatively (though less preferably), a portion of the single bulk sample is taken for that purpose. The LOI sample is placed in a hermetically sealed container for transfer to the laboratories of the refiner, the customer, and usually an umpire or referee. The LOI samples are not analyzed for precious-metal content, but are ignited to an agreed-upon temperature (usually around 900°C) to remove volatiles and oxidizables. This provides a reproducible basis at the laboratories and a settlement weight for the lot. The LOI concept is discussed in some detail by Helm, et al. (5), and will be treated in more detail in the final article of this series.

The other bulk sample (or the balance if there was only one) is then ground to a particle size suitable for further splitting to approximately 1 kg. For a typical catalyst, 100% –40 mesh is suitable and can be achieved in a ball mill. A rotary divider can then safely split –40-mesh material down to about 1 kg. Before dividing the material into 120-g laboratory samples, it should be ground to approximately 100-mesh, typically in a ring-and-puck mill. Several of these so-called quality samples are packaged and sealed for the customer, the refiner, an umpire, and for reserves.

The owner of the material and the refiner usually assay the quality samples (on an LOI basis) independently for the precious metals of concern. If these assays agree to within predetermined limits, they are simply averaged to arrive at the payable settlement. If they do not agree, then the sealed sample (called the umpire sample) is sent to an independent laboratory (the umpire). The three resulting assays are used (again by an agreed-upon procedure) to determine the settlement. Many times this involves averaging the two closest assays or using the middle assay to determine the final settlement. The reserve samples (usually sealed by both the catalyst owner and the refiner) are held in reserve to cover any possible irregularities during the sampling procedures.

Keep in mind that throughout the sampling procedure, the refiner must adhere to all applicable environmental codes and standards with regard to effluent disposal and atmospheric emissions. Therefore, a typical ideal sampling system is enclosed for dust control and evacuated under a low-volume flow into a dedicated baghouse. In addition to the obvious reasons for preventing atmospheric discharge of toxic and/or noxious fumes, the dust collected during this sampling process is also recovered and sampled separately, with its value — which is often substantial — returned to the catalyst owner.

catalyst recovery - part 2

Figure 2. In an automated sampling system enclosed for dust control, dust generated during the sampling process is captured and sampled later to recover the precious metals.

catalyst recovery - part 2

Figure 3. Electric arc furnaces are utilized in the recovery of precious metals in spent ceramic-supported catalysts.

A sampling system built to these specifications is shown in (Figure 2). On the right are loading facilities for totes, drums and bags; to the left are vibrating screens to remove oversize tramp material and fines, which are treated separately. Three automatic samplers cut dual bulk samples of 1% each. One of these is further split into approximately 1-lb portions and retained in hermetically sealed aluminum cans for LOI determination. The other 1% sample, which may be as large as several hundred pounds, is ground in a batch ball mill to 100% –40-mesh. The–40-mesh material is split by another auto-sampler followed by a 24-pan carousel. This approximately 2-lb portion is ground in a ring-and-puck mill to 100% –100-mesh and split on a small rotary sampler into eight laboratory-sized samples. These approximately 150-g samples are packaged and sealed for the customer, the refiner, the umpire and reserves.

Note that the entire bulk sample of 1% of the batch is ground to –40-mesh before any further splitting occurs. This is quite important if there is any significant inhomogeneity in the sampled material (3). A secondary grinding in the ring-and-puck mill is performed later for the same reasons. The system is enclosed for dust control and evacuated into a baghouse, and the customer is credited with the metal values contained in the dust.


In conjunction with this comprehensive sampling process, accurate and repeatable assaying procedures also play a major role in determining precise values of precious metals remaining in spent catalysts. Once the final samples are obtained, sophisticated instrumentation is used to measure their precise precious-metals content.

Among the equipment and methods used in a well-equipped analytical laboratory to determine the approximate grade of recoverable precious metals is X-ray fluorescence. X-ray fluorescence helps to determine the amount of copper to be added to the mix in order to obtain the desired bullion grade and to provide information on the matrix or non-precious-metal constitution. Other assay procedures employ atomic absorption (AA) and inductively coupled plasma (ICP) emission spectroscopy, and classic volumetric, gravi-metric and fire assay techniques.

What may seem to be a simple analysis of a few elements in an inorganic matrix can be deceptively complex. Remember that the goal is to determine the precious-metal content to an extremely high accuracy at analyte levels of around 0.1% and many times much lower than that. Interferences and matrix effects are very common, and therefore only a laboratory with extensive precious-metals experience should be used.


When sampling is complete, the spent catalyst lot is blended with a mix of flux and a carrier metal such as copper or iron. The proportions in this mix are determined by the calculated concentration of recoverable precious metals in the lot and the desired slag chemistry, which takes into account its electrical conductivity, corrosivity, morphology, melting temperature and other parameters.

This mix, called the charge mix, is then smelted, usually in an electric arc furnace (Figure 3). The furnace produces a two-layer molten pool. One layer is slag, which is the reaction product of the catalyst’s ceramic support and the added fluxes. This layer floats on a pool of collector metal in which the precious metals have dissolved. After the slag is poured off, the remaining layer of metal is poured into molds of a suitable size. The resulting ingots are then sent to a conventional hydrometallurgical or electrorefining facility for separation and production of the market grade precious metals. The slag, which contains trace amounts of precious metals, is usually also processed for further recovery and refining.


The keys to obtaining maximum value — that is, recovering all possible remaining precious metals — from spent catalysts are detail, detail, and more detail: the thoroughness and accuracy of the materials sampling process, with assaying of the sample lots close behind. When you are seeking a precious-metals refiner for your spent catalyst materials (or working with one presently), you must look carefully into these areas, and work closely with the refiner whenever possible. These steps, as well as the refiner’s overall policies with regard to environmental compliance, should provide you with the knowledge and confidence to select (or work with) the proper precious-metals refiner. Ultimately, your relationship with the refiner must be viewed as a partnership, and must be based upon mutual trust and fair treatment.


  1. Jacobsen, R. T., “Catalyst Recovery —Part 1: Recovering Precious Metals from Catalysts — The Basics,” Chem. Eng. Progress, 101 (2), pp. 20–23 (Feb. 2005).
  2. Pitard, F. F., “Sampling and Process Control for Precious Metals,” Francis Pitard Sampling Consultants, LLC, 14800 Tejon St., Broomfield, CO 80020, www.fpscsampling.com (2001).
  3. Gy, P. M., “Sampling of Particulate Materials, Developments in Geomathematics 4,” Elsevier Scientific Publishing Co., Amsterdam, The Netherlands (1982).
  4. Merks, J. W., “Sampling and Weighing of Bulk Solids,” Trans Tech Publications, Clausthal-Zellerfeld, Federal Republic of Germany (1985).
  5. Helm, P. S., et al., “Loss on Ignition Sampling and Determination,” Precious Metals 1999, Proceedings of the Twenty-Third International Precious Metals Conference, Acapulco, Mexico, International Precious Metals Institute, Pensacola, FL, www.ipmi.org, pp. 351–358 (1999).

ROBERT T. JACOBSEN is vice president of Sabin Metal Corp., (East Hampton, NY; Phone: (585) 538-2194; Fax: (585) 538-2593; E-mail: rtj@sabinmetal.com).

He has an extensive background in the precious metals industry, starting in the mid-1960s when he served at Sp rague Electric Co. in research, development, engineering and production of precious metals, ceramics and electronic components. He joined Sabin Metal in 1980 and has served in a variety of technical and management positions, including general manager and corporate technical director of the company’s Scottsville (Rochester), NY, refining facility. Over the years, he has been involved in development and production of pyrometallurgical and hydrometallurgical activities for recovering maximum values from recyclable precious metals. He taught chemistryat a number of American institutions, including Clarkson and Cornell Universities in New York, and North Adams state and Williams Colleges in Massachusetts. He is on the board of the International Precious Metals Institute (IPMI) and serves on that organization’s Environmental and RegulatoryAffairs Committee. He has served as chairman of the Precious Metals Committee of the American Society for Testing and Materials (ASTM). He is a member of AIChE, the American Chemical Society, Sigma Xi and the New York Academy of Sciences. He holds a BA in chemistry from the Univ. of Rochester, a Master’s degree in education from Columbia Univ., and a PhD in chemistry from Clarkson Univ.