Lanthanum Bromide Advantage
Lanthanum bromide scintillators are gaining popularity due to their superior resolution properties. Typical resolution of a LaBr3(Ce) scintillation detector @ 661.7 keV is between 2.5 and 3.5% depending on manufacturer and crystal grade. Thus, LaBr3(Ce) detector resolution is 2-3 times better than a typical 6-7% resolution of a high-quality NaI(Tl) detector.
However, NaI(Tl) detectors pose other more subtle problems: because NaI(Tl) scintillators are inexpensive and relatively easy to manufacture numerous new companies (Chinese and otherwise) are now offering such detectors. Unfortunately, often times the resolution of these detectors is not optimal and may reach into 8-9% FWHM @ 662 keV. The lower than expected resolution is not necessarily caused by poor crystal quality. The problem could be due to the use of a cheap PMT, primitive voltage divider, or imperfect light tightness of the assembly.
More often than not the suboptimal resolution of most NaI(Tl) detectors is due to moisture. NaI is highly hygroscopic and will absorb water from the environment. Therefore NaI(Tl) detector resolution tends to decline with age, but this decline could be a lot faster than anticipated due to poor assembly quality of the new scintillators. E.g. the initial resolution of 6% may rapidly decline to 8 or even 9% when the detector is used in a high-humidity environment because the NaI crystal was not properly sealed in it’s housing.
In this regard using cheap NaI(Tl) scintillators in critical radiation detection hardware is a risky practice. In the same time quality manufacturers such as Saint Gobain and Scionix have begun offering reasonably priced LaBr scintillators, which render NaI is all but obsolete.
Here I am not going to repeat every advantage of LaBr over NaI. From an end-user perspective, assembly quality (e.g. assurance that the crystal resolution will not degrade with time), low sensitivity to temperature variation, and above all – highest possible resolution – are strong enough reasons to quit using sodium iodine in favor of lanthanum bromide.
Enhanced LaBr3(Ce+Sr) Detector
To prove my point, Saint Gobain was very kind to provide their top of the line 1.5″ x 1.5″ (38 x 38 mm) enhanced LaBr3(Ce+Sr) 38S38 detector for evaluation – Fig. 1.
Because this is an enhanced lanthanum bromide it’s specs deserve a little bit more attention – see below.
The enhanced lanthanum bromide is brighter, has higher resolution and faster decay time than ordinary lanthanum bromide. The enhanced lanthanum bromide is available only from Saint Gobain; all other manufacturers offer standard lanthanum bromide scintillators only.
The 38S38 detector came complete with a AS20 8-stage voltage divider – Fig. 2. – and was priced just under $9,000 USD.
The declared resolution of this detector was a stunning 2.19% @ 662 keV. This is 3x higher than the resolution of the best NaI(Tl) scintillator. In fact, the resolution of the Saint Gobain 38S38 detector is better than a typical resolution of spectrometer-grade CZT crystals and is on par with the resolution of the top of the line CZT spectrometers such as Kromek Spear. Needless to say, typical CZT crystals are quite small. Therefore a fairly large 5x5x5mm CZT spectrometer cannot compete with a 38x38mm LaBr scintillator in terms of total sensitivity especially in the high-energy range.
Yet, the surprisingly good performance of 38S38 detector merits a reasonable question: can this phenomenal 2.2% resolution be realized in practice? Clearly, to achieve the maximum resolution of the scintillation crystal the following conditions must be met:
- HV bias power supply must be low ripple and low noise;
- ADC and preamp electronics must be low-noise and very well shielded from an EM interference;
- ADC sampling rate and the input channel bandwidth must be high enough to match the detector fast rise time;
- energy recovery algorithm and spectrum processing software must be precise enough to provide sufficient shaping time and support high count rates.
These may not be easy requirements to meet. In fact, when working with low-resolution detectors these requirements do not matter nearly as much because a little bit of noise, low bandwidth and a slack in coding will not impact resolution that is low enough to begin with and is already limited by the scintillator itself. Not so in Saint Gobain 38S38 case. So how did the 38S38 detector fair in our system?
38S38 Resolution Evaluation Results
I shall pat myself on the back here – our MCA-PRO 16 hardware system and PulseCounter software have fared extremely well when coupled to the 38S38 detector. Fig. 3 presents a Cs-137 spectrum captured using the Saint Gobain detector.
Without making any special provisions to optimize the PulseCounter software or MCA-PRO 16 hardware for the LaBr detector I was able to achieve 2.3% FWHM resolution at 662 keV, which is very close to the manufacturer-specified value of 2.2%. The minor difference in the FWHM figure could be both due to a different resolution computation technique (PulseCounter uses an iterative fitting to Gaussian). Also, I did not make any provisions to eliminate background and did not pursue a long enough count time to get a perfectly smooth peak.
Fig. 4 compares the Cs-137 spectrum captured using the MCA-PRO 16 and PulseCounter to the reference spectrum provided by Saint Gobain.
From Fig. 4 it is evident that the main Cs-137 peaks in both spectra are virtually identical. It is also obvious that my spectrum had a lot more background counts in it as witnessed by a prominent bump over the Compton shelf: the ~200 keV background peak in my spectrum is 2x taller compared to the reference spectrum.
To investigate the spectrum closer Fig. 5 and 6 zoom in on 662 keV and 32 keV peaks.
As one can see from Fig. 4, the 662 keV peaks are essentially identical in both spectra. Hence 0.1% between the measured and the manufacturer-specified FWHM value are likely to be due to the differences in the FWHM computation method.
However Fig. 5 reveals a clear superiority of the manufacturer-specified spectrum over the captured spectrum in resolving the low-energy 32 keV peak. This is not surprising because the PulseCounter software was using an energy recovery algorithm tuned for a very high count rate: the shaping time was only 0.7 us (the sampling rate was 10 MHz and the ADC bandwidth was 100 MHz). The short shaping time reduced the pulse height recovery precision at lower energies.
38S38 As a Reference System
The 38S38 detector makes an excellent reference system for optimizing and tuning gamma spectrometry hardware and software. Outstanding resolution of the lanthanum bromide scintillator makes it well suited for detecting and eliminating noise in the electronic system as well as optimizing the signal processing algorithm for optimal energy recovery vs. count rate.
In fact I think that all labs and equipment manufacturers should have at least one lanthanum bromide detector to be used as a reference system for hardware and software optimization.
Encouraged by the initial results I have experimented with the 38S38 detector some more to see how the resolution will behave when I use different power supplies and lower sampling rates. When I reduced the sampling rate to 5 MHz, limited the signal bandwidth to 5 MHz as well, and used a 3 us shaping time my typical FWHM @ 662 keV value averaged at 2.5%. Thus, even with a cheap low-bandwidth / low-sampling rate ADC I was able to obtain the excellent resolution.
I have also tried a transistor-stabilized voltage divider VD14-E1-X35 by Scionix and was able to obtain even higher resolution of 2.1% – Fig.6. Unfortunately, the Scionix divider was 10-stage as opposed to 8-stage necessary for the lanthanum bromide detector coupled to the R6231-100 photomultiplier and therefore was occasionally glitching. Still it looks like the transistor stabilization coupled with pulse-height resolution optimized divider design is able to push the 38S38 resolution even higher.
Some More Spectra
The interesting thing about lanthanum bromide gamma detectors is that as soon as you start using one it is very hard to go back to sodium iodine. In fact I see no reason for using sodium iodine any more for general-purpose gamma spectroscopy. With lanthanum bromide the peaks are much more defined and you see a lot more of them. Fig. 7 shows a Ba-133 reference spectrum captured using the Saint Gobain 38S38 LaBr3(Ce+Sr) detector compared to the spectrum captured using a Scionix 38B57 NaI(Tl) detector; note additional peaks resolved by the lanthanum bromide system.
Fig. 8 shows a depleted uranium spectrum captured using the Saint Goban 38S38 and Scionix 38B57 detectors. Once again, additional peaks are resolved by the lanthanum bromide system.
Conclusion
It is a given that lanthanum bromide detectors are vastly superior to sodium iodine systems. So unless one is totally strapped for cash one should go with lanthanum bromide, which in my experience is pretty much a drop-in replacement for an older sodium iodine probe. Granted, you may not be able to achieve the ultimate 2.2% FWHM @ 662 keV resolution in your gamma spectrometer without taking additional steps, but you will get an excellent 2.5% FWHM @ 662 keV easily, even with a cheap 5 MHz ADC.
Last but not least, lanthanum bromide is a must-have detector for system design verification and software tuning.
Acknowledgements
I thank Saint Gobain and Mayra Gonzalez Saldana for providing their top of the line 38S38 detector for evaluation and thus making this blog post possible.