AUGUST, 1966 THE ANALYST Vol. 91, No. 1085 Rapid Automated Determination of Biphenyl in Citrus Fruit Rind* BY F. A. GUKTHER AND D. E. OTT (Department of Entomology, University of California Cztrus Research Center and Agricultural Experiment Station, Riverside, California) A totally automated analytical procedure for determining the fungistat biphenyl in citrus fruit rind has been developed. Small pieces of hand-peeled rind are automatically homogenised in water and steam-distilled to liberate the biphenyl, which is trapped in cyclohexane solution ; this solution is exhaustively extracted to remove interfering steam volatiles, and the biphenyl remaining is read a t 246 mp in a continuous flow recording spectrophotometer. Time required from the introduction of rind samples to read out of biphenyl present is 9 minutes, with about 15 minutes for the first sample.The useful range is from 1 to about 150 p.p.m. on a whole-fruit basis with a reproducibility of about 3 per cent. The method has been applied most extensively to Valencia oranges. THE fungistat biphenyl has been used commercially since about 1935l to protect citrus fruits from the normally extensive post-harvest decay caused principally by the so-called blue- green moulds, PeniciZZium digitaturn and P. notatztm. Despite the objectionable odour of biphenyl it is still used in immense tonnages, wherever citrus is grown, as a mould inhibitor on the multi-billion dollar citrus crops in storage and transit around the world. A com- bination of fortuitous properties makes the chemical biphenyl unique for this purpose: it is non-toxic, non-carcinogenic,2 inexpensive, its vapour tension - temperature relationships1 are ideal for the purpose, it can be easily handled and is a sublimable solid, and its vapour is highly fungistatic to the omnipresent blue and green moulds in their rind locales.With any of the usual methods of application to citrus fruits it does not penetrate through the ntact rind but resides in the wax “layers” and oil sacs of the rind until the fruits are adequately aired, allowing it to escape slowly by vaporisation. Because of its strong and persistent odour, however, much effort during the past 20 years has been expended to find a less offensive but otherwise satisfactory substitute for biphenyl. Thousands1 (Souci, S.W., private communication ; Eckert, J. W., private communication) of candidate chemicals have been screened in many laboratories for this purpose, but an equally efficient and acceptable substitute citrus fruit fungistat or fungicide has not yet been found. Because of its unusually low mammalian toxicity2 the United States’ legal tolerance for biphenyl on and in citrus fruits and products is 110 p.p.m. on a whole-fruit basis; this is the highest specific tolerance value yet assigned to any pesticide by the U.S. Food and Drug -Administration, except for some organobromine fumigants calculated as inorganic bromide. To stimulate efforts to find a less offensive substitute, however, some other countries have accorded lesser tolerance values to biphenyl in this usage.For example, Great Britain and France allow a still realistic maximum of 70 p.p.m., whereas in 1965 The Netherlands’ government legislated3 the completely unrealistic and fungistatically ineffective (Eckert, J. W., private communication) maximum of 30 p.p.m. Absolute enforcement of any pesticide tolerance restriction requires the extensive use, with confidence, by both the producing agency and the ultimate consuming agency-whether * Universitj- of California Citrus Research Ccntcr and A4gricultural Experiment Station Paper No. 1698. This material was presented in part at the Scptember, 1985, Technicon Symposium “Automation in Analytical Chemistry,” New York. 475476 GUNTHER AND OTT : RAPID AUTOMATED DETERMINATIOK ;Analyst, TTol. 91 it be wholesale buyer, importer or regulatory official-of an adequate residue analytical method.Adequacy in this instance relates to satisfactory reproducibility in many different laboratories, to a minimum detectability that will comfortably bracket the lowest amount permissible, to simplicity of operation because of the world-wide shortage3 of trained residue analysts and to rapidity because of the thousands of samples that should be involved. I t should also relate to the ability of the operation and required apparatus to be standardjsed so as to circumvent the unfortunate errors4 almost always introduced when a residue analytical method produced by one laboratory is adapted to the always slightly different equipment, supplies, conditions and personnel training in another laboratory.Because of the importance of biphenyl over three decades, there are dozens of residue analytical methods for this fungistat; Rajzmanl has reviewed in detail the most acceptable ones, and has pointed out the advantages and disadvantages of the many analytical approaches involved. Because it was readily adaptable, one5 of these methods has now been totally automated, from fruit or fruit rind to p.p.m. of biphenyl present, to provide a method that is fast, reliable, operationally simple, and unique in that it is a standardisable biphenyl-residue screening m e t h ~ d . ~ This method represents the near-ultimate in pesticide residue analysis. It is the first example of complete automation, from homogenisation in water of the fruit or fruit rind to recorder read out 9 minutes later of the amount of biphenyl present in that sample relative to fortified control samples, within the range of 1 to about 150 p.p.m.on a whole-fruit basis. METHOD APPARATU s- as shown in Fig. 1, and are listed as follows. AutoAnalyzer modules (Technicon Controls Inc., Ardsley, Xew York) are connected Solidprep sampler. Proportioning pzwzps, two. Digestor, with accessories f o r d i s t i l l a n h use. 6 0 PS4 @ 0 Proportioning Waste Proportion i ng Pump I 1 Ultraviolet spectrophotorneterT IO-mm quartz rectangular f j c 246 mp I Recorder Fig. 1. Automated system for biphenyl residues in citrus fruit rind. to pump 45ml of water as homogcnising fluid, and the rate of sampling is 7 per hour. in inches = Acidflex tubing; S = Solvaflex tubing) The Solidprep sampler is set Tube sizes are-lugust, 19661 OF BIPHENYL I N CITRUS FRUIT RIND 477 Evacuated separator (see Fig.2)--For simplicity in drawing this figure, the waste digestant liquid a t the end of the helix is shown as being aspirated into the vacuum source; in practice, however, the most convenient disposal is to pass to waste through an all-glass or plastic water aspirator. Distillables plus added cyclohexane Pump cyclohexane phase $i n sleeve ,lip j o i n t vacuum trap t 1 Pump aqueous 5 cm phase t o waste Fig. .2. Details of cxracuated separator essential to the biphenyl-residue automated system I'ltraviolet s#ectro#hotometer-The spectrophotometer is dual beam, with a 10-mm light Strip-chart recorder, 10 mV. Glassware and tubing (see Fig. 1). ASSEMBLY OF APPARATUS- The detailed flow diagram for the apparatus is schematically presented in the standard manner in Fig.1 ; a few minor changes may be needed in a particular system for smooth and reliable operation. Thus, the flow-rates controlled by some of the tubings in the two pumps may need to be: varied from those shown in the figure in order to achieve optimum flow characteristics. In particular, the diameter of the pump tubing at position 11 on the left-hand pump may need to be increased for maximum flow, or decreased to keep air bubbles out of the spectrophotometer cell. Also, the rate of flow of material pumped from the bottom of the evacuated separator may have to be varied to achieve good separation within the 5eparator. Acidflex sleeving over joints and transmission tubing is used throughout the system where there is possibility of contact with concentrated sulphuric acid, except where indicated otherwise.,411 joints (glass to glass, flexible-tubing to glass, or flexible-tubing to flexible- tubing) associated with the sampling system from the homogeniser vessel to the pipette which introduces sample and digestant into the digestor helix must be butt joints (no con- necting nipples) to avoid blockage by particles of orange rind. For the same reason the tubing from the mixing coil in this same stream to this introduction pipette should be Acidflex sleeving rather than transmission tubing. The introduction pipette should be long enough to introduce the sample directly into the heated zone ; accumulation of rind and other particles occurs if introduced into an unheated zone.Biphenyl-exhausted particles will accumulate in the cool downstream end of the helix, however, but here there is no risk of their subsequent dislodgement and contamination of another sample; the helix must be cleaned after about 1 week when in constant use to remove these gross accumulations. path rectangular quartz flow-cell in the sample side, and nothing in the reference side.478 GUNTHER AND OTT: RAPID AUTOMATED DETERMINATIOK [ArtdySt, VOl. 91 PROCEDURE- This system utilises either analytical-reagent grade or spectrograde cyclohexane” for extrac- tion of the distillate, and analytical-reagent grade sulphuric acid for the cyclohexane wash series. The Solidprep sampler is set to pump 45 ml of water as homogenising fluid, and is operated with a standard factory set programmer; the homogeniser motor is set for medium speed during the sampling cycle. The vacuum pump connected to the evacuated separator (Fig.2) is set for about 2 inches of mercury, the requirement for the separator, which in turn supplies the vacuum to collect the vapours in the collection funnel. This funnel is inserted into the exit end of the digestor helix which rotates a t 20 r.p.m. and is heated by setting the heater controls 1 and 2 at 3.0 and 5-0 amps, respectively. The ultraviolet spectrophotometer is set at the pre-determined absorption maximum for biphenj-1 in cyclohexane solution (246 mp under the present conditions, although Gunther et aZ.5 specify 248 mp). A 0.062-inch horizontal aperture slit is used in the air path of the reference side of the dual-beam spectrophotometer.To keep the measuring cell clean, the following procedure when shutting down the system is suggested: as rapidly and nearly simul- taneously as possible, the transmission tubing that is connected to the cell and to the pump tube which pulls the continuous stream of cyclohexane through the cell, is disconnected at the nipple joint from that pump tube, while a small pinch clamp is tightened over the three tubings which are connected collectively to the spectrophotometer (see Fig. 1) ; finally, a similar pinch clamp on the waste line connected to the bottom of the BO glass fitting is loosened to by-pass the contents in that fitting to waste. M7hen starting the system up, and after equilibrium conditions are reached, a continuous stream of cyclohexane pours into the BO fitting, and the foregoing procedure is performed in reverse. The chart drive of the recorder is turned on, and when the base-line becomes stable after about 15 minutes the Solidprep sampler is started with an empty cup in the first position followed by samples of previously chopped fruit rindt weighed into sample cups in every third successive cup, with two intervening empty cups between each sample for “wash” purposes; in this mode of operation the rate of sampling is 7 per hour.For routine screening analyses one empty “wash” cup will probably suffice, thus decreasing the time required per analysis by one-third. 1Vith series of samples containing both very high and very low levels of biphenyl, however, two cups for washing are necessary (see diffusion on Fig.4). DISCUSSIOS OF THE METHOD A simplified flow diagram of the basic steps in this biphenyl analytical system is shown in Fig. 3. The rind previously chopped into approximately Q-inch or smaller cubes is auto- Citrus rind Solidprep sampler - Digestor E Condenser Digestion-distillation Condensation of biphenyl and other distillables in cycloh exan e (C y c I o h e xan e p h as e) Ultraviolet spectrophotorneter (measurement at 246 mp) Fig. 3. Simplified flow diagram of the basic steps in citrus fruit rind c Separator Extractors Conc. H2S04 extraction of cyclohexane phase t HzS04 t o waste in the automated system for biphenyl residues matically homogenised in water, and then steam-distilled in the presence of sulphuric acid to liberate the biphenyl PLUS citrus oils (mostly terpenoids) and waxes.These steam volatiles are trapped in cyclohexane, the oils and waxes are quantitatively extracted into concentrated sulphuric acid, and the biphenyl left in the cyclohexane is determined at 246 mp. In the original manual method” some p-cymene escaped the room-temperature acid washing or was * The manual method5 requires thc iisc of spectrogradc cyclohexane. t $-inch or smaller cubes prepared by a Hobart or similar food chopper.August, 19661 OF BIPHENYL I N CITRUS FRUIT RIND 479 formed6 from other terpenoids ; to minimise this highly variable interference it was oxidised with permanganate, then washed out with more sulphuric acid.The present continuous acid extraction step at 45" C either removes @-cymene without the necessity for an oxidation step, or else the small-sized starting sample (2 g of rind against 150 g of whole fruit for the manual method) effectively reduces this particular background contribution to nil in this measuring system. One of the major obstacles in the development of this automated procedure was the problem of continuously separating an aqueous phase from an immiscible solvent phase, while at the same time having these two phases under reduced pressure. This was overcome with a special glass evacuated separator. As it is new as well as essential for the biphenyl system it has been shown in detail in Fig. 2. This device does not guarantee an absolutely non-aqueous stream out of the top.After the slip joint has been optimally set under equili- brium and fixed conditions for vacuum pump and digestor temperature, the occasional small amounts of aqueous phase and occasional air bubbles are entrained in the emergent cyclo- hexane solution; these are either sufficiently constant, or in such small amounts that the net reproducibility of results is not affected. Typical chart recordings are produced in Fig. 4, and were obtained from 1 or 2-g starting samples of chopped Valencia orange rind. These recordings are from the system operated for 7 consecutive, noise-free hours according to the flow diagram in Fig. 2. Thirty-four samples and 8 controls and fortified controls were run that day with no special (single) operator attention.The average time interval was therefore 9 minutes per test, exclusive of about 30 minutes for warm-up and 15 minutes for shut-down, Note the absence of significant background (Fig. 4) from even 2 g of control rind, and also the close agreement between '"I 20 I- - F Time Chart recordings obtained from the biphenyl-residue automated sytem: A, 1 g of Valencia orange rind control; B, 2 g of rind; C, 1 g of rind fortified with biphenyl a t 350 p.p.m.; D, 2 g of rind fortified with biphenyl a t 350 p.p.m.; E, 1 g of rind from the biphenyl-treated carton of fruit; F, sampling cycle origin of last peak Fig. 4. replicates by comparison of ordinate intercepts of peak maxima alone. For greater precision, peak height (or area) in absorbance units should be used: this is the difference in absorbance units between the peak maximum and a base-line drawn in the usual manner beneath the peak.Absorbance unit - peak height measurements were obtained in this way for a series of controls (rind from untreated samples) fortified in the range 35 to 700 p.p.m. to establish a fortified-control standard curve. To 1 or 2-g portions of chopped control rind, weighed into every third sampler cup, were added appropriate volumes (up to 2 ml) of a standard solution of 350 pg per ml of biphenyl in 95 per cent. ethanol, or a standard volume of solutions of varying strengths. There were no measurable differences between the 1 or 2-g samples of control rind run alone through the system and the 1 or 2-g samples of control rind each containing 1 ml of 95 per cent.ethanol. The resultant standard curve had a slope of 195 p,p.m. of biphenyl per 0.1 absorbance unit. Reproducibility results were as follows. The lowest point, 35 p.p.m. in the rind (a value which is roughly equivalent to 7 pp.m. based on the480 GVNTHER AXD OTT RAPID AIJTOMATED DETERMINATION [,4?2dySt, lrO1. 91 U.S. Food and Drug Administration practice of weight of whole fruit,” and well below The Netherlands’ tolerance of 30 p.p.m. mentioned earlier), was replicated six times jielding a mean value of 0.017 i 0.001 absorbance unit; a next higher value at 70 p.p.m. was replicated four times, with a mean value of 0.037 t 0.004 absorbance unit; the mean value at a 175 p.p.m. level, from three replicates, was 0.082 0.005 absorbance unit; other absorbance values averaged 0-200 at 350 p.p.m.and 0-350 at 700 p.p.m. These values were not corrected for a background of 0.003 absorbance unit per gram of control rind. For greater minimum detectability, 2-g samples of rind can be used routinely, or a recorder range expander can be added with l-g samples: the 2X position is feasible but noise and base-line shifting usually exclude the 4X and 1OX positions. Two-gram samples of grapefruit and lemon rind are necessary for adequate sensitivity and precision when AutoAnalyzing these citrus fruits. To examine samples representative of commercial practice, half of a field box of tree- ripened Valencia oranges was stored for 6 days at 25” to 30” C in a standard vented citrus fruit shipping carton. Standard, commercial, biphenyl-treated liners were placed in the bottom and top of the carton, and for extreme dosage one was placed in the middle with oranges above and below it.The other half box of oranges was kept in a separate room’as a control. Fruits were sampled from the top of the “treated” carton for AutoAnalysis, producing the two peaks labelled E in Fig. 4. From the fortified-control standard curve, 175 8 p.p.m. of biphenyl were in and on the rind of this treated fruit, equivalent to 33 & 2 p.p.m. on a whole-fruit basis. The primary problem that has occurred with the present system is that occasionally there is excessive signal noise to the recorder associated with the physical composition of the stream flowing through the measuring cell. However, if the cell tubing connections are all clamped off when this noise appears, and the cyclohexane stream is momentarily by- passed (according to directions in the shut-down procedure) to leave in the light path a full and static-condition cell, the noise immediately stops, yet there are no visible air bubbles or other contaminants.This problem is more annoying than serious, however, as peaks recorded during these periods quantitate comparably to smooth peaks from equivalent samples if a curve drawn midway through the noise is used for measurement purposes. The noise rarely lasts longer than 30 minutes at a time; if it starts in the middle of a peak that peak will probably be lost for quantitative purposes because there is a marked base-line shift at the start of the noisy period. At present it is not possible to relate directly and simply the slope of the fortified control standard curve to that of a primary standard curve with biphenyl alone.Without a “keeper,” (e.g., orange oil), biphenyl alone shows large and variable losses during the Solidprep homo- genisation cycle. The reason for lower responses from the few exploratory samples of fortified grapefruit and lemon rind, compared (Table I) with fortified l’alencia orange rind, may be due to lower amounts of oils and waxes as “keepers” in the first two varieties. Thus, standardisation must always be in terms of fortified controls of the same variety as the unknowns. Both automated and manual5 methods are compared in Table I. Background values from untreated (control) samples are collated in Table I1 in com- parison with corresponding samples run by the manual5 method.I t is clear from these results that variable backgrounds from grapefruits and lemons were not a cause of the lower recoveries shown in Table I. Despite these minor disadvantages, the advantages of speed, convenience, precision and reproducibility, coupled with more than adequate sensitivity for monitoring purposes, result in an automated system that should merit further evaluation for eventual adoption bj- control laboratories around the world as a screening method for biphenyl residues in citrus fruit rind. With this system, speed of analysis is reduced to minutes-per-sample rather than the hours-by-sample required by the several manual methods at present being used both in the United States and in Europe. With cups in the Solidprep sampler arranged as recom- mended, the time lapse between successive samples is 9 minutes.This was the loading used in the present investigation to achieve the maximum accuracy from assured adequate purging of the system between samples containing, at random, from none to more than 175 p.p.m. of biphenyl each on a whole-fruit basis. In screening operations routine samples * Mature Valencia orangcs contain 18.7 1 6-3 per cent. rind based upon 297 measurements, Navel oranges contain 22.1 5 7.3 per cent. rind (567 mcasurements), lemons contain 30.0 1 8.5 per cent. rind (632 measurements), and grapefruit contain 2 3 4 & 3.2 per cent. rind (47 measurements).August, 19661 OF BIPHENYL IN CITRUS FRUIT RIND 481 would contain from about 30 to about 120 p.p.m., and this meticulous purging might not be necessary; in this situation every other cup could contain a sample with consequent reduction of average time per analysis to 6 minutes.In its present form this method should also work with whole fruit that has been made into a purke and commercial fruit products, but these applications should be checked with fortified samples for reproducibility, sensitivity and efficiency before routine application. Applications to grapefruits and lemons should incorporate enough fortified controls to establish a reliable recovery value as related to Valencia oranges at 100 per cent. TABLE I ILLUSTRATIVE COMPARATIVE RECOVERIES AND REPKODUCIBILITIES O F AUTOMATED AND MANCAL BIPHENYL METHODS ON COMMERCIALLY TREATED FRUITS I N THE APPROXIMATE RANGE OF 30 TO 150 P.P.M.ON A WHOLE-FRUIT BASIS Avcrage recovery, per cent. Reproducibility, per cent. (- Automated* Manual Automated Manual h _7 r--- 7 Variety (2-g sample) 1. ( 150-g sample) (2-g sample) 7 (1 50-g sample) Grapefruit . . . . . . 38 (44 p.p.m.) 100: 1.2 1 3 : Lemon . . . . . . 52 (58 p.p.m.) 99 1 2 5 1 Orange, Valencia . . . . 100 98 f 3 *1 Orange, Navel . . . . c3 100 8 5 1 * From standard curves prepared from fortified controls, compared to Valencia oranges at t Rind only. assumed 100 per cent. recovery. Unpublished results developed in 1959 by Gunther and co-workers in routine application 5 Navel oranges were not in season at the time of this study; the manual results illustrate to commercial shipments in Hamburg, %‘.Germany. recoveries for possible comparisons by others in routine applications of the present method. TABLE I1 ILLUSTRATIVE BACKGROUND VALUES FROM CONTROL SAMPLES WITH AUTOMATED AKD MANCAL METHODS OX A WHOLE-FRUIT BASIS Background Absorbance units Equivalent, p.p.m. I h > 7 <----- 7 ----- Automated M anii a1 Automated Manual I’arie t y (2-g samples) * (1 50-g samples) t (2-g samples) * (150-g samples) t Grapefruit . . . . . . 0-004 0.001 0.005 0.003; 1.9 & 0.3 1.5 0.6: Lemon . . . . . . 0.004 i 0.001 0.008 & 0.002 2.6 f 0.5 1.8 5 0.4 Orange, Valencia . . . . 0.003 & 0-001 0.010 0.004 1.1 f 0.2 3.1 0.9 Orange, NaXTel . . . . § 0.004 & 0.001 § 1.0 i 0.2 * Kind only. t \:hole fruit. L-npublishcd results developed in 1969 by Gunther and co-workers in routine application S Navel oranges were not in season a t the time of this study; the manual results illustrate to commercial shipments in IIamburg, FV. Germany. recoveries for possible comparisons by others in routine applications of the present method. REFERENCES 1. 2. 3. 4. 5. 6. Gunther, F. A., Adv. Pest Control Res., 1962, 5 , 191. Rajzman, X., i9t Gunther, F. A., Editov, “Residue Reviews,” Springer-T’erlag, Heidelberg and New York, Volume 8, 1965, p. 1. van Stratum, I?. G. C., “The Toxicity of the Citrus Fungistat Diphenyl,” Central Institute for Nutrition and Food Kesearch T.N.O., Report No. R1838, The Netherlafds, November, 1964. Gunther, F. A., and Ott, D. E., “Automation in Analytical Chemistry, 1965, Technicon Sym- posium, New York, N.Y., 1965. National Academy of Sciences-National Research Council, Pesticide Residues Committcc, Report on “no residues” and “zero tolerance,” Fiashington, D.C., June 1965. Gunther, F. A., Rlinn, R. C., and Barkley, J . H., Analysf, 1963, 88, 36. Received February 14th, 1966