Precision half-life measurement of the \beta^{+} decay of 37K

Precision half-life measurement of the decay of 37K

P.D. Shidling    D. Melconian    S. Behling    B. Fenker    J.C. Hardy    V.E. Iacob    E. McCleskey    M. McCleskey    M. Mehlman    H.I. Park and B.T. Roeder Cyclcotron Institute, Texas A&M University, College Station, Texas, 77843-3366 Department of Physics, Texas A&M University,College Station, Texas, 77843-4242 Department of Chemistry, Texas A&M University,College Station, Texas, 77843-302
July 26, 2019July 26, 2019
July 26, 2019July 26, 2019

The half-life of 37K has been measured to be , a value nearly an order of magnitude more precise than the best previously reported. The decay of 37K occurs mainly via a superallowed branch to the ground-state of its mirror, 37Ar. This transition has been used recently, together with similar transitions from four other nuclei, as an alternative test of CVC and method for determining , but the precision of its value was limited by the relatively large half-life uncertainty. Our result corrects that situation. Another motivation for improving the value was to determine the standard-model prediction for the -decay correlation parameters, which will be compared to those currently being measured by the Trinat collaboration at Triumf. The new value, , is now limited in precision by the ground-state branching ratio.

21.10.Tg, 23.40.-s, 27.30.+t

I Introduction

Many precision measurements of the correlation parameters and values of -decaying nuclei have been used to help form our understanding of the fundamental symmetries of the weak interaction. Experiments of this kind continue to be performed to search for physics beyond the standard model. For example, the measured values for superallowed pure Fermi transitions, have been used to verify the conserved vector current (CVC) hypothesis to one part in ; to determine most precisely the value of , the up-down element of the Cabbibo-Kobayashi-Maskawa (CKM) matrix; and to set limits on scalar and right-handed currents HardyPRL:05 (); Hardy:05 (); Hardy:09 ().

Transitions between isospin doublets in mirror nuclei are also useful weak-interaction probes because there are relatively few corrections required to describe their decay. The most thoroughly studied and theoretically cleanest example of such a transition is the decay of the neutron mendenhallPRC2013 (); plasterPRC2012 (); PERKEO (), but a number of other cases have attracted attention recently: the decays of 19Ne, 21Na, 29P, 35Ar and the subject of the present work, 37K. Naviliat-Cuncic and Severijns NavPRL09 () used the values for these five transitions to determine , though less precisely than has been achieved from the transitions. With more precise measurements, however, the result for could be improved enough to provide a valuable cross-check on the result from the transitions. Our measurement is a first step on this path.

For mixed Fermi/Gamow-Teller transitions, such as the decay of 37K to the ground state of 37Ar, the measured value can be related to via the corrected value, which incorporates calculated corrections for radiative and isospin-symmetry-breaking effects, as follows SevPRC08 ():


where , the Fermi constant , and the vector coupling constant . The parameters , , and are the usual radiative and isospin-symmetry-breaking correction terms calculated for the vector current Hardy:09 (). The statistical rate functions for vector and axial-vector currents are denoted and . In these mirror nuclei their ratio generally is within a few percent of unity according to shell model calculations SevPRC08 ().

The Fermi matrix element in the limit of strict isospin-symmetry is


for a transition from a nucleus having protons, neutrons, and isospin . It takes the value 1 for =1/2 mirror transitions.

Finally, the Gamow-Teller to Fermi mixing ratio is given by


where is the Gamow-Teller matrix element, is the axial-vector coupling constant, and the correction terms with superscripts are equivalent to those with superscripts but must be evaluated for the axial-vector component. Since, in general, cannot be precisely calculated, must be taken from experiment. This also makes it unnecessary to calculate the axial-vector correction terms.

The decay of 37K was one of the five cases used in Ref. NavPRL09 () to determine . The relatively large uncertainty of its contribution was overwhelmingly dominated by the precision of , which was obtained from a single measurement of the neutrino asymmetry parameter,  MelconianPLB2007 (). The value is known much better, to  SevPRC08 (), with the largest contribution to its uncertainty being the lifetime of the decay. A survey of mirror transitions and the theoretical corrections to their values was made in Ref. SevPRC08 (), in which three 37K half-life measurements Sc58 (); Ka64 (); Az77 () were averaged to arrive at . This uncertainty is much larger than the and uncertainties on the other two contributors to for 37K: the branching ratio and value, respectively.

The aim of the present work is to improve the lifetime of 37K so that it no longer dominates the uncertainty in the deduced value. A more precise value on its own will not improve the value of obtained from 37K decay. For that purpose, an improved measurement of via a correlation parameter such as or is necessary.

Ii Experiment

A high purity radioactive beam is essential for performing a precision half-life measurement. We have measured the half-life of 37K at the Cyclotron Institute of Texas A&M University using a continuous-gas-flow proportional counter and a fast tape-transport system. We produced 37K via an inverse kinematics reaction (fusion-evaporation reaction) by accelerating a primary beam of 38Ar to in the K500 superconducting cyclotron, and impinging it on a -atm H gas target cooled to liquid nitrogen temperature. The forward focused reaction products were analyzed using the Momentum Achromat Recoil Separator (MARS) mars (); tamu-beams+MARS (). Selection and focusing of the desired ions in MARS was done according to their value. At the focal plane of MARS, a 16-strip position-sensitive detector of thickness with an active area of could be inserted to detect the secondary reaction products, which were identified according to their position and energy loss in the strip detector. The position of the reaction products on the strip detector was related with its ratio. This detector was initially inserted to help us tune the spectrometer and was then removed before the measurement began. During the measurement, it was reinserted once a day to ensure that no changes had occurred. In this way we determined that the 22.7- 37K beam passing through the spectrometer extraction slits was pure, the remaining contaminants being 35Ar, 34Cl and 33Cl.

This separated beam then exited the vacuum system through a thick Kapton foil. Immediately following this foil was a -thick BC-404 plastic scintillator, which counted the number of ions. The beam then passed through a stack of Al degraders, and finally was implanted into the 76--thick aluminized Mylar of the fast-tape transport system. The degrader thickness was adjusted so as to place the implanted 37K at a depth mid-way through the tape. At this setting (m of Al), the combination of electromagnetic separation in MARS and range selection in the degraders resulted in the 37K beam reaching the tape being pure.

For part of the measurement we used two different thicknesses of degrader: m, which resulted in the 37K being placed at the front of the tape; and m, which put it at the back of the tape. The former led to a higher proportion of impurities; the latter to a much lower proportion. In both cases the intensity of 37K was reduced. In the best case, with the 37K deposited near the back of the tape, the sample purity was .

A sample of 37K was collected by implanting the beam into a section of aluminized Mylar tape for a time interval of . The beam was then turned off and the tape-transport system moved the sample in to a well-shielded location, stopping it in the center of a proportional gas counter. There, the particles from the decay were counted for half-lives (). This cycle was repeated continuously, with each sample being collected on a fresh region of tape, until the desired statistics had been accumulated.

The gas counter and the data acquisition system used in the present measurement were similar to those originally described in Ref. koslowskyNIMA1997 (). The current system has also been described in detail in reports of previous half-life measurements of superallowed emitters performed at Texas A&M University (see, for example, iacobPRC2006 (); iacobPRC2010 ()). Briefly, the preamplified signals from the gas counter are passed to a fast-filter amplifier with a high gain (). At this high gain many of the pulses would saturate the amplifier so, to ensure that the amplifier recovers quickly, large pulses are clipped with a Schottky diode inserted after the first stage of amplification. The amplified and clipped pulses are then sent to a discriminator with very low threshold ( mV).

Since deadtime is a serious concern for half-life measurements, the discriminator signals were split and sent to two fixed-width nonretriggering and nonextending gate generators, which established different dominant deadtimes in the two separate streams, both of which were multiscaled into 500-channel time spectra.

The total measurement was split into 13 runs with different experimental conditions: three discriminator threshold settings (150, 200 and 250 mV), two combinations of dominant deadtimes (4 and 6 s; 3 and 8 s), three detector bias voltages within the detector’s plateau region (2600, 2650 and 2700 V) and three degrader thicknesses (13, 51 and 70 m). The initial activity of 37K for each cycle was in the range of . All runs were composed of  cycles, which yielded a total of counts in each. In addition, a background measurement was performed for which all conditions were the same as for normal data taking except that the tape motion was disabled. The background rate was found to be four orders of magnitude lower than the initial count rate for each collected sample.

Figure 1: (Color online) Typical summed decay curve of the deadtime-corrected data obtained from a single run (top) and corresponding residuals of the fit (bottom). The 37K decay curve accounts for 99.5% of the data, with 0.27% from the 35Ar and 33,34Cl contaminants, and the rest from background. The reduced of the fit was 1.05 with 488 degrees of freedom.

Iii Data Analysis

The selection of data for final analysis was made according to three criteria. First, cycles with too few counts – less than 5000 – were rejected. Second, the number of ’s recorded in each cycle was compared to the corresponding number of heavy ions recorded in the scintillator at the exit of MARS. If their ratio was abnormally low, it meant that the tape did not stop with the activity positioned correctly within the proportional gas counter. In that case, the cycle was rejected. Finally, a cycle was rejected if a half-life fit of the recorded data resulted in a very poor reduced , which corresponds to a goodness-of-fit statistic less than .

We analyzed the decay data from accepted cycles using two different methods: one based on summing all accepted cycles in a run (“summed” analysis) and the other on performing a simultaneous fit on all the cycles of a run (“global” analysis). All fits employed a fortran code based on the Marquardt algorithm for minimization. The program assumes Poisson rather than Gaussian statistics in the data.

The fit function contained a constant background plus four exponentials corresponding to the decays of 37K and the main contaminants 35Ar, 34Cl and 33Cl. The half-lives of the contaminants were fixed at their well-known measured values: for 35Ar SevPRC08 (), for 34Cl Hardy:09 () and for 33Cl SevPRC08 (). The intensity of the contaminants could not be fit separately because of the high correlations between them, so we fixed the relative normalization of the contaminants at the values measured in the MARS focal-plane detector (corrected for losses in the degraders), and allowed only the total amount of contaminant activity to vary relative to 37K in the fitting procedure.

Figure 2: (Color online) Simultaneous fit to the 21 decay curves of a single run containing a total of 311 cycles combined into groups of 15 (top) and corresponding residual plot (bottom). The data here are the same as shown in Fig. 1 except in this case the data are not summed to a single decay curve. Here the reduced is 1.025 with 6858 degrees of freedom.

In the “summed” method of analysis, the first step was to correct the measured decay spectrum in each cycle channel-by-channel for deadtime, based on the measured rate in each channel. Next, the cycles of a given run were summed into two decay curves, one for each imposed dominant deadtime. Finally, these spectra were fitted as already described. Figure 1 shows a typical deadtime-corrected decay curve for a single run analyzed using the summed analysis technique. Note that the fitted background level in the figure agrees well with the value obtained in our dedicated background run.

In the “global” analysis, the fitting function itself is adjusted to account for deadtime effects. In this method, many separate decay spectra within a single run were fit simultaneously, all with the same lifetime for 37K but with separate contamination levels and backgrounds allowed for each spectrum. Runs had between 120 and 750 cycles; so, to create a more manageable number of spectra and also to improve the statistics in the individual spectra, we summed the cycles together in groups and for each group took the effective count rate to be the average of the rates in its component cycles. We then performed a global fit on the resulting summed spectra. The 37K half-life extracted turned out to be insensitive to the number of cycles summed together for each spectrum. Figure 2 shows a simultaneous fit using the global-analysis method; it is of the same run for which the summed analysis is illustrated in Fig 1. The extracted values for the half-life obtained from the two analysis methods are well within uncertainties, as one can see from the results shown in Figs. 1 and 2.

Figure 3: The half-life obtained for 37K with different experimental conditions using the two analysis methods. The labels for each run are the detector bias, discriminator setting, degrader thickness and imposed deadtime. The solid line shows the best-fit half-life with the dashed lines representing the statistical uncertainty in the fit.
Method half-life [s]
Summed analysis
Global analysis
Average 37K half-life
Table 1: Half-life values for 37K obtained from two different analysis techniques. All of the uncertainties shown are purely statistical. The adopted half-life is taken to be the unweighted average of the two methods.

iii.1 Systematic Uncertainties

The changing of parameters from run to run allowed us to test for potential systematic effects that could contribute to the uncertainty in the final results. In Fig. 3 we plot the half-life obtained for 11 different run conditions using both analysis techniques. Since two of the 13 runs had the same conditions as another run, each has been averaged with its twin and plotted as a single half-life. Each plotted result is labeled by the four measurement parameters that pertained to that result: detector bias, discriminator setting, degrader thickness and imposed deadtime. As already mentioned, each run produced two data streams, each with a different imposed deadtime. In all cases, the half-lives obtained from the two streams agreed very closely. Since both contain the same primary data, only the result from one of the data streams is presented in Fig. 3. No systematic dependence on measurement parameters is evident.

In Table 1 we present the half-lives, averaged over all runs, obtained from the two different analysis techniques, “summed” and “global”. We take our final result to be the unweighted average of these two half-lives, with the statistical uncertainty on the average conservatively chosen to be the larger of the two individual values.

Figure 4: Test for otherwise undetected short-lived impurities or for possible rate-dependent effects. Each point is the result of a separate fit to one of the runs as the starting time of the fitting range is increased. The solid and dashed lines correspond to the average half-life value and uncertainties given in Fig. 3.
Source Uncertainty
Analysis method
Fitting range
Bias voltage
Discriminator threshold
Total systematic
Total uncertainty
Table 2: Error budget for the 37K lifetime measurement. All of the values listed are in milliseconds.

Although we regularly monitored impurities with the detector at the focal plane of MARS and accounted for the observed impurities in our fitting procedures, we nevertheless made an additional check at the analysis stage for unidentified short-live impurities. We did this by artificially removing up to the first (60 channels) of the counting period in steps of . At each step the remaining channels in the reduced data set were fitted and a half-life extracted. The results are plotted in Fig. 4, from which it is clear from that the derived half-life is stable against these changes. This test also demonstrates our system’s independence of counting rate. We examined the sensitivity of our fit results to all the potential sources of systematic errors described above. The outcome is the error budget given in Table 2. We found the greatest variation in the extracted lifetime when we fixed the value of the total impurity level to the amount observed in the MARS focal-plane detector rather than leaving it free to vary in the fits. The next largest source of uncertainty was the counting statistics. Our final result for the half-life of 37K is:


This measurement is nearly an order of magnitude improvement over the previously accepted value SevPRC08 () and now completely dominates the world average of the 37K half-life.

Iv -value

We follow the convention used in Refs. NavPRL09 () and SevPRC08 () by defining the corrected value for mirror nuclei which includes the variation due to the ratio of the Fermi to Gamow-Teller matrix elements. Specifically, the value is defined as


where experimentally has already been defined in Eq. (1) in terms of the measured value and the small radiative and isospin-mixing corrections as calculated in Ref. SevPRC08 (). The mixing ratio defined by Eq. (3) is experimentally determined from a measurement of one of the angular correlations of the decay.

When calculating the value, we take the branching ratio and total decay energy of 37K from the survey of Ref. SevPRC08 (): and .

Using our new value for the half-life of 37K, we find


This result is four times more precise than the previous value quoted in Ref. NavPRL09 (), and is now comparable to the precision of the three most precisely measured mirror nuclei. The branching ratio, no longer the half-life, is what now limits the precision of the value. We have already made a new measurement of this branching ratio and are currently in the process of analyzing the data. This should reduce the uncertainty in the value even further.

The only 37K correlation measurement published to date,  MelconianPLB2007 (), can be used to determine a value for . Using the recoil-order corrections of Holstein holstein (), we find , where the asymmetric error bars arise from the fact that the uncertainty in is so large that the uncertainty on the extracted value of is not Gaussian. Using this value of and the value above leads to


With our new lifetime measurement, the central value of has changed and is now better aligned with other mirror transitions.

Figure 5: (Colour online) Values of deduced for the four most precisely measured mirror transitions as a function of mass number. The smaller error bars with the filled in region represent the component of the total uncertainty arising from just (the total error bars are all dominated by the correlation measurements). The hatched region represents the fit to the average () yielding . For comparison, the solid band shows the value of derived from the pure Fermi transitions.

Finally, noting that for a mixed transition, is related to by NavPRL09 ()


we calculate that from 37K alone


This can be compared to the older value of 0.985(12). The uncertainty is barely affected by our improved half-life measurement because it is dominated by the large uncertainty in the correlation parameter measurement.

In Fig. 5 we plot values for the four most precisely measured nuclear mirror transitions. The point for 19Ne has been updated from the survey of Ref. SevPRC08 () to include recent lifetime results triambakPRL2012 (); ujicPRL2013 (); broussardPRL2014 (). The figure shows that the main effect of the present 37K lifetime measurement is to shift its value of into better agreement with the other mirror nuclei and the transitions.

V Conclusion

We have measured the half-life of 37K, the decay of which proceeds primarily via a mirror transition to its analog, the ground state in 37Ar. Our new result for the half-life is . The precision of this result is nearly an order of magnitude improvement over the previously accepted world average. The value for the 37K mirror transition is now determined to precision, and the corresponding values of and have shifted into better alignment with values determined for the other well-measured mirror transitions. Nevertheless, the uncertainties of and remain completely dominated by the correlation measurement for the transition. The results from improved correlation measurements, currently in progress by the Trinat collaboration at TriumfmelconianLASNPA (), are required before the uncertainty on the mirror-transition value for can be improved.

Vi Acknowledgements

We are grateful to the support staff of the Cyclotron Institute, especially Don May and George Kim for providing the primary beam. This work was supported by the U.S. Department of Energy under Grant No. DE-FG02-93ER40773 and Early Career Award ER41747, as well as the Robert A. Welch Foundation under Grant No. A-1397.


Comments 0
Request Comment
You are adding the first comment!
How to quickly get a good reply:
  • Give credit where it’s due by listing out the positive aspects of a paper before getting into which changes should be made.
  • Be specific in your critique, and provide supporting evidence with appropriate references to substantiate general statements.
  • Your comment should inspire ideas to flow and help the author improves the paper.

The better we are at sharing our knowledge with each other, the faster we move forward.
The feedback must be of minimum 40 characters and the title a minimum of 5 characters
Add comment
Loading ...
This is a comment super asjknd jkasnjk adsnkj
The feedback must be of minumum 40 characters
The feedback must be of minumum 40 characters

You are asking your first question!
How to quickly get a good answer:
  • Keep your question short and to the point
  • Check for grammar or spelling errors.
  • Phrase it like a question
Test description