Anal. Chem. 1998, 70, 5029-5036
Design and Operation of a Capsule-Based Microwave Digestion System Guy Le´ge`re† and Eric D. Salin*
Department of Chemistry, McGill University, 801 Sherbrooke Street West, Montreal, P.Q., Canada, H3A 2K6
A high-pressure microwave digestion system has been designed based on the concept of using large-bore tubes as digestion vessels and capsules as a vehicle for sample introduction. Many of the design aspects are dictated by the use of a relatively large (8.4 mm o.d., 25 mm long) capsule, which in turn dictates the inner tube dimensions. A variety of materials were studied for use as the tubing material. PFA was selected as the best material for a demonstration arrangement. The cycle of operation involves insertion of the capsule by a flexible rod followed by addition of digestion reagent and then a heating/ cooling/venting cycle for removal of gases. When the digestion is completed, the system removes the liquid with the same flexible rod and then cleans itself. The system is highly automated with computer-controlled venting, cooling, and reagent addition. Data indicate that performance is similar to that expected of a conventional microwave oven operated at the same temperatures.
The conversion of solid samples to a liquid format is traditionally called digestion. The same term is also often used for the conversion of a liquid sample from one form to another (e.g., organic to aqueous). The conversion process is usually enhanced and sometimes only possible with the addition of heat to the system. The reagents most often used, acids, often boil off at atmospheric pressure below the temperature levels that would be most useful. The boiling point limitation has led to the use of pressured systems, which allow higher temperatures to be reached. These pressurized systems are usually called bombs and are very sturdy. Bombs are most often sealed with screw mechanisms, which do not lend themselves well to automated processing. Furthermore, the shape of the vessels is not well suited to automated liquid removal, rinsing, and cleaning. For many years, samples were heated by radiative means ranging from hot plates through furnaces. This method was technologically simple yet slow since it required that the environment (e.g., furnace) be heated, followed by the vessel (e.g., bomb), followed by the sample. Microwaves, however, can pass through suitable materials and transfer the majority of their energy directly into the sample. The advantages are 2-fold. Less energy is used, and the sample is heated more quickly. The desire for higher throughputs and automation led to some experiments with † Present address: Sciex Inc., 71 Four Valley Drive, Concord, ON, Canada L4K 4V8.
10.1021/ac9710689 CCC: $15.00 Published on Web 11/04/1998
© 1998 American Chemical Society
pressurized tube based systems for the digestion of slurried samples.1,2 While the system did demonstrate the viability of the tube concept, it was felt that this was an intermediate step in the development process due to certain limitations in the design. Foremost of the limitations was the need for valves in the traditional flow injection sense. Valves for this type of work must be able to withstand a corrosive environment at elevated temperatures and pressures. Most of these materials are relatively soft. The passage of slurried sample through the valves results in sample entrainment with a resulting degradation of the valves and memory effects. A further inconvenience of an arrangement with narrow tubing, which is much more suited for higher pressures, results from the generation of bubbles from the decomposition process. In a narrow tube, these bubbles can cause local inhomogenity in the heating process and digestion reagent concentrations. Of more serious concern, one cannot do multiple-step chemistries in the tubing because the bubbles act as barriers to the addition of new or different reagents in subsequent steps. Finally, long lengths of narrow-bore tubing cannot be physically cleaned, although they can be chemically treated. Our early work2 demonstrated that materials can be deposited on the walls of the tube. Chemical treatments were ineffective in removing these coatings. Although one could wonder whether the deposits were a risk from a contamination point of view, they certainly present a threat of blockage with prolonged use. Finally, one must consider the desirability of slurries as a sample format. Most samples do not present themselves in a slurry format, but as solids. They must then be treated, often by grinding, to a size and distribution that renders them suitable for handling in a flow injection type of apparatus. Our experience has demonstrated that slurries can be very different, ranging from geologicals, which fall in solution like the proverbial stone, through botanicals, which creep up walls. Some slurries may require the addition of surfactant, and the use of ultrasonification has become common. These suggest that sample handling as well as preparation can be cumbersome for the preparation of slurries. In the system discussed below, we have completely bypassed the slurry-generation step. This has led to a design that is not only different from those presently in use but appears to offer significant advantages for automated, low-cost sample handling. (1) Liu, B.; Salin, E. D. Flow Injection Microwave Solids Sample Decomposition for ICP-AES. Third Chemical Congress of North America, Toronto, 1988. (2) Karanassios, V.; Li, F. H.; Liu, B.; Salin, E. D. J. Anal. At. Spectrom. 1991, 6 (6), 457-463.
Analytical Chemistry, Vol. 70, No. 23, December 1, 1998 5029
Figure 1. Side view of the microwave oven and U tube.
DESIGN AND CONSTRUCTION The Capsule. The design of the instrument is, to a large extent, dictated by the use of capsules as vehicles for sample introduction. In turn, the use of capsules was dictated by the weakness of a conventional tube-slurry approach. The primary advantage of capsules is that the sample is introduced into the system) without any potential for entrainment in the first valve. The capsule can also be filled in the field, reducing samplehandling steps. For identification, the capsule can be color coded and bar coded. Internal standards can be impregnated into the capsule body, thereby enhancing accuracy and precision. Capsulehandling technology has been well developed by the pharmaceutical industry as has the technique of manufacturing capsules with a variety of materials. We found in our initial investigations that the traditional capsule manufacturing materials, designed for mammal digestion tracts, were replete with measurable concentrations of a large number of elements. For this reason, we elected to use polyacrylamide gel as a capsule material. There were no available procedures for the construction of capsules with this material so a technique was developed which is described in considerable detail by Le´ge`re.3 In summary, the technique involves placing a measured amount of the gel onto matching pin molds followed by a careful drying procedure. The resulting final capsule has an 8.4-mm outer diameter and is 25 mm long and clear. Thin capsules may be slightly brittle, but most capsules are quite robust. Microwave Oven. The microwave oven used for all experiments was a domestic oven 1 ft3 700 W Eaton model RE-778TC. The back of the oven was replaced with a heavy-gauge aluminum plate to hold the flange valves and digestion tube. The replacement of the back and the addition of controls for computer control could have compromised the oven safety interlocks or produced microwave leakage, so it was arranged that the oven was turned on manually and then controlled by computer. Microwave leakage was found to be below measurable levels. The magnetron efficiency remained constant throughout testing. Digestion Tube Design. The use of a capsule dictated that the minimum diameter roughly match or exceed the diameter of the capsule. The final diameter of the tube was 3/8 in. (9.53 mm) with an outside diameter of a different configuration, which is illustrated in Figure 1. Rather than a coil, the tube is arranged in (3) Le´ge`re, G. Capsule-Based Microwave Digestion. Ph.D. Thesis, McGill University, 1995.
5030 Analytical Chemistry, Vol. 70, No. 23, December 1, 1998
Figure 2. Capsule-based microwave digestion system schematic.
a tilted “U” arrangement. The tube is held in place by a “flange valve” (Figure 2) which is described by Le´ge`re.3 The valve is very strong and opens to provide unobstructed access to the tube as well as convenient cleaning of all mating surfaces. The only other common type of valve that opens to provide unobstructed access is the ball configuration, which will entrain material when the ball is rotated. The flange valve is also interfaced to a vent/feed system as illustrated in Figure 3, a more complete schematic of the system. The vent/feed system uses smaller tubing which is much stronger than that used in main U tube. The vent/feed system connects directly to a flow injection type valve (vent valve) which is operated by a pneumatic actuator. The pneumatic actuator allows the vent system to open and close quite rapidly. This allows the system to vent small amounts of gas very rapidly. The valve system is also linked to a loading valve which is used to load reagent into the U tube through the flange valve. It is important to note that the vent valve is connected to both flange valves so that gases are vented simultaneously from both sides. If this arrangement is not used during venting, the solution will boil out through the single open vent. The large tube is especially interesting because it does allow the system to vent, unlike narrowtube systems. This allows organic materials or other gas-releasing compounds to be digested without risk of overpressurizing the system. Automated depressurization also allows the system to add more reagent, opening the door for multistep chemistries. To minimize the loss of volatile species, the digestion tube (U tube) is normally cooled before venting. Tap water is used for cooling and is administered under computer control. The computer monitors the exhaust stream of the cooling water to determine when cooling is complete. The digestion tube is constructed of 1/2-in.-o.d., 3/8-in.-i.d. Teflon PFA 350 (DuPont). This fluoropolymer is inert to most reagents and transparent to microwave energy and visible light. It has a melting point of 306 °C and a burst pressure of 200 psi at 200 °C. Other materials were considered; however, both Pyrex and quartz
Figure 3. Flange valve concept and actual design.
would not respond well to being constrained during thermal expansion. The PFA proved to be more than adequate for demonstration purposes, although the temperature/pressure limits preclude the use of temperatures that would be necessary for difficult organic materials. Experiments were conducted with a woven glass sheath which allowed the PFA tubing to operate up to3 400 psi; however, the sheath slowed the response of the temperature monitoring and cooling system, and so its use was dropped for the moment despite its effectiveness. A variety of configurations are possible for the digestion tube; however, the inclined U has several important features. Solution naturally gravitates to the bottom of the U. When boiling, the solution is kept relatively far from the valves. Any splattering that occurs does not deposit material on the valve surface, although it may deposit material somewhat above the solution surface. Rising vapor reaches the cooler valve surface, condenses, and flows down the walls, washing material back into the solution. This is very convenient for obtaining a thorough digestion. Physical contact between a scrubbing tool and the tube walls seems to be necessary to avoid the buildup of organic digestion byproducts.2 While not as easy to work with as a straight tube, a U shape lends itself to convenient cleaning. The cleaning device, a “squeegee”, is actually a multipurpose tool and will be discussed below. Squeegee. The cleaning device is called the squeegee after the soft rubber blades used to wash windows. The squeegee is a plug of room-temperature vulcanizing (RTV 700) silicon formed
Figure 4. Squeegee cleaning system configuration.
at the end of a flexible 1/8-in. PTFE plastic rod (Figure 4). The plug is soft and can conform to the tube contours as it is pushed through the tube. To avoid degradation of the RTV by acids, a thin Teflon foil (0.005-in. thickness, 2.0-in. diameter) is placed over the plug. The Teflon foils are considered disposable so as to minimize any potential for memory effects. The foil-encased plug performs two functions. The first is that it picks up any particulates that may have become lodged on the walls. Since the seal formed by the plug and foil is gastight, the passage of the squeegee also forces all liquid out of the system without the use of pumps and any risk of dilution. Sensors. With the primary exception of the flange valves, most of the system is automated as shown in Figure 3, which illustrates the sensor arrangement as well as giving a general Analytical Chemistry, Vol. 70, No. 23, December 1, 1998
5031
schematic of the system. As reported above, the valve system can vent very rapidly using a pneumatically operated valve. When the pressure/temperature limits of the material are approached, the valve system allows small amounts of gas to be released. Small pressure releases are important, because they eliminate the possibility of violent boiling of the solution, which would, in turn, cause particles and analyte solution to be transported far up the digestion tube and even into the vent tubing system. After considerable experimentation with other designs, a flush stainless steel front diaphragm in-line pressure-monitoring system was developed. The flush face of the sensor is covered with a thin PTFE disk.3 The PTFE disk protects the sensor from hot acid vapors and has almost zero volume displacement. The system was calibrated from ambient to 200 psi using a gas supply with a readout gauge. The sensor is claimed to be accurate to 1 psi, which was more than sufficient for our needs. Temperature monitoring is essential in the central digestion tube and useful in several other places in the system. To prevent overheating, the magnetron is factory-equipped with a thermal device that cuts power when a preset temperature is reached. This is a fail-safe mechanism to minimize the possibility of selfdestruction of the system. For our work, a stainless steel jacketed E-type thermocouple was placed underneath the thermal cutout switch. This thermocouple was calibrated with ice and boiling water. The temperature of the magnetron was a reflection of the load in the microwave oven. Small loads, which reflect back more energy, produced higher operating temperatures than large loads. The sensor also allowed the system to avoid an automatic magnetron thermal shut-off, which would have been inconvenient. In fact, as long as there was a load in the cavity, magnetron thermal shut-off conditions never occurred. Cooling water flows through a 1/8-in. tube wrapped around the digestion tube. The inner diameter of the tube is 1/16 in. so a conventional thermocouple could not be used. Instead, a thermocouple was imbedded in a small hypodermic needled and inserted into a “T” connection of the 1/8-in. tubing. The low mass of the thermocouple gives fast response and is not affected by microwave radiation. Monitoring the water temperature allows the system to “know” when the cooling process has been completed. The digestion tube temperature was the most difficult to monitor. It might initially seem that one could determine the temperature by monitoring the pressure; however, the system is not in thermal equilibrium. One must also keep in mind that the integrity of the digestion tube must be maintained to preserve its strength. Our first approach was to use a wide variety of thermocouple arrangements. Shielding turned out to be a major problem as the electromagnetic fields would often cause arcing between conducting solutions and the grounded shielding cable. Large amounts of shielding would produce zones in the tube that were not as exposed to microwaves as others. A further problem was the time lag involved with the transfer of heat from the inside to the outside of the PTFE tube. Finally, the thermocouples themselves had a tendency to heat in the microwave field. Eventually, a better solution was found, infrared viewing. An Omega (OS36-E-240-GMP) thermopile (called an infrared thermocouple, IR/TC) was mounted with the viewing window flush with the floor of the microwave cavity, approximately 0.3 5032 Analytical Chemistry, Vol. 70, No. 23, December 1, 1998
Figure 5. Capsule insertion step. The capsule is pushed to the lowest point in the U tube.
cm from the bottom of the digestion tube. The tube was viewed through a small hole in the floor. The digestion tube was held in a PTFE bracket on the floor to minimize expansion and movement during heating. The thermopile was calibrated by heating 10 mL of nitric acid inside the digestion tube and monitoring the temperature of the nitric acid directly up to 200 °C by using a J-type thermocouple fitted through a pressure adapter.3 System Control. To have a control system that would have a high degree of flexibility yet isolate the user from the mechanics of the program, a high-level language called MICRO2 was developed in Turbo Pascal. It was felt that this language would provide a number of control constructs that were not available or convenient in commercial packages such as Labview. MICRO2 uses an ASCII text instruction file as input (i.e., its program). The instructions read very much like English statements (e.g., “R 1 ON” means Relay 1 on). In addition to providing a readable program format, MICRO2 provides a real-time split-screen display while operating. One portion of the display is the instruction presently being executed and the nearest five lines of code above and below this line. The other section consists of data. A complete description of MICRO2 is available,3 and copies of the source code may be requested from G.L. Several other projects in our laboratory are now operated under MICRO2 due to its flexibility and ease of use. Operating Cycle. The operating cycle of the system is illustrated in Figures 5-9. Figure 5 indicates that the capsule has been pushed by the squeegee to the bottom of the digestion tube. The squeegee is then removed and the flange valves are closed by hand. Reagent is then added through the loading valve and vent valve, and the system appears as illustrated in Figure 6 with 10 mL of solution. The vent valve is then closed by computer, completely sealing the system. At this point, the computer will initiate the digestion cycle which starts with the application of microwave energy as illustrated in Figure 7. The solution will boil at a temperature dependent on the pressure, which rises relatively rapidly as will be demonstrated. The amount of gas released is highly dependent on the reagents used and the material digested. Organic-based materials obviously will release large amounts of carbon dioxide if the conditions are strong enough to break the bonds in that particular compound. After each application of energy, a cool-and-vent cycle (Figure 8) is executed to release the decomposition gases and, if the
Figure 6. Reagent addition step. The flange valves have been closed and solution is pumped through the smaller tubes.
Figure 7. Heating cycle. All valves are closed and the system can be run at higher pressures and temperatures.
Figure 8. Cool and vent cycle. Water is flowing through the cooling tubes. Pressure is released in small increments through the small tubes while under computer control.
digestion is complete, reduce the temperature of the liquid for removal with the squeegee (Figure 9). A typical digestion cycle is presented in Figure 10 for a sucrose sample, which will release large amounts of carbon dioxide. The bottom trace indicates when the microwave oven is on. The next trace up is the pressure, which starts close to zero. The axis (from 0 to 200) serves as both a temperature and a pressure scale. The units for temperature are degrees centigrade and the pressure units are psi. Note that when the pressure reaches the preset pressure threshold of 100 psi, the magnetron is shut off and the cooling cycle is initiated (Figure 7). In this case, pressure continues to rise to 155 psi over the next 15 s. Note that the tube temperature has risen slightly during this period, partially due to the action of the nitric acid (which caused an elevated starting temperature) and partially due to the microwave energy. Cooling water temperature, the lowest trace, which also starts just above 0, rises and then drops off during the cooling cycle. When the solution temperature reaches 75 °C, the system is vented. Note
Figure 9. Solution removal step. The squeegee is pushed through the system removing the liquid and any remaining solids.
that this cycle can be repeated several times. In the case of this type of material, which releases large amounts of gas, the first pressure maximum was set lower for the first vent cycle to minimize the possibility of rupture. In the second cycle, the pressure trigger was set to 170 psi. The burst pressure is a function of both pressure and temperature. Experiments have demonstrated that 215 psi and 180 °C will cause a rupture with this material. After the second venting, the digestion temperature was maintained for 5 min without exceeding the maximum pressure trigger of 170 psi. If the system were sufficiently robust, venting would not be necessary; however, the pressure requirement could exceed 1000 psi even for samples as small as 200 mg. It is also important to realize that it is possible to add additional or new reagents when the system is depressurized. This allows multistage chemistries to be used, e.g., nitric acid followed by perchloric acid or hydrogen peroxide; however, considerable care must be taken with reagents of this type. After as many cooling and venting cycles as necessary, the solution can be removed by the squeegee as illustrated in Figure 9. Note that the snug fit of the squeegee head appears to remove all undigested material with the single pass that removes the solution. Rinse solutions can then be run through the system, and the system can also be cleaned by running a “blank” acid solution. OPERATION AND CHARACTERIZATION OF THE SYSTEM Experimental Details. For demonstration purposes, all of the following digestions were performed at 180 °C for 5 min using 10 mL of concentrated nitric acid. The acid was added volumetrically with a calibrated dispenser. The capsule, sample + capsule, digestate bottle, digestate bottle + digestate, and 1 mL of the digestate were all weighed. The weights were used to calculate the total recovered volume of digestate removed from the digestion tube with the squeegee. The recovered volume, sample weight, and concentrations in the digestate were then used to correct the sample concentrations. All digestates and reagent blanks were diluted 10 times in 15% HNO3 for analysis. Two multielement standards (QC-19, QC-7 SCP Science, LaSalle) were used and diluted in 15% HNO3. A phosphorus, yttrium, and carbon standard was made in-house using potassium phosphate, yttrium nitrate, and mannitol diluted in 15% HNO3. The majority of the analyses was performed on a PE/Sciex 5000 ICPMS (Canadian Geological Survey, Ottawa, ON, Canada). Analytical Chemistry, Vol. 70, No. 23, December 1, 1998
5033
Figure 10. Typical digestion cycle for a gas releasing material, sucrose in this case. See text for details.
Other analyses were carried out on a Sciex 250 ICPMS, JarrellAsh 25 ICP-AES (scanning), and ARL Quantometric Analyzer ICPAES (direct reading). ICPMS was used for all trace element determinations while ICP-AES was used for major elements that were too concentrated to run by ICPMS. All equipment, all procedures, and some experimental results are discussed in greater detail in ref 3. Capsule Material. (a) Commercial Capsules. It is essential that the capsule construction material have minimal levels of any of the analytes. Due to their availability in a wide variety of sizes and uniformity of construction, commercial capsules were analyzed to determine their suitability for this system. The analytical results are available in ref 3. The capsules were selected to represent the different commonly available types; a red capsule contained an opaque pigment, a black capsule used an opaque dye, and the Vegecap capsule was made of cellulose. The capsules (made for human consumption) are devoid of all toxic trace elements determined, with the exception of lead, which is just detectable. Na, Mg, Al, Ca, P, Fe, and, Ni, however, were present in all three capsules well above the detection limit. Vegecap, the cleanest of all three capsules, is still not clean enough for use in trace analysis. (b) Polyacrylamide Capsules. To determine the levels of contamination that one might expect from polyacrylamide capsules, six capsules were run as samples. They revealed the presence of trace quantities of Fe, Ca, Na, Al, and Mg. As discussed earlier, the digestion temperatures never reach the 230 °C needed to break down the polyarcylamide and remove it from solution as CO2. The analysis by density of the capsule digestate revealed that almost all of the capsule material is still present in solution after the digestion. This is consistent with the low volumes of decomposition gases generated during the digestion of an empty capsule. It was initially thought that aspirating the digestate solution directly would simplify the analysis process, reduce the risk of contamination, and minimize dilution; however, the digestate solution would not nebulize well. Upon investigation 5034 Analytical Chemistry, Vol. 70, No. 23, December 1, 1998
it was determined that the polyacrylamide was responsible. The same quantity of acrylamide monomer in solution does nebulize well, suggesting that the polymerized form of acrylamide is responsible for the abrupt change in nebulization efficiency. When an equivalent quantity of polyacrylamide gel (5 mg/mL) is placed in water or nitric acid, the solution does not nebulize. This is believed to be due to the long polyacrylamide molecule that does not allow the liquid to shear apart during nebulization. Diluting to 0.5 mg/mL polyacrylamide allows the solution to be nebulized with the same efficiency as an aqueous standard so a 50-mg capsule digested in 10 mL of nitric acid must be diluted 10 times to avoid problems with nebulization. Therefore, all digestates were diluted 10 times in 15% HNO3 for analysis. Certified Reference Material Studies. (a) TORT-1. To study the capability of the system in a high-speed 5-min run, several materials were studied. The first of these, National Research Council of Canada (NRCC) Certified Reference Material (CRM), NRCC TORT-1, lobster hepatopancreas, was digested seven times. This particular reference material was chosen because it was used to evaluate the CEM SpectroPrep tube digestion system.5 In that evaluation, the lobster material required a predigestion before it could be pumped as a slurry into the tube digester. Apparently this material is very stringy, clogging the inlet valve if not predigested 6 when concentrations greater than 1% (w/w) were used. Obviously, predigestion was not required for the capsule system. The results for the seven digestions analyzed are listed in Table 1 as are the results from the other CRMs. The elements selected and accepted values are those employed in an evaluation study of the CEM system by Sturgeon et al.5 The elemental concentrations were obtained without using internal standardization, mo(4) Knapp, G.; Pichler, U.; Michaelis, M. Microwave Assisted Flow Digestion of Liquid Samples and Slurries at High Temperatures. Pittsburgh Conference, 1996; Paper 1250. (5) Sturgeon, R. E.; Willie S. N.; Methven B. A.; Lam, W. H. J. Anal. At. Spectrom. 1995, 10, 981-986. (6) Sturgeon, R. E. Telecon, March 1994.
Table 1. Percent Recoveries for CRMs Using 5-min Total Microwave Heating Time
element Be Na Mg Al Si P K Ca Ti V Mn Fe Co Ni Cu Zn As Se Mo Cd Sb Ba Tl Pb av
Lobster Tissue TORT-1
102 94 104 126 126
Bovine Liver 1577
Orchard Leaves 1571
95 99
127 87
91 102 91
113 90 113
146 97 100
92 73
97 104 106 104
123 100 116 110 100 114 93
137
108
98
115
103
103
Soil SO-2
Sediment MESS-1
43 1.6 76 31 0.1 80 4 22
47 31 55 23 0.1 100 14 53 3.4 44 58 80 96 84 93 107 93 88
40 77 70 32 86 45 86 83
119
16 3.1 33
92
44
65
lecular ion corrections, or adjustment for matrix effects. The first element listed, 75As, has a high recovery. The high levels of Cl in this marine sample produce an ArCl molecular ion (40 + 35) interference giving consistently higher results for 75As. The accepted value for Cd is debatable; the material selected may have come from a different lot than that actually used for the evaluation study.6 Previous work done on this material by the same author yielded a result of 23 µg/g and originally certified TORT-1 material had a value of 26.1 µg/g for Cd (NRCC certificate). Ni, Cu, and Zn all have excellent RSDs and recoveries. Examination of the raw results indicates that there appear to be two concentration levels of Pb present in the sample. This could be due to sample inhomogenity or contamination. At this time not enough information is available to make an evaluation.6 The TORT-1 sample is organic and generated decomposition products during the digestion which required the system to be cooled and vented three times. The recovered volumes removed from the digestion tube were all within 3% of the initial volumes added. This result clearly demonstrates that the venting did not result in a significant loss of analyte or reagent as well as indicating that it is not necessary to calculate the recovered volume. One can simply remove the digestate from the digestion tube, dilute, and analyze. The tube-based digestion system required the addition of an internal standard to the sample slurry input and dilution to a known volume on exit from the digestion tube, although the latter is performed automatically.6 (b) Bovine Liver Digestion and Analysis. Bovine liver is an example of a sample type that is difficult to convert into a uniform slurry. It has a tendency to float and creep up container walls. As a result, this sample type is not easily digested in a narrow-tube digestion system that requires the sample to be pumped in as a slurry. Liver is an organic material requiring a
venting sequence in our system to complete the digestion. When bovine liver is digested in a traditional microwave bomb, a twostep digestion (in which the bomb is opened) is usually needed to remove gaseous decomposition products that form during the first digestion. Table 1 contains the analysis summary of 0.5-g bovine liver digestions. Generally the results are quite good although there is strong evidence that indicates that one of the samples was contaminated by stainless steel.3 (c) Orchard Leaves Digestion and Analysis. Orchard Leaves 1571 was chosen because it does not totally dissolve under normal digestion conditions. The Si matrix of these leaves contains Fe which is not released into solution unless HF acid is used. The results of the analysis in Table 1 show that while Fe is low, Mn and Mg are at approximately 90% of the certified value. Other workers7 using only HNO3 for digestion of Orchard Leaves 1571 also obtained low results for Mg and Mn but gave no specific reason. Visual observation of the digestates against a light revealed thin strands that settled after 24 h. Unlike the Bovine Liver 1577, there is very little Cl present in the Orchard Leaves 1571 and relatively accurate values were obtained for As. Arsenic is an important indicator element, because it can volatilize and be lost during venting. The performance demonstrates that As remains in solution after venting and suggests that similar volatile species will remain in solution. (d) Soil SO-2. Several more difficult materials were studied with 5-min digestion times using nitric acid (Table 1). As expected, the digestions were much more incomplete than those previously presented. The recoveries are low for most elements as expected in SO2, because HNO3 does not totally dissolve this material.8 This is the case for soils in general8-10 which have a silicoalumino matrix that remains after the digestion. (e) Marine Sediment MESS-1. MESS-1 is soil with a botanical component. This type of sample is not only abrasive but also generates gaseous decomposition products when treated with acid. The digestion required a venting sequence to complete the digestion. The recoveries are low for most elements. Those that are close to 100% recovery are probably from the botanical fraction of the sediment. The Pb and Zn show full recovery and fall within the certified value range. Precision. Figure 11 presents the precision data from the CRM study. The average RSD is about 8%; however, this number is strongly affected by several large outliers as the figure indicates. In general, the performance indicates that the precision will be similar to that of other systems and that volatile elements are not lost during the process. CONCLUSION As configured, capsule/tube-based microwave digestion would appear to offer several advantages. Foremost, there are fewer human handling steps involved with this technique. The only step (7) Andrasi, E.; Dozsa, A.; Bezur, L.; L Ernyei.; Molnar, Z. Fresenius’ J. Anal. Chem. 1993, 345, 340-342. (8) Kammin, W. R.; Brandt, M. J. Spectroscopy (Eugene, Oreg.) 1989, 4, 4950, 52-53. (9) Xu, L.; Shen, W. Fresenius’ Z. Anal. Chem. 1989, 333, 108-110. (10) Hewitt, A. D.; Reynolds, C. M. At. Spectrosc. 1990, 11, 187-192.
Analytical Chemistry, Vol. 70, No. 23, December 1, 1998
5035
Figure 11. Histogram of elemental determinations vs RSD (%) for the CRMs presented in Table 1.
appearing to require human handling is the initial sampling process. Tube-based microwave digestion requires both weighing of the sample and generation of a slurry. There would also appear to be safety advantages in having a system that can be run completely without human intervention. Two engineering questions remain. The most difficult of these is the development of a more robust tube arrangement. This could be simply a stronger material or perhaps a more clever configuration like that of Knapp et al.4 If higher pressures can be tolerated, then higher temperatures can be used. Certainly, it would be desirable to reach 250 oC, because this would allow the digestion of most organic materials. The flange valve arrangement is, at present, threaded for strength and hand-operated. It seems to us that the system could be neatly automated with a small hydraulic arrangement; however, other arrangements may have benefits that we have not foreseen. The flange valve design proved to be the simplest, safest, and most reliable valve for the digestion tube. All parts that came into contact with sample could easily be cleaned, eliminating any memory effects or abrasion of the valve. A large-diameter digestion tube design was found to be the best choice for several reasons. First, a large-inclined-diameter tube in the shape of a U allowed the effective venting of decomposition gases. Second,
5036 Analytical Chemistry, Vol. 70, No. 23, December 1, 1998
the large internal diameter allowed samples to be transferred directly into the tube. This avoided any need to generate slurries. Capsule-based sample introduction also guaranteed 100% transfer into the digestion vessel. Samples that have a large particle size can be introduced without grinding. The third major advantage of the large-internal-diameter tube is that digestate can be pushed out with a squeegee. The liquid volume transferred out with the squeegee was always 97% efficient. This eliminated the need for an internal standard to account for loss of sample. Two washings were sufficient to remove any memory effect from the digestion tube. The large tube design can also be cooled rapidly, which is convenient when venting is required. Traditional microwave digestion bombs are bulky and take almost 10 times longer to cool. The ability to vent the system under software control allowed digestions to progress without overpressurizing the system or loss of sample. It was found that no analyte was lost during venting. Cooling of the digestion tube to prevent loss of analyte also prevented loss of reagent. With respect to the future, a multitube design could increase the sample throughput. It might be possible for a single magnetron with waveguides to feed individual tubes, although one could argue that magnetrons are not a particularly expensive component. The material PTFE-TFM11 might be a good replacement for the PFA. This material has a melting point above 340 °C. In general, capsule-based microwave digestion appears to offer the promise of the potential for significantly improving throughput and reducing analysis costs.
Received for review September 26, 1997. Accepted July 8, 1998. AC9710689 (11) Lautenschlager, W.; Schweizer, T. LaborPraxis 1990, 376-382.
