Background In Flow-Through Detection: An Update

As a sequel to our discussion, "Background Considerations In Radio Flow-Through Detectors", let's talk a bit more about the "background" of a flow-through detector during a chromatography run. While we risk being repetitive, recent developments offer the prospect of improved counting. We also feel that extending our previous remarks can provide further understanding of what background is and what can and cannot be done about it.

As we've previously suggested, if one looks into the subject even in the most superficial way, the inescapable conclusion is that the "background" is not really background, i.e., ambient radioactivity and cosmic ray activity make up only a part, sometimes a small part, of the counts typically observed in the absence of known radioactive components.

The sample cell of a flow-through detector is small, presenting a poor target for ambient activity and cosmic rays. For work with liquid scintillator, cell volumes of 500-1000 ul are usual; for modern Fast-LC, the tendency is toward liquid scintillator cells with volumes in the 50-100 ul range and even smaller packed cells.

Prior to the introduction of our liquid scintillator, IN-FLOW 2:1, liquid scintillator:mobile phase ratios were typically from 3:1 to 5:1; the mobile phase occupies perhaps 15-25% of the cell volume with the remainder being scintillator solution. When comparable resolution is sought from packed cells, they are smaller. After packing an empty coil, about half the volume remains void, about half is the scintillator; cell size is expressed on the basis of void volume. For our discussion, we might think of a packed cell with a stated volume of one-fourth the size of a liquid cell as containing about the same volume of HPLC mobile phase; with equal mobile phase flow rates we can expect comparable residence times and more or less the same resolution.

In either case, liquid or solid, we use small cells and they are small targets for background radiation. If cell volume is changed within normal limits, the observed "background" does not change significantly. When liquid scintillator is employed, the target is the total cell volume, the combination of liquid scintillator plus mobile phase. But, for packed cells, it is the scintillator, not the mobile phase, that provides the target.

Referencing our discussion to discrete sample liquid scintillation counters where 10 ml samples might be considered typical, the background of modern systems for 3H or 14C counting falls in the range of 0.5-2 cpm/ml. It is made up of two gross components. So-called "quenchable" background, about half the total, is derived from ambient activity, including cosmic rays, which affects the counting vial contents; inside the vial the scintillation process is at work. "Non-quenchable" background, the other half, is derived from photomultiplier crosstalk, photomultiplier noise, 40K in the photomultiplier faceplate and in the walls of the counting vial, and other phenomena unrelated to happenings within the sample.

With a flow-through detector under static conditions, i.e., where we needn't be immediately concerned about HPLC column eluate, or liquid scintillator in flow, or mixing, the division is markedly different, possibly only 5-10% "quenchable" and 90-95% "non-quenchable". The substantial reduction of "quenchable" background is a direct consequence of greatly reduced sample size, probably on the order of 90-95%. Referring to the discrete sample counter with its overall background of 0.5-2 cpm/ml with about half being "quenchable", if we assume that "quenchable" background in a flow-through detector is not very different, i.e., in the range of 0.25-1 cpm/ml, then with our small cells we are dealing with "quenchable" backgrounds of 1 cpm or less.

On the other hand, unless something is done about it, "non-quenchable" background is greater than in the sample counter. Photomultipliers might be identical to those of discrete sample counters but they are brought appreciably closer together to maximize light collection with the unfortunate result of elevating crosstalk as well. Further, individual sample counters are never used with solid scintillators which, because their photon output is limited, would otherwise encourage the instrument manufacturer to employ higher voltages on the photomultipliers. As we will see, this can be disadvantageous.

An interesting point is the marked difference between the backgrounds for liquid scintillator vs. solid scintillator cells. Comparing a liquid scintillator cell having four times the volume to a packed cell, the packed cell will have the higher background. With ambient activity plus cosmic rays being the same, and with the crosstalk and other external factors also about the same, and even with the liquid scintillator solution occupying the entire cell while the solid scintillator occupies only a part of it, how is the higher background for packed cells explained?

Solid scintillators exhibit essentially no "quenchable" background at all. There are, however, added "non-quenchable" characteristics above those of liquid scintillators. Solid scintillator is more dense than is liquid scintillator solution and therefore has more stopping power for the gammas and cosmic rays that generate true background. But, that is only a part of the total; also important contributors to the background of solid scintillators are ambient light and especially heat activation.

Neither is particularly a problem with liquid scintillator. While light may activate liquid scintillator, decay time is fast and relatively short periods of dark adaptation combined with coincidence counting are sufficient to eliminate the problem. Exposing solid scintillators to ambient light results in long-lived phosphorescence, sometimes lasting days; the phosphorescence is a single-photon phenomenon, largely, but not completely eliminated by the coincidence circuitry. However, if sufficient dark adaptation has taken place, and there is no inadvertent light exposure, we may assume that there are no lasting effects of light activation.

Ambient heat is another story. It is always present and it stimulates solid scintillators to produce single photons; there is nothing comparable with liquid scintillators. That is the principal cause of the elevated background of solid scintillators. As a practical matter, not a great deal can be done about it. Cooling is costly and something of a nuisance: it imposes problems on liquid flow. Therefore, in pursuit of the lowest backgrounds (as well as the highest efficiencies, which is another topic) liquid scintillation counting is preferred.

With so much of the "background" not really background, the reality is that for both liquid scintillation and solid, shielding is not an important consideration in flow-through detector design. In fact, most flow-through detectors employ little or no shielding and, even in those that do, its presence is more a matter of conscience than necessity. Unless the environment is absolutely terrible, it is difficult to justify the expense and complexity of more than perfunctory shielding for a cell which is not more than a tenth the size of a discrete sample vial, and often less. The path to improve flow-through detector performance must take another direction.

Various means have been used to reduce "non-quenchable" background -- smaller photomultipliers with less glass and lower phototube noise, a reduced aperture between the two photomultipliers to minimize light interactions between them, more careful attention to operating high voltage, anti-crosstalk discriminators, data treatment to eliminate count bursts.

Scintillating guard disks with long decay times (>100-200 nanoseconds) placed on either side of the cell are responsive to 40K gammas and hard betas and are probably also stimulated by light produced within the photomultipliers. Whereas the scintillators that produce sample counts have decay times of 5-50 nanoseconds, and with fast circuitry show few afterpulses, the guards continue to produce pulses relatively long after the initial stimulus. By examining afterpulse characteristics, much of the time it is possible to distinguish between events originating within and without the sample and then to reject those which have their origin outside the cell. It is an elegant approach though the guards do absorb light and may reduce the counting efficiency of low-energy emitters such as 3H as well as preclude the use of slower solid scintillators which sometimes have advantageous properties.

It seems preferable to eliminate the problem of "non-quenchable" background as much as possible rather than to correct for it after the fact. It is well-known that the light produced within a photomultiplier, the principal cause of crosstalk, is a direct function of photomultiplier size. Such light has several origins, all equally important. Cerenkov light is produced by the passage of energetic electrons, primarily derived from 40K, through the glass of the phototube envelope. The smallest photomultiplier with the thinnest faceplate and the least weight of glass has the least 40K content and exhibits the least internal light generation. Also, electrons from the multiplication process which deviate from the normal trajectory may strike the glass and cause it to scintillate. Cosmic rays that also cause scintillations, are more apt to strike a large tube than a small one. Such light output is somewhat reduced by the application of "HA Coating" (which has other purposes as well), black conductive paint and then a black insulating cover applied along the length of the tube.

Reducing phototube light output is but a part of the solution. The other part, when cell size permits, is to reduce the size of the opening between the two tubes. For example, the exposed face of a 1-1/8" photomultiplier is 3.98 sq. in. Upon placing a 3/4" dia. mask over each tube, the exposure is reduced to 1.77 sq. in., a reduction of 55%. Measuring crosstalk, we find it to be reduced even more, perhaps by 75%, both because light piped up the barrel of the tube and concentrated at the periphery of the faceplate tends to be intercepted and also because many more angular components are stopped. While reducing the aperture between the phototubes in this manner is not possible with conventional cells as it would mask the cell coil and reduce counting efficiency, when it comes to Fast-LC and cells of less than 100 µL, it is now accepted practice.

Smaller photomultipliers also operate at substantially lower voltages than larger ones resulting in reduced noise rates. Residual gases inside the tube can be ionized by collision with electrons. When these ions strike the photocathode or earlier dynode stages, secondary electrons may be emitted, thus resulting in noise pulses. The smaller tube, of course, has less residual gas while the lower voltage makes these ionizations less likely. A corollary is that the lowest backgrounds are seen at the lowest voltages, an indication that the conventional method of operating photomultipliers at the voltage required for H-3 counting and then attenuating the output signal for more energetic isotopes must be reconsidered.

Reducing the high voltage is unquestionably the one best means to reduce background for isotopes other than 3H where, in fact, it is already quite low. Remember, that for every other more energetic isotope, a wider energy range must be examined and so, if the same high voltage is employed, the background -- "quenchable" added to "non-quenchable" -- is that for 3H plus an additional amount depending upon how much broader is the region being examined. But, when the high voltage is turned down, the "quenchable" background is reduced as the smallest pulses are not sufficiently amplified to produce counts. Far more important is the reduction in "non-quenchable" background for reasons that have just been suggested and also because pulses derived from Cerenkov light, as is the case for quenchable background, may not be sufficiently amplified to produce counts.

It should be made clear that reducing the high voltage for energetic isotopes need not cost a significant number of counts. It may be regarded as the equivalent of signal attenuation practiced in older instrument designs. Perhaps a few of the lowest energy pulses, those barely at the limits of detection, will be lost but for energetic isotopes there are not many of these. On the other hand, with almost no loss in counting efficiency, the background may be reduced 50-70% depending upon the isotope, not even considering the performance gain to be had by reducing the aperture through which one phototube views the other.

Thusfar, we have been describing static conditions and have ignored the sample itself, including the chromatography. That is rather a mistake. While neither the chromatography nor the sample can properly be said to contribute to "background", they certainly can contribute to the baseline.

Sometimes, especially with liquid scintillator, there is chemiluminescence and insufficient time between the scintillator reservoir and the measurement cell for dark adaptation. While we do incorporate delayed coincidence circuitry which is capable of coping with modest levels of chemiluminescence, it is not really meant for higher rates where there are apt to be accidental coincidences between chemiluminescence pulses from one photomultiplier and noise pulses from the other.

There must be concern of the possibility that vestiges of previous runs are being slowly washed from the HPLC column to elevate the background of a current run. Then, there is peak tailing and the effect of a long tail from a large peak on the background as a subsequent small peak appears. And it goes without saying that contamination of solid scintillator by activity from a previous run, or even the current one is unacceptable.

Again, as with "non-quenchable" background, the only good solution is preventive -- good chromatography with sharp well-defined peaks, good sample clean-up prior to the chromatography, choice of the best scintillator solution for the mobile phase at hand, and, of course, complete decontamination of solid scintillators. Without these things, one can never know the "background" no matter how carefully measurements are made.

The overall situation cries for some realism. We all want low backgrounds in order to see and quantitate small peaks. Even more, no matter what the background, we'd like to know it accurately. But, there are too many unknowns, especially with regard to the sample itself. While we can take substantial steps to reduce instrument contributions to background, and by so doing reduce the uncertainty of low-level measurements, until we eliminate the sample (and the reasons for making the measurement) the background will likely vary from run to run; we must be ever cognizant of that possibility.