2431
Anal. Chem. 1983, 55,2431-2434
I
0
Diphenylacetylene
0
Azobenzene
I
2
4
6
I I
provides adequate signal while ensuring that the spot CM be conveniently located entirely within the measuring beam profile. The suggested protocol, detailed in the Experimental Section, provides a simple and convenient method for determining the sensitivity of any slit scanning HPTLC densitometer. Typical values for the Shimadzu CS-910 scanning densitometer are given in Table I. The proposed method ensures that detection limits and other sensitivity-dependent parameters are measured under conditions of known sensitivity. It is also hoped that the protocol will be widely adopted by other scientists and instrument manufacturers to provide a practical and consistent method of comparing the performance of different densitometers. Registry No. Azobenzene, 103-33-3;diphenylacetylene, 50165-5. LITERATURE CITED
8
I
1
I
0
TIME (min) Flgure 1. Relationship between the time of development and the
distance migrated by the standard to give a desired spot width. direction of scanning and the slit height in the orthogonal direction. Instrument resolution is usually at a maximum for slit widths in the range 0.5 mm to 0.8 mm (3). The signalto-noise ratio is slightly greater at 0.8 mm and this value is suggested for use. Slit height selecton is more critical. Sensitivity increases as the slit height is reduced and reaches a maximum for very small slit heights (3). However, as small variations in the position of the measuring beam with respect to the spot center produce relatively large variations in the signal, for practical convenience, a larger slit height should be used. A slit height equal to 110% of the spdt diameter
(1) Pollack, V. Adv. Chromatogr. 1979, 17, 1-52. (2) Coddens, M. E.; Butler, H. T.; Schuette, S. A.; Poole, C. F. Llq. Chromatogr. HPLC Mag. 1983, 1, 282-289. (3) Butler, H. T.; Schuette, S. A.; Pacholec, F.; Poole, C. F. J . ChromatOgr. 1983, 267, 55-63. (4) Coddens, M. E.; Khatib, S.; Butler, H. T.; Poole, C. F. J . Chromatogr.,
In press. ( 5 ) Butler, H. T.; Pooh C. F. HRC CC,J . High Res. Chromatogr. Chromatogr. Commun. 1983, 6 , 77-81. (6) Butler, H. T.; Pacholec, F.; Poole, C. F. HRC CC,J . H/gh Res. Chromatogr. Chromatogr. Commun. 1982, 5 , 580-581. (7) Yamamoto, H.; Kurlta, T.; Suzukl, J.; Hlra, R.; Nakano, K.; Makabe, H.; Shbata, K. J . Chromatogr. 1978, 776, 29-41. (8) Cheng, MA.;Poole, C. F. J . Chromatogr. 1983, 257, 140-145.
RECEIVED for review June 20,1983. Accepted August 24,1983. M.E.C. thanks the Michigan Heart Association for the provision of a graduate student fellowship in support of this project.
Cryogenic Homogenization of Biological Tissues Rolf Zeisler,* John
K. Langland, and Sally H. Harrison
Center for Analytical Chemistry, National Bureau of Standards, Washington, D.C. 20234 The quality of the analytical sample, as opposed to the quality of the chemical analysis, is too often a disregarded factor in analytical chemistry. The sample on which the results of a chemical analysis are finally based depends very much on the properties of the bulk sample and the analytical test portion taken from the bulk sample. Hence, sampling and subsampling procedures prior to analysis critically influence the precision and accuracy of the result (I). Subsampling of inhomogeneous bulk samples and loss or contamination during sample collection and pretreatment may bias measurements of trace levels of elements and organic compounds, resulting in imprecise, inaccurate data that do not represent the original sample. The sample quality is particularly important in the characterization of biological matrices, where often only small analytical test portions are used to determine the trace components in the bulk sample. For example, despite the macroscopic homogeneous appearance of human liver, large inhomogeneities have been reported within a single liver when 1-g test portions were used (2). The analyst is confronted with this inhomogeneity problem when a precise and accurate characterization of the bulk sample is desired. To obtain trace
data on the bulk sample and to make results from different test portions of the same sample equivalent, homogenization of the bulk sample is necessary. In this contribution a method of homogenizing biological tissues is described which ensures the quality of the subsamples. EXPERIMENTAL SECTION A variety of devices are in use for particle size reduction and homogenization of tissue. These include glass tube-and-pestle grinders, hydraulic or cutting shear homogenizers, Waring type blenders, ball mills, and so on. Each has drawbacks. Some homogenization devices, such as the tube-and-pestle grinder, generate heat that may alter the sample; furthermore,they can accommodate only small amounts of material at a time. The construction materials in these devices, such as, i.e., stainless steel and/or various plastics, may introduce trace contaminants. Iyengar et al. (3) introduced a cryogenic homogenization (brittle fracture) technique for sample preparation of biological materials whereby the tissue is ground at near liquid nitrogen temperature in an oscillating ball mill of Teflon. This method produces contamination-free samples, however, it was only evaluated and used for small freeze-dried samples (up to 20 g). To meet the need for homogenization of larger samples (e.g., whole organs), new brittle fracture mills have been designed and evaluated at
This article not subject to US.. Copyrlght. Published 1963 by the Amerlcan Chemlcal Society
2432
ANALYTICAL CHEMISTRY, VOL. 55, NO. 14, DECEMBER 1983
Table I. Cryogenic Homogenization Machines Used in This Study Teflon milling chamber vo 1 total, cm ball mill A ball mill B disk mill no. 1 disk mill no. 2
920 1000 1400 2500
disk mill no. 3
6500
milling body
capacity for soft tissue, g
ball, 65 cm3 ball, 200 cm3 disk and 1ring, 700 cm3 disk and 1ring, 900 cm3 disk and 2 rings, 1300 cm3 disk and 3 rings, 2800 cm3
60 150 100 250 400 1000
Spex Industries, Inc., Metuchen, NJ.
drive unit motor power/revolutions
type NBSshaker
0.25 kW, variable
shatter box by Spex Industriesa Disk Mill TS 250 by Siebtechnikb
0.25 kW, 825 rpm 0.75 kW, 825 rpm
Siebtechnik GmbH, Mulheim (Ruhr), Federal Republic of Germany.
the National Bureau of Standards. To preserve the integrity of the samples, all parts of the mill that are in contact with the sample are constructed of PTFE Teflon. Teflon has several essential advantages: (1)Teflon is generally very pure, therefore the danger of contaminating the sample with trace elements (other than carbon and fluorine) and organic compounds by abrasion and leaching is minimal. (2) Parts made from Teflon are chemically inert and withstand thorough cleaning procedures with acid and organic solvents; this also reduces the risk of contamination greatly. (3) Teflon provides the specific heat capacity (1.05 kJ/(kg-K)) and insulation (thermal conductivity = 0.07 W/(mK)) to maintain mill and sample at cryogenictemperatures throughout the homogenization process, thereby preserving the physical and chemical state of the sample. (4) Teflon has the density necessary for the grinding balls and disks and, at liquid nitrogen temperature, the hardness and strength (without becoming brittle) to efficiently grind frozen biological tissues. The experimental mill designs used in this study were of two types, a ball mill design patterned after the design of Iyengar et al. (3) and a disk mill design based on laboratory disk mills as marketed by Siebtechnik, Spex industries, and others. The different experimental designs, ranging in capacity from 50 g to 1000 g, are listed in Table I. Table I also includes information on the machines that drive the mills. The milling chambers, balls, disks, and rings (examples are shown in Figure 1)were machined from virgin PTFE Teflon round rod stock and flat sheet stock. Construction is straightforward, using either a standard metal lathe or a numerical controlled turning machine. The chamber is cut from one piece and the ring and disk are cut from smaller stock. Generallythe ring and disk are cut from a single piece. The inside corner of the disk mill chamber is rounded to help prevent stress cracking; therefore, the ring must have a matching radius on this corner. Teflon has a few properties that are unfavorable for this use: (1) During machining, close tolerances cannot be maintained becasue PTFE Teflon is an unstable material, Le., it will cold flow, or warp. (2) The linear expansion coefficient of the PTFE Teflon was determined to be about 1.2 X OC-' resulting in about 1 mm increases in the diameter of the heads during operation; both properties were considered in the design of the disk mills. The seals between mill chamber and lid are flat surface joints that are strongly clamped together with an aluminum backing of the lid. The beveled bottom of the mill head fits into a beveled aluminum holder which is adapted to the respective machine. This ensures a firm support of the mill, without play, during its operation. To aid the homogenization process, a Teflon mortar is used to prefracture large solid blocks of tissue. The mortar is precooled to liquid nitrogen temperature and the tissue is broken into small pieces (1to 2 cm3) by direct impact. After homogenization,the mill is opened in a cold box under nitrogen atmosphere to protect the tissue powder from moisture, contamination, and melting during subsampling.
RESULTS AND DISCUSSION The cryogenic homogenization process transforms the solid frozen tissue into a particulate powder. For subsampling the homogeneity should improve as the particles are ground to
b
r I 26 cm
Figure 1. (a) Ball mill: (1) milling body, (2) grinding ball. (b.) Disk mill: (1) mllling body, (2) grinding disk and rlng, (3) adapter plate.
a finer consistency. The particle size distribution of the homogenate obtained from the milling procedures was measured. In general, sieving experiments are an easy way to determine particle sizes; however, cryogenic sieving of frozen samples is much more difficult. The sieving experiments were carried out in a large liquid nitrogen vapor phase freezer which provided the cold temperatures and the dry atmosphere necessary to prevent melting and particle agglomeration due to moisture condensation. A standard set of metal sieves with ASTM mesh sizes of 40,60,80,100,140, and 200 were used to determine the particles size distribution of the homogenized tissues. The performance of the mills was evaluated by several different homogenization and sieving experiments with different tissues. The results of these experiments are summarized in Figure 2. The disk mill is much more effective in producing a small particle homogenate from beef liver than the ball mill. This matrix was selected to simulate the behavior of human liver tissue, an important target tissue for biomonitoring for environmental health. With the disk mill, virtually all of the material passes the 40 mesh sieve; Le., the particles are less than 0.46 mm in diameter. The results shown in Figure 2 are reproducible within about *lo% (1s) for constant weight, temperature, homogenization, and sieving time. Therefore, only the disk mills were considered for further use.
ANALYTICAL CHEMISTRY, VOL. 55, NO. 14, DECEMBER 1983
2433
Table 11. Homogeneity of Two Cryogenically Homogenized Livers, Mean Concentrations Based on Wet Weighta disk mill no. 2, liver ID L1M0344B ball mill B, liver ID LlB0014B u, % R,% B R, % u, % N N element method
x
C N
H
P K K S c1
c1
Na Fe Mg
Zn
Rb cu Br Cd Cd Mn Se B Mo Hg co Cr La cs Sb sc
PGAA PGAA PGAA PGAA PGAA INAA PGAA PGAA INAA INAA INAA INAA INAA INAA INAA INAA PGAA INAA INAA INAA PGAA INAA INAA INAA INAA INAA INAA INAA INAA
8 8 8 8 8 10 8 8 10 10
Major and Minor Components 143 5.0 3.0 29.5 4.1 2.0 20.2 2.6 0.3 3.01 7.0 16 2.95 3.1 1.0 2.92 3.6 2.3 2.13 3.0 2.5 1.08 3.0 1.5 1.00 3.3 1.0 0.84 1.4 0.2
(z,mg/g)
11
2.8
6.0
9
11 11
1.38 1.10
2.2 1.2
1.0 1.2
(Ub
1.4 11 1.2 1.2
0.4 9 0.3 0.7
2.49
1.2
1.6
0.98 1.31
4.9 3.2
6.5 2.3
1.9
1.0
7.2 11 1.8 17 25 5.4 4.6 14
10 5.0 0.9 15 25 3.8 6.6 10
Trace Components, (E, rglg) 0.85 0.7 10 20 10 11 0.87 0.2 10 2.2 1.3 10 16 10 1.5 0.8 10 3.4 1.5 7.0 6.0 10 5.0 3.5 11
10 10 10 10 10 10 8 10 10
196 150 41.9 9.52 5.88 2.75 1.31 1.34 1.23
10 8 10 10 8 8 10 10 10 10
Ultratrace Components (X, ng/g) 518 1.5 0.7 10 486 3.4 6.0 180 7.0 7.0 10 140 6.0 2.5 10 1.0 10 47.1 4.1 30 25 10 25 12 13 10 21 7.0 5.0 10 (1) 16.7 5.0 4.0 10 0.17 12 13 10
43 0 150 59.5
51 2 5.28 107 46.5 26 20 14.9 6.1 0.40
a The analysis was carried out on freeze-dried samples. Dry weight factors of the individual samples ( f = 0.2626 to 0.2703) were used for conversion to wet weight. N , number of subsamples analyzed; X,mean concentration; R , relative sample standard deviation; u, relative uncertainty of individual determination due to counting statistics; PGAA, neutron Relative values of specific capture prompt gamma activation analysis; INAA, instrumental neutron activation analysis. activity only (no standards used).
~-
Two more tissues were used to test the performance of the disk mills: pork fat (adiposusabdominalis)to simulate human adipose tissue, which is used for monitoring of organic pollutants, and beef muscle to include a fibrous tissue. The adipose tissue was the most difficult to homogenize. In all experiments, a fraction of up to 16% remained on top of the 40 mesh sieve. Often, the remainder included several 1cm3 chunks which were not broken up during the homogenization process. These larger pieces may have hindered the action of the mill. It was necessary to reduce the weight of the sample to about 120 g for the mill to function properly, since the volume of the adipose tissue homogenate is considerably larger than the volume of a liver homogenate sample having the same weight. In addition, the temperature of the mill had to be maintained as close as possible to liquid nitrogen temperature so that the fat remained brittle. Therefore, the aluminum support plates were also precooled in liquid nitrogen. Although the mill still did not perform as well with the adipose tissues as with the liver tissue, the properties of the adipose tissue homogenate appeared to be suitable for analysis. The fibrous muscle tissue also was not completely homogenized during the selected 4-min cycle. About 10 to 15% of the material remained on the 40 mesh sieve. This fraction consisted mostly of fibers about 1 cm long and 1 mm thick, Numerous smaller fibers were found in the 40 to 60 mesh fraction. The fibers were broken up by further milling. After an additional 4 min homogenization,only 2% of the material was left in the coarse fraction above 40 mesh. The particle
^ -
--
-
size distribution of the fibrous tissue showed, as with adipose tissue, that the bulk of this material was ground somewhat finer than the liver tissue. In addition to demonstrating the reduction in paritcle size, data supporting the homogeneity of the material was obtained by elemental analysis of the homogenate. Two differently prepared human liver homogenates were each subsampled into 20 jars with each jar containing about 7.5 g (fresh weight). The subsamples were lyophilized and randomly selected test portions of 250 mg of lyophilized material (corresponding to about 1g fresh weight) were analyzed for 26 elements using neutron activation analysis. The analytical procedures are described elsewhere (4). The results are summarized in Table I1 including the observed standard deviation R from the set of 10 samples and the uncertainty u due to counting statistics of the individual measurements. If the analytical uncertainty (in this case a) is small or insignificant relative to R, then R is an estimate of the inhomogeneity in the material. None of the elements determined showed any inhomogeneous distributions exceeding the analytical uncertainty when either the ball mill or the disk mill was used. The sample standard deviation R depends mainly on the dominant analytical uncertainty u which is due to the counting statistics. Therefore, no significant difference can be observed between ball mill and disk mill. However, the disk mill provides a finer particulate homogenate as shown in the sieving experiments and, because of mechanical requirements, can homogenize larger samples. For the majority of elements, R is smaller than 5 %
Anal. Chem. 1983, 55.
2434
2434-2436
Though handling of the parts of the mill and of the materials a t low temperatures is inconvenient, the Teflon disk mill is recommended as an effective, contamination-freedevice for size reduction and homogenization of biological tissue. Operation a t cryogenic temperatures reduces loss of volatile components and changes in composition during the size reduction step. Samples as large as 1000 g can he homogenized with this technique. The mill uses a commercially available drive unit and is fast and simple to operate. The quality and quantity of the samples produced should make this technique useful for the sample preparation for the analysis of biological tissues. LITERATURE CITED
40
so
so
Id0
sieves
1o;
( mesh
260
I
LIVER, ball mill LiVER
ADiPOSE
$%6
MUSCLE
Flgw 2. Pallick size UWibulbn of frozen Wue after hcmogenizatbn In Teflon milk approximately 150 g of tissue, grinding time 4 min. liver
in ball mill B, and liver, adipose, and muscle tissue in disk mill no. 2.
and less then 2% for several essential trace elements, for which large differences among 1-g test portions were observed previously (2). This suggests that subsampling errors due to inhomogeneity can be confined to less than 2%.
(1) Kratochvii. 8.: Taylor, J. K. Anal. Chem. 1981, 53. 924A-938A. (2) Lievens. P.: Versieck. J.; Comeiis, R.: Hoste, J. J . Radioanal. Chem. 1977. 37. 483-496. (3) lyengar, 0. V.: KaSperek, K. J . Radloanal. Chem. 1977. 39.
301-316. (4) Bailey, J.; Filrpalrick. K. A,; Harrison. S. H.: Zeisler, R. NBS Spec. Publ. ( U S . )1983,No. 656.
RECEIVED for review June 20,1983. Accepted August 25,1983. This work was supported in part by the Office of Research and Development, U.S. Environmental Protection Agency. Certain commercial equipment, instruments, or materials are identified in this paper to specify adequately the experimental procedure. Such identification does not imply recommendation or endoresement by the National Bureau of Standards nor does it imply that the materials or equipment identified are necessarily the best available for the purpose.
Coulometric Determinatlon of Sub-Part-per-Million Levels of Sulfur in Volatile Liquids I. J. Oita Standard Oil Company (Indiana), Research Department, Analytical Services Diuision, Naperuille, Illinois 60566 In petroleum refining technology, the presence of subpart-per-million levels of sulfur has become increasingly important. The Wickbold method ( 1 ) requires complete volatilization and comhustion of a large sample over a period of several hours. It has not heen widely accepted because of the hazards involved. Coulometric methods (2) are generally not precise enough below 1ppm. Drushel(3) reports a precision of about 10% for samples containing less than 1ppm, using the Houston-Atlas analyzer, which is a reductive method. The newly developed method described in this paper is a modification to oxidative coulometry which permits sulfur analysis below 0.1 ppm. In oxidative coulometry, the sample is burned in oxygen and the generated sulfur dioxide is coulometrically titrated with iodine. However, there are three main problems with this method. First, although the S02/S08equilibrium in comhustion reaction is generally 90% in favor of SO, a t the 1000 OC comhustion temperature, it varies with temperature, flow rate, and the type and amount of sulfur. For trace samples, this ratio has been observed to he as low as 70%. For this reason, it was essential to develop a method that would convert all of the sulfur in the sample to SO,. The second protilem with the roulometrir approarh is that reaction of iodine with SO? is nut selective. I f olefins are formed during the comhustion, they cnnsume iodine leading to high results. In order to minimize olefin formation, the sample must be burned slowlv. Howrver, this is unmtistkctog. in two respects: la) the large snmple size required ior trace analysis makes the combustion time excessive and rb) the iodine generation is so drawn out that accurncy of its measurement is greatly reduced. On the other hand, tw, rapid 0003-270018310355-2434SO1 5010
comhustion can lead to soot formation which can adsorb SO, and cause low results. Thus, depending upon the comhustion conditions and type of sample, results could be high or low. Although many of these errors are not important above the 1ppm sulfur level, they can become quite significant helow 1ppm. The present development ensures that all sulfur enters the coulometric cell as SO, and is based on the following reactions (4):
SO,
- + - + so, +
f 3CuO
Cu,O
so, f CUO
CuSO,
cuso,
cuso, (900 "C) CUO so, + 1/,o2 so,
-
(1) (2)
1/20,
(3) (4)
Reaction 3 was incorporated into a recent gas chromatography method for sulfur (5) wherein combustion products were passed over CuO at 900 "C in a single step yielding only SO,. Reaction 4 does not take place to any appreciable degree since SO, is effectively removed by the CuO. An excellent study of this reaction was published hv Robinson and Kusakabe (6). I n order to apply these rearfions to rnulometry, a two-step process was devi4oped. f i r g t , the sample is burned i n oxygen and the combusti~nprodurts arc passed over ('110 at 700 "C. a temperature nt which both SO2 and 5 0 , fbrm stable C'iiSO,. Partially oxidized hydrocarhons IP oxidized hy the CuO that eliminntps hoth d d i n and soof formation. In the second step. the CuO CuSO, zone is rapidly heated to YO0 "C which hherates unly SO? according to reaction 3 nnd ns R slug u,hich is easily measured. The 700 "C trupping tempcrnturc was Z 1983 American Chemical Soc ely
