X射線熒光光譜儀XRF分析誤差的來源探討

X射線光譜分析儀的好壞常常是以X射線強度測量的理論統(tǒng)計誤差來表示的，X射線熒光儀的穩(wěn)定性和再現(xiàn)性，已足以保證待測樣品分析測量的精度，被分析樣品的制樣技術(shù)成為影響分析準確度的至關重要的因素，在樣品制備方面所花的工夫?qū)从吃诜治鼋Y(jié)果的質(zhì)量上。X射線熒光儀器分析誤差的來源主要有以下幾個方面：

1． 采樣誤差：

非均質(zhì)材料

樣品的代表性

2． 樣品的制備：

制樣技術(shù)的穩(wěn)定性

產(chǎn)生均勻樣品的技術(shù)

3． 不適當?shù)臉藰樱?/p>

待測樣品是否在標樣的組成范圍內(nèi)

標樣元素測定值的準確度

標樣與樣品的穩(wěn)定性

4． 儀器誤差：

計數(shù)的統(tǒng)計誤差

樣品的位置

靈敏度和漂移

重現(xiàn)性

5． 不適當?shù)亩繑?shù)學模型：

不正確的算法

元素間的干擾效應未經(jīng)校正

l 顆粒效應

純物質(zhì)的熒光強度隨顆粒的減小而增大，在多元素體系中，已經(jīng)證明一些元素的強度與吸收和增強效應有關，這些效應可以引起某些元素的強度增加和另一些元素的強度減小。圖1列舉了強度與研磨時間的關系：①粒度的減小，引起鐵、硫、鉀的強度減小，而使鈣、硅的強度增加。②隨著粒度減小至某一點，強度趨

于穩(wěn)定。③較低原子序數(shù)的元素的強度隨粒度的減小有較大的變化。

l 礦物效應

圖2中樣品為用不同礦物配成的水泥生料。標為“I"的樣品是用石灰石、頁巖和鐵礦石配成的。標為“F"的樣品含有相同的石灰石和鐵礦石，但硅的來源是用砂巖代替了頁巖。兩組原料用同一設備處理，用同一研磨機研磨，每一個樣品約有85%通過200目。圖2表明這種強度—濃度上的變化首先反映了硅的來源不同，“I"的硅來自頁巖，“F" 的硅來自砂巖。然而兩組樣品的進一步研磨指出這僅僅是一個粒度效應問題。圖3表明在全部樣品經(jīng)研磨機粉研到325目（44μm）以后，兩組樣品的實驗點均落在同一曲線上。

l 元素間吸收—增強效應

任何材料的定量X射線熒光分析要求元素的測量強度與其百分含量成正比，在巖石和礦物（由兩種或兩種以上礦物的組合）這類復雜的基體中，由于試樣內(nèi)其它元素的影響，元素的強度可能不直接與其含量成正比。一般認為，多元素體系中這種非線性是由元素間效應引起的。元素間效應可以是增強效應或吸收效應，也可以是同時包括這兩種效應。仍以圖2、圖3實驗為例，圖4表明通過簡單的研磨可以改進CaO的分析結(jié)果。圖5表明校正鉀對鈣的干擾后，兩組樣品的實驗點均落在同一曲線上。

Analysis on the error source of XRF spectrometer

The stability and reproducibility of XRF spectrometer has little influence on the

precision, and the performance of XRF spectrometer is characteristic of the theoretic statistic error of  intensity of XRF spectrometer . the sample preparation technique plays a important role and  will be reflected in the results. the error source of XRF spectrometer is as follows:

1, Sampling error:

Heterogeneous Material

Representative of sample

2. Sample preparation

Stability of the preparation technique

Technique of preparing homogeneous samples

3. Irrelevant standard samples.

Whether the content of unknown samples is in the cover range of the standard samples or not

Accuracy of chemical analysis result of  the standard sample

Stability of the samples and standard samples

4. Instrument error

statistic error of count

location of samples

sensitivity and drift

5. Irrelevant quantitative mathematical models

Irrelevant algorithm

Element interference effect is not calibrated

● granular effect

the fluorescence intensity of pure material increase with the decrease of the granularity. In multi-element system, fluorescence intensity of some element has been proved to be related with matrix effect, which can increase some element

intensity and decrease some element intensity.The relationship between intensity and grinding time is as showed in Fig1:①the decrease of element granularity decrease the intensity of Fe, S, K, and increase the intensity of Ca and Si. ②The intensity tend to be stable when granularity decrease to a certain degree. ③The intensity of low Z element changes a lot with decrease of granularity.

●Material effect

There are some cement raw prepared by different material(Fig 2), the samples marked I are composed of limestone, shale, iron ore. the samples marked F contains the same limestone and iron ore, but contains different silicon, in which, shale is replaced by sandstone. Two kinds of raw material are treated with the same equipment ,and grinded by the same grinding machine. And 85% of them passed 200 mesh. It’s obvious that change of intensity-content relationship show the different resource of silicon. In fact, it is just a matter of granularity, when the sample pass 325 mesh by increasing the grinding time, all points are located in the same curve.

●Matrix effect

All materials measured by XRF are required that the intensity of element is proportional to its content. in the complicated matrix such as rock s and ores(the

composite of at least two kind of materials), for the influence of other elements, the intensity may not be proportional to its content. generally, this nonlinear relationship is caused by elemental matrix effect.

Matrix effect is absorption effect or enhancement effect, or the combination of them. Longer grinding can improve the CaO analysis result(Fig 4). After the calibration of interference of K, the points of two series of samples locate in the same curve.