 |
Description  |
|
|
BACKGROUND OF THE INVENTION
The invention herein relates to an improved form of cross-correlation
frequency domain fluorometry and/or phosphorimetry. This process is
well-known per se, being as described, for example, in Gratton U.S. Pat.
No. 4,840,485, and also in a large body of technical literature on the
subject. Also, instruments for performing this process are sold by I.S.S.
Inc., of 309 Windsor Road, Champaign, Ill. 61820, among others.
Instruments for performing the above processes are utilized for measurement
of the fluorescence decay, phosphorescence decay, anisotropy decay of
fluorescence or phosphorescence, and other known uses. These instruments
differ from the more common steady-state spectrofluorometers since they
provide a means to record the time evolution of the deactivation of
molecules or atoms after excitation with light. Typical times involved in
these processes span from 1 millisecond to 1 picosecond. Such frequency
domain fluorometers (which term is intended to include corresponding
phosphorimeters) are also utilized for the measurement of times involved
in other molecular dynamic processes such as the rotations of molecules or
parts of large molecules. Also, the apparatus may be used for the
resolution of excitation/emission spectra of different fluorescence or
phosphorescence molecules in a mixture; for the determination of
time-resolved spectra; for the resolution of kinetics decays of
fluorophores in a mixture; or for the measurement of reactions occurring
in the electronic excited state.
In a frequency domain fluorometer, the excitation light beam causing
fluorescent emission is amplitude-modulated by a light modulator, such as
a Pockels cell, or it is intrinsically modulated when the source is a
mode-locked laser or synchrotron radiation source. The fluorescence
emission is phase-shifted and demodulated with respect to the excitation
light beam. The shift in the phase and the demodulation are both related
to the lifetime of the excited electronic level of the emitting molecule
or atom, providing a means to determine the modalities of the decay.
Two types of frequency domain fluorometers are commercially available at
this time:
In a first type of instrument, the excitation light beam is modulated at a
certain frequency F, generally in the 0.1 KHz to 300 MHz range. The phase
shift and the demodulation of the fluorescence or phosphorescence are
measured using the cross-correlation technique. Measurements are repeated
at different modulation frequencies, usually 10 to 20 different
frequencies which are logarithmically spaced in a desired frequency
interval which depends on the characteristic decay time of the fluorescent
or phosphorescent molecule under investigation. This type of instrument is
referred to in the literature as a "serial" frequency domain fluorometer,
since the various measurements at different modulation frequencies are
made in a sequence of time, one after the other.
Several models of such serial fluorometers are commercially available, for
example the K2 introduced by I.S.S. in 1989 and the SLM 48000, marked by
SLM Instruments. Data acquisition with instruments belonging to this class
usually take from one-half hour to one hour for the collection of 10 to 20
different frequencies. These instruments offer the best sensitivity, which
is an important factor when working with substance having a low
fluorescence quantum yield or substances in low concentration such as
proteins or other biological materials. Similarly, these instruments
measure in a differential way the rotational rates of molecules without
the necessity of deconvolution techniques.
A second type of instrument has also been introduced to the market, as
described by Mitchell U.S. Pat. No. 4,937,457 and entitled Picosecond
Multiharmonic Fourier Fluorometer. This instrument si referred to as a
"parallel" frequency domain fluorometer, since data are acquired
simultaneously at different modulation frequencies. Usually, about 100
different modulation frequencies are acquired simultaneously. This type of
instrument can potentially reduce the acquisition time by an order of
magnitude, but as a disadvantage it has very low sensitivity. The
advantage obtained by the reduction in data acquisition time is thus
offset by the fact that the system is only capable of studying systems
with a very strong fluorescent signal. When the signal is low, which is
the case encountered in most applications involving biological materials,
the only way to get reasonable data from this kind of instrument is to
increase the data acquisition time. Therefore, in many instances the
instrument does not offer any tangible advantage over a standard serial
instrument.
Also, the parallel type frequency domain fluorometer is inherently more
expensive, which provides further disadvantage.
Parallel frequency domain fluorometry is described in the article by B. A.
Feddersen et al. entitled Digital Parallel Acquisition in Frequency Domain
Fluorometry, Rev. Sci. Instrum. Vol. 60 (1989) page 2929-2936.
By this invention, a new type of cross-correlation frequency domain
fluorometer and/or phosphorimeter is provided which has a significantly
reduced time required for the acquisition of a good signal having a high
signal to noise ratio, when compared with the standard serial-type
fluorometers. However, the apparatus of this invention also retains the
high sensitivity to faint signals of serial fluorometry, while providing a
speed of signal acquisition which is comparable to parallel fluorometry.
DESCRIPTION OF THE INVENTION
This invention relates to a method and apparatus for cross-correlation
frequency domain fluorometry and/or phosphorimetry. The apparatus
comprises a source of electromagnetic radiation, which is typically light,
as well as means for amplitude modulating the electromagnetic radiation at
a first frequency. Means are also provided for directing the
amplitude-modulated electromagnetic radiation at a sample for testing.
Means are also provided for detecting the luminescence (or phosphorescence)
of the sample. Means are present for providing a signal coherent with
amplitude modulated signals produced by the amplitude modulating means, at
a second frequency, to the detecting means.
Means are also provided for modulating the gain of the detecting means or
multiplying the output of the detecting means, by said signal. The gain
modulating means typically comprise photomultiplier tubes. The alternative
output multiplying means may comprise photodiodes and/or microchannel
plates for equivalent function.
The second frequency is different from the first frequency. Means are
provided for deriving a resultant signal from the electromagnetic
radiation and the detecting means at a frequency of the difference between
the first and second frequencies (which difference is the
cross-correlation frequency), to detect phase shift and modulation changes
of the luminescence from that of the source of electromagnetic radiation.
In accordance with this invention, means may be provided for sequentially
performing runs of the cross-correlation frequency domain fluorometry
and/or phosphorimetry by the apparatus described above at sequentially
differing first and second frequencies. For example, each sequentially
differing first and second frequency may differ in logarithmic order, each
successive first and second frequency being for example 10 times larger
than the immediately preceding first and second frequency, while,
typically the cross-correlation frequency remains constant throughout the
sequential performing runs. Means are provided for detecting the
intensities of signal responses of the respective runs at the respective
frequency which is the difference of the respective first and second
frequencies used in each run, i.e., the cross-correlation frequency of
each run. Means are also provided for prolonging the detecting of each
said signal response at each of said differing first and second
frequencies, until an integrated signal with a specific standard deviation
has been acquired for each of said differing first and second frequencies.
Significant advantage is achieved by the above, since the noise associated
with the measurements is not expected to be the same at all the modulation
frequencies. By this invention, the measurement is performed in such a way
that more time is spent when measuring at frequencies where the signal is
weak, and, importantly, less time can be spent at frequencies where the
signal is strong. Thus, significant savings of time can be achieved, since
the measurement at each frequency is only for that necessary amount of
time to achieve the desired signal to noise ratio, for a desired degree of
measurement accuracy. That is to say, the acquisition is "adaptive" at
each frequency in that it is possible to specify an acceptable standard
deviation for the measurement. The instrument acquires data at the
cross-correlation frequency of each of the differing first and second
frequencies, until the specified standard deviation has been reached. Then
it automatically moves on to the next set of frequencies.
Accordingly, any desired accuracy of data acquisition can be automatically
obtained at the minimum time necessary for such acquisition, contrary to
any of the systems of the prior art.
Preferably, means are provided for automatically executing a program of
said sequentially performed runs, to reduce the time required for
collecting the desired data to near its theoretical minimum for the
particular apparatus used.
Also, means may be provided for synchronizing the acquisition of data
waveforms sensed by the resultant signal deriving means in each run, with
the phase of the signal modulating the electromagnetic radiation at the
first frequency. The above means also causes the superimposing of
corresponding segments of the waveforms thus sensed, to obtain an average
waveform value for each run having an increased signal to noise ratio over
the individual waveform segments. This process permits the linear increase
of the signal to noise ratio over time in a manner which is more rapid
than techniques used in the prior art.
It is also preferred for the resultant signal deriving means to comprise
variable frequency digital filter means. Particularly, the preferred
digital filter means is set to filter signal responses at substantially
the frequency which is the difference of the respective first and second
frequencies used, i.e., the cross-correlation frequency. Most preferably,
the digital filter means is capable of filtering with a band which narrows
over time as a signal response is detected. Such a preferred digital
filter can be provided in the program of a personal computer that controls
the operation of the apparatus. This digital filter starts out with wide
band filtering width, and narrows as the process proceeds, as compared
with an analog bandwidth filter which stays at one bandwidth forever and
is not adjustable. By this invention, the variable digital filter proceeds
to its filtering operation much faster due to an initial acquisition at
wide bandwidth, and then narrowing down to the desired cross-correlation
frequency.
Additionally, as a significant improvement, the filtering frequency at
which the digital filter is set can be selected by the user through the
computer software, and, if desired, can vary with different operations of
this invention. Thus, if one chooses a set of first and second frequencies
for the practice of this invention, the set of frequencies may typically
number 10 or 20 different first and second frequencies for testing. For a
single exponential decay sample one may select just a couple of
frequencies. Alternatively, one may select up to 50 frequencies or more if
desired. The apparatus of this invention has the capability of selecting
any desired number of frequencies to measure at, and as one does so, the
ratio between the time the instrument acquires data and the time the
measurements take to be completed (the duty cycle) increases, contrary to
instruments that are presently in the prior art. Furthermore, the
frequencies can be selected on a linear, or preferably a logarithmic
scale, for a better pattern of frequencies for analysis.
Additionally, the signal from the resultant signal deriving means may be
automatically amplified by automatic gain means without phase and
modulation changes, in those circumstances when a digital filter means is
used in accordance with this invention. Such is not deemed possible when
analog filter means are used, as in the prior art.
Typically, the first and second frequencies as described above are
generated by frequency synthesizer means, typically phase-locked loop
frequency synthesizers. While any difference between the first and second
frequencies may be used, it is generally preferred to use a
cross-correlation frequency of 100 to 1000 hertz. Higher cross-correlation
frequencies make it possible to obtain a larger number of superimposed,
corresponding segments of the waveforms, so that the average waveform
value for each run having an increased signal to noise ratio is more
rapidly acquired, to provide an overall increase in the speed of data
acquisition.
As another advantage of this invention, one can simply set the desired
cross-correlation frequency, as provided by the digital filter, to a
frequency where the signal is clearly received. For example, if an
instrument in accordance with this invention is installed close to a radar
station, a radio station, or a laboratory where an NMR instrument is
working, one can reset the first and second frequencies, and the
cross-correlation frequency on the digital filter, to avoid interference
problems. Thus, measurements can be performed at a cross-correlation
frequency of 10 hertz to 100 kilohertz or above, with ease.
DESCRIPTION OF THE DRAWINGS
FIG. 1 is a diagrammatic view of a multi-frequency cross-correlation
frequency domain fluorometer in accordance with this invention;
FIG. 2 is a block diagram of certain hardware and functions of the
fluorometer of FIG. 1;
FIG. 3 is a block diagram of the software Monitor routine used in the
fluorometer of FIG. 1;
FIG. 4 is a block diagram of the software Acquisition routine used in the
fluorometer of FIG. 1;
FIG. 5 is a schematic diagram of the synchronous acquisition circuitry of
FIG. 1;
FIG. 6 is a schematic diagram of the automatic gain control circuitry,
typically found in the personal computer of FIG. 1; and
FIG. 7 is a typical printout of data acquired by this invention.
DESCRIPTION OF SPECIFIC EMBODIMENT
Referring to FIG. 1, the fluorometer of this invention is similar in
structure and operation to prior art type cross-correlation frequency
domain fluorometers, except as otherwise indicated herein.
A light source 10 may be a continuous wave laser or a collimated coherent
or incoherent DC light source such as an arc lamp. Light from the laser 10
passes through a light modulator 12 such as a Pockels cell to provide a
beam of light 14 that is amplitude modulated at a first frequency (as
previously discussed). The amplitude modulated light then passes through a
beam splitter 16 and into a rotating turret 18 to irradiate the sample 20
held therein. The turret can then shift by 180.degree. to irradiate a
reference sample 22.
First frequency synthesizer 24 is locked in phased relation with second
frequency synthesizer 26 as shown, and imposes the first frequency on the
Pockels cell 12 which, in turn, produces the beam of light 14 at said
first frequency. Beam 14 may be carried by a fiber optic bundle, if
desired.
Second frequency synthesizer 26, communicating through amplifier 28,
modulates the gain of light detectors 30, 32 at the second frequency,
which is different from the first frequency. Detectors 30, 32 may be
photomultiplier tubes, photodiodes, microchannel plates, a diode array
detector, a charge coupled device detector, or an avalanche photodiode
system.
The signal of light beam 14 is sent by beam splitter 16 to light detector
30, while light detector 32 picks up the fluorescent light emitted by the
irradiated sample 20 or 22 in turret 18, optionally through a fiber optic
bundle.
The signal from light detector 30 is sent via wire 34 to a digital
acquisition card circuit 36 through automatic gain circuitry card 35,
which typically resides in a personal computer 38. Similarly, the signal
from light detector 32 is sent via wire 40 to the same automatic gain
circuitry card 35 and digital acquisition card 36. Digital acquisition
card 36 may be a commercial circuitry card, such as model A2D-160 from DRA
Laboratories of Sterling, Virginia, or, alternatively, the Metrabyte DAS20
card. Such a card must have at least two channels of data acquisition for
connection with the respective wires 34, 40 as well as the possibility of
changing the gain under computer control, a digitizer with at least 12 bit
resolution, a digitization rate on the order of 100 KHz, and the
possibility to start the digitization cycle and setting the sampling rate
under control of an external trigger. The circuitry of automatic gain card
35 may be as shown in FIG. 6.
A synchronization signal from the frequency synthesizers is fed to card 36
through synchronous acquisition card circuit 42 as shown in FIG. 5. The
purpose of this module is to provide a synchronous signal which is phase
locked to the synthesizer master oscillator 24 or 26. Such synchronization
greatly improves the signal to noise ratio of the measurement. Card 36 can
accommodate two modules as shown by the circuitry of FIG. 5, which feature
current-to-voltage converter means and computer-controlled instrumentation
amplifiers for each channel of card circuitry 36.
A single wire addition to the card circuitry 36 allows to obtain the 5 volt
supply to pin number 9 of the DB-connector to power the synchronization
module. Card 36 fits into an 8 bit slot of the personal computer 38 and
has two independent sample-and-hold circuits and one 12 bit digitizer. The
maximum sampling rate is 160 KHz. As preferably operated in accordance
with this invention, card 36 uses one of the computer's direct memory
access channels, to relieve the central processing unit of the computer of
the computer from processing data during the acquisition, so that data
collection and storage occurs in the background.
As previously stated, one great advantage of this invention lies in the
ability to sweep a predetermined frequency range by varying the time of
measurement at each modulation frequency depending on the noise at that
frequency. A set of frequencies is first selected. The frequencies are
typically logarithmically spaced in the frequency range of interest. This
possibility is provided by this invention as compared with the
multiharmonic frequency method of the prior art. It has been previously
demonstrated that the best way to sample a decay process of fluorescence
or phosphorescence emission in the frequency domain is to logarithmically
space frequencies around the frequency corresponding to the reciprocal of
the characteristic decay time of the sample under investigation.
It is also been shown in the literature that measurement at 10 to 20
frequencies often provides the best comprise between the time of data
acquisition and the information recovered. The improvement of the signal
to noise ratio depends on the cube root of the number of frequencies.
Therefore there is only a marginal improvement in using a 100 frequencies
instead of 20. The estimated improvement, assuming that all frequencies
are measured with the same signal-to-noise ratio, is about 1.7.
The signal-to-noise ratio is not constant at each frequency in the
multi-harmonic techniques, since in the technique of the prior art the
same acquisition time is allocated for all frequencies, but the detected
signal is much weaker at higher frequencies. Instead, by this invention,
great amounts of time can be saved, since in each of the serial
measurements performed by this invention, less time will be spent at those
frequencies where the signal is stronger, resulting in a net saving of
time.
An additional advantage of this invention relates to the digital processing
of the signal. A first operation performed on the digitized waveform is
the "folding" operation by which successive periods of the
cross-correlation frequency waves are arranged exactly in phase, as a part
of the monitor routine shown in FIG. 3, and also FIG. 2. Such a software
process is available to the prior art, and is discussed for example in
Malmstadt, et al., Digital and Analog Data Conversions, Part III, W. A.
Benjamin, Inc. (1973).
As more waves are averaged, the signal to noise ratio increases linearly
with the number of waves rather than with the square root of the number of
waves averaged. This is due to the fact that every signal which is not
exactly in phase with the cross-correlation signal will be cancelled out
as more waves are averaged, so that a digital filter function is provided
to the Monitor routine, particularly by steps, 50, 52 and 54 of the
monitor routine (FIG. 3). The equivalent bandwidth of this digital filter
is a function of time, and the signal to noise ratio will increase
rapidly.
For example, assuming that the basic waveform to be measured is at 100 Hz,
after folding for one second, all the frequency components higher than 1
Hz will be averaged out, while those having frequencies below 1 Hz will
remain. For the same reason, after 5 seconds integration, only frequency
components below 0.2 Hz will contribute to the signal. The equivalent Q of
this digital filter, (defined as the value of center frequency divided by
the bandwidth) is then 500, and the Q increases if the cross-correlation
frequency is increased. The new digital acquisition mode makes the
selection of the cross-correlation frequency very simple. Therefore, very
high Q filters, with no center frequency drift and gain distortion, can
easily be implemented.
It is clear from the above that a synchronous (to the cross-correlation
frequency) signal is desirably available to trigger the digitization
process, as provided by synchronous card circuitry 42.
Another advantage that the digital filter function of this invention has,
compared with an equivalent analog filter, is that data acquisition can
start immediately after the new frequency has been selected, since the
filter Q is very low at early times. Instead, using an analog filter with
a Q of 80 at 40 Hz requires at least 2-3 seconds before the signal has
reached a steady-state value. Thus, in this situation, a fast frequency
sweep cannot be efficiently performed.
Referring further to the monitor routine, the waveforms of the signal input
56 are folded (reference numeral 50) by a known software routine. It is
determined whether enough waveforms are obtained, which also determines
the filter band pass 52. If enough waveforms have not been obtained, the
process is recycled back 58 to fold more waveforms 50. If enough waveforms
have been obtained, there is a checking process 60 for signal overflow and
the best gain. If the answer is no, after automatic gain control 62, a
known routine is provided (64) for reinitializing variables, i.e.
providing an automatic gain search. The collected data is discarded, and
the system restarted (66) by further signal input 56 to the waveform
folding 50.
If, however, the answer is yes to the check signal overflow and check best
gain 60, a fast Fourier transform is calculated (54). The AC/DC phase
modulation 68 is performed, and the values are displayed on a monitor 70.
Also, the adequacy of the noise level 72 is determined. If not, through an
exit 74 the signal input 56 is reactivated for more waveform folding. If
the answer is yes, a data-ready flag 76 is set. The variable are
reinitialized 78 to change frequencies and go on to the next step of the
process, which may be operation at a different frequency, or activation of
turret 18 to switch from one sample to the other 20, 22, or vice versa.
And then the next step 80 proceeds.
Typically, the process of this invention can proceed as follows:
1. The light shutters of the apparatus (conventional equipment) are closed,
and a background reading is acquired for about 5 seconds, which is
typically optimal.
2. The sample 20 is then illuminated, and data acquisition starts
immediately. At every second the data acquired are transferred to a
working vector without interrupting the acquisition process which proceeds
in the background. Noise monitor 72, a conventional software expedient,
estimates the amount of noise in the waveform acquired and compares it
with a preselected value. Acquisition can be as short as 1 second for
bright samples.
3. The acquisition continues until the estimated noise is below an
acceptable value. Then, a new frequency is selected per steps 76, 78, 80
from a preferably logarithmically-spaced frequency series, and the process
starts again, continuing step by step until all frequencies of the set
have been measured.
4. Then a reference compound 22 is illuminated, and the same process as for
the sample is performed, step-by-step, until all frequencies have been
measured.
5. The phase and demodulation ratio of the sample are calculated with
respect to the phase and demodulation ratio of the reference compound.
The entire process is very efficient, and lasts for typically about 3
minutes for medium intensity sample of 10 frequencies. If samples are very
bright, the entire process can terminate in less than 1 minute for the
acquisition of 10 frequencies.
To take advantage of the new capabilities offered by this invention, it is
preferred for data collection to proceed without loss of synchronization,
and data are collected using a large number of points for each waveform.
In addition, the waveforms should be at the highest possible frequency
compatible with the speed of the digital card 36 used herein. Preferably,
cross-correlation frequencies up to 500 Hz are particularly desirable.
The minimum number of points per waveform that provides accurate phase and
modulation determination is on the order of 128 points. The power of 2 is
necessary for the application of the Fast Fourier transform 54. Since at
least two different signals must be acquired, one from the sample
photomultiplier 32 and the other from the reference photomultiplier 30,
the number of points to be sampled per second is typically about 126,200.
This is approximately the maximum digitization speed of the digital
acquisition card 36 used herein and described above.
It is also desirable to have the capability to continuously display on the
screen the values of the voltages at each detector 30, 32, the values of
the modulation of the sample and reference 20, 22, and the phase
difference between sample and reference as computed by the software. This
feature is important for the setting of the instrument prior to each
measurement and for monitoring the measurement during data acquisition,
since no other information regarding the amount of light reaching the
detectors may be available. In addition, it is desirable to have a way to
monitor the noise of the signal in order to select the proper integration
at each frequency.
In order to provide a monitor of the instrument signals and a noise monitor
during data acquisition, among other reasons, the software used in this
embodiment preferably also utilizes the following features:
1. Data acquisition proceeds at constant speed in the background using the
direct memory access capabilities of the IBM personal computer (an
IBM-compatible computer can be utilized as well; specifically, every
computer utilizing a CPU of the Intel's iAPx86 family microprocessor
starting with the 82086 and including the 80386 and the 80486.)
2. The digitized data are stored in a "circular buffer" that contains a
maximum of 64000 points.
3. At each computer clock tick (18 times per second) an interrupt routine
is activated that checks how much of the data buffer has been filled.
4. If more than half of the data buffer has been filled, half of the buffer
is copied and folded in a working array that contains 256 points. The
signal is folded in such a way that each period of the waveform is added
exactly in phase to the previously stored waveform. When the second half
of the buffer is filled, then data are processed from this part of the
buffer while the first part is receiving the new data from the digitizer.
5. When a certain number of waveforms have been folded then the Monitor
routine (FIG. 3) is called. Generally the number of waveforms to be folded
is chosen in such a way that the Monitor routine is called every 0.5
seconds.
6. The Monitor routine performs a series of tests on the signal; it checks
for signal overflow, determines the most appropriate gain for the
amplifiers connected with each channel, and calculates the fast Fourier
transform of the signal to determine the value of the phase and modulation
of the signal from the two detectors.
7. Depending on the operation condition of the instrument, the Monitor
routine passes data to the Acquisition routine (FIG. 4) of the main
program for accumulation and storage and display.
8. The folding of 64000 data points, the fast Fourier transform (FFT)
calculation, and the screen display of the different instrument parameters
require about 0.4 to 0.6 second on a 386 computer with math coprocessor.
Since this operation should be performed every 0.5 to 1 second, clearly
there is very little time for performing any other task such is keyboard
entry and display, driving the instrument motors and reading or writing
disk files.
9. For monitoring purpose only, it is not necessary to collect so many data
points and to calculate the FFT on 128 points. During data acquisition the
computer is not performing other operations such as moving motors or
writing disk files. Therefore two modes of operations have been
implemented: one that read one every 8 data points and performs a FFT on
16 points only and a second mode of operation in which all data points are
processed.
10. The Monitor routine (FIG. 3) co | | |
