Many fine papers have been delivered and written about chocolate temper by authors far more knowledgeable than this writer as regards to the chemistry and processes involved in producing the crystalline structure required for good temper.
This paper deals with obtaining a repeatable measure of temper, its history and its future. Before addressing the measurement process, it may be advantageous to define the benefits of good temper and the processes used to obtain it.
Seguine defines good chocolate temper as "The largest number of the smallest possible crystals of the right crystalline form (polymorphic form)." ¹
The largest number of crystals is desirable, since the amount of cocoa butter set as chocolate is a function of the seed crystals. This is important because crystallization continues after the chocolate leaves the cooling tunnel.
The smallest possible crystals are desirable for a number of reasons. Smaller crystals do not affect chocolate's flow properties as much as larger crystals. Flow properties in turn affect the coating or molding of a product and control its weight. Seguine also points out that a larger quantity of smaller crystals reduces the mean free path between crystals, which reduces the chocolate set time. Finally, the finer crystalline structure produces a product with high gloss, and hence consumer appeal.
Much has been written about proper crystalline structure. Of the four crystal forms commonly cited-Gamma, Alpha, Beta Prime, and Beta (there may be more)- only Beta exhibits the characteristics desired. Gamma and its derivative, Alpha, crystallize and melt at temperatures far too low to be of practical use to the confectioner. These crystal forms are both unstable. The contraction (important in demoulding) of the Alpha is a little more than half that of the desirable Beta form. The Beta prime form is somewhat stable, but its low melting point makes it unuseable for consumer products. The Beta form crystals may not be ideal, but they are the best choice available and their formation is imperative.
A bibliography of excellent articles describing the chemistry, processes, handling and pitfalls encountered in tempering chocolate is provided at the end of this paper. A variety of tempering methods, including batch, drip feed and automatic machine tempering, have been and are currently being used. For our purposes, these methods can be considered very similar. The processes are virtually identical, with only the mechanical implementation changing.
All methods begin by heating the chocolate supply to a temperature that melts all existing crystals. If the crystal nuclei are not removed, large crystals will form, resulting in fewer of the desirable smaller crystals. Insufficient temperature at this step in the process can have a longer term effect, since the crystals remaining will gradually increase in size over time causing an over temper condition. Over tempered chocolate thickens and is difficult to transport and deposit. On the other hand, well tempered chocolate remains at a relatively constant viscosity
During the next stage, the chocolate is cooled to promote nucleation and crystal growth. The chocolate is then gently reheated slightly to melt unwanted Alpha and Beta Prime form crystals and to transport the chocolate to the enrober or to mold it. Low-speed agitation is used to mix the crystals, and scraping is used to remove undesirable crystal forms which may have formed on the cooler container walls. Ideally tempered chocolate will contain a large number of small Beta form crystals that promote further growth of a like kind. Seguine indicates that even with good temper, only 70 to 85 percent of the cocoa butter is crystallized when the product leaves the cooling tunnel. Excess chocolate is heated and returned to the detempering tank.
Temper meters do not measure the quality of temper. Temper meters quantify the crystal growth process for an identical product formulation at the same location on a production line. The determination of good temper must be made by evaluating the product quality discussed earlier.
The function of a temper meter is to repeatedly measure some consistent function of the heat of crystallization. Whether the temper meter measures the heat of crystallization exactly is irrelevant. As long as the measured function shows the effects of process changes that cause the product to deviate from optimum, the temper meter is doing its job.
The same can be said for the assessment curves commonly used to evaluate temper. With all the different chocolate formulations that are used, the desired shape of the curves to provide good temper may be over, under, or good using the assessment curves that are discussed later in some detail. A number of confectioners now believe initial slope is a better indicator of temper.
Since measuring chocolate temper consists of nothing more than measuring the heat of crystallization of the chocolate as it cools and solidifies, it should be simple. Implementing this measure in an accurate and repeatable manner is not so simple.
The first method of determining temper relied on "OL' Frank." "Ol' Frank" may have had some other name, but his job was to ensure good temper. He could dip his finger into the chocolate, hold it to his lips and tell if the chocolate was over, under or good temper. "Ol' Frank" learned to make this determination by virtue of his many years of working with "Ol' Joe," who taught him the trade before retiring. There are still a number of "Ol' Franks and Joes" out there. During the last year, our people met one at a well-known Chicago land confectionery. He would call "over," "under," or "good" temper, and when the chocolate was checked using a temper meter he was always right.
When the "Ol' Franks" and "Ol' Joes" inevitably left their companies, a method of temper measurement was required that could be performed by a variety of people.
To the best of this writer's knowledge, the next temper measurement method employed a sample holder with a heat exchanger, a temperature measurement probe, an ice bath, and a temperature recorder as shown in Figure 1. Many of these systems are still in use today. Their operation is quite simple. A cup-like opening at one end of a long metallic (copper) tube is filled with a chocolate sample. The other end of the tube is inserted in a mixture of ice and water held in an insulated container. A temperature probe (thermocouple in early versions) is then inserted into the chocolate. Since the chocolate sample is above 85°F and the sample holder is attempting to reach the temperature of the ice bath (a questionable 32°F), the chocolate gives up heat to the holder and is cooled to solidification.
The temperature probe monitors this heat loss, but more importantly it records the temperature fluctuations caused by the formation of crystals. When this combination of heat loss and heat gain is recorded on a strip chart over time, the familiar Greer Assessment Curves (shown in Figure 2) are generated.
It is unknown, at least to this writer, whether Greer deliberately chose a sample volume, a heat conducting mechanism and a cold sink temperature that would generate these curves, or whether it was pure serendipity. We will further investigate this later in this paper. Whatever the cause, the curves are valuable in providing a rapid visual evaluation of temper and have become a standard tool of the industry.
In an effort to better quantify these curves, a procedure was used where a scientist or technician constructed the initial temperature decrease slope and the slope during the most visible temperature rise due to heat of crystallization. A single numerical value for this performance was obtained by subtracting an arbitrary temperature value from the temperature value at the intersection of the two slopes as can be seen in Figure 2. The resultant number was given the name Greer Temper Units or GTU.
The introduction of the microprocessor in the middle-to-late 1970s revolutionized all industries. The power of the computer became available in a small-sized package and, even more importantly, at a cost that allowed it to perform an almost limitless array of functions.
A very progressive major confectionery manufacturer contacted TRICOR Systems in 1977 because of the experience in the fledgling microprocessor application arena that TRICOR Systems possessed. The confectioner wanted to improve the quantification of temper measurement data by allowing the computer to determine the aforementioned slopes and GTU computation. Today that task is trivial, but in 1977, without the aid of the peripheral tools, higher-level languages, etc., it was not easy.
Data collecting methods and computational software were developed along with an operator display presentation. The confectioner developed the actual cooling mechanism consisting of an ice bath with magnetic stirring, a thermocouple probe and sample holders of the same design as previously used.
The real impetus for today's modern temper meters (Figure 4) came after the first computational unit was delivered. The confectioner engaged in a test program to determine if the developed software was accurately calculating the results from the data provided. Computations from multiple tests of the same batches of chocolate did not repeat, and the system delivered was suspect. After some period of agonizing, TRICOR was convinced that the software was correct. Subsequent testing using the original manual computation techniques revealed that the deviations in results had been masked by a human's ability to estimate the slopes from the data. The computer's accuracy was bringing the measurement deviations into focus!
In 1980, after being convinced the system was providing accurate data, the same confectioner wanted to purchase more computational units. During the course of investigating the computational accuracy, a number of contributors to measurement deviations were noted. These were reported at the PMCA Symposium in January 1986. The purchase order for more computational units was declined, with TRICOR offering instead to build a more repeatable unit if an order for multiple units were placed. The confectioner placed an order for three units, and development began in earnest.
Errors in cooling source temperature caused by differences in ice bath temperature and gradients in the Dewar were easily eliminated by using a precisely controlled thermoelectric cooler and a large thermal mass. Errors caused by probe temperature and position in the sample were also eliminated by fixing the position of a heated probe in the sample.
The major problem encountered was finding a combination of sample volume and shape, sample cup heat transfer characteristics and a cooler temperature that would provide the familiar assessment curves in a 5- minute test time.
The problem was compounded by the lack of an automatic tempering machine that could provide chocolate for testing. Hand tempering by someone other than "Ol' Frank" was not necessarily consistent. Each time a variable (volume, sample cup size, or temperature) was changed, it was extremely difficult to determine if the result was caused by the variation or the temper state of the chocolate.
It is interesting to examine the mechanism by which the assessment curves are generated. The same crystal growth pattern is found in the sample cup as is described by Seguine for chocolate in a cooling tunnel. The flow of heat is outward from the center of the chocolate to the cool walls of the sample cup, just as it is from the center to the cool air of the tunnel. A temperature gradient across the chocolate from the center to the edge is established. As cooling continues, cocoa butter crystals form on the periphery. These crystals yield their heat of crystallization, which slows the cooling process.
It should be noted that a material with the same thermal characteristics as that of liquid chocolate under identical conditions would cool to a specified temperature in a shorter time than chocolate. This is partially due to the lower thermal conductivity of the solid (crystallized) chocolate layers being formed. As these layers form from the outer boundary toward the center, the rate of cooling of the liquid chocolate remaining is reduced by both lower thermal conductivity of the solid chocolate and the reduced difference in temperature between the liquid center and the cooling source (tunnel, ice, thermoelectric). At some point, the liquid center reaches a critical volume with respect to its ability to transfer heat, and the heating effects of the crystal formation are more prominent.
Depending upon the extent of crystal growth at this time, the heat generated can cause a temperature rise until the remaining cocoa butter crystallizes (under temper); it can cause the curve to exhibit a constant temperature for a period of time (good temper); or it can cause a rate change of the temperature decline (over temper) as was shown in Figure 2. No curve disturbance indicates that there is no crystallization taking place, hence no temper.
The words under, over and good used to describe chocolate temper actually refer to the state of crystallization prior to the inflection. Since there is a temperature rise during the final crystallization of the liquid center, under temper indicates that an insufficient number of crystals were formed prior to this event. Likewise, the lack of sufficient temperature rise indicates too many crystals were formed (over temper) before the center crystallized. Although zero slope or, more correctly, a temperature rise equal to heat transfer, is generally accepted as good temper, there is no scientific basis for this judgment. The heat of crystallization has only a secondary effect on the total heat transfer process. The primary heat transfer is a function of cooler temperature, volume and shape of the sample and thermal conductivity of the sample cup/holder.
To this point, we have assumed crystallization process caused by the temperature-sensing probe. This is not true. The original ice bath units used a thermocouple mounted in a relatively large diameter stainless steel housing. No care was taken to ensure that the thermocouple and housing were maintained at constant temperature prior to the measurement to increase its repeatability.
The heat flow from the center of the sample to the cooler periphery was described earlier. But now a probe, presumably at room temperature (72°F), is introduced into the chocolate center that is initially above 85°F. The probe acts as a cold sink and complicates the crystallization process. Heat transfer is both to the periphery and to the probe until the chocolate and the probe temperatures are equal. The larger the mass of the probe and the more the thermal coupling to the world outside, the greater the effect of the probe on the measurement.
If the probe were at room temperature with good thermal coupling to the outside, it would act as a cold sink until both chocolate and probe were at the same temperature and then as a heat source when the cooler reduced the chocolate temperature below ambient.
Since the heat of crystallization temperature rise is being measured in the locale of this thermodynamic disturbance, it is desirable to keep the probe's thermal mass as small as possible. Use of a probe made of a low thermal conductive material thermally decoupled from outside influences minimizes the disturbances. This was accomplished in automatic temper meters using a fiberglass probe housing and a small thermistor for temperature measurement. Additionally, great pains were taken to decouple the probe from the cooling mass. The probe was heated to a constant temperature above ambient to remove variations caused by ambient temperature fluctuations.
On the surface it would appear that the disturbance caused by the probe presents a major problem. However, earlier we stated temper measurement need not be exact. We also said the temper meter must measure the heat of crystallization or a function thereof repeatability and with sufficient sensitivity to detect a process variation. Therefore, a change of an absolute value does not mean there is a measurement error as long as the change is consistent.
The problems involved with inline temper measurement are formidable. Five steps are required:
The steps outlined above appear on the surface to be relatively straightforward. Their implementation is not as simple.
Selection of a sample volume is not straightforward. The size of the volume must take into consideration a number of factors. Isolating the sample should not alter the state of the temper of the chocolate to be measured. The sample volume defines the sample chamber, and the sample chamber must not present a thermal load that cannot be rapidly changed to the specified temperature. The sample volume must be selected so that its measured temper curves approximate curves measured with an off-line temper meter.
Thermally isolating the samples from heat sources also provides a formidable problem. An in-line system provides a sample by using input and output valves with the chocolate sample contained between them. The warm chocolate supply and discharge lie beyond the valves, providing a heat source via the valves and thermal insulators.
The temperature-sensing probe must be metal due to safety considerations (metal detection, etc.). Metals possess high thermal conductivity, so the probe easily stabilizes to the temperature of the flowing chocolate prior to a measurement. The probe acts as a very minor heat sink during this condition, since the sample chamber and thermal mass through which the probe must pass closely approximate the chocolate temperature. The situation changes dramatically when the sample and chamber thermal mass are rapidly cooled prior to measurement. Even though the temperature probe is insulated from the cold surfaces that it passes through, the large temperature differential between the chocolate and these surfaces causes the probe to act as a heat sink for the chocolate. The chocolate cooling is now affected by two heat sinks. The cold chamber walls are responsible for the primary heat flow from the chocolate, but the probe also provides a lesser path creating the condition discussed earlier. The geometry of the cooling system has been disturbed.
The stable temperature of the heat sink thermal mass to which the heat from the chocolate flows must be rapidly established. If the thermal mass transition from near chocolate temperature to its stable temperature (approximately 45°F) is slow, the chocolate will crystallize during this interval. This situation would be acceptable if thermal mass temperature transitions from higher to lower values were always constant in time. This additional potential variable with its effect on repeatability can be eliminated if the mass can be cooled rapidly. Fortunately, the metal thermal mass temperature can be changed rapidly using a boost cooler. After the target temperature has been reached, a thermoelectric cooler similar to those in the off-line temper meters can be used to ensure precise temperature control.
Once the sample has been thermally isolated (as well as possible) from the chocolate source and discharge, and the thermal cooling mass is at its design temperature, the measurement of heat of crystallization is very similar to that of the off-line temper meters.
When the measurement is complete, a solid "plug" of chocolate that must be removed is contained in the sample volume. Hot water can be used to remelt the chocolate or, if time is not important, reversing the voltage to the thermoelectric cooler causes the cooler to function as a heater and melt the chocolate. In either case, the probe in the chocolate measures temperature, providing an indication as to when the input and output valves can be opened.
With the valves open, the chocolate is again allowed to flow until temperatures are again stable, at which time the valves close and the process is repeated.
Although implementation of a number of the steps necessary to provide in-line temper measurement is formidable, perhaps the most difficult task is developing a combination of volumes and temperatures that provide the familiar temper curves produced by automatic temper meters and their predecessors. During the course of this paper, we have attempted to show that the assessment curves are a function of the measurement parameters and not intrinsic. The thermodynamics, however, are really not important. If in-line temper measurement is to gain the confidence and acceptance of the confectionery industry, the familiar curves must be present.
The original development of the automatic temper meter taught the user to compare instruments. At some point, in-line measurement will be compared to off-line measurement. It is difficult to predict whether the measurements will ever be identical due to the different thermal and time paths.but they better be close!
Temper measurement started with "Ol' Joe" and has been automated to the point that it can be used as a feedback element to control temper on a continuous basis. It is essential, however, to remember the words of Timson:
"You have to set up operating standards for your chocolate, for your application, and you must set operating standards for each product that you make. I am a believer in temper meters and would suggest that you consider them."
"The curve created on a temper meter should be personalized to each product and application in your kitchen or factory. Once you have established how the very best temper can be obtained, a temper meter will enable you to check the condition of your chocolate prior to molding or enrobing."²
Phillip Allen - TRICOR Systems Inc.